Dietary milk polar lipids modulate gut barrier integrity and lipid metabolism in C57BL/6J mice during systemic inflammation induced by Escherichia coli lipopolysaccharide

The focus of this work is the role milk polar lipids play in affecting gut permeability, systemic inflammation, and lipid metabolism during acute and chronic inflammation induced by a single subcutaneous injection of lipopolysaccharide. Three groups of C57BL/6J mice were fed: modified AIN-93G diet with moderate level of fat (CO); CO with milk gangliosides (GG); CO with milk phospho-lipids (MPL). The MPL did not prevent a gut permeability increase upon LPS stress but increased the expression of tight junction proteins zonula occludens-1 and occludin in colon mucosa. The GG prevented the gut permeability increase upon LPS stress. The MPL decreased absolute and relative liver mass and decreased hepatic gene expression of acetyl-CoA carboxylase 2 and 3-hydroxy-3-methylglutaryl-CoA reductase. The GG increased hepatic gene expression of acetyl-CoA acyltransferase 2. In conclusion, milk GG protected the intestinal barrier integrity but had little effect on systemic inflammation and lipid metabolism; milk MPL, conversely, had complex effects on gut permeability, did not affect systemic inflammation, and had beneficial effect on hepatic lipid metabolism.


INTRODUCTION
Obesity is a serious health concern across the globe (Tsur and Twig, 2022).It is well accepted that obesity is associated with a state of low-grade, chronic metabolic inflammation, which is a link between adiposity and the complications of obesity, such as a leaky gut, dyslipidemia and non-alcoholic fatty liver disease (NAFLD) (Clemente-Postigo et al., 2019).Chronic feeding of high fat diets results in the increased intestinal permeability, endotoxin absorption, and low-grade metabolic inflammation (Rohr et al., 2020).Gut permeability plays an important role in metabolic inflammation.Nutrient metabolism and inflammatory pathway activation are linked by critical energy metabolism regulators, such as Toll-like receptors and peroxisome proliferator-activated receptors (Johnson et al., 2012).A milk extract, rich in polar lipids, prevented the increase in intestinal permeability to fluorescein isothiocyanate (FITC)-dextran in mice stressed by intraperitoneal LPS (Snow et al., 2011).A subclass of milk polar lipids, sphingolipids, have lipidlowering and anti-inflammatory properties (Norris and Blesso, 2017).Sphingomyelin (SM), one subfraction of the sphingolipids, plays an important role in neonatal gut maturation during the suckling period in rats (Motouri et al., 2003).Milk SM reduces adipose tissue inflammation in high fat diet induced obese mice (Norris et al., 2017).Another subfraction of the milk sphingolipids, gangliosides (GG), inhibit the degradation of gut occludin tight junction (TJ) protein during the LPS-induced acute inflammation (Park et al., 2010).Additionally, GG were shown to affect the intestinal immune system maturation in mice during weaning (Vazquez et al., 2001).Taken together, dietary milk polar lipids may directly reduce systemic inflammation and may also decrease inflammation by preventing the increase of gut permeability.
High fat diets contribute to the development of NAFLD (Hebbard and George, 2011).According to the "two-hit" hypothesis by Day et al. (Day and James, 1998), saturated fatty acids may be the first hit and LPS the second (Csak et al., 2011).The second hit leads to hepatic inflammation and non-alcoholic steatohepatitis (NASH) (Guo et al., 2018).LPS stress in combination with high fat diets may result in NAFLD/NASH (Csak et al., 2011).NAFLD is associated with low-grade systemic inflammation (Haukeland et al., 2006).NAFLD, gut leakiness and systemic inflammation may hasten the development of NASH.Prior work has shown that milk polar lipids reduce hepatic steatosis in mice fed a high fat diet (Wat et al., 2009).The phospholipid concentration and species composition of the intestinal mucus barrier are significantly altered when the intestinal barrier is compromised (Braun et al., 2009).Maintenance of the phosphatidylcholine (PC) in the hydrophobic surface may play an important role in promoting health and preventing disease (Dial et al., 2008).It may be hypothesized that dietary milk polar lipids help maintain the intestinal barrier integrity and therefore reduce the endotoxin absorption and the metabolic inflammation associated with moderately high fat feeding.
The mouse model of gut leakiness and systemic inflammation induced by lipopolysaccharide (LPS) is well established.LPS at 2 mg/kg (Mathan et al., 1988) and 10 mg/kg (Wang et al., 1998) body weight injected subcutaneously induces endotoxemia and intestinal stress in mice.Chronic models mimicking human sepsis are more clinically relevant than rapidly fatal models (Deitch, 1998).LPS elevates the serum concentrations of various cytokines in mice, including interleukin-6 (IL-6), monocyte chemotactic protein-1 (MCP-1), and tumor necrosis factor-α (TNF-α) (Nandi et al., 2010).Approximately 30% of the subcutaneously LPS injection leaves the injection site within an hour and the injected LPS is retained at the injection site for more than a month (Yokochi et al., 1989).Thus, both acute and chronic inflammation may be achieved by injecting LPS subcutaneously.
In addition to the aforementioned classes, milk polar lipids also contain phosphatidylcholine (PC), phosphatidylethanolamine(PE), and phosphatidylserine (PS) (Zhou and Ward, 2019).This study was designed to test the hypotheses that dietary milk polar lipids from milk fat globule membrane (MFGM; (Zhou et al., 2012)), including MPL and GG (at similar doses as in previous studies, 1% and 0.02% by wt, respectively (Zhou and Ward, 2019)), may prevent an increase in gut permeability and plasma inflammatory cytokines, reduce liver lipids, and affect the expression of genes associated with fatty acid synthesis and cholesterol regulation in the liver during the acute inflammatory response induced by the subcutaneously injected LPS.We further hypothesized that dietary MFGM polar lipids may facilitate the recovery of the aforementioned endpoints during the chronic inflammation induced by the LPS stress.

Diet Formulation
Three moderate-fat (34% by energy) diets based on the AIN-93G rodent diet were formulated: control (CO), gangliosides (GG), and milk polar lipids (MPL) (Zhou and Ward, 2019).The diets were identical in macronutrient and micronutrient content, except for the composition of the fat.The MPL and GG were sourced from Fonterra USA (Rosemont, IL).The composition of the MPL and GG were analyzed via HPLC and P31 NMR.The GG was supplemented at 0.2 g/kg diet and the MPL was supplemented at 10 g/kg diet, and the polar lipid and sterol composition are listed in Table 1.The overall macronutrient and micronutrient composition and fatty acid profiles of the diets are available in a previous publication (Zhou and Ward, 2019).One percent of MPL in the diet is equivalent to about 106.5 mg phospholipids or 35.3 mg SM kg -1 of body weight in a human.The human equivalent dose for GG is about 2.7 mg GG kg -1 of body weight (Milard et al., 2019a).

Animals and Stresses
Power analysis indicated that a sample size of 4 -6 can be used to detect changes in inflammatory cytokines.Five-week-old male C57BL/6J mice (n = 18; Jackson Laboratory) were randomly assigned by using Microsoft Excel to one of the following treatments: 1) CO diet (n = 6); 2) GG diet (n = 6); 3) MPL diet (n = 6).To avoid introducing confounders, mice were given alphanumerical identifier which were used throughout the duration of the study.Mice were housed singly and maintained as previously described (Zhou and Ward, 2019).Feed intake and body weight were measured every other day.The body composition was assessed at baseline, d 2, 5, 11, 23, 32, 39, and 54 by using magnetic resonance imaging (MRI; EchoMRI-700, EchoMRI, Houston, TX).Whenever possible, measurements were made randomly across groups so as not to introduce confounding.The mice were fed the experimental diets for 2 weeks before they were challenged with subcutaneous LPS (Escherichia coli 0111:B4; Sigma-Aldrich) at 5 mg/kg body weight.The concentration and injection location was selected to induce intestinal stress and endotoxemia as prior work has shown that LPS injected subcutaneously at 2 mg/kg (Mathan et al., 1988) and 10 mg/kg (Wang et al., 1998) induced intestinal stress and endotoxemia but did not cause animal deaths.The site of subcutaneous injection was the loose skin over the upper back.After LPS injection and oral gavage of sugar probes, 0.5 mL PBS was injected subcutaneously to facilitate urine production.The animals were observed twice a day during the following week to monitor the gross pathological changes, including eye discharge.After the LPS challenge, all animals were fed the experimental diets for another 6 weeks.
Mice were sacrificed by CO 2 asphyxiation after a 4-h fast.One mouse from each group was killed in a row before moving to the second mouse.Blood, liver, quadriceps muscle, intestinal and colonic mucosa, feces, and adipose tissue samples were collected and every tissue mass was recorded.The tissue samples were flash frozen and stored at −80°C until further analysis.
The experimental timeline is shown in Figure 1a.The experiments were conducted in conformity with the Public Health Service Policy on Humane Care and Use of Laboratory Animals and were approved by the Utah State University Institutional Animal Care and Use Committee.The protocol number was 1507.

Assessments of Intestinal Barrier Integrity
Intestinal permeability was assessed by the FITCdextran absorption test (baseline, d 34 and 57) and by the differential sugar absorption test (DST) every 2 weeks (baseline, d 16, 29, and 44) (Zhou and Ward, 2019).Briefly, the mice were gavaged with sugar probes and were housed in metabolic cages to collect the 24-h urine.Urinary sugars were quantified using gas chromatography with flame ionization detection to assess gut permeability.To reduce stress to the animals, blood was not collected until 2 weeks after the LPS challenge.In addition, the DST and blood withdrawn were not applied in the same week.For the DST after the LPS challenge, the LPS was injected in the afternoon and the DST was initiated in the morning of the following day.Gut permeability may reach its peak during 12 -24 h after the LPS stress (Snow et al., 2011).By the end of the study, the jejunal, ileal and colon mucosa were collected, the mucosal total protein was extracted using tissue protein extraction reagent (Zhou and Ward, 2019) and the tight junction proteins (zonula occludens-1, ZO-1 and occludin) were analyzed by Western blot (Zhou and Ward, 2019).

Biochemical Analyses of Plasma and Liver
Blood samples were collected by cheek bleeding and plasma samples were analyzed to measure glucose, insulin, leptin, resistin, MCP-1, IL-6, TNF-α, tissue plasminogen activator inhibitor-1 (tPAI-1), and endotoxin (Zhou and Ward, 2019).Hepatic expression of genes associated with lipid metabolism in the liver was analyzed by realtime quantitative polymerase chain reaction (RT qPCR).

Tissue Lipid Profiling
Lipid profile of skeletal muscle, liver, and adipose tissue were analyzed by using high performance thin layer chromatography (HPTLC) (Zhou and Ward, 2019).The following lipid classes were quantified via relative band density: PE, PC, PS, phosphatidylinositol (PI), SM, diglycerides (DG), free fatty acids (FFA), triglycerides (TG), and cholesteryl ester (CE).The total GG content in the intestinal mucosa was determined by measuring the gangliosides bound sialic acid with gas chromatog- raphy-mass spectrometry (GC-MS) (Zhou and Ward, 2019).

Statistical Analyses
The investigators were aware of the group allocation since the initiation of the experiment.One-way or mixed model ANOVA (ANOVA) was performed by SAS 9.2.For one-time measurements, one-way ANOVA was performed and the group means were compared by the Ryan-Einot-Gabriel-Welsch Multiple Range Test.For repeated measurements, mixed model ANOVA was performed to assess diet, time and diet*time effect, and the group means were compared by the Least Squares Means Contrast.The data were reported as mean ± standard error of the mean (SEM).

Food Intake, Body Weight and Body Composition
One mouse each from the GG group and the MPL group died 3 d after the LPS stress.Another mouse each from the GG group and the MPL group died 5 d after the LPS stress.For the data analysis after the LPS challenge, the following number of mice was used: CO, n = 6; GG, n = 4; MPL, n = 4.
There were no significant differences among the groups regarding daily diet intake (4.17 ± 0.07, 3.95 ± 0.13, 4.04 ± 0.15 g/d for CO, GG and MPL, P = 0.36, Figure 1b) and total weight gain (5.02 ± 0.21, 6.03 ± 0.62 and 4.85 ± 0.58 g for CO, GG and MPL, P = 0.20).LPS stress, blood sampling, and housing in metabolic cage decreased food intake (Figure 1b).There was no significant difference among groups for body weight at any time point (Figure 1c).Overall, MPL decreased body weight compared with GG (23.56 ± 0.20 vs 24.53 ± 0.19 g, P = 0.01), and there was no significant difference in body weight between GG and CO groups (24.53 ± 0.19 vs 23.91 ± 0.17 g, P = 0.077).The mice gained body fat during the first 2 weeks and there was no significant body fat increase after the LPS stress (Figure 1d).The GG and MPL groups had higher body fat percentage during the first 2 weeks after the dietary treatments compared with the CO group and the difference tended to remain during the rest of the study (diet effect: P = 0.03, Figure 1e).The fat mass increase was 1.28 and 1.88 g more in the GG and MPL groups compared with the CO group during the first few days.The faster initial increase in body fat in GG and MPL groups may be due to the slightly higher food intake during the first 2 d in GG and MPL groups.But the absolute fat mass increase was small.There was no significant difference among groups in final body fat percentage.The disappearance of the difference over time could be due to adaptation to the diets since the differences in food intakes decreased across groups.
The final body fat percentage in CO, GG and MPL group was 16.50 ± 1.24%, 17.71 ± 1.64%, 15.42 ± 1.50%, respectively.There was no significant difference in body fat mass among groups at any single time point.In general, there should be a linear increase in body fat with age in C57BL mice as a function of feeding (Fenton, 1956).While the C57BL/6J mice are prone to high fat diet-induced obesity (Takahashi et al., 1999) and the mice were fed a moderately high fat diet in the present study, the mice did not gain body fat during the 6 weeks after the LPS stress (Figure 1d and e).The mice gained considerable amount of the body weight during that period of time (Figure 1c).The MPL group gained less body weight compared with the GG group.
Compared with CO, MPL decreased body lean mass, especially after LPS stress (17.92 ± 0.23 vs 17.07 ± 0.24, P = 0.057, Figure 1f).MPL group had lower body lean mass at d 23 compared with CO group (P = 0.049).This finding is consistent with a previous report showing that MFGM polar lipids suppresses high-fat dietinduced body weight gain in C57BL/6J mice (Milard et al., 2019a).That high-fat diet had 48.2% calories from fat, which was much higher than the dietary fat level in the present study.The smaller effect size in the current study may be due to the much lower fat content in the diet.The LPS stress could be another factor contributing to the discrepancy.MFGM material comprising mostly of protein and a small amount of polar lipids attenuates high-fat diet-induced obesity by inhibiting adipogenesis (Li et al., 2018).It cannot be ruled out that the small amount of polar lipids may have suppressed weight gain in the study done by Li et al.MFGM proteins may have different effect on weight gain compared with casein and whey proteins (Price et al., 2021).Therefore, both MFGM polar lipids and proteins have the potential to suppress weight gain during diet-induced obesity.MFGM has no significant effects on food intake and weight gain during early life and adulthood of normal mice (Snow et al., 2011, Bhinder et al., 2017), but improves growth performance and alleviates the intestinal damage induced by LPS challenge during the early life of low birth weight mice (Huang et al., 2019).Data in the current study further support that MFGM phospholipids and gangliosides may have different effect than MFGM proteins on the growth performance in mice.Further study is needed to compare the effect of MFGM polar lipids and proteins on weight gain.Dietary PC at 5.65% (wt/wt, choline content equivalent to 1.0% choline chloride) suppressed food intake and weight gain compared with 1.0% choline chloride (Sugiyama et al., 1987).The difference in PC content between the MPL and CO diets in the current study was only 0.5%.The smaller effect size of the current study could also be due to the smaller dose of PC.

Tissue Lipid Profiles and Gene Expression
Semiquantitative Analysis of Tissue Lipids by HPTLC.The MPL group had lower liver mass (P = 0.01) and lower liver index (as liver weight/body weight × 100%, P = 0.024) compared with the CO group while the GG group was in the middle (Figure 2a).This is consistent with previous reports of MPL's effect on liver weight in C57BL/6J mice (Wat et al., 2009, Milard et al., 2019a).The following lipid classes were measured in adipose, muscle and liver: PE, PC, PS, PI, SM, DG, FFA, TG, and CE.Those classes with significant differences are shown in Figure 2b.MPL lowered TG in the muscle (Figure 2b) and in the liver.There were no differences in the following lipid classes among groups: adipose DG, TG, PC, PE, SM; muscle DG, CE, PC, PE, PS, SM, FFA; liver DG, CE, PC, PE, PS, SM.The MPL group had less liver TG compared with the CO group (posthoc contrast P = 0.04).Milk SM at 0.1% (wt/wt) lowers TG in the liver but not in the skeletal muscle or the serum of C57BL/6J mice (Norris et al., 2017).Dietary SM level (0.3%, wt/wt, for the MPL group) was much higher in the current study, the lower liver TG level in the MPL group could be due to the higher amount of SM.Choline chloride and soy PC at similar dosage as in the current study lower liver TG, raise plasma CE while the increase by PC is smaller in Wistar rats (Sugiyama et al., 1987).The GG and MPL groups had more FFA in the liver compared with the CO group (Figure 2b).The MPL group had more PI in the skeletal muscle compared with the CO and GG groups (Figure 2b).GG and MPL increased PI in the liver (Figure 2b).The milk polar lipids contained a small amount of PI.Dietary PI at 2% (wt/wt) suppresses the increase of inflammatory cytokines and prevent the pathogenesis and development of liver injury in C57BL/6 mice injected with concanavalin A (Inafuku et al., 2013).The PI content was much lower (<0.13%,wt/wt) in the current study.The significance remains to be elucidated for the increase of PI in the muscle and liver by dietary polar lipids.
Of all the lipid classes measured in the mucosa, the MPL group had less PE and PC in the ileum mucosa and less PE in the colon mucosa on a per organ basis compared with the other 2 groups (group P = 0.0085, P = 0.016, P = 0.067, respectively, Figure 2c) given the fact that the MPL diet had much higher level of these polar lipid classes.This indicates that dietary phospholipids may be negatively associated with mucosal phospholipids.The MPL group had lower level of gangliosides in colon mucosa compared with the GG group (P = 0.026, Figure 2d).It is not clear through which mechanism did the dietary polar lipids affect the mucosal polar lipids.The influence of intestinal length may be ruled out since there was no difference in intestinal length (31.80 ± 0.72, 31.28 ± 1.02, 29.13 ± 0.77 cm for CO, GG, and MPL, P = 0.10).Semiquantitative Analysis of Plasma Lipids by HPTLC at Three Time Points.The following lipid classes were analyzed: CE, TG, FFA, DG, PC, PI, PE and SM.Those classes with significant differences are shown in Figure 3. Plasma CE was lower in the GG and MPL groups at d 34 compared with the CO group (P = 0.04, P = 0.003, respectively, Figure 3a).Plasma CE decreased between d 34 and 57 (P < 0.001, Figure 3a).In the current study, we observed an increase of plasma CE in the CO group but not much in the GG and MPL groups (Figure 3a).The plasma CE levels decreased in all groups to below the baseline level by the end of the study.The increase of plasma CE in the CO group should have been due to the choline bitartrate added in the diet.The total choline content was balanced across all 3 diets for the current study.Milk PC affected plasma CE differently from choline bitartrate.The relevant mechanism remains to be elucidated.Plasma TG decreased over time and there was no significant treatment effect (time effect P < 0.0001, Figure 3b).Plasma free fatty acids decreased over time (P < 0.0001) and the MPL group had higher level at d 34 compared with the CO and GG groups (P = 0.04, Figure 3c).Plasma DG increased slightly toward d 34 and decreased significantly from d 34 to 57 (P < 0.0001, Figure 3d).There was no diet effect for plasma DG.There was no treatment effect on either plasma PC (Figure 3e) or SM (Figure 3f).Plasma SM increased from baseline to d 34 (Figure 3f).

Gut Permeability Analyses.
Western Blot Analysis of TJ proteins.The MPL increased the expression of ZO-1 and occludin in the colon mucosa compared with the CO and the GG (Figure 4a).The GG and the MPL decreased occludin expression in the jejunum mucosa compared with the CO (Figure 4a).In jejunum mucosa, PE and PC content were negatively associated with ZO-1 expression (r = -0.71,P = 0.004; r = -0.74,P = 0.003, respectively).The increased amount of the TJ proteins in colon mucosa of the MPL group could increase the integrity of the colon but the MPL group did not have lower plasma LPS (Figure 4c).Not only the amount of the TJ proteins but more importantly the distribution of the TJ proteins affects the intestinal permeability (Ferraris and Vinnakota, 1995).Plasma LPS was negatively correlated with the occludin in the jejunum mucosa (r = −0.84,P = 0.038) in the CO group but not in the GG and MPL groups.The decreased jejunal occludin in the MPL group may be compensated by other mechanisms provided by the MPL that may increase the gut integrity.Sphingomyelin, at similar dose as in the MPL diet, increases tight junction expression in male C57BL/6 mice (Milard et al., 2019b).There are limitations of total protein abundance measures for evaluating tight junctions compared with immunofluorescence staining, which may demonstrate the change of TJ proteins in tight junction complexes and the architectural continuities in the inner lining of the intestinal epithelium (Woo et al., 2016).There was no difference among groups for ZO-1 in the jejunum mucosa, which was negatively associated with PE and PC content (r = -0.71,P = 0.004; r = -0.74,P = 0.003, respectively).

Intestinal Permeability Assessment by FITC-dextran Absorption Test at Three Time Points.
Gut permeability to FITC (as measured in plasma after oral gavage) decreased in all groups relative to the baseline level and there was no treatment effect among groups (Figure 4b).The general trend of the decreased gut permeability to FITC in all groups may be explained by the maturing of the gut barrier during development since the mice were entering adulthood (3 mo old) by the end of study (Flurkey et al., 2007).Plasma FITC at d 35 was lower than the level at baseline.In the study by Snow et al., the gut permeability of FITC was assessed at 24 h and 48 h after the LPS stress (Snow et al., 2011).Plasma FITC in the current study may have increased within 48 h after the LPS stress.To reduce stress level, plasma FITC was not measured during that time window when the mice were kept in the metabolic cages for 24h immediately after the LPS injection.Additional factors may contribute to the difference between the Snow's result and the current result.First, the LPS was administered intraperitoneally and subcutaneously, respectively.Second, the milk polar lipids-rich material used in the Snow's study contained milk proteins and nonpolar lipids.That material was more complex than the milk polar lipids concentrate used in the current study.

Plasma Endotoxin Measurement by Fluorescence
Assays.The presence of bilateral purulent discharges in all mice was a good indicator of successful systemic inflammation induced by the bacterial LPS.At d 34, the MPL group had higher plasma LPS compared with the GG group while plasma LPS was not at a high level 3 weeks after the LPS challenge (Figure 4c).There could be 6% of the injected LPS still retained at the injection site at d 34 (Yokochi et al., 1989).The low plasma LPS may indicate that the injected LPS did not contribute to plasma LPS at this time point.Five weeks of feeding plus the LPS challenge 2 weeks after dietary treatment did not increase plasma LPS significantly in the CO and GG groups but increased plasma LPS significantly in the MPL group (Figure 4c).The difference disappeared after the LPS were normalized by the body fat mass (Figure 4d).This indicates that plasma LPS may be directly associated with the body fat mass.After 3 more weeks of experimental feeding post the LPS stress, plasma LPS increased significantly in all 3 groups (Figure 4c).By the end of the study, plasma LPS reached a very high level and the injected LPS should have been gone by this time.The increased plasma LPS at d 57 may not be contributed significantly by the injected LPS.There was no accompanying increase of proinflammatory cytokines at d 57.It may be deduced that LPS from the gut was much less proinflammatory compared with external LPS.There are differences in lipid A portion of LPS from different bacteria (Caroff and Karibian, 2003).As a result, there are notable differences in endotoxin activity/toxicity of Bacteroides LPS and Escherichia coli LPS (Johne et al., 1987).The plasma LPS by the end of study could be mainly contributed by bacteria such as Bacteroides, which is more abundant than Escherichia coli (Coats et al., 2011).The Escherichia coli LPS injected subcutaneously could also trigger different immune response compared with the symbiotic counterpart (Coats et al., 2011).

Intestinal Permeability Assessment by DST at Four Time Points.
The excretion of sucrose and lactulose increased upon the LPS stress and then returned back to baseline level by d 29 (Figure 5a and b).Sucrose is rapidly hydrolyzed by sucrase-isomaltase once it moves beyond the gastroduodenal region.The increase of sucrose excretion indicated permeability increase of the gastroduodenal region after LPS stress (Fihn et al., 2000).There was no treatment effect for intestinal absorption of sucrose (Figure 5a).The increase of lactulose excretion implied an increase of paracellular permeability for the small intestine (Norman et al., 2012).GG tended to decrease lactulose excretion after the LPS stress but the difference did not reach statistical significance (Figure 5b).GG could prevent the increase of paracellular permeability for the small intestine upon LPS stress.The excretions of mannitol and sucralose increased upon the LPS stress and then returned back to baseline level by the end of the study (Figure 5c &d).Although the permeating pathway of mannitol has not been fully understood, transcellular permeation is involved (Norman et al., 2012).The increase of mannitol absorption after LPS stress was an indication of increased mannitol permeation through transcellular pathway.Lactulose and mannitol are degraded by the bacterial flora of the colon while sucralose is stable throughout the gastrointestinal tract (Arrieta et al., 2006).Compared with CO, GG prevented the increase of mannitol excretion (56.4% of control) and sucralose excretion (35.6% of control) after the LPS stress (Figure 5c &d), indicating that GG prevented the increase of small intestine permeability and colon permeability, respectively, by LPS stress.Therefore, GG may prevent the increase of both paracellular and transcellular permeability of the small intestine upon LPS stress.A MFGM supplementation, containing phospholipids and gangliosides, prevents shortening and histological damage of the colon in rat pups challenged with Clostridium difficile toxin (Bhinder et al., 2017).The differential effects of protein and lipid components from MFGM on gut permeability await further investigations.
The urinary lactulose/mannitol ratio increased significantly after the LPS challenge at d 18 and returned to the baseline level by d 44.There was no treatment effect (Figure 5e).The increase of the urinary lactulose/ mannitol ratio indicates the increase of small intestinal permeability (Meddings and Gibbons, 1998).The LPS stress should have resulted in villous atrophy and therefore decreased the absorptive area (Iba et al., 1998).The milk phospholipids increased the permeability of small intestine as indicated by the increased urinary sucrose/ lactulose ratio compared with the CO at d 29 and 44, led to a higher level of plasma LPS, and did not affect plasma inflammatory cytokines (Figure 4c, 5f &6).The milk gangliosides protected gut barrier integrity as indi-cated by preventing the increase of mannitol and sucralose excretions after the LPS stress (Figure 5 c and d) but did not affect systemic inflammation (Figure 6 b and f).
The urinary sucrose/lactulose ratio decreased after the LPS challenge and increased during the rest of the study.The MPL increased the urinary sucrose/lactulose ratio compared with the CO at d 29 and especially at d 44 (Figure 5f).The decrease of the urinary sucrose/lactulose ratio reveals distal damage of the small intestine while the damage became proximal when the sucrose/lactulose ratio increases (Meddings and Gibbons, 1998).The rise of urinary sucralose/mannitol ratio without the lactulose/ mannitol ratio being affected indicated that the permeability of the colon increased (Meddings and Gibbons, 1998).The DST indicated an increase of colon permeability at d 29 (Figure 5 d and f).The decrease of the jejunal occludin in the MPL group was not associated with the increased gut permeability as indicated by plasma LPS/FITC and the DST.The decrease of sucrose/lactulose ratio should have been mainly due to the increase of lactulose absorption.After LPS stress, the urinary lactulose increased to 3-4.5 times of baseline while the sucrose/lactulose ratio decreased to 40-90% of baseline (Figure 5b and f).So, there should have been also an increase of sucrose absorption, which may be accounted for by several factors, including permeability increase of the proximal small intestine and the decrease of sucrase activity.The LPS stress and experimental feeding increased the permeability of the proximal small intestine, which was coupled with the decreased expression of the TJ protein occludin in the jejunum mucosa.Taken together, the high-fat feeding in the context of systemic inflammation may have resulted in the colonic damage while the LPS-induced systemic inflammation may have caused the permeability increase of the small intestine.A low fat control group will be needed to confirm this observation.
Oral single dose soy PC at 100 mg/kg prevents LPSinduced intestinal permeability changes in 150-250g Male Sprague-Dawley rats (Dial et al., 2008).The PC dosages of the GG and MPL groups were 97.3 mg/kg and 774.1 mg/kg, respectively.Since the choline content was balanced across 3 diets, it's possible that the effect of PC on intestinal permeability was compensated by the added choline.
Systemic inflammation and plasma cytokines.All mice developed severe conjunctivitis 16h after the LPS stress as indicated by the bilateral purulent discharges.Plasma insulin, leptin, MCP-1, IL-6, TNF-α and tPAI-1 were measured at 3 time points during the study.Plasma insulin decreased from baseline to d 34 and appeared to remain stable until d 58 (Figure 6a).The lack of body fat accumulation may be due to the decrease of plasma insulin since insulin plays an important role in increas- ing body fat storage (Dimitriadis et al., 2011).The decrease of plasma insulin (Figure 6a) may be caused by the injected LPS.The mean retention of the LPS injected subcutaneously is 73.4% after 6h, 49.1% after 3 d, 23.1% after 14 d, and less than 6% after 32 d in the injection site (Yokochi et al., 1989).Intraperitoneally injected LPS damages pancreatic cells and decreases the expression of the insulin mRNA in the pancreatic tissues (Saitoh et al., 2004).Much higher dose of LPS was injected subcutaneously in this study, it may induce similar pancreatic damages.Further morphological assessment of the pancreas may confirm the damage.The intravenous injection of LPS decreased plasma insulin from 0.5 ng/ml at baseline to 0.2 ng/ml at d 11 in C57BL/6 mice fed the chow diet at 4 mo of age (Tweedell et al., 2011).The proinflammatory cytokine, TNF-α, decreases serum insulin in C57BL/6 mice at 10-11 weeks of age (Endo et al., 2007).
Plasma IL-6 increased (measured at d 34) upon LPS stress and then decreased (measured at d 57) almost to the baseline level.There was no treatment effect for IL-6 (Figure 6b).The increase of IL-6 at d 34 should be due to the LPS challenge.In low birth weight female C57BL/6J mice at postnatal d 21, intraperitoneal LPS at 10 mg/ kg BW increased IL-6 level, which was higher than the IL-6 level in adult mice of the current study (Huang et al., 2019).MFGM presupplementation prevented the increase of IL-6 upon intraperitoneal LPS stress in C57BL/6J mouse pups (Huang et al., 2019) and milk polar lipids presupplementation prevented the increase of IL-6 upon oral LPS stress in SD rat pups (Yang et al., 2020).The MPL in the current study at similar dosage to the aforementioned MFGM dosage did not affect IL-6 level.It's possible that the proteins and other components in the MFGM may have different effect from MPL.When rat puts were stressed by LPS at 5 mg/kg through oral gavage, milk polar lipids had a dose-dependent effect on IL-6 level in intestinal tissue (Yang et al., 2020).The intestinal IL-6 level was the lowest when milk polar lipids was supplemented at 100 mg/kg compared with 50 and 200 mg/kg, indicating that too much milk polar lipids may facilitate the inflammatory response.Plasma IL-6 was not correlated with the plasma LPS at the end.This could mean that the LPS absorbed from the gut is not as proinflammatory as the injected LPS.The intestinal LPS may have been deactivated.The intestinal alkaline phosphatase (IAP) secreted by the enterocytes detoxifies the intestinal LPS through dephosphorylating the lipid A moiety.The IAPs reduce the serum LPS content in Wistar male rats (Koyama et al., 2002).The IAPs affect both the toxicity and the concentration of the LPS in plasma.The expression of IAP in the intestine needs to be measured to reveal whether the IAPs reduced plasma LPS in this study.The LPS is absorbed gradually from the gut into the circulation.The tolerance to the LPS could also help explain the lack of proinflammatory response by the LPS absorbed from the gut (Andrade et al., 2019).The tolerance to LPS-induced increase of serum colony-stimulating factor develops in C57BL/6J mice after either intravenous or intraperitoneal injection of LPS and the tolerance occurs after one to 2 preinjections (Quesenberry et al., 1975).The inhaled low-dose LPS can induce adaptation to subsequent higher doses for cellular and inflammatory parameters in bronchoalveolar lavage from young and old C57BL/6J mice (Elder et al., 2000).A third explanation is that plasma LPS may be deactivated by anti-LPS antibodies (Poxton, 1995).Further studies may reveal whether the anti-LPS antibodies were involved in the process of deactivating plasma LPS that were absorbed from the gut in the current study.
Plasma tPAI-1 increased slightly during the first 34 d and then increased at a higher rate toward the end of the study (Figure 6c).The plasma tPAI-1 level was much lower than that of male C57BL/6J mice fed a diet with 42% fat by energy for 12 weeks without LPS stress (Wahlang et al., 2013).The adipokine tPAI-1 could be positively associated with adipose tissue mass.The LPS stress did not change plasma tPAI-1 level.The increase of tPAI-1 could be due to the moderately high fat diet feeding.
Plasma resistin stayed stable during the first 34 d and then increased (Figure 6d).The plasma resistin level (Figure 6d, 3 ~5 µg/L) was much lower than that (29.3 ug/L) for male C57BL/6J mice not stressed with LPS (Zhou et al., 2015).The lower resistin level should be due to the LPS stress, which inhibits the expression of resistin (Xiang et al., 2013).Leptin increased toward the end of the study (P = 0.05) but there was no treatment effect (Figure 6e).GG and MPL may have suppressed leptin production compared with CO (Figure 6e).Leptin is directly associated with the mass of white adipose tissue and inflammation (Otero et al., 2006).The increase of leptin overtime should be due to the accumulation of body fat mass.There was no time or treatment effect on plasma TNF-α (Figure 6f).The plasma TNF-α was slightly higher than that of male C57BL/6J mice fed diets with lower fat (Zhou et al., 2015) and was similar to that of C57BL/6J mice fed diets with 41% fat by energy (Andersson et al., 2010).
This study confirmed some findings reported by other groups and provided a relatively comprehensive assessment of the effects of dietary supplementation of milk polar lipids on gut permeability, systemic inflammation and lipid metabolism during LPS induced acute and chronic inflammation.Although a significant number of endpoints were monitored, more further studies are needed to fill the gaps for understanding the effects of milk polar lipids in this model.It is important to assess different organ systems when dietary supplements are studied.This paper provides strong evidence that endpoints should be assessed at multiple time points and in all relevant organ systems during assessment of dietary supplements to generate integrative and unbiased opinions.
The moderately high fat feeding and the LPS stress together resulted in increased gut permeability to small molecules, increased body weight and decreased body fat content, and increased plasma LPS and inflammatory cytokines.The data from the current study indicated that the acute and chronic inflammation induced by the subcutaneously injected LPS is a valuable animal model for studying systemic inflammation and gut permeability.The animal death in this study may be due to the LPS challenge and the metabolic cage housing.Certain group differences might haven't been detected due to the decreased sample size.However, significant findings on bioactivities of milk polar lipids should be explored further in follow-up experiments in acute and chronic inflammation models and in humans.The LPS dosage may be decreased (e.g., at 2.5-3 mg/kg body weight) to prevent animal death for chronic studies.The increased LPS absorption from the gut into plasma was not accompanied by the increase of plasma proinflammatory cytokines, but the injected LPS increased systemic inflammation.Further studies are warranted to explore the mechanisms for the different effect of intestinal LPS and exogenous LPS.

CONCLUSIONS
During the acute and chronic inflammation induced by subcutaneous LPS challenge, the dietary MPL decreased the liver mass, affected hepatic expression of genes associated with fatty acid and cholesterol synthesis, increased the permeability of the small intestine, and did not affect plasma inflammatory cytokines; the GG protected intestinal barrier integrity by preventing the permeability increase of both small intestine and the colon, slightly affected lipid metabolism, and did not affect the systemic inflammation.The data suggest that milk GG and MPL have potential in promoting gut health and improve lipid metabolism upon inflammatory stress.

NOTES
This research was supported by the Western Dairy Center.Robert E. Ward was partially supported by the Utah Agricultural Experiment Station (#9214).
The data used for this article are available upon request.
The authors declare that there is no conflict of interest.

Figure 1 .
Figure 1.Effects of the polar lipids on body weight and body fat content.(a) Experimental timeline.(b) Food intake.(c) Body weight.(d) Body fat increased during the study.(e) Body fat change.(f) Lean mass.*P = 0.049 for MPL vs CO.The data represent mean ± SEM.
Zhou and Ward: MILK LIPIDS MODULATE GUT INTEGRITY & LIPID METABOLISM Lipid Metabolism Analyzed by RT qPCR.Those genes affected by diet treatment are shown in Figure 2e.The GG increased hepatic expression of the β-oxidation gene Acaa2 (vs the CO & the MPL, P = 0.01, P = 0.0.014,respectively).The MPL suppressed hepatic expressions of the fatty acid synthesis gene Acacb (vs the CO, P = 0.002) and the cholesterol synthesis gene Hmgcr (vs the GG, P = 0.023).It is not clear which fraction(s) of the polar lipids led to the aforementioned changes in gene expression.The following genes associated with lipid metabolism in the liver were not affected by the treatments: fatty acid synthesis: Elovl5, Slc27a5, Me1 and Scd1; fatty acid oxidation: Acox3 and Cpt2; cholesterol regulation: Acat2, Ldlr, Scarb1 and Cyp7a1.
Zhou and Ward: MILK LIPIDS MODULATE GUT INTEGRITY & LIPID METABOLISM

Figure 2 .
Figure 2. Effects of milk polar lipids on liver mass, tissue lipid profile, and gene expression associated with lipid metabolism in the liver.(a) Liver mass.(b) Muscle and liver lipids.(c) Mucosa lipids.(d) Mucosa GG.(e) Hepatic expression of Acaa2, Acacb and Hmgcr.The data represent mean ± SEM.

Figure 3 .
Figure 3. Effects of milk polar lipids on plasma lipid profile at baseline, d 34 and d 57.(a) Plasma CE. *P = 0.003 for MPL vs CO, # P = 0.04 for GG vs CO.(b) Plasma TG.(c) Plasma FFA.# P = 0.04 for GG vs CO.(d) Plasma DG.(e) Plasma PC.(f) Plasma SM.The data represent mean ± SEM.

Figure 4 .
Figure 4. Effects of milk polar lipids on gut permeability.(a) ZO-1 and occludin expression in mucosa.(b) Plasma FITC.(c) Plasma LPS.*P = 0.04 for MPL vs GG.(d) Plasma LPS normalized by body fat mass.Means in a row with different superscripts are significantly different (P < 0.05).The data represent mean ± SEM.
Zhou and Ward: MILK LIPIDS MODULATE GUT INTEGRITY & LIPID METABOLISM
Zhou and Ward: MILK LIPIDS MODULATE GUT INTEGRITY & LIPID METABOLISM

Table 1 Polar lipids and sterol composition of diets (unit: g/kg) 1
Zhou and Ward: MILK LIPIDS MODULATE GUT INTEGRITY & LIPID METABOLISM