Acylated Ghrelin and The Regulation of Lipid Metabolism in The Intestine

Acylated ghrelin (AG) is a gastrointestinal (GI) peptide mainly secreted by the stomach that promotes cytosolic lipid droplets (CLD) hypertrophy in adipose tissues and liver. However, the role of AG in the regulation of lipid metabolism in the intestine remains unexplored. This study aimed at determining whether AG influences CLD production and chylomicron (CM) secretion in the intestine. The effects of AG and oleic acid on CLD accumulation and CM secretion were first investigated in cultured Caco-2/15 enterocytes. Intestinal lipid metabolism was also studied in Syrian Golden Hamsters submitted to conventional (CD) or Western (WD) diets for 8 weeks and continuously administered with AG or physiological saline for the ultimate 2 weeks. In cultured Caco-2/15 enterocytes, CLD accumulation influenced CM secretion while AG reduced fatty acid uptake. In WD hamsters, continuous AG treatment amplified chylomicron output while reducing postprandial CLD accumulation in the intestine. The present study supports the intimate relationship between CLD accumulation and CM secretion in the intestine and it underlines the importance of further characterizing the mechanisms through which AG exerts its effects on lipid metabolism in the intestine.

The alarming increase in the incidence of obesity is a major issue for social health authorities worldwide. Among the various metabolic complications of obesity, lipid and lipoprotein disorders are prevalent and predispose to cardiovascular diseases. In particular, postprandial dyslipidemia constitutes a significant risk factor for the development of atherosclerosis, ischemic stroke and other chronic diseases 1 . Although insulin resistance and amplified intestinal chylomicron (CM) production are considered important processes for the increased postprandial lipemia 2,3 , additional studies are needed to unravel the underlying mechanisms. This is mostly important since an increasing proportion of individuals frequently consume lipid-enriched meals throughout the day and therefore spend most of their time in the postprandial state.
As recently reviewed by our group 4 , apolipoprotein B-48 (apoB-48)-containing lipoproteins are formed in the endoplasmic reticulum (ER), moved to the Golgi apparatus to complete their maturation and are secreted as chylomicrons (CM) across the basolateral membrane of enterocytes in the postprandial state. Once CM particles reach the circulation, their triacylglycerol (TAG) content is hydrolyzed by lipoprotein lipase (LPL) to provide peripheral tissues with non-esterified fatty acids (NEFA). This process reduces TAG content of CM to produce CM remnants (CMR). Thereafter, CMR particles are internalized by the liver or accumulated in the arterial endothelium following chronic postprandial dyslipidemia, thereby increasing the risk of developing cardiovascular diseases. Considerable efforts are currently directed at discovering the factors, which contribute to the pathophysiological processes leading to postprandial dyslipidemia. The rate at which CM are processed is intimately correlated with the accumulation of cytosolic lipid droplets (CLD) 4 , which contain a hydrophobic core of TAG and cholesteryl ester surrounded by a phospholipid monolayer comprising few free cholesterol molecules. CLD are formed by the clustering of neutral lipids within the hydrophobic region of the ER 4 . CLD accumulation and hypertrophy occur in organs and tissues with key metabolic functions such as adipose tissues and liver 5 . In the latter, excessive CLD accumulation is intimately associated with the development of non-alcoholic fatty liver disease (NAFLD) 6 . As recently reviewed by our group, CLDs are key lipid reserves that undergo hydrolysis/ re-esterification to provide the ER with sufficient TAG for CM assembly and secretion 4

Results
Oleic acid treatment and CLD accumulation in Caco-2/15 cells. This section was carried out to determine the conditions promoting optimal CLD growth in cultured enterocytes. In Caco-2/15 cells incubated with OA, CLD expansion was stimulated from 1 h to 6 h while reaching its peak at 24 h (Fig. 1D,E). Also, in response to OA, total CLD area was increased in a dose-response manner (Fig. 1A) while maximal average CLD size was reached at concentrations of 600 and 900 µM (Fig. 1B). For both experiments, CLD numbers were not influenced by the incubation time or OA concentrations (Fig. 1C,F). These results were analyzed from images obtained by confocal microscopy and examples are shown in Fig. 1G. As illustrated, the optimal OA concentration/incubation time enhanced total CLD area by a factor of 2X when compared to control treatments. OA (600 μM for 24 h) also significantly increased Plin-2 ( Fig. 1H) and TAG (Fig. 1I) levels by a factor of 2X in Caco-2/15 enterocytes.

Inhibition of chylomicron secretion and CLD growth in Caco-2/15 cells.
To examine the interaction between CM output and CLD accumulation, we used Lomitapide Mesylate, an inhibitor of microsomal triglycerides transfer protein (MTP), which blocks CM formation. Importantly, the presence of the MTP inhibitor did not alter cell viability and functionality since its addition to Caco-2/15 cells did not affect viable cell count (according to Trypan Blue Dye Exclusion Assay), transepithelial electrical resistance (TEER), sucrase and villin as biomarkers (results not shown). Lomitapide Mesylate significantly reduced CM secretion by 69% (Fig. 1K) while increasing total CLD area per microscopic field (Fig. 1J).
Acylated ghrelin and fatty acid uptake by Caco-2/15 cells. Caco-2/15 cells were incubated with different concentrations of AG (10 pM, 100 pM, 1 nM, 10 nM). AG significantly reduced FA uptake (kinetics and area under the curve, Fig. 2A,B) at all concentrations tested (P < 0.01 vs control treatment) but not at 10 pM. Interestingly, a plateau effect was reached in response to AG at concentrations above 100 pM.

Acylated ghrelin and lipoprotein formation in enterocytes.
In response to AG (100 pM) with or without a GHSR-1a antagonist, [D-Lys 3 ]-GHRP-6 (50 μM) on the apical and basolateral side, no modulation of lipoproteins, TAG or cholesteryl ester (CE) levels was noted in Caco-2/15 cells (Fig. 3). Similarly, no differences were detected in total area and number of CLD per microscopic field (Fig. 3).
Influence of acylated ghrelin on syrian golden hamsters. Since results obtained in cultured cell lines are not always representative of physiological conditions in living organisms, a Syrian Golden Hamster model was subsequently used to test AG's effects on lipid metabolism in the intestine. As presented in Table 1, when compared with CD, WD significantly increased fasting plasma TAG, total cholesterol, free cholesterol, cholesteryl ester and non-HDL cholesterol concentrations, while having no significant effect on HDL cholesterol. This indicated that WD induced hyperlipidemia. With respect to CD treatment, WD rats also displayed increased liver weight. This observation is supported by an elevation of hepatic TAG and cholesterol content, along with an alteration in the visual aspect of their livers (Fig. 4A). Importantly, the quantity of TAG doubled, but the level of cholesterol increased by 6 times more in the liver under the influence of WD. Accordingly, liver histology demonstrated a marked accumulation of cholesterol crystals along with the development of a micro and macro steatosis, and an abnormal height of the hepatocytes in WD-fed animals (Fig. 4B,C). On the other hand, the small intestine was characterized by a significant rise in weight and cholesterol content (Table 1). A reduction in colon weight was also observed in the WD group (Table 1) and this was also underpinned by the decrease of its overall length (Fig. 4D). Only a non-significant increase in glucose, insulin and HOMA-IR 2 was observed without an increment in body weight and epididymal fat ( Table 1). As for AG, it significantly increased body weight in treated hamsters  under the two diets (Table 1). However, no significant difference in the caloric intake was detected between hamsters submitted to the different treatments. Nevertheless, WD + AG animals displayed significantly increased liver and epididymal fat pad weights compared to WD hamsters. AG infusions further increased TAG and total cholesterol levels in the plasma of WD animals, which therefore amplified systemic dyslipidemia. Finally, insulin resistance was slightly increased by AG under CD and WD, but the differences relative to fasting glucose, insulin and HOMA-IR 2 did not reach statistical significance.

Acylated ghrelin and ERK1/ERK2 activation in the small intestine.
To further characterize the increase in the weight of the small intestine, measurements of ERK1/ERK2 and phospho-ERK1/ERK2 levels (proteins known to regulate cell proliferation) were performed 21 . Whereas no significant difference was noted in ERK1/2 protein expression among animal groups (Fig. 5A,B), an increase was observed in phosphoERK1/ERK2 in response to WD feeding (Fig. 5A,C). The phosphorylation was accentuated for animals submitted to WD + AG (Fig. 5D) Acylated ghrelin and chylomicrons output in Syrian golden hamsters. After a 16h-fasting, animals were given an olive oil gavage, and CM catabolism was inhibited by poloxamer 407. Two hours later, the triacylglycerol-rich lipoprotein (TRL) fraction containing CMs and VLDLs was isolated from the plasma. This fraction mostly contained CMs (in view of dietary lipids given to the animals), and to a lesser extent, VLDL particles (due to similarities in their respective density values: d < 1.006 g/L). To ascertain that the lipid contents are representative of intestinal TRL, the analysis of apoB-48 levels in the plasma was also included. Compared to the CD, the WD diet increased TAG (Fig. 6A) and cholesterol content (Fig. 6B) in the TRL fraction while also raising apoB-48 levels (Fig. 6C). On the other hand, AG had no effect on the plasma moieties of TRL, e.g. TAG (

Acylated ghrelin and lipid accumulation in the small intestine. Most of the lipids internalized by
enterocytes are secreted as CM or accumulated in the cytosol as CLDs 7 . Since jejunum is one of the most critical site for lipid absorption in the intestine 22 , we investigated how the diets and AG influence its capacity to generate CLDs. A potent reduction of CLD total area and numbers was observed in the jejunum of WD hamsters ( Fig. 7A-C). Hence, AG significantly decreased CLD total area and numbers both in WD and CD animals ( Fig. 7A-C).

Discussion
Cytosolic lipid droplets are important organelles containing a core of TAG and cholesteryl ester surrounded by a phospholipid monolayer 23 . They are found in many tissues where they are proposed to play distinct physiological roles. For instance, CLD hypertrophy is associated with energy storage and the progression of cardiometabolic disorders in adipose tissues and liver, leading to obesity and NAFLD, respectively 24,25 . The interest towards understanding the physiological relevance of CLDs in the intestine has grown over the recent years and the mechanisms underlying their production and synthesis are under tight regulatory control. Once lipolytic products are absorbed from the intestinal lumen, they are transported to the ER membrane for lipid re-esterification and are used for the synthesis and secretion of CM into the circulation. The lipolytic products may also serve for the CLD production 26 . As reviewed, only a few GI factors are known to modulate CLD synthesis and accumulation in the intestine 4 . One potential regulator of CLD production is AG, a GI peptide known to increase appetite while promoting lipid storage in adipose tissues and liver 27 . Surprisingly, the effect of AG on lipid metabolism in the intestine has been overlooked over the last years. The first objective of the present study was, therefore, to appraise the relation between CM production/secretion and CLD accumulation using cultured Caco-2/15 enterocytes and employing Syrian Golden Hamsters. The second objective was to identify how AG modulates lipid metabolism in the intestine. The major findings show that CLD growth was increased in response to prolonged exposure to high lipid concentrations and this effect was even higher in response to CM inhibition in Caco-2/15 enterocytes. In WD hamsters, CLD content was reduced while postprandial CM secretion was increased. Prolonged AG treatment amplified the dyslipidemia induced by WD, reduced CLD accumulation in the intestine and increased CM output. Overall, our data emphasize a strong connection between CM biogenesis and CLD accumulation.
The present results show that CLD production and hypertrophy were increased in response to prolonged exposure to higher concentrations of oleic acid (OA). The optimal conditions were met when Caco-2/15 cells were incubated with 600 μM of OA for 24 h, using two fluorescent markers for nuclei (DAPI) and neutral lipids/ CDLs (LipidTox), confocal microscopy and image analysis. The latter method was instrumental in characterizing the effects of OA on CLD accumulation in Caco-2/15 cells. In fact, it allowed determining the outcomes of the different treatments on CLD total area and numbers per microscopy field. Characterization of CLDs was always performed using a monolayer of differentiated Caco-2/15 cells plated on transwell filters at 100% confluence. This was critical for standardizing the method. Confocal microscopy was previously proposed as one of the most efficient tools for characterizing CLDs 28 . Using a method similar to the one reported in the present study, OA (600 µM) was shown to increase CLD production by 100% in Caco-2 cells 29 . This effect is potentially more important in Caco-2 enterocytes since these cells do not secrete as much lipoproteins as primary enterocytes. Perilipin www.nature.com/scientificreports www.nature.com/scientificreports/ 2 (Plin-2) and intracellular TAG levels were also measured, which confirmed the results obtained with the fluorescent markers/confocal microscopy.
In our hands, Lomitapide Mesylate inhibited MTP activity and decreased CM production/secretion by 69%, which concomitantly increased total CLD area per microscopy field. Following their uptake by enterocytes of the jejunum, NEFAs are transported to the ER by both I-FABP and liver FABP (L-FABP) 30 . Once transported to the ER, NEFAs are re-esterified as TAG and enter the ER where CM or CLD synthesis is achieved 31 . CM synthesis requires TAG assembly with apoB-48 under the action of MTP 22 . This suggested that MTP inactivation or blockage could prevent CM synthesis and promote CLD production. The present results support the hypothesis that the inhibition of CM production/secretion increases CLD synthesis to prevent intracellular lipotoxicity in the intestine. Our data clearly evidence the intimate relationship between CLD accumulation and CM secretion.
There is little information about the effects of AG on intestinal functions. AG was previously reported to promote enterocyte proliferation in the intestine. However, the present study is the first to investigate its role in lipid transport in enterocytes 32 . According to our findings, AG down-regulated FA uptake in enterocytes. This effect was not initially expected since our group previously reported that AG potentiates NEFA uptake in differentiated 3T3-L1 cells 16 . However, the present data are supported by the observation that AG reduces palmitic acid transport in undifferentiated myoblasts 33 . This indicates that AG exerts tissue-specific actions on NEFA uptake. However, in the conditions tested, AG did not influence lipoprotein secretion or CLD accumulation in Caco-2/15 cells. Treatment with AG significantly reduced NEFA transport at 100 pM, a concentration that was previously shown to be "optimal" according to our group 16 . This treatment with 100 pM of AG can also be considered physiological since plasma concentrations of AG in humans range from 50-300 pM 34,35 . Several factors could have influenced the present results in Caco-2/15 cells. For instance, some peripheral outcomes of AG were reported www.nature.com/scientificreports www.nature.com/scientificreports/ to be mediated by factors related to the CNS 36 . Also, it is possible that AG increased the rate of CM production/ secretion but this effect was no longer noticeable after 24 h of incubation. Previously, Caco-2/15 cells provided critical information on how enterocytes absorb lipids and release them as CMs. Although they were originally isolated from the colon, their capacity to metabolize lipids following full differentiation is closely related to the capacity of enterocytes in the small intestine 37 . However, they display a reduced capacity to secrete lipoproteins (approximately 15-20% of all TAG molecules) and are proposed to synthesize more CLDs than native enterocytes of the small intestine 38 . Based on these factors, it was critical to validate these results in vivo.
The Syrian Golden Hamster is considered as one of the best rodent models for human lipoprotein production in physiological and pathophysiological conditions. In contrast to mice and rats, hamsters also express cholesteryl ester transfer protein (CETP) responsible for TAG and cholesteryl ester exchange between VLDL/LDL and HDL particles 39 . Such a model was previously used to characterize the effects of GI peptides, e.g. glucagon-like peptide 1 (GLP-1) and 2 (GLP-2), on CM secretion 10,40 . When submitted to WD, hamsters developed obesity, dyslipidemia and severe hepatic steatosis, as well as inflammation and reduced weight of their colon 41 . In the present study, hamsters submitted to WD did not develop obesity when compared to CD animals. However, dyslipidemia was present along with elevated lipid content and weight in the small intestine and liver in response to WD. In contrast, reduced colon weight was observed in WD animals. Although not initially expected, resistance to diet-induced obesity and the absence of insulin resistance were previously reported in hamsters and rats by other groups [42][43][44][45][46] .
As mentioned above, the weight of the small intestine was higher in the WD group compared to control diet and was even more increased with AG administration. These results were surprising given the lack of differences in TAG content. To understand these observations, we examined whether other lipid species and cell proliferation may contribute to the weight increment of the small intestine. Indeed, cholesterol content and phosphorylated ERK expression (as a biomarker of cell proliferation) were both elevated, which may play a role in intestinal weight gain. Noteworthy, ERK is an important modulator of cell proliferation in different tissues such as the small intestine [47][48][49][50] and AG was shown, in several studies, to positively influence cell proliferation in many tissues such as intestinal epithelial IEC-6 cells [51][52][53][54] .
In hamsters submitted to CD, continuous AG treatment significantly reduced intestinal CLD total area and numbers per microscopic field. However, AG did not induce any other change in the metabolic phenotype of CD hamsters. In response to WD, AG infusions significantly increased body weight, epidydimal fat and liver weight. Surprisingly, this effect was not associated with an increased caloric intakes, but AG was previously proposed to promote obesity through other mechanisms 55,56 . The most striking differences observed in hamsters submitted to WD and AG infusion were related to the development of dyslipidemia (increased fasting plasma TAG and cholesterol levels). Since AG was previously reported to promote hepatic steatosis 57 , liver TAG and cholesterol levels www.nature.com/scientificreports www.nature.com/scientificreports/ were measured. In contrast to what was observed in the mouse model in physiological and pathophysiological conditions, AG infusion did not increase hepatic TAG and cholesterol concentrations.
In WD hamsters, olive oil gavage promoted CM secretion, this was validated by an increase in TRL lipid contents and apoB-48 levels in plasma, while decreased CLD production in the jejunum was also observed. Our results confirm previous data showing that lipid-rich diets increase CM secretion through an adaptive mechanism intimately associated to insulin resistance 58,59 . In the present study, the higher weight of the small intestine observed in WD animals, suggests an adaptation of the small intestine to dietary fat and sucrose consumption, potentially increasing not only the absorptive surface of the small intestine but also its weight as pointed out above 60 .
In contrast, CD animals displayed reduced CM secretion and increased CLD production by the jejunum. This indicates that hamsters on CD accumulated TAG in CLDs rather than releasing them for the formation and secretion of CM particles. In the conditions tested, AG did not influence CM secretion but it significantly reduced CLD accumulation in CD and WD hamsters. These results are intriguing and indicate that, in response to AG, TAG could be hydrolyzed and utilized through other mechanisms than those involved in lipoprotein production. For instance, AG could prevent TAG re-esterification and direct NEFAs towards β-oxidation 61 rather than CLD or CM production. Since AG reduces lipid absorption in cultured enterocytes, its obesogenic effects could be mediated via mechanisms which bypass intra-enterocyte lipid uptake and involve increased intestinal permeability. Approximately 40% of lipids are directly transported via the portal circulation instead of being re-esterified and sequestered in the ER for CM production/secretion 62 . However, the mechanisms underlying these effects are still unknown. If proven right, AG could potentiate fat absorption through the portal circulation, which promptly stimulates lipid uptake in the liver and adipose tissues 16,57 . This effect could also be explained by the negative correlation between the concentration of ghrelin and certain bile acids 63 . If confirmed, AG could reduce lipid emulsification and absorption in the intestine. Although it is not supported by the present in vitro data, it is also possible that AG promptly enhanced CM secretion while this effect was no longer detectable 2 h after the olive oil gavage. Finally, AG could also influence other yet undefined mechanisms regulating lipid metabolism in the intestine. It is therefore critical to conduct further investigations on the role of AG in lipid metabolism in the intestine. Importantly, we cannot exclude the possibilities that, under the influence of a Westernized type of diet, lipoprotein lipase could be altered. Figure 6. Effect of acylated ghrelin on chylomicron secretion output. Syrian Golden Hamsters were submitted to conventional chow (CD) or Western (WD) diets for 8 weeks. For the last 2 weeks of treatment, animals were continuously administered with acylated ghrelin (AG, 100 nmol/kg/day) or physiological saline. Before sacrifice, hamsters were fasted for 16 h, followed by an olive oil gavage (200 µL). After 20 minutes, they were peritoneally injected with 0.5 mg/kg of Pluronic F-127 to inhibit LPL-induced chylomicron (CM) catabolism. Two hours after gavage, hamsters were sacrificed by an intracardiac exsanguination. Blood was collected and processed to isolate and measure triglyceride-rich lipoproteins (TRL) content. Triacylglycerol (A) and total cholesterol (B) concentrations were determined. The apoB-48 content (C) was obtained by Western blots and results are illustrated as densitometric ratios of apoB-48 to protein content (stain-free gels). The apoB-100 (D) and apoB-48 (E) content of the fasted plasma taken by the tail was also analyzed by Western blots. A representative blot is shown, illustrating an experiment in triplicate on the same gel and at the same exposure. The results of every experiment are shown as mean ± SEM of n = 3-6 animals * P < 0.05; ** P < 0.01; *** P < 0.001. (2019) 9:17975 | https://doi.org/10.1038/s41598-019-54265-0 www.nature.com/scientificreports www.nature.com/scientificreports/ The high levels of plasma TAG and cholesterol in fasting conditions indicate the presence of dyslipidemia following AG administration. As the liver is usually a significant contributor of VLDL, it was important to determine apoB-100, the most characteristic protein of VLDL. Our findings stress no increment in apoB-100, which suggests the lack of VLDL overproduction by the liver, thereby ascribing the dyslipidemia to intestinal CM output. An explanation for the failure of the liver to enhance VLDL secretion in response to WD and WD + AG may be associated with the extremely high level of cholesterol crystals, the macro and micro steatosis, and the reduction of hepatocytes size. Additional efforts are needed to delineate the mechanisms.
In conclusion, the present results indicate that AG plays a role in fat absorption, uptake or metabolism through yet undefined mechanisms in the intestine. This report is first to show that in addition to its adipogenic effects, AG also promotes dyslipidemia via intestinal lipoprotein secretion under the fasted state. This suggests the critical need to use an integrative approach to determine how GI peptides regulate lipid metabolism in key metabolic organs such as the intestine.

In vitro lipoprotein output and isolation by ultracentrifugation. Radiolabeled [ 14 C]-oleic acid
(Perkin Elmer, USA) was added to unlabelled oleic acid and both were conjugated to fatty acid-free bovine serum albumin (BSA). This preparation (600 µM of unlabelled + 0,45 µCi of radiolabeled oleic acid) was added to the apical side of differentiated Caco-2/15 cells for 24 h. For investigating the role of ghrelin on lipid metabolism, enterocytes were also incubated 24 h with AG (100 pM) on basolateral and apical sides with or without 50 µM of a GHSR-1a antagonist [D-Lys 3 ]-GHRP-6 (Tocris Bioscience, UK). Following the incubation period, enterocytes were scraped in cold PBS containing 1 mM of pepstatin A, PMSF, and aprotinin. The basolateral medium was collected and lipid-rich plasma (as a carrier for the lipoproteins synthesized by Caco-2/15 cells) was added (4:1 ratio). The lipoproteins were isolated by ultracentrifugation (OptimaMax Ultracentrifuge, Beckman Coulter, USA): chylomicrons (CM) at 25,000 rpm for 40 min, VLDL and LDL at densities of 1.006 g/mL and 1.063 g/mL, respectively, at 100,000 rpm for 2h30, HDL at density of 1.21 g/mL and at 100,000 rpm for 6 h. Scintillation liquid (Ecolite, MP Biomedicals, USA) was added to the isolated fractions in order to determine the radioactivity The selected concentration of AG was previously shown to increase food intake in rats 66 . After the implantation of mini-pumps, hamsters were injected with a subcutaneous dose (5 mg/kg) of Ketoprofen (Merial, Canada).
Liver histology and electronic microscopy. For hematoxylin-eosin staining, tissues were fixed in formalin 10% and embedded in paraffin. Thereafter, they were processes for light microscopy with hematoxylin-eosin. Images of stained tissues were captured by a Zeiss Imager A1. Measurements were taken with the axiovision software. For electronic microscopy, the liver was cut into small pieces and fixed with 3% glutaraldehyde. Afterwards, it was post-fixed with 1% osmium tetroxide, followed by a consecutive incubation in an ascending acetone series (50%, 70%, and 90%), with a final incubation in 100% ethanol for hydration. After incubation, the tissues were embedded in Epon. 1-nm sections were cut and stained using uranyl acetate and lead citrate. Electron microscopy images were taken with the Phillips EM208 electron microscope. estimation of in vivo postprandial chylomicron production. Two weeks after mini-pumps were inserted, hamsters were fasted for 16 h and 500 µL of blood was collected from the tail vein under anesthesia with isoflurane/O 2 , before receiving an oral gavage of pure olive oil (200 µL). Twenty minutes after gavage, animals were injected with an intraperitoneal (ip) dose (0.5 mg/kg) of Pluronic F-127 (poloxamer 407; 10% in saline, Millipore Sigma, USA) to inhibit lipoprotein lipase activity and prevent CM catabolism. Two hours after gavage, animals were sacrificed by cardiac exsanguination and blood was transferred into EDTA-containing tubes (Vacutainers, B&D, USA). Blood was centrifuged at 6000 rpm for 10 minutes at 4 °C. Plasma and isolated tissues were flash-frozen and kept at -80 °C for further analyses. Fresh plasma was used to isolate triglyceride-rich lipoproteins (CM and VLDL) by ultracentrifugation. Briefly, 250 µL of plasma was added to 4 mL of a sodium chloride buffer (density of 1.006 g/L) and centrifuged at 35,000 rpm for 70 minutes at 4 °C. The upper fraction was removed and its lipid moieties, TAG and cholesterol, were respectively determined by commercial kits from Randox Laboratories Ltd (United Kingdom) and Wako Diagnostics (USA). www.nature.com/scientificreports www.nature.com/scientificreports/ Lipid analysis in the jejunum. Approximately 0.1 g of jejunum tissue or liver was homogenized in an EDTA buffer and lipids were extracted in a 2: 1 chloroform/methanol solution for 18 h. After evaporating the inferior phase with N 2 gas, lipids were resuspended in 800 µL of nanopure H 2 O (Millipore, USA). TAG and total cholesterol were analyzed with the same kits used for the biochemical analysis. For phospholipids quantification, tissue homogenates were analyzed using the Bartlett method 68 . In short, homogenates were diluted in perchloric acid and the solution was heated at 160 °C for 3 hours. Water, ammonium molybdate, and 2.4-Diaminophenol Dihydrochloride were incorporated and heated another 7 minutes at 96 °C. Phospholipid concentrations were determined at an absorbance of 830 nm.

Biochemical analyses in
Western blot analysis. For Western Blots, 1 µL of plasma or 20 µg of proteins was heated for 10 min at 80 °C with a buffer containing sodium dodecyl sulfate (SDS) and dithiothreitol (DTT). Samples were loaded and migration was achieved on 6% SDS-PAGE gels while protein transfer was performed on a nitrocellulose membrane (GE Healthcare Lifescience, Canada). The membrane was then blocked for 1 h in TBS-Tween buffer containing 5% of skimmed milk and washed in TBS-Tween buffer. Incubation was then achieved overnight with the appropriate primary antibody. The antibodies used are: apoB-48 primary antibody at 1/1000 in milk (apoB, MAB012, Millipore Sigma, USA), Plin-2 primary antibody at 1/250 in milk (Plin-2, 690102, Progen, Germany), ERK1/ ERK2 primary antibody at 1/1000 in milk (ERK1/ERK2, 61-7400, ThermoFisher, Canada) and phospho-ERK1/ ERK2 (phospho-ERK1/ERK2, MA5-15173, ThermoFisher, Canada). A specific secondary antibody conjugated with horseradish peroxidase was used to detect the relative amount of primary antibody. Bands were analyzed with the ImageLab software (Biorad). For apoB-48, total protein content was determined using stain-free gels (Bio-Rad, USA) and sample normalization was carried out using these values, as previously described 69 . For Plin-2, β-actin (1:250 000, Sigma-Aldrich, USA) was used for sample normalization while GADPH (1:1000, Abcam, USA) was used for ERK and phosphoERK.

Statistical analysis.
Results are presented as mean ± SEM. Statistical analyses were conducted using ANOVA and the differences between means were analyzed using post-hoc Turkey tests. If the experiment did not pass the Shapiro-Walk normality test, the Kruskal-Wallis test was used. Significance is considered at P < 0.05.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.