3-Monochloropropane-1,2-diol reduced bioaccessibility of sn -2 palmitate via binding with pancreatic lipase in infant formula during gastrointestinal digestion

Infant formula contains 3-monochloropropane-1,2-diol esters (3-MCPDE), which are formed during the deodorization step of vegetable oil refining. The European Food Safety Authority stated that 3-MCPDE can be hydrolyzed in the gastrointestinal tract to free-form 3-monochloropropane-1,2-diol (3-MCPD), which has potential toxicity and can be rapidly absorbed. Evaluating the effect of 3-MCPD on nutrition absorption is a prerequisite for establishing effective management strategies. A total of 66 crucial lipid molecules associated with 3-MCPD were identified based on debiased sparse partial correlation analysis. 3-MCPD affected triglyceride hydrolyzation and increased the concentration of undigested sn -2 palmitate (9.57 to 17.06 mg kg −1 ). 3-Monochloropropane-1,2-diol reduced the bio-accessibility of fatty acids, and more short-(31.42 to 58.02 mg kg −1 ) and medium-chain fatty acids (17.03 to 26.43 mg kg −1 ) remained unabsorbed. Lipidomic profiles of infant formula models spiked with different 3-MCPDE levels were investigated, and the results were consistent with the experiments with the commercial formula indicating lipid alteration was mainly affected by the digestive 3-MCPD. The formation of 3-MCPD ester-pancreatic lipase with the binding energy of −4.9 kcal mol −1 was more stable than triglyceride-pancreatic lipase (−4.0 kcal mol −1 ), affecting triglyceride hydro-lyzation. 3-Monochloropropane-1,2-diol was bound to Glu13 and Asp331 residues of the pancreatic lipase via hydrogen bonds, which resulted in a conformational change of pancreatic lipase and spatial shielding effect. The


INTRODUCTION
Infant formula as a substitute for breast milk provides nutrition to babies by adjusting the composition (carbohydrates, proteins, and lipids) to match the nutrient profile of breast milk.Various fat sources with different proportions are added to the animal milk to mimic the fatty acid composition of breast milk.Vegetable fats as the most common fat source are mainly composed of coconut oil, corn oil, palm oil, soybean oil, and sunflower oil.Due to a broader range of triglycerides and fatty acids and more in line with the composition of human milk lipids, bovine milk fat is commonly added to infant formula (Kloek et al., 2023).Historically, with the development of infant formula, animal fat has gradually been replaced by vegetable fats for several reasons, including the avoidance of cross-contamination with specific compounds, such as dioxins, and the high cost of incorporating bovine milk into infant formula, although, because of several benefits, today bovine milk lipids are used more in infant formula (Hageman et al., 2019).Heat-contaminants 3-monochloropropane-1,2-diol esters (3-MCPDE) are formed in refined vegetable oils and are present in infant formula along the production chain (Huang et al., 2020;Xie et al., 2021).The Joint FAO/WHO Expert Committee on Food Additives (JECFA) established a provisional maximum tolerable daily intake (PMTDI) for 3-MCPDE of 4 μg kg −1 BW per day (Joint FAO/WHO Expert Committee on Food Additives, 2016).Recent studies investigated the occurrence of 3-MCPDE in infant formula and evaluated the concentration of 3-MCPDE.The content of 3-MCPDE in 40 infant formula products purchased in Brazil varied from <80 to 600 μg kg −1 (Arisseto et al., 2017).Analysis of 23 infant formula samples of both Chinese 3-Monochloropropane-1,2-diol reduced bioaccessibility of sn-2 palmitate via binding with pancreatic lipase in infant formula during gastrointestinal digestion Wei Jia, 1,2,3 * Xixuan Wu, 1 Jing Shu, 2 and Lin Shi 1 and imported brands showed a 100% positive rate of 3-MCPDE, and the contents ranged from 1.7 to 362.9 μg kg −1 (Li et al., 2022).The contents of 3-MCPDE in 5 infant formula products purchased from the Malaysian market were determined as between <50 to 138 μg kg −1 (Goh et al., 2019).Growing evidence recently suggested the occurrence of 3-MCPDE was relatively high in infant formula, more than a PMTDI established by JECFA.3-Monochloropropane-1,2-diol (3-MCPD) is released by the intestinal lipase efficiently from 3-MCPD monoesters and 3-MCPD diesters.The enzymatic hydrolysis of diesters yields 2-monoacyl esters, which is similar to 2-monoacyl-sn-glycerols hydrolyzed by triacylglycerols.2-Monoacyl esters of 3-MCPD were re-esterified subsequently, incorporated into cellular membranes or deposited in adipose tissue (Beekman et al., 2020;Shen et al., 2023).Furthermore, the isomerization of 2-monoacyl esters of 3-MCPD occurs and is degraded to 3-MCPD completely.Consumption of 3-MCPD contributes to testicular and renal toxicity, and the International Agency for Research on Cancer recognized 3-MCPD as a potential carcinogen to humans (Arisseto et al., 2017).
Among the various infant formulas available, the market share of goat milk-based products continues to climb.The higher content of phospholipids and smaller partial size of fat globules contribute to more easily digestible goat milk-based products than other products (Zhang et al., 2020;Wang et al., 2023).Lipids as the dominant component of goat milk contain fatty acyls, glycerides, glycerophospholipids, and sphingolipids.Triglycerides provide energy and functional fatty acids for infant growth and development (Luo et al., 2023).The structure of triglycerides, including the composition and position of fatty acids, affect lipid digestion and absorption (Pereira et al., 2023).Triglycerides composed of 3 fatty acids esterified at the middle or secondary (sn-2) and outer or primary (sn-1 and sn-3) locations of triglycerides make up around 98% of the lipids in human milk (Fabritius et al., 2020).Palmitic acid is the primary fatty acid in human milk.The proportion of palmitic acid esterified at the sn-2 position of triglycerides is about 70% in human milk, and palmitic acid is supplied to the infant in a unique configuration (Chen et al., 2022).Infant formula frequently contains bovine milk as the component.Approximately 40% to 45% of palmitic acid in bovine milk is present at position sn-2.Vegetable oils, such as coconut oil, corn oil, palm oil, soybean oil, and sunflower oil are the common fat source of infant formula.Only 10% to 20% of palmitic acid is esterified at the sn-2 position of triglycerides in vegetable oil (Smith et al., 2021).Goat milk contains about 20% to 30% of palmitic acid, and the content is similar to human milk.Approximately 40% of palmitic acid in goat milk is present at position sn-2 of triglycerides (Prosser et al., 2010).To mimic the levels of palmitic acid found in human milk, infant formulas are currently supplemented with palmitic acid in the form of palm oil.Supplementation of vegetable oil leads to the release of 3-MCPD in goat milk-based infant formula during gastrointestinal digestion.Little is known about whether the 3-MCPD release is associated with lipid bioaccessibility and whether it obstructs the acquisition of lipid nutrients in goat milk-based infant formula.
This study aimed to investigate the effect of released 3-MCPD during gastrointestinal digestion on lipid molecule bioaccessibility and assess the nutrition absorption from goat milk-based infant formula in response to 3-MCPD.Leveraging the power of lipidomics based on UHPLC-Q-Orbitrap HRMS was to depict dynamic patterns in the lipidome of infant formula after in vitro digestion.Molecular docking analysis was applied to reveal the 3-MCPD-pancreatic lipase binding mechanism and decipher the molecular mechanism of 3-MCPD driving low absorptivity of lipid nutrients.Results from this study provided a holistic absorption overview of lipid nutrients in response to 3-MCPD released from infant formula after in vitro digestion and assisted in further efforts in directing the mitigation of 3-MCPD release in infant formula.

MATERIALS AND METHODS
Because no human or animal subjects were used, this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.

Sample Collection
A total of 120 infant formula products with similar expiration dates were stochastically purchased from local supermarkets in Xi'an, Shaanxi Province, China.These products were intended for infants aged 0 to 6 mo old (stage I, 40 samples), infants aged 7 to 12 mo old (stage II, 40 samples), and infants aged 13 to 36 mo old (stage III, 40 samples).These samples covered Chinese domestic brands (80 samples) and foreign brands (40 samples) to ensure generalizability.Goat milk is the primary ingredient in all formulas.The detailed sample information, including distribution and vegetable oil ingredients, was presented in Table 1.All samples were stored in their original packaging at 20°C ± 3°C and protected from light.
Vegetable oil proportions and other ingredient proportions in the infant formula models are shown in Supplemental Table S1 (https: / / data .mendeley.com/datasets/ 26jsdz32hc; Wu, 2023).Fresh goat milk as a base ingredient was heated to 60°C ± 1°C, and other ingredients were sequentially added and mixed.The OPO structural lipid and vegetable oils were premixed according to the following procedure.Add them sequentially to a stainless-steel container and then mix them with a magnetic stirrer.Control the pH of the final wet mixture at 6.7 to 6.9.Homogenization of the mixture was performed in HST homogenizer at 60°C ± 1°C (HST Maschinenbau GmbH, Dassow, Germany).Primary and secondary pressures were set at 12.5 and 2.5 MPa, respectively.Partial moisture of the wet mixture was removed by Hipex falling film evaporator (Melbourne, Australia).Concentrated samples were spray-dried in a spray-drying tower HYB-5L.Inlet and outlet temperatures were 150°C and 90°C, respectively.The centrifugal frequency of the atomizer was set at 420 Hz.The composition and content of lipid molecules in infant formula models were the same.Infant formula models were spiked with different 3-MCPDE levels (0,50,100,150,200,250,300, and 350 μg kg −1 ).

Determination of 3-MCPDE in Infant Formulas
Standard Solutions.Chloropropanol and deuterated chloropropanol stock solutions (1,000 mg L −1 ) were prepared by dissolving 10.0 mg of 3-MCPD, 2-MCPD, 1,3-DCP, 2,3-DCP, 3-MCPD-d 5 , 2-MCPD-d 5 , 1,3-DCPd 5 , and 2,3-DCP-d 5 in 10 mL of ethyl acetate, respectively.Chloropropanol intermediate standard solutions (10 mg mL −1 ) were prepared by transferring 0.1 mL of 3-MCPD, 2-MCPD, 1,3-DCP, and 2,3-DCP stock solution to 10 mL volumetric flask and diluting by n-hexane to the mark.Chloropropanol spiking solution (1 mg mL −1 ) was prepared by transferring 1.0 mL of 3-MCPD, Rapeseed oil, sunflower oil 2-MCPD, 1,3-DCP, and 2,3-DCP intermediate standard solutions to 10 mL volumetric flask and diluting by nhexane to the mark, respectively.Deuterated chloropropanol mixed standard working solution was prepared by transferring 0.1 mL of 3-MCPD-d 5 , 2-MCPD-d 5 , 1,3-DCP-d 5 , and 2,3-DCP-d 5 stock solutions to 10 mL volumetric flask and diluting by n-hexane to the mark, respectively.Chloropropanol fatty acid ester stock solutions (1,000 mg L −1 ) were prepared by dissolving 53.1 mg of 1,2-bis-palmitol-3-chloropropanediol and 58.1 mg of 1,3-distearoyl-2-chloropropanediol in 10 mL of ethyl acetate, respectively.Chloropropanol fatty acid ester mixed spiking solution was prepared by transferring 0.1 mL of 1,2-bis-palmitol-3-chloropropanediol and 1,3-distearoyl-2-chloropropanediol stock solutions to a 10-mL volumetric flask and diluting by n-hexane to the mark, respectively.Deuterated chloropropanol fatty acid ester stock solutions (1,000 mg L −1 ) were prepared by dissolving 30.6 mg of 1-palmitoyl-3-chloropropanediol-d 5 and 56.0 mg of 1,3-distearoyl-2-chloropropanediol-d 5 in 10 mL of ethyl acetate, respectively.Deuterated chloropropanol fatty acid ester mixed standard working solution was prepared by transferring 0.1 mL of 1,2-bis-palmitol-3-chloropropanediol-d 5 and 0.1 mL of 1,3-distearoyl-2-chloropropanediol-d 5 stock solutions to 10-mL volumetric flask and diluting by n-hexane to the mark, respectively.Chloropropanol calibration solutions were pipetting chloropropanol standard working solution (1 mg mL −1 ) in aliquots of 0.01, 0.05, 0.10, 0.20, 0.40, 0.80, and 1.60 mL, and 0.32 mL of chloropropanol intermediate standard solution (10 mg mL −1 ) into a 5-mL capped test tube and 20 μL of deuterated chloropropanol mixed spiking solution (10 mg mL −1 ) was added.The solution was diluted by n-hexane to 2 mL and chloropropanol calibration solutions with masses of 10, 50, 100, 200, 400, 800, 1,600, and 3,200 ng were obtained.Sample Preparation.Determination of 3-MCPDE in infant formula samples was performed according to the national standard (National Health and Family Planning Commission of the People Republic of China, State Administration of Food and Drug Administration, 2023;GB5009.191-2016).A total of 2 g of infant formula samples was precisely weighed.After spiking with 20 μL of deuterated chloropropanol fatty acid ester mixed with standard working solution (10 μg mL −1 ), 4 mL of n-hexane was added and sonicated for 20 min.After standing for stratification, the upper layer was transferred.The extraction step was repeated twice and the collected upper layer was combined.After adding 20 μL of 1,3-DCP-d 5 and 20 μL of 2,3-DCP-d 5 standard working solutions, the sample was concentrated to 1 mL under a gentle stream of nitrogen.The sample was added with 0.5 mL of methyl tert-butyl ether/ethyl acetate solution (8:2, vol/vol) and 1.0 mL of sodium methylate-methanol (0.5 mol L −1 ) and then vortexed for 30 s.After 4 min of reaction at 20°C ± 1°C, 100 μL of glacial acetic acid was added to terminate the reaction.The mixture was added with 3 mL of sodium bromide solution (20%, wt/vol) and 3 mL of n-hexane and vortexed for 30 s (20°C ± 1°C).After standing for stratification, the upper layer was discarded and 3 mL of n-hexane was used for extraction.The aqueous phase was transferred into an Extrelut column (diatomaceous earth based solid phase) and equilibrated for 10 min.Then 15 mL of ethyl acetate was applied for elution and the eluate was collected.The eluate was added with 4 g of anhydrous sodium sulfate and filtered after standing for 10 min.The filtrate was transferred to a transparent glass tube and concentrated to 0.5 mL under a gentle stream of nitrogen.The residue was dissolved into 2 mL of n-hexane and vortexed for 1 min.The purification solution was added with 0.04 mL of heptafluorobutyryl imidazole and vortexed for 1 min.The mixture was held at 70°C ± 1°C for 20 min and added with 2 mL of sodium chloride solution (20%, wt/vol) after cooling to 20°C ± 1°C.The upper organic phase was removed and dried by adding 0.3 g of anhydrous sodium sulfate.The solution was transferred to the injection vial for determination by gas chromatography-mass spectrometry.

In Vitro Digestion System
Static in vitro digestion was conducted using the harmonized INFOGEST protocol (Brodkorb et al., 2019).Infant formula emulsions were prepared by dissolving infant formula into 50 mL of pre-heated water according to the product instruction.All solutions were preheated to 37°C ± 1°C before adding to the in vitro digestion procedures and all digestion procedures were performed at 37°C ± 1°C applying a circulating water bath.
Oral Phase Condition.Simulated saliva fluid (SSF) composed of 15.10 mmol L −1 KCl, 3.70 mmol L −1 KH 2 PO 4 , 13.60 mmol L −1 NaHCO 3 , 0.15 mmol L −1 MgCl 2 , 0.06 mmol L −1 (NH 4 ) 2 CO 3 , and 1.10 mmol L −1 HCl was prepared.Samples were diluted with SSF at the ratio of 1:1 (vol/vol).The sample (15 mL) was mixed with SSF (12 mL) and added with CaCl 2 achieving a final concentration of 1.50 mmol L −1 .The pH was adjusted to 7.0 and the volume was topped up to 30 mL with distilled H 2 O.The resulting mixture was placed on a linear shaking water bath (Grant GLS 400, Grant Instruments, England) and incubated at 37°C ± 1°C for 2 min to simulate the physiological process of infant oral digestion.

Jia et al.: PANCREATIC LIPASE IN INFANT FORMULA
KH 2 PO 4 , 25.00 mmol L −1 NaHCO 3 , 47.20 mmol L −1 NaCl, 0.12 mmol L −1 MgCl 2 , 0.50 mmol L −1 (NH 4 ) 2 CO 3 , and 15.60 mmol L −1 HCl was prepared.The obtained digests from the oral phase were added with SGF at the ratio of 1:1 (vol/vol).After adding CaCl 2 to achieve a final concentration of 0.15 mmol L −1 , the pH was adjusted to 3.0 using HCl (2.00 mol L −1 ) and NaOH (1.00 mol L −1 ).Porcine pepsin and gastric lipase were added to a final concentration of 2,000 U mL −1 and 60 U mL −1 , respectively.The volume was topped up to 60 mL with distilled H 2 O and the resulting mixture was incubated at 37°C ± 1°C for 120 min.
Intestinal Phase Condition.Simulated intestinal fluid (SIF) composed of 6.80 mmol L −1 KCl, 0.80 mmol L −1 KH 2 PO 4 , 85.00 mmol L −1 NaHCO 3 , 38.40 mmol L −1 NaCl, and 0.33 mmol L −1 MgCl 2 was prepared.Gastric chyme was added with SIF at the ratio of 1:1 (vol/vol).The resulting mixture from the gastric phase (60 mL) was added with 16 mL SIF and 2 mL bile salts (10 mM) and stirred continuously at 37°C ± 1°C to achieve the complete solubilization of bile.After adding CaCl 2 to reach a final concentration of 0.60 mmol L −1 , trypsin, chymotrypsin, pancreatic α-amylase, and pancreatic lipase were added to a final concentration of 100, 25, 200, and 2,000 U mL −1 , respectively.The mixture was incubated at 37°C ± 1°C for 120 min and the pH-stat was applied to maintain pH at 7.0 using 0.1 mol L −1 NaOH.The intestinal digestion was terminated in a boiling water bath for 10 min.
Sample Preparation.Determination of 3-MCPD in postdigestion samples was performed according to the method published by Association of Official Analytical Chemists (Brereton et al., 2001).For 3-MCPD analysis, 4 g samples were precisely weighed and spiked with 20 μL of 3-MCPD-d 5 internal standard working solution (10 μg mL −1 ).Then, 4 g of sodium chloride solution (20%, wt/vol) was added and followed by sonicating for 5 min.The homogenized sample was gently mixed with the Extrelut NT20 refill pack.The mixture was transferred to a glass chromatographic column (40 × 2.0 cm) and a 1 cm layer of anhydrous sodium sulfate was placed on the top of the column packing.A total of 40 mL of n-hexane/anhydrous ether solution (9:1, vol/ vol) were used to elute the mixture.The column was rinsed with 150 mL of anhydrous ether at a flow rate of 8 mL min −1 .The eluate was collected and mixed with 15 g of anhydrous sodium sulfate for dehydration.Then the filtrate was concentrated to 0.5 mL using a rotary evaporator at 35°C ± 1°C.The flask was washed with 2 mL of n-hexane and together with concentrated solvent transferred to a 10 mL-capped test tube.Reconstitution fluid was added with 0.04 mL of heptafluorobutyric anhydride using a gas tight syringe.The mixture was vortexed for 30 s and kept at 70°C ± 1°C for 20 min.After cooling to 20°C ± 1°C, 2 mL of sodium chloride solution (20%, wt/vol) was added and vortexed for 1 min.The organic phase was transferred to a new test tube and dried with anhydrous sodium sulfate.The solution was transferred to a vial for further GC-MS analysis.
GC-MS Analysis.Gas chromatography was performed on a DB-5MS fused-silica capillary column (30 m × 0.25 mm, 0.25 μm; Hewlett-Packard, Avondale, PA).The injection port temperature was set at 250°C and the spitless injection mode was performed with an injection volume of 1.0 μL.The carrier gas was high-purity helium (≥99.999%) at a flow rate of 1.0 mL min −1 .The GC temperature was programmed as follows: initial temperature 50°C, holding for 1 min, followed by a 2°C min −1 ramp to 90°C and then a 40°C min −1 ramp to 270°C, holding for 5 min.The mass spectrometer with an electron impact ion source was operated in selected ion monitoring mode at 70 eV of electron energy.The temperatures of the transfer line and ion source were 250°C and 280°C, respectively.

Lipidomics Analysis
Lipid Extraction.The digestion solution (60 μL) was added with 340 μL of H 2 O and vortexed for 1 min.
After adding 960 μL of MTBE/MeOH solution (5:1, vol/ vol) and 20 μL of internal standard mixture solution, the sample was vortexed for 1 min and sonicated for 10 min in an ice-water bath (Jia and Di, 2023).After centrifugation at 12,000 × g for 15 min at 4°C, the supernatant (400 μL) was transferred to a fresh tube.The original sample was added with 400 μL of MTBE/MeOH solution (5:1, vol/vol) and the extraction step was repeated twice.The supernatants were pooled together after 3 extractions and dried under a gentle stream of nitrogen.The dried sample was reconstituted with 100 μL of ACN/IPA/H 2 O solution (65:30:5, vol/vol/vol) and filtered through 0.22 μm membranes (Millipore, Bedford, MA).Quality control (QC) samples were prepared by pooling equal volumes (10 μL) of each sample.

Data Analysis
Identification of lipids was performed based on precursors and characteristic fragments matching.The mass tolerances of precursors and fragments were 5 and 10 ppm, respectively.The retention time of the compound and the mass-to-charge ratio of parent ions were used for qualitative analysis (Fan et al., 2023).The drift tolerance of retention time was set at 0.25 min.Lipid assignments and nomenclature were in ac-cordance with Lipid Metabolites and Pathways Strategy.The individual species of glycerophospholipids, lyso-phospholipids, phospholipids, and sphingolipids were quantified by internal standards.The concentration was calculated according to the following equation (Equation 1): where C Lipid and C IS represent the concentration of the individual lipid molecule and corresponding to the internal standard, respectively.RF Lipid represents the ratio of the slopes of lines for the individual lipid molecule and internal standard.A Lipid and A IS represent the peak area of the individual lipid molecule and corresponding to the internal standard, respectively.Data sets were imported to MetaboAnalyst 5.0 to complete data preprocessing, including log 2 -transforming and pareto scaling.Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were performed.Permutation testing was conducted to evaluate the quality of the PLS-DA model and validation parameters including R 2 Y (goodness-offit) and Q 2 (predictive ability) were inspected (Jia et al., 2023b).Differential compounds were screened based on criteria of P-value < 0.05 in the ANOVA and variable importance in the projection (VIP) value >1.0 in the PLS-DA model.Differential compounds were linked to perturbated biochemical pathways using Kyoto Encyclopedia of Genes and Genomes database.Pathways were identified by the hypergeometric test and the topology was represented by relative-betweenness centrality (Jia and Wang, 2023).The P-value was adjusted through the false discovery rate estimation and pathways with P-value <0.05 were regarded as significantly impacted pathways.

Statistical Analysis
Data were statistically analyzed using SPSS statistics (version 22.0, SPSS-IBM).One-way ANOVA was used for assessing means of 3 or more samples and P-value < 0.05 was regarded as significant.Experimental results were expressed as sample mean values ± standard deviation.

Molecular Docking Analysis
The crystal structure of pancreatic lipase (PDB code: 1LPB, resolution: 2.20 Å) was downloaded from Research Collaboratory for Structural Bioinformatics Protein Data Bank.The 3-dimensional structure of ligand 3-MCPD (PubChem CID: 7290) was retrieved from PubChem database.The charge energy minimiza- tion was implemented with Merck molecular force field of 30,000 steps.The Protein Data Bank with partial change Q and atom type t (PDBQT) format files were generated by AutoDockTool-1.5.6 after merging nonpolar hydrogens, adding polar hydrogens, and computing Gasteiger-Huckle charges.The internal conformational search was conducted using the Lamarckian Genetic Algorithm with 100 docking runs.Dimensions of the grid box were set as 100 × 100 × 100 Å (x, y, and z) and the grid spacing was 0.375 Å. Exhaustiveness was set to 2000 for improving the docking precision.Blind docking was performed for pancreatic lipase and 3-MCPD by using AutoDock Vina 1.1.2(The Scripps Research Institute Molecular Graphics Laboratory, La Jolla, CA) with default parameters.The docking model with the lowest binding free energy and bonding condition was chosen for further analysis.Intermolecular combinational patterns between protein AA residues and ligands were visualized using Discovery Studio 4.5 (Dassault Systemes Biovia, San Diego, CA) and PyMol 1.7.6 (DeLano Scientific LLC, San Carlos, CA).

Content Variation of Released 3-MCPD Driving Shifts in Lipid Composition
To investigate the effect of 3-MCPD release during gastrointestinal digestion on lipid bioaccessibility and acquisition of lipid nutrients in goat milk-based infant formula, postdigestion infant formula samples released different levels of 3-MCPD (0 μg kg −1 -100 μg kg −1 ) during in vitro digestion were analyzed with lipidomics.Acquired spectra from postdigestion infant formula samples were processed including peak detection, background noise subtraction, retention time correction, peak alignment, and gap filling.Exponentially Modified Gaussian (EMG) fitting algorithm was applied to model the peak shape as an EMG curve (Wahab et al., 2021).The area and height of saturated and cutoff peaks could be interpolated and closer to true values by fitting peaks to EMG distribution.Identified features were normalized to the mean peak area of QC to mitigate variance in the data set due to sample processing and instrument stability and retain biological variability (Jia et al., 2023c).Background noise subtraction including chemical noise of column, instrument contaminants, and the mobile phase was performed and the signal reproducibility was filtered.Retention time shifts were corrected by determining shift as a function of retention time within a sample run.Calibrants uniformly distributed a large portion of the chromatogram were selected and accurate estimation of retention time shifts at different elution times was gained.Ordered bijective interpolated warping (OBI-Warp) algorithm was employed for peak alignment (Yin et al., 2019).Gap filling was designed to search and reintegrate the missing peaks in the experimental data set.

Quantitative Lipidomics Analysis Reveals 3-MCPD Induced the Alteration of Lipid Molecules
To decipher the alterations in the pathways leading to dynamic changes of lipid molecules, lipidomics analysis of digestion samples releasing diverse 3-MCPD concentrations was performed.Differences in lipid molecules among infant formula samples released different levels of 3-MCPD (0 μg kg −1 -100 μg kg −1 ) after in vitro digestion were investigated using multivariate statistical analysis.A clear distinction in the PCA score plot implied the significant differences among the 6 groups and the independent biological replicates from the same group clustered together suggesting high reproducibility of the obtained data set.The variability of the first 2 principal components (PC1 and PC2) were 47.06% and 21.91%, respectively (Figure 2A).The supervised PLS-DA result revealed the lipidomics profiles of the control and experiment groups were entirely separated, indicating that released 3-MCPD from infant formulas during in vitro digestion induced significant biochemical changes.Two components explained 71.33% of the total variance (component 1 for 48.25% and component 2 for 23.08%; Figure 2B).The permutation test (n = 100) was utilized to access the adaptability and robustness of the PLS-DA model (Figure 2C).High values of goodness of fit (R 2 ) = 0.999, goodness of prediction (Q 2 ) = 0.949 and accuracy = 0.950 verified the fitness of PLS-DA score plot efficiently.According to VIP value >1.0 and P-value < 0.05, a total of 85 lipid molecules were identified as the differential lipid molecules (Figure 3B).Due to the different formulation strategies of collected infant formulas, infant formula models spiked with different 3-MCPDE levels were applied for analysis to clarify the influence factor of lipid alteration in postdigestion infant formula.In vitro digestion of infant formula models was performed and the lipidomics profiles of postdigestion infant formula models releasing diverse 3-MCPD concentrations were investigated based on lipidomics analysis.Postdigestion infant formula models with different 3-MCPD levels showed obvious trends of separation in the PCA plot (Figure 2D).The 6 different model formula per 3-MCPDE concentration refers to replicate measurements (n = 6).Postdigestion infant formula model with low-level 3-MCPD was distributed in an area closer to the postdigestion infant formula model with 0 μg kg −1 3-MCPD, which was consistent with the tendency in postdigestion infant formula samples (Figure 2A).The tendency of lipid alteration in postdigestion infant formula model was in line with postdigestion infant formula samples, indicating digestive releasing 3-MCPD level was the main influence factor of lipid alteration.
Untargeted lipidomics method was validated by analyzing the sample spiked with the internal standard.The validation results including recovery, LOD, LOQ, and precision are listed in Supplemental Table S4 (https: / / data .mendeley.com/datasets/ 26jsdz32hc; Wu, 2023).The acceptable recoveries of all lipid standards varied from 86.36% to 102.10%.Excellent correlation coefficients (r) were from 0.9934 to 0.9988.Limit of detection and LOQ were defined as the minimum identified and quantified concentrations of each analyte and were determined as 3 and 10 times of the signal-to-noise ratio (S/N), respectively.The LOD and LOQ values ranged from 0.003 to 0.018 and 0.009 to 0.047 μg mL −1 , respectively.The matrix effect was evaluated as the following equation (Equation 2) and the result of the matrix effect was 3.26% to 7.82%.The precision was evaluated at low (1 × LOQ), medium (2 × LOQ), and high (4 × LOQ) concentrations, and the values were expressed as the relative standard deviation (% RSD).The intraday precisions were performed 6 parallel experiments in the morning, afternoon, and evening on the same day and the interday precisions were determined in 6 consecutive days.The intraday and interday precisions were below 4.75% and 6.40%, respectively.

Matrix effect %
where A Matrix and A Solvent represent peak area of matrix standard solution and solvent standard solution, respectively.The circular bar plot was used to visualize the specie changes of lipid subclasses in infant formulas containing different levels of 3-MCPD (0-100 μg kg −1 ) after digestion (Figure 3A).From the perspective of the total number of species, there is no significant difference among the 6 groups.The influence of different 3-MCPD levels on the dynamic changes of lipid molecules concentra-tion in gastrointestinal digestion was shown in Figure 1D.The significant changes in lipid molecules showed a concentration-dependent trend.Absolute quantification of the annotated lipid molecules with VIP value >1.0 in the analyzed samples was achieved (Table 2).Sphingomyelins as the major source of choline for infants can promote the rapid development of organs and biofilm synthesis in neonates (Pan et al., 2022).As shown in Figure 1E, more SM were not digested (from 1.10 to 1.82 mg kg −1 ) with the elevations in 3-MCPD.As the component of pancreatic lipase, bile salt-stimulated lipase can hydrolyze a series of substrates including tri-, di-, and monoglycerides, cholesteryl esters, phospholipids, and ceramides (Binte Abu Bakar et al., 2022).Ceramides are further hydrolyzed to sphingosine and fatty acids.The concentration of unhydrolyzed Cer increased from 1.21 to 1.99 mg kg −1 with the elevations in 3-MCPD (Figure 1F).Ceramides have been demonstrated to have nonoverlapping biological effects in inflammation, regulation of cell growth, and induction of apoptosis (Nilsson, 2016;Jia et al., 2023d).

Correlation Network Analysis Reveals 3-MCPD Affected Triglyceride Bioaccessibility
A total of 66 crucial lipid molecules derived from the control and experiment groups were involved in the construction of correlation network.Debiased sparse partial correlation (DSPC) algorithm was implemented to support data-driven correlation network analysis (Bayraktar et al., 2020).After log-transformation and appropriate normalization, peak areas of p lipid molecules from n samples were stored in an n × p matrix X. X ij represents the peak areas of the jth lipid molecules in the ith sample.Assuming peak areas of p lipid molecules come from a multivariate Gaussian distribution N(0, R)    and lipid molecules m and n have nonzero partial correlations if and only if Θ mn ≠0 (Wang et al., 2022a).Apart from the data set of lipid molecules, the content of 3-MCPD was also incorporated into the DSPC network.Nodes of the weighted network represent the identified lipid molecules and edges represent the associated P-values.As shown in Figure 4A, DSPC network was dominated mainly by TG species and DG species which were significantly correlated with each other.3-MCPD was positively correlated with TG.The altered lipid molecules formed a synergistic correlation network.
The location distribution of fatty acids in infant formula affects the bioaccessibility of lipid molecules.Pancreatic lipase as the predominant lipase hydrolyzes triglyceride in the small intestine and specifically acts on ester bonds between fatty acids and glycerol (Kumar and Chauhan, 2021;Jia et al., 2023e).Triglyceride digestion results in the hydrolysis of fatty acids from the sn-1,3 position and the release of one sn-2 monoglyceride and 2 unesterified fatty acids (Wang et al., 2022c).Short-chain fatty acids of total carbon number from 2 to 6 are easily digestible and transported to the liver to metabolize via β-oxidation (He et al., 2020).As shown in Figure 4B, the concentration of undigested short-chain fatty acids was higher in infant formula released high level of 3-MCPD (58.02 mg kg −1 ) than at low level of 3-MCPD (31.42 mg kg −1 ).Medium-chain fatty acids are preferentially hydrolyzed in the intestine and directly transferred to the portal circulation.The absorption and metabolism of medium-chain fatty acids supply rapid energy to the infant body and reduce the tendency for the adipose formation and the content of circulating total and low-density lipoprotein cholesterol (Balthazar et al., 2017).The concentration of undigested medium-chain fatty acids (Figure 4C) increased significantly in infant formula releasing the high level of 3-MCPD (26.43 mg kg −1 ) compared with the low level of 3-MCPD (17.03 mg kg −1 ).Long-chain fatty acids constitute energy supply and maintain normal physiological function for infants (Nguyen et al., 2015;Jia et al., 2023f).Figure 4D showed the concentration of undigested long-chain fatty acids in samples and the concentration was higher in infant formula released high level of 3-MCPD (232.72 mg kg −1 ) than at low level of 3-MCPD (131.34 mg kg −1 ).The digestion of palmitic acid depends on its stereospecificity and the level of sn-2 palmitic acid is proportional to its absorption.Palmitic acid at the sn-2 position facilitates intestinal absorption as sn-2 glycerol monoester (Murota, 2020).The concentration of undigested sn-2 palmitic acid in TG increased (9.57 to 17.06 mg kg −1 ) with the elevations in 3-MCPD (Table 3).Meta-analysis revealed the relationship between the high level of sn-2 palmitic acid and health outcomes in infants.Higher proportions of sn-2 palmitic acid feeding was associated with better skeletal mineralization and enhanced physical development (Wei et al., 2019).As shown in Table 3, the concentration of esterified C16:0 was lower in infant formula released high level of 3-MCPD (32.06 mg kg −1 ) than in low level of 3-MCPD (56.11 mg kg −1 ), indicating the more unesterified C16:0 were released.Unesterified C16:0 increases the tendency to bind with calcium and other divalent cations and form insoluble soaps (Zhang et al., 2022).The generation of saturated fatty acids soaps prevents the absorption of calcium and saturated fatty and excretes through feces resulting in stool hardening and constipation (Béghin et al., 2019).

Molecular Docking of the Interaction Between Pancreatic Lipase and 3-MCPD
Triglycerides are emulsified by a mix of mechanical and shear forces in the stomach.When emulsions are exposed to the highly acidic environment of the stomach, the stability of micelles is significantly reduced, while proteins reach their isoelectric point, leading to the flocculation of fat globules, which are then rehomogenized by mechanical action through the gastric sinus and pylorus to form a new form.The ultimate morphology of the emulsion is determined by the surface characteristics and composition of the fat globules as well as the lipid composition.Lipids containing high proportion of medium and short-chain fatty acids are more prone to form a stable and homogeneous form.This homogeneous emulsion with a smaller particle size provides more surface binding sites for gastric lipase, which facilitates the digestion of lipids in the stomach.Lipids containing a high proportion of long-chain saturated fatty acids are more prone to coalesce into clumps of larger particle size in the stomach due to a melting point higher than 37°C.This emulsified form has a reduced contact area with gastric lipase, leading to lesser free fatty acid release and lipolysis in the stomach (Wang et al., 2022d).Previous research has reported that the composition of triglycerides in infant formula is characterized by long-chain fatty acids (Wang et al., 2022b).This result indicates the limited stomach digestion of triglycerides.
Dietary fat starts to hydrolyze partially in the stomach by gastric lipase.Large fat molecules emulsified with bile salts form a complex structure with cholesterol, denatured proteins, free fatty acids, oligosaccharides, phospholipids, and polar lipids (Nilsson et al., 2021).Pancreatic lipase is the predominant lipase for hydrolyzing triglycerides in the small intestine.It primarily acts on the ester bonds formed by glycerol and fatty acid molecules to produce diglycerides, monoglycerides, and free fatty acids.The sn-1 ester bond of triglycerides is first acted upon by pancreatic lipase, followed by the sn-3 ester bond.Pancreatic lipase is anchored at the interface of lipid micelles and forms a Michaelis-Menten adsorption complex leading to enzyme acylation (Wang et al., 2022c).Serine residue located in the catalytic site is capable of recognizing the interface and further controlling lipase adsorption.The conformational change of pancreatic lipase contributes to the movements of the lid domain and surface loops and hence paves the way for the substrate to arrive at the catalytic site.Ser153 of pancreatic lipase attacks the carboxyl group of triglycerides ester and catalyzes the hydrolysis of triglycerides.Pancreatic lipase releases diglycerides, monoglycerides, free fatty acids, cholesterol, lysophosphatidic acid, and fat-soluble vitamins after hydrolysis.Previous studies indicated 3-MCPDE with a similar structure of triglycerides is cleaved by lipase in the gastrointestinal tract to the free-form 3-MCPD (Liu et al., 2021;Eisenreich et al., 2023).

CONCLUSIONS
Dynamic patterns in the lipidome of postdigestion infant formula samples with different levels of released 3-MCPD were identified.The clustering approach was applied to different levels of released 3-MCPD sample groups and validated in an independent set.The network of lipid nutrients-3-MCPD interactions was constructed which illustrated the biological effect of 3-MCPD on altered lipidome.3-MCPD released during digestion was associated with TG hydrolysis and the bioaccessibility of fatty acids particularly sn-2 palmitic acid in TG.The formation of 3-monochloropropane-1,2-diol ester-pancreatic lipase with the binding energy of −4.9 kcal mol −1 was more stable than triglyceride-pancreatic lipase with the binding energy of −4.0 kcal mol −1 , affecting triglyceride hydrolyzation.The existence of the spatial shielding effect reduced the accessibility of pancreatic lipase and further affected lipolysis.This study provides information on the effect of 3-MCPD on lipid molecules bioaccessibility by the simulated in vitro gastrointestinal tract and lays the foundation for the subsequent nutrition enhancement design and 3-MCPD regulation.

Table 1 .
Jia et al.: PANCREATIC LIPASE IN INFANT FORMULA Distribution of collected infant formula samples and vegetable oil composition according to the package information Jia et al.: PANCREATIC LIPASE IN INFANT FORMULA 4 + (C 54 H 104 O 6 N) as an example, 4 major peaks at m/z 845.75926, 589.51904,  575.50339, and 565.51904were observed (Figure1A).
Jia et al.: PANCREATIC LIPASE IN INFANT FORMULA