Post-prandial secretion of glucagon-like peptide-2 (GLP-2) after carbohydrate-, fat- or protein enriched meals in healthy subjects

Glucagon-like peptide 2 (GLP-2) is an important regulator of intestinal growth and function. In adherable mixed meals the macronutrient composition with the best potential for stimulating GLP-2 secretion is not known. We compared the effect of 3 iso-energetic meals, where approximately 60% of the energy ratio was provided as either carbohydrate, fat, or protein, respectively, on the post-prandial endogenous GLP-2 secretion. The responses were compared to secretion profiles of peptide YY (PYY), and glucose-dependent insulinotropic peptide (GIP). Ten healthy subjects were admitted on three occasions, at least a week apart, after a night of fasting. In an open-label, crossover design, they were randomized to receive a high carbohydrate (HC), high fat (HF) or high protein (HP) meal. The meals were approximately ~3.9 MJ. Venous blood was collected for 240minutes, and plasma concentrations of GLP-2, GIP and PYY were measured with specific radioimmunoassays. Mean GLP-2 levels peaked already at 30minutes for the HC meal, however the HP meal induced the highest mean GLP-2 peaking levels, resulting in significantly higher mean GLP-2 area under the curve (AUC) from baseline of 7279 pmol*min/L, 95%-CI [6081;8477] compared to the HC meal: 4764 pmol*min/L, 95%-CI [3498;6029], p=0.020 and the HF meal: 4796 pmol*min/L, [3385;6207], p=0.011. Findings were similar for the PYY. The HC meal provided a greater AUC for GIP compared to the HP- and HF meals. The HP meal was most effective with respect to stimulation of the postprandial GLP-2 and PYY secretion, whereas the HC meal was more effective for GIP.


Introduction
Glucagon-like peptide 2 (GLP-2), a single chain polypeptide of 33 amino acid residues, is produced by a tissue-specific posttranslational processing of the precursor, proglucagon [1,2], in the intestinal L-cells [3] and secreted following meal-stimulation.The open-type L-cells are most numerous in the ileum and the colon [4] and are among the most frequently occurring of at least a dozen of different types of enteroendocrine cells (EECs) scattered in the intestinal epithelium among the enterocytes [5].Typically, the open-type EECs are conically in shape with a small apical pole with presumably sensory microvilli facing the gut lumen and a broader base, from which the peptide hormones are released by exocytosis of secretory granules.Thus, many gut hormones, GLP-2 included, are secreted in relation to luminal sensing after meal stimulation as well as in response to indirect enteroendocrine and neural mechanisms [6].The EECs function both through classical endocrine mechanisms and frequently also through activation of afferent, sensory nerves, in addition to their local, paracrine functions [7].Multiple, especially metabolic functions are regulated by the gut hormones, which send information via multiple pathways about the condition in the various segments of the gastrointestinal tract to a wide spectrum of organs and tissues in the body [5,8].
Secretion of the gut hormones is characteristically regulated by the arrival of and presence of food components (carbohydrate, fat and protein) in the lumen of the gastrointestinal tract, and special attention has been attributed to glucagon-like peptide 1 (GLP-1) and peptide YY (PYY) from L-cells, and glucose-dependent insulinotropic polypeptide (GIP), which are secreted from the K-cell in the proximal duodenum, due to their roles in the treatment of obesity and type 2 diabetes [9].GLP-2, which is co-secreted with GLP-1, has attracted less attention, but it is now established that it has a multifaceted role in the regulation of intestinal growth and function.Thus, GLP-2 promotes expansion of the intestinal mucosa via stimulation of crypt cell growth and inhibition of enterocyte apoptosis in relation to exogenous administration [10].In addition, GLP-2 inhibits gastric acid secretion and gastric emptying [11,12], stimulates intestinal blood flow [13,14] increases intestinal barrier function [15,16], opposes inflammatory insults [17,18], enhances nutrient and fluid absorption [19][20][21] and has an anti-catabolic effect on the bone tissue by inhibiting bone resorption [22].With these diverse effects, exogenous administration of native GLP-2 as well as GLP-2 analogues has been implemented as treatment for short bowel syndrome [23] and suggested for inflammatory bowel disease [18], chemotherapy induced mucositis [24] and osteoporosis [22].However, an alternative strategy would be to identify and target substances, dietary or others, that would enhance the postprandial GLP-2 secretion.To clarify the effect of diet composition on postprandial endogenous GLP-2 secretion, we compared the effect of 3 isoenergetic (~3.9 MJ) mixed meals, possible to adhere to, containing approximately 60 % of energy from either carbohydrate, fat or protein, respectively, in healthy subjects.In order to relate our findings to secretory patterns of other important gut hormones, we also measured PYY, co-secreted from at least the distal L-cells, and glucose-dependent insulinotropic polypeptide (GIP) from the more proximally located K-cells, thought to be preferentially secreted in relation to high carbohydrate diets [25].

Participants
Ten healthy subjects were recruited for the study.None of them had any medical history, and none had physical or mental illness or current pregnancies.The 4 male and 6 female participants were 33 years in average (range 24-55) and had a mean body mass index (BMI) of 23.4 kg/m 2 (range 19.6-28.6 kg/m 2 ).

Study protocol/Experimental design
The 10 healthy subjects were admitted to the Department of Intestinal Failure and Liver Diseases, Rigshospitalet, on three occasions at least a week apart after a night of fasting.In an open-label, crossover design, they were randomized to receive a high carbohydrate (HC), high fat (HF) or high protein (HP) meal.The meals were isoenergetic, as illustrated in Table 1, and the dietary components of the meals are presented in Table 2.
Venous blood was collected at 15 min before and 10, 20, 30, 45, 60, 120, 180 and 240 min after the start of the meals.The meals were completed within 15 min.The blood samples were collected in icechilled 10 ml EDTA tubes containing the DPP-4 inhibitor valinepyrrolidine and stored at − 20 • C until analyzed.The immunoreactivity of the GLP-2, PYY and GIP were measured with specific radioimmunoassays using ethanol-extracted plasma as described previously [26][27][28].The GLP-2 assay is directed against the intact N-terminus of GLP-2 and therefore measures the concentrations of active GLP-2; the assay was extensively validated [26].The detection limit was below 2 pmol/L and the intraassay coefficient of variation < 6 %.Radioimmunoassay of PYY was performed using antiserum code no.8412-211 [27].Detection limit of the assay was below 1 pmol/L and the intraassay coefficient of variation was < 5 %.Antibody R65 which reacts with the C-terminal tripeptide, was used to measure total GIP [28].The detection limit was below 5 pmol/L and the intraassay coefficient of variation was 8.8 ± 1.0 %.All samples from the same individual were measured in the same assay run.

Statistical analysis
The mean change in peptide concentrations (GLP-2, PYY and GIP respectively) from baseline value (at − 15 min) in response to the three different isoenergetic meals (HC, HF and HP respectively) were estimated together with 95 % confidence intervals (CI).CI for the mean change difference between the meals were also estimated, by taking account the pairing of the data.The area under the curves (AUC) of change in GLP-2 from baseline for each patient were computed.The mean difference in AUCs between meals were compared using Welch's ttest and a matching 95 % CI.This approach was chosen since the F-test can even be rather counterproductive, since the F-test can lead to paradoxical situations in which a difference between the three groups is shown (i.e., the F-test is significant) while none of the pairwise comparisons shows sufficient evidence of a difference in mean between any two groups [29].CI are important for a thorough interpretation of the results to distinguish between statistical results that suggest lack of evidence from those who suggest lack of effect.R (version 4.2.1) was used for the statistical analyses.A value of p < 0.05 was considered significant.

Ethics
Procedures followed were in accordance with ethical standards of the Helsinki declaration of 1975, as revised in 2013.The protocol was approved by the Ethics Committee for Medical Research in Copenhagen, Denmark, identification H-17038155.Informed consent was obtained from all participants before study entry.

Results
Prior to the three meals, no differences were observed regarding baseline mean concentrations of GLP-2, PYY and GIP, respectively.Mean values and standard deviations are provided for each hormone in the supplementary material.

GLP-2
All three meals elicited a significant and rapid rise from baseline in mean plasma concentrations of GLP-2 at all measured time points (10-240 min), see Fig. 1A.
The HC meal stimulated an abrupt increase in the mean plasma GLP-2 concentration with a significant elevation compared the HF, but not the HP meal, at 10, 20 and 30 min and peaking levels at 30 min.The GLP-2 mean concentration in the HC meal then subsequently declined.
The early GLP-2 response to the HC meal was also illustrated by the highest mean AUC difference from The onset and increase in the GLP-2 mean concentration was more modest in relation to the HF meal, but it showed a slower decrease, resulting in a significantly higher mean concentration compared to the HC meal at 240 min.Mean GLP-2 for the HP meal, showed a plateau from 30 to 240 min with a peak at 120 min resulting in significantly higher levels compared to the HC and HF meal at 120, 180 and 240 min.This late mean GLP-2 response for the HP meal was also illustrated by the highest mean AUC from baseline from 60-240 min of 5826 pmol*min/L, 95 %-CI [4979;6673] compared to 3124 pmol*min/L, 95 %-CI [2095;4154] for the HF meal, p < 0.001, and 3735 pmol*min/L, 95 %-CI [2571;4899], for the HF meal, p < 0.001, see Fig. 2B.There was no difference between the HC and the HF meal, p = 0.215.
All raw data, spaghetti plots, and figures for comparisons of the mean change from baseline are included in the supplementary material as an open data policy and to justify conclusions.

PYY
All three meals showed a mean rise in PYY from baseline for all time points (10-240 min), although the rise was not significant for the HF meal at 10, 20 and 240 min, see Fig. 1B.
The HP stimulated the most rapid mean increase in PYY, peaking at 120 min with a sustained plateau until 240 min, whereas the HC and HF mean increase was more modest and for the HC meal significant lower, compared to the HP meal, at 60, 120 and 180 min.At 240 min the PYY mean increase from baseline for the HF meal was significantly lower compared to the HP meal.

GIP
All three meals significantly increased the mean GIP levels from baseline for all time points (10-240 min), see Fig. 1C.
With a prompt rise in mean GIP and a peak at 120 min, the HC meal demonstrated the highest values at all measure points compared to the HF and HP meals.This rise was significant at all times compared to the HP meal which caused the smallest mean GIP response.Intermediary mean GIP responses were seen for the HF meal peaking at 30-45 min and then decreasing.

Discussion
The aim of this study was to evaluate the effect of adherable isoenergetic carbohydrate-, fat-or protein enriched meals on the endogenous secretion of GLP-2 in healthy, adult subjects, where PYY and GIP served as internal controls.
Our results show that the HP-meal elicited the highest GLP-2 secretion when measured as the change of mean AUC from baseline (from − 15 to 240 min) as well as peaking levels from baseline which were seen from 120 to 180 min after meal ingestion.The HC-meal provided an abrupt increase in GLP-2 secretion peaking already at 30 min, but the high level was not maintained.The early GLP-2 response to the HC meal was also illustrated by the highest mean AUC from baseline from − 15 to 60 min.The GLP-2 response to the HF meal was more even.Similar AUCs from − 15 to 240 min were seen for the HC and HF meals.Likewise, PYY secretion was highest in the HP meal as indicated by the mean AUC difference from baseline from − 15 to 240 min.Lower levels of PYY compared to GLP-2 were seen in the early phase in accordance with the possibility that some L-cells in the proximal small intestine have no coexpression of PYY [30].As expected, the endogenous GIP secretion with the largest mean AUC from baseline from − 15 to 240 min was provided by the HC meal and the lowest by the HP meal.
The composition of the mixed meals with approximately 60 % of the studied ingredient was chosen in consideration of the fact that in everyday life, it is very difficult to compose and adhere to a diet with an even higher content of any of the three macronutrients.Little is published on the endogenous secretion of GLP-2 in relation to changes in macronutrients in the diet, and therefore, this was investigated in healthy subjects.
We know from previous studies, that the presence of luminal nutrients is the primary stimulus for GLP-2 secretion.The physiological effects of GLP-2 could be modified by changes in the dietary macronutrient composition.The biological half-life of GLP-2 is relatively short (approximately 7 min) [31], which could limit the effects of exogenous supplementation.In this study, we therefore examined which meal-composition that best maintained high plasma concentrations of GLP-2.
Xiao et al. tested the effect of individual macronutrients as 300 ml liquid meal of 1.67 MJ with nearly 100 % carbohydrate (5.6 fold GLP-2 increase), fat (2.7 fold GLP-2 increase) but no altered GLP-2 by protein intake in contrast to our findings [33].In Xiao's study the protein was administered as a 300 ml liquid meal containing 95 g whey protein which was consumed over a 5-minute period.This was in contrast to our study where the meal bulk was larger, and the consumption time was longer.Further, our meals were mixed, and it was not the individual component being studied.
As for human studies, animal studies have also shown that the circulating GLP-2 concentration increases after an oral feeding and is positively correlated with the level of energy in the enteral intake [1,32,[38][39][40].Part of the variation in published GLP-2 results is probably related to difficulties with respect to reliable measurement of GLP-2.The    current measurements were made with a carefully validated assay against the intact N-terminus of GLP-2 using terminal wrapping antibodies.Only assays including such antibodies will be able to measure GLP-2 accurately without interference by other fragments of proglucagon [26].No available commercial assays fulfill this requirement so far.
GLP-2 is synthesized and secreted from the L-cell in a 1:1 stoichiometric ratio with GLP-1 [1,26], therefore, results of measurements of GLP-1 would be expected to resemble measurements of GLP-2 secretion.However, GLP-1 is metabolized much more rapidly and extensively by DPP-IV than GLP-2, to the extent that intact GLP-1 levels are often immeasurable.The metabolite of GLP-1 (GLP-1 9-36amide), however, has a half-life in the circulation which resembles that of intact GLP-2, and indeed plasma profiles of intact GLP-2 and total GLP-1 are often rather similar [41,42].
Looking into various protein sources, it is evident for both GLP-1 and GLP-2 that whey protein may generate the highest levels of the active form of the peptides, perhaps due to its ability to reduce DPP-IV activity [40,43].However, the focus of this study was about adherable mixed meals and our HP meal was not especially rich in whey protein compared to other protein sources.
Like GLP-2, studies on healthy humans show increased total plasma GLP-1 peak concentrations and AUC positively correlated with increasing amounts of energy in solid mixed meals [44,45].
If a mixed meal of 2.09 MJ is blended before ingestion higher peak concentration, seen at 15 min, and AUC over 180 min have been shown [46].In relation to our study, solid food ingestion of 1.57 MJ almost 100 % glucose, fat or protein, evaluated for 180 min, showed peaking GLP-1 plasma levels at 30 min postprandial for glucose ingestion compared to 150 min for fat and 30 and 180 min for protein, which is similar to our findings; the AUC was not calculated [47].With liquid meals of 0.84 MJ, consisting of > 90 % glucose, fat or protein, evaluated for 240 min, the HF meal showed significantly higher AUC compared to HC meal, and no difference compared to the HP meal [48].This relatively strong effect of the liquid HF meal could be due to a faster gastric emptying and transit time compared to that for solid fat.However, it is presumably difficult to adhere to a liquid diet containing > 90 % of one of the macronutrients.In obese humans HF mixed meals of 2.47 MJ increased both peak concentrations and AUC for GLP-1 compared to HC meal evaluated for 180 min [49].
GLP-2 responses to enteral feeding in patients with altered gut anatomy or disease may differ from those of healthy subjects.Short bowel intestinal failure patients with less than 150 cm small bowel ending in a jejunostomy showed no significant GLP-2 increases to 1.88 or 3.92 MJ mixed meals [34], consistent with the L-cell localization in ileum end colon [50].Patients with short bowel syndrome with less than 140 cm small bowel, maximum 10 cm preserved ileum and at least half the colon in continuity showed significantly higher levels of fasting GLP-2 and increases to a mixed meal of 3.92 MJ compared to healthy subjects [35], suggesting a large degree of passage of undigested nutrients to L-cells in the colon.
Regarding the GLP-2 physiological effects on GLP-2 receptor expressing cells, it is not clear whether the GLP-2 peak concentration or AUC are the most important for the various functions.
GLP-2 may exert both endocrine and paracrine effects [51] and the rapid increasing GLP-2 levels seen in response to nutrient intake suggest that the peak concentrations cold be important for coordinating assimilation, i.e. gastric hypersecretion, gall-and pancreas secretion, blood flow and transit time [11][12][13][14].GLP-2 plasma AUC/time (i.e. the average concentration in the observation period) on the contrary could be more important for intestinal trophism.Thus, the short-acting GLP-2 analog teduglutide administered once daily (i.v.half-life approximately 30 min; 2-4 h exposure after s.c.administration) shows apparently similar increases in intestinal function in patients with short bowel syndrome as long acting GLP-2 analogues glepaglutide and apraglutide [52][53][54], although direct comparisons have not been made.Thus, it is likely that the acute as well as chronic high concentrations are highly relevant in these patients.
When studying the effects of inflammatory bowel disease and the endogenous secretion of GLP-2 results have been somewhat different, probably due to different levels of disease activity.Patients hospitalized for active intestinal inflammation due to ulcerative colitis and Crohn's disease showed increases in fasting plasma GLP-2 to 229-317 % of levels in healthy subjects suggesting an adaptative response to intestinal injury [55].However, again compared to healthy subjects, unchanged fasting and postprandial levels of circulating GLP-2 have been found in both active ulcerative colitis and Crohn's disease and in inactive ulcerative colitis [56] (same assay as the one used here).In paediatric patients with acute ileal Crohn's disease fasting GLP-2 levels were not altered, but decreased postprandial GLP-2 response were detected [57].
In conclusion adherable iso-energetic mixed meals, with approximately 60 % of the energy coming from one of the three macronutrients, the high protein meal was most effective with respect to stimulation of the postprandial endogenous GLP-2 secretion regarding both AUC and peak.The high carbohydrate meal had a fast peak and then decreased whereas the response to the high fat meal was more even.These findings could reflect a slower gastric transit time with a more prolonged exposure of the L-cells in the case of the high protein and fat meal compared to the high carbohydrate meal.
Future studies of specific patient-groups are also needed to provide information about the specific secretory profiles in relevant diseases and also to measure the possible effects of the GLP-2 secretion and its relation to clinical effects e.g., in patients with injuries after chemotherapy and radiation therapy, inflammation in inflammatory bowel disease and nutrient absorption in patients with short bowel syndrome.

CRediT author statement
All authors contributed to the conception and design of the study, or acquisition and analysis of data, drafting the article or revising it critically for important intellectual content.All authors approved the final revised version to be submitted.

Fig. 1 .
Fig.1.A, B and C. Mean change from baseline with 95 % confidence intervals in plasma GLP-2, PYY and GIP following three 3.9 MJ isoenergetic meals with approximately 60 % of the energy coming from one of the three macronutrients.Venous blood was collected at 15 min before (baseline values) and 10, 20, 30, 45, 60, 120, 180 and 240 min after the start of the meals.The meals were completed within 15 min from time 0. GLP-2: glucagon-like peptide 2, PYY: peptide YY, GIP: glucose-dependent insulinotropic polypeptide.

Table 1
Energy distribution.

Table 2
Dietary components of the three isoenergetic meals.