LIPO-PROTEIN EMULSION STRUCTURE IN THE DIET AFFECTS PROTEIN DIGESTION KINETICS, INTESTINAL MUCOSA PARAMETERS AND MICROBIOTA COMPOSITION

Scope Food structure is a key factor controlling digestion and nutrient absorption. We tested the hypothesis that protein emulsion structure in the diet may affect digestive and absorptive processes.


INTRODUCTION
Pre-processed or ultra-processed food has become a prominent part of the diet in western countries [1,2]. This rise in their consumption may partly account for the increased prevalence of a number of metabolic diseases [3,4]. Observational nutritional studies have shown that the impact of diet on health parameters cannot be explained only by the chemical composition itself, but that the food structure, which can be substantially altered by food processing, should also be taken into account [5,6]. Along with other food constituents, the functional properties of proteins in food matrixes can be drastically modified by the processing methods [7,8]. The emulsifying properties of the proteins are particularly sensitive to thermal treatments, which reduce protein emulsion droplet size and may affect their digestibility [9]. Other (food) processing procedures currently used in human and animal food industry such as freezing, extrusion or high pressure treatment also alter the emulsion properties of proteins [7], thereby potentially affecting food digestion and nutrient absorption.
In addition, according to different protein sources, heat treatments and gelation may influence the micro-and macrostructure of food matrices, and consequently, dietary protein digestion and absorption [10,11]. In humans, different dietary nitrogen (N) sources can display different coagulation properties in the stomach acidic environment. For instance, casein is retained longer in the stomach than the whey proteins [12] [13], impacting the speed of amino acid (AA) absorption in the gut and affecting the whole body protein anabolism [14]. When compared to milk, dairy product gel structure delays gastric emptying rate and subsequently decreases the kinetics of AA absorption in minipig [15]. Similarly, the way milk coagulates (acidification versus renneting) affects the AA digestion, release and apparent bioavailability in the same animal model [10]. In humans, a semi solid meal decreases the gastric emptying rates and improves satiety feeling, when compared to liquid meal of similar composition, meals (3g, dry matter) which had the same composition as experimental diets but whey proteins were intrinsically labeled with 15 N. Rats were euthanized 15 min (n=10), 1 h (n=10) or 5 h (n=10) after the end of the test meal (5 rats in each group for each time point). Five rats from the LFE and 5 from GCE groups were fasting (not fed the test meal) when euthanized.
The rats were anesthetized with pentobarbital sodium (40 mg/kg body weight ip) before collecting portal and peripheral blood into tubes containing EDTA and protease inhibitor as described before [24]. Plasma were frozen after centrifugation and stored at -80°C for peptide and biochemical analysis. Gastric, intestinal, cecal and colonic contents were collected, weighed and frozen at -20°C for 15 N enrichment measurements. An aliquot of cecal contents to be used for SCFA analysis was immediately frozen in liquid N and stored at -80°C.
Rinsed jejunum and ileum segments were sampled and stored at -80°for gene expression analysis. This article is protected by copyright. All rights reserved. 8 derivatized and analyzed with a gas chromatograph equipped with a capillary column (30 m, 0.32 mm ID; RestekRtx 502.2) and fitted with a flame ionization detector [29]. SCFA concentrations were determined by external standards with reference to internal standards and SCFA were expressed as percentage of total SCFA.

Intestinal mRNA extraction and gene expression analysis
Total RNA was isolated from jejunum and ileum segments using mirVana® miRNA isolation kit (Ambion, France). Reverse transcription was performed with 2 µg total RNA using High capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA). RT-PCR was performed based on TaqMan gene expression assays with predesigned Taqman primers and probes for the Rat (Assays-on-Demand TM , Gene Expression Products; Applied Biosystems) (Table S1). β-actin, a housekeeping gene, was used to normalize the mRNA abundance of each target gene as described previously [24]. Reads with length ≥ 400 bp were kept and checked for quality and length using sickle [31] and corrected for known sequencing error using SPAdes [32]. Reads were clustered at 97% This article is protected by copyright. All rights reserved. 9

Microbiota analysis
of identity using Vsearch pipeline. Chimeric OTUs (operating taxonomic unit) were identified using UCHIME and discarded. OTU classification was performed using RDP classifier [33]. OTU sequences were aligned using ssu-align. OTUs with abundances lower than 0.005% were removed from the analysis [34]. Supplementary Table S3 summarizes sequencing results for each sample. Sequencing data were analyzed using the Phyloseq package in R [35].

Statistical analysis
Effect of food structure on microbiota composition was tested using linear model with the DESeq2 analysis pipeline, using Benjamini Hochberg multi-testing correction for the pvalues. Alpha-diversities (within sample microbial diversity) were verified using observed OTU numbers, Chao1 and Shannon indexes. Digestibility differences were tested using a ttest to compare the mean of the two groups at each time. Comparisons of the incorporation of dietary N in the studied N pools were analyzed using a two-way analysis of variance (structure of the emulsion and digestion time), performed with the GLM procedure of SAS.
Bonferroni's posthoc tests were used for pairwise comparison (SAS 9.1, SAS Institute, Cary, NC, USA: Mixed Procedure). Differences were considered statistically significant at P< 0.05. For all the other data two-way ANOVA was performed using GraphPad Prism Version 6.02 for Windows (GraphPad Software, La Jolla California).

Effect of the lipo-protein emulsion structure of the diet on food intake, body weight, body composition and plasma metabolite concentration
Rats fed LFE or GCE diets displayed similar growth rates and food intake levels ( Figure S1).
The decrease in food intake at day 30 matched the habituation of the rats for prompt test meal This article is protected by copyright. All rights reserved. 10 groups, but TG and NEFA levels changed over time after the test meal ( Table 2). The NEFA levels were elevated 1h and 5h after the test meal compared to the levels in fasting rats. The plasma concentrations of TG were higher 1h after the test meal but returned to the fasting levels 5h after, in both groups.

Transfer of dietary N to serum proteins, amino acids and urea
The transfer of dietary N to serum proteins, amino acids and urea at different time points after the test meal is shown in Figure 4. Significant effects of time and emulsion structure-by-time interaction were observed in these three N pools. The incorporation of dietary N into serum proteins progressively increased over the post-prandial period and was faster with the diet containing the LFE than the GCE ( Figure 4A). One hour after the test meal, the amount of dietary N transferred to serum proteins was higher in the LFE group than in the GCE group  continued to increase until 5h ( Figure 4B). Five hours after the test meal, the transfer of dietary N to serum AA pool was lower for the LFE than the GCE group (P=0.03). The transfer of dietary N to body urea increased until 1 h for both groups, then slowly declined for the LFE group but remained stable for the GCE group ( Figure 4C). The amount of 15 N was significantly higher for the LFE group than for the GCE group after 1 h (P=0.02), but the residual amount of dietary N in body urea was not different among groups after 5 h and represented 6% of the ingested N. Total urea level increased overtime after the test meal.
However, 5h after the test meal, the urea level dropped 30% in the LFE group but was maintained in the rats fed GCE diet. The 5-h AUCs for the transfer of dietary N to serum proteins, amino acids and urea were identical in both groups.

Amino acid and peptide transporter expression in the small intestine
In the jejunum, none of the expression levels of AA and peptide transporters tested were affected by the protein-lipid emulsion structure of the diet (Table 3). In the ileum, the expression levels of two genes (Slc6a14 and Slc3a1) encoding the cationic AA transporter systems, system B (0,+) and b (0,+) , were higher in the LFE group than in the GCE group. A similar tendency was also observed for the expression levels of the genes encoding the system X -AG and the basolateral system A. Surprisingly, a strong time effect was observed in the jejunum for four transporters (Slc15a1, Slc6a19, Slc36a1 and Slc3a1) and in the ileum for two transporters (Slc15a1 and Slc6a14). The expression level of these genes was significantly reduced 5h after feeding in both GCE and LFE groups. Possibly, this reduction in AA transporter expression was the return to normal levels after an adaptive response to the This article is protected by copyright. All rights reserved. 12 overnight fast. Indeed, expression levels of Slc15a1, Slc6a19, Slc3a1 and Slc1a2 were upregulated by fasting (Table S2).

Impact of the lipo-protein emulsion structure on the gut peptide circulation levels
Changes

Bacterial metabolite composition in the lumen of the caecum
Since changes in the release of nutrients from the diet could potentially modify the substrates available for the microbiota, we evaluated the caecal SCFA composition before and 5h after the test meal. The feeding status did not modify the SCFA composition (Table 4). However, the percentages of acetate and butyrate (products of bacterial metabolism of carbohydrates and AA) varied according to the lipo-protein emulsion structure of the diet. In the caecum of LFE fed rats, the proportion of acetate was lower while the percentage of butyrate increased compared to GCE fed group (Table 4). In addition, the percentage of isovalerate was approximately 3 times higher in the rats fed LFE when compared to GCE. The total proportion of SCFA specifically derived from AA (isobutyrate, isovalerate and valerate) was This article is protected by copyright. All rights reserved. 13 also higher in LFE than in GCE, due to the higher proportion of isovalerate in the caecum of LFE fed rats.

Impact of the lipo-protein emulsion structure on gut microbiota composition
Fecal microbiota composition, in terms of phyla ( Figure 6A) and alpha diversity, was similar whether the diet was GCE or LFE (data not shown). At lower taxonomic levels, important differences were observed in the three most abundant phyla. In the Firmicutes phylum, abundances of OTUs from Clostridium Cluster XIV, and from Streptococcaceae family, were lower in feces from LFE fed rats compared to GCE ( Figure 6B). In contrast, OTUs assigned to the Parabacteroides genus (Bacteroidetes phylum), were significantly more abundant in LFE fed rats ( Figure 6C). Moreover, OTUs assigned to Bifidobacterium genus (Actinobacteria phylum) were significantly less abundant in LFE fed rats compared to GCE fed rats. Interestingly, OTUs from the Proteobacteria phylum were strongly impacted at almost all taxonomic levels ( Figure S2, A, B, C.1, C.2 and C.3). The most impacted OTUs were from the Gammaproteobacteria class, being more abundant in LFE than in GCE fed rats, and from the Betaproteobacteria class, conversely less abundant in LFE fed rats ( Figure   S2, A). Among the Betaproteobacteria, , the abundance of Sutterella and Parasutterella genera was decreased in LFE relative to GCE fed rats ( Figure S2, B, C.1, C.2, C.3).

DISCUSSION
Our results clearly indicate that the structure of the lipo-protein emulsion in the diet markedly impacts dietary protein digestion kinetics and the subsequent absorption of amino acids and oligopeptides. Indeed, when compared to the GCE diet, the LFE diet was found to accelerate protein digestion and incorporation of the 15 N label into blood proteins and urea. These  with two similar lipo-protein emulsions, can be extrapolated to an in vivo physiological situation.

Digestion kinetics and endocrine response of the gut.
We observed more rapid digestion kinetics with LFE diet than with GCE diet. This likely results in faster nutrient (AA and peptide) release into the lumen and their subsequent absorption. Solid foods are known to be emptied more slowly than liquid meals when matched for caloric content [36] and macronutrient composition [37], due to the role of the pylorus that allows only small particles to leave the stomach. The increase in consistency and viscosity has also been shown to impact gastric emptying [38], as reported with milk gelation [15] or fermentation [39] that markedly slowed the gastric emptying of proteins. Accordingly, in our study, the texture triggered important differences in N kinetics. It is noticeable that 1 h after the meal, 80 % of N from LFE has already been absorbed subsequently to faster digestion kinetics. This is in agreement with previous studies in humans showing that whey protein is completely emptied from the stomach within 2h hours after ingestion, displaying a marked peak of dietary N appearance in plasmatic AA [13,40]. Protein digestibility is mainly driven by intrinsic protein characteristics, such as chemical composition and tridimensional structure that are responsible for resistance to hydrolysis [41]. In the present study, protein This article is protected by copyright. All rights reserved. 15 digestibility did not differ between GCE and LFE groups despite important kinetic differences, likely resulting from differences in tridimensional structure between the two lipoprotein emulsions. This discrepancy between digestion kinetic and digestibility had been shown previously in humans in whom delaying gastric emptying by adding sucrose did not modify ileal digestibility of milk protein [42]. In addition, milk protein, both casein and whey, have been found to be the most digestible among dietary proteins [43,44], whatever their digestion speed, fast for whey and slow for casein. Thus, despite a more rapid kinetic of protein digestion in the LFE diet compared to the GCE diet, we did not observe any difference in dietary N absorption 5h after the meal, this being consistent with the similar weight gain of animals in the two groups. increasing numbers from the jejunum to the colon [46]. This suggests that by inducing a faster release of nutrients into the lumen, LFE diet impacted the nutrient-sensing by the GIPproducing entero-endocrine cells in the proximal part of the small intestine, but not the entero-endocrine cells located in the distal GI tract. This may be linked to the absence of any significant impact of the emulsion structure on the overall digestibility 5h after the test meal as observed in our study. GIP release is well known to be stimulated by carbohydrates and lipids [47]. Despite a less clear mechanism of the K-cell activation by AA or proteins, a previous human study showed a pronounced postprandial increase in GIP as soon as 15 min after the ingestion of whey drink compared to other protein sources [48]. This was associated with the marked raise in leucine, isoleucine, valine, lysine, and threonine in plasma suggesting the potentially crucial role these AA play in whey-induced GIP release [48].
However, subsequent study showed that ingestion of a mixture of these AA failed to stimulated GIP release [49] suggesting that GIP secretion is affected differently by free AA vs AA released from dietary protein, perhaps because of different absorption kinetics.
Chronic effects of the emulsion structure in the distal region of the gut.
While the lipo-protein emulsion structure did not have acute effects on the endocrine response in the distal region of the gut following the test meal, the chronic ingestion of the two diets during 3 weeks differentially affected the ileum AA transporter expression profile and caecum microbiota metabolic activity and fecal bacteria communities. Despite a similar N absorption between groups after 5h, the impact of emulsion structure on digestion kinetics likely affected the nature of proteins or peptides reaching the lower part of the gut and thus the bioavailability of specific AA for the microbiota metabolic activity. This is supported by This article is protected by copyright. All rights reserved. 17 the effects that we observed of the emulsion structure on the SCFA composition in the caecum, especially the relative increase in isovalerate under the LFE diet. Since isovalerate is a branched SCFA exclusively produced by deamination of leucine, our results suggests an increased protein fermentation in the caecum after LFE consumption (branched-chain fatty acids are markers of protein fermentation [50]). This effect may also be ascribed to the longer residence time of residual proteins in the caecum that arrive earlier with LFE than with CGE.
In addition to isovalerate, the relative amount of butyrate was also higher in the caecum of the LFE fed rats while acetate was decreased compared to the GCE group. Thus, in this study fermentation processes led to various bacterial metabolite profiles associated with, according to the chemical structure and luminal concentrations of metabolites, both detrimental and beneficial effects on the colonic epithelium [51].
The composition of bacterial metabolites in the caecum could also be affected by changes in the microbiota composition and metabolic activity following ingestion of the two different diets: the LFE-fed rats showed an increase in OTUs from the Parabacteroides genus (isovalerate producers) which could also partly explain the increase of the isovalerate percentage measured in the caecum of those rats [52,53]. Among the other bacterial groups significantly impacted by the lipo-protein emulsion structure, Sutterella and Parasutterella proportions significantly increased under the GCE diet. Both genera contain bile acids resistant species [54,55]. We previously proposed that lipids in GCE diet are less digestible This article is protected by copyright. All rights reserved. 18 However, total gastric contents were not significantly different between the two groups 15min and 1h after the test meal (data not shown). Therefore, since starch is the main component of the diet, it is unlikely that starch digestive kinetics were altered by the emulsion structure.
The two diets also differentially affected gene expression of specific AA transporters in the ileum (Slc6a14, Slc3a1), an effect that was not observed in the jejunum. This confirms our  [56], but is less clearly established for most AA transporters. However our data (table S2) clearly demonstrate a significant impact of overnight fasting on Slc6a19, Slc3a1 and Slc1a2.
This suggests that the difference in the expression of genes corresponding to the AA transporter may be affected by the nutritional status of the rats.

Conclusions and perspectives
In conclusion, our study highlights the impact of lipo-protein emulsion structure on gastric emptying, kinetics of protein digestion, and nitrogenous compound absorption and metabolism. The LFE diet, when compared to the GCE diet, accelerated dietary N disposal and metabolism and likely as a consequence, modified the microbiota and bacterial metabolite composition in the large intestine. However, these differences were not associated with any difference in the food consumption and body weight. Thus, the modifications in food structure tested here, although impacting important parameters of GI physiology and metabolism, are not sufficient per se to affect such rough endpoints. This is likely explained by the fact that LFE and GCE diets were both well-balanced diets. In the future, in order to evaluate the impact of lipo-protein structure on metabolic and physiological parameters related to the health status it will be important to make lipo-protein emulsion structure in the context of unbalanced diets using different sources or amount of lipids or carbohydrates. To distinguish between deleterious and beneficial effects of the metabolites resulting from bacterial proteolysis, the impact of the lipo-protein emulsion structure should be evaluated on renewal, homeostasis, and barrier function of the colonic epithelium in a long-term perspective [57]. Finally, our data support the idea that food structure and its impact on the digestion kinetics should be taken into account to fully understand the impact of the food on health, and to develop future food products with optimal structure from these points of view.     This article is protected by copyright. All rights reserved. 30    This article is protected by copyright. All rights reserved. 32    This article is protected by copyright. All rights reserved. 34 This article is protected by copyright. All rights reserved. 36 This article is protected by copyright. All rights reserved. 37 This article is protected by copyright. All rights reserved. 39 This article is protected by copyright. All rights reserved. 41