Glucagon like peptide 1 (GLP-1) is an intestinal hormone that plays an important role in glucose metabolism. GLP-1 is released from mucosal L cells following nutrient ingestion and contributes to the incretin effect, with the enhancement of insulin secretion occurring with enteral compared with intravenous glucose administration. The mechanisms linking nutrient absorption and GLP-1 secretion are unknown, and studies addressing this topic, particularly in small animal models, have been hampered by the relatively low concentrations of GLP-1 in the circulation. We hypothesized that GLP-1 levels would be higher in samples of intestinal lymph compared with plasma and could provide a novel system in which to study meal-induced hormone secretion. We addressed this hypothesis in conscious rats with indwelling catheters in the portal vein and distal intestinal lymph duct. These animals had plasma and lymph sampled before and for 240 min after instillation of a liquid meal in the gastrointestinal tract. Lymph contained detectable concentrations of glucose, insulin, and GLP-1 that were reliably measured using our assays. Before and after the Ensure feeding, plasma insulin levels were approximately two times as high in portal plasma as intestinal lymph. In marked contrast, GLP-1 levels were five to six times higher in lymph relative to portal plasma following nutrient administration. This relative difference in GLP-1 levels was even greater when lymph was compared with peripheral plasma and dramatically exceeded the ratio of lymph to plasma peptide tyrosine-tyrosine concentrations. This is the first observation of a gastrointestinal hormone being disproportionately transported in lymph. The remarkable levels of GLP-1 in intestinal lymph demonstrate the potential for lymphatic sampling as a more sensitive means of studying the secretory physiology of this hormone in vivo. In addition, these data raise the possibility that intestinal lymph may serve as a specialized signaling conduit for regulatory peptides secreted by gastrointestinal endocrine cells.
- gastrointestinal hormone
- glucose tolerance
the importance of gastrointestinal hormones to augment insulin secretion following nutrient ingestion, termed the incretin effect, has generated renewed interest in recent years (5, 6, 13). Of the many putative insulin-stimulatory factors released by the gut, glucagon-like peptide 1 (GLP-1) has received considerable attention because of its potency to lower blood glucose in persons with diabetes and the potential of therapeutics based on the GLP-1 signaling system (6, 13). Studies using targeted gene deletion of the GLP-1 receptor in mice (21, 27) and specific receptor antagonists or neutralizing antibodies in other mammalian models (7, 15, 31, 34) have demonstrated that signaling by GLP-1 is necessary for normal oral glucose tolerance.
Given the key role of incretins in glucose homeostasis, understanding the regulation of their secretion is an important issue. Secretion of the other major incretin, glucose-dependent insulinotropic polypeptide, from upper intestinal K cells appears to be tightly coupled to glucose absorption by enterocytes (16, 25, 30). Less is known about the mechanisms by which nutrients, primarily carbohydrate and lipid, stimulate GLP-1 release from L cells (6, 25). Recent findings suggest that GLP-1 release is not stimulated by direct contact of nutrients in the intestinal lumen with L cells, and there is evidence to suggest that the GLP-1 response to nutrients is mediated by neural signals (8, 24).
Much of the in vivo work on GLP-1 secretion has been done in rats. One of the major difficulties confronting investigators studying the mechanism of GLP-1 release has been that the active peptide circulates in low levels and that, relative to the sensitivity of current RIAs, the volume of plasma that can be removed from a small animal is frequently limiting. We hypothesized that concentrations of GLP-1 would be elevated in intestinal lymph since this fluid drains the intestinal lamina propria and the pool size of this compartment is smaller compared with the blood pool. Sampling from cannulas in the primary intestinal lymphatic is technically feasible in conscious animals, and this model has been successfully used to study lipid absorption (32, 33). We therefore measured GLP-1 in portal venous plasma and intestinal lymph during intragastric administration of a liquid meal to conscious rats.
Animals and surgical preparation.
All procedures were approved by the University of Cincinnati Internal Animal Care and Use Committee and complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Male Sprague-Dawley rats (Harlan, Indianapolis, IN) were maintained on a 12:12-h light-dark cycle with lights on at 0600. Animals had free access to food and water except where noted. For surgery, rats were fasted overnight and anesthetized with halothane. A laparotomy was performed, and the superior mesenteric lymph duct was cannulated with a vinyl tube (medical grade; 0.50 mm ID and 0.80 mm OD; Dural Plastics and Engineering, Dural, Australia); the tube was fixed in place with a drop of cyanoacrylate glue and externalized through a stab wound in the right flank (12). A second cannula (Silastic medical grade 602-205, 0.75 mm ID and 1.6 mm OD; Dow Corning Medical Products, Midland, MI) was placed in the fundus of the stomach, where it was secured with a purse-string suture and externalized through a second flank wound. Groups of rats with feeding tubes in the stomach had silicone rubber tubes (1 mm OD) placed in the portal vein starting just proximal to the confluence of subportal vessels, or the jugular vein; these tubes were passed through the abdominal wall, tunneled subcutaneously, and exteriorized at the dorsum of the neck. A third group of rats had gastric cannulas passed through the pylorus 2 cm in the duodenum; the cannulas were infused with 140 mM saline/0.28 mM glucose solution at 3 ml/h until the start of experiments. These animals also had vascular cannulas placed in the jugular vein. Following surgery, rats were placed in restraint cages in a temperature-regulated chamber at 30°C and allowed to recover for 24 h.
Administration of a liquid meal and blood and lymph sampling.
In the first experiment, 12-h-fasted rats were sampled before and after an intragastric bolus of 5 ml Ensure Plus (10 kcal distributed as 1.1 g carbohydrate, 0.23 g fat, and 0.27 g protein). Single blood samples were taken from the portal (n = 6) or jugular (n = 5) veins, and lymph collection was commenced from the cannula in the superior mesenteric lymphatic vessel. Following the instillation of the gastric nutrients, portal or jugular venous blood was sampled at specific intervals over the next 4 h. Lymph was collected continuously for 60 min before and for the entire 4 h after the intragastric meal and separated into 0–30, 30–60, 60–120, and 120–240 min aliquots.
Rats with intraduodenal tubes had 60-min infusions of either 5.5 mM glucose (total of 1.8 g; n = 3) or 10% liposyn (total of 18 g; n = 4). Intestinal lymph and blood from the jugular vein were sampled over 60 min. A separate group of rats had intraduodenal perfusion of 10% liposyn (n = 6) or saline (n = 6). Plasma and lymph were sampled at baseline and hourly after the start of the intraduodenal infusion for measurement of dipeptidyl peptidase IV (DPP-IV) activity.
Measurement of insulin, glucose GLP-1 and peptide tyrosine-tyrosine in plasma and lymph.
Blood and lymph samples were collected in 0.5 M EDTA and 500 kallikrein inhibitor units/ml aprotinin, stored on ice until centrifugation, and frozen at −20°C until assayed. Glucose was measured using a glucose oxidase method, and insulin and GLP-1 were measured using previously published RIA (29). The GLP-1 RIA uses an antiserum reacting to an epitope including the COOH-terminal amide and recognizes bioactive GLP-1-(7-36) as well as its precursor GLP-1-(1-36) and metabolite GLP-1-(9-36). With these RIA, insulin is measured directly in plasma, whereas GLP-1 is assayed in ethanol extracts of plasma. For both assays, lymph was measured with and without extraction. Linearity of the RIA in samples of lymph was assessed by assaying serial dilutions of plasma and lymph. Known amounts of insulin and GLP-1 were added to lymph samples of known concentration and assayed for calculation of recoveries. Nonspecific contribution to RIA results were estimated by “stripping” samples with the addition of charcoal and adsorption of peptides followed by filtration and assay of the remaining supernatant. Plasma and lymph concentrations of peptide tyrosine-tyrosine (PYY) were measured using a commercial RIA kit according to the manufacturer's specifications (Millipore, Billerica, MA).
The activity of DPP-IV was measured by the determination of the rate of release of dipeptide conjugated to p-nitroanilide, according to the method as described by Gotoh et al. (11).
Glucose and hormone responses to the intragastric meal were calculated as area above the fasting level using the trapezoidal rule. Comparisons of hormone levels in plasma and lymph taken from the same animals were made with unpaired t-tests. Data are presented as means ± SE.
Assay performance in plasma and lymph.
Assessments of assay linearity, nonspecific immunoreactivity, and recovery were made for the insulin and GLP-1 RIA in plasma and lymph (Table 1). Serial dilutions of lymph and plasma yielded concentrations of insulin and GLP-1 near the expected 50%, demonstrating similar degrees of linearity. Samples of lymph assayed directly or after extraction with ethanol, for GLP-1, gave similar results (data not shown). Recovery of synthetic insulin and GLP-1 added to lymph was reliable, with values ∼90% for insulin and slightly greater than 100% for GLP-1. Assay of lymph treated with charcoal to remove peptides showed minimal nonspecific contribution.
Plasma and lymph glucose and hormone concentrations.
Fasting levels of portal vein glucose were 95 ± 3 mg/dl and increased to a peak concentration of 151 ± 14 mg/dl 30 min following Ensure administration (Fig. 1). Portal plasma glucose levels returned to near basal by 120 min and were stable for the remainder of the experiment. Intestinal lymph glucose concentrations were slightly higher in the fasting state, 115 ± 3.5 mg/dl (P < 0.001), and reached peak levels of greater than 180 mg/dl from 30 to 90 min following Ensure administration (Fig. 1A). In contrast to portal vein plasma glucose concentrations, glucose levels in the lymph remained elevated for a longer period and returned to the basal level at a slower rate. The glucose area under the curve (AUC) over the 240-min sampling period was significantly greater in lymph than plasma (plasma 6,403 ± 1,014 vs. lymph 13,590 ± 970 mg·dl−1·min−1, P < 0.001; Fig. 1B).
Insulin levels in the portal vein were 77 ± 14 pM in the fasting state, increased promptly to peak levels of 293 ± 60 pM at 30 min, and remained elevated for the 2 h of the experiment (Fig. 2A). In contrast, lymph insulin levels were lower in the fasting state (55 ± 10 pM) and peaked later, at 60 min after the glucose load, than plasma insulin. Intestinal lymph insulin concentrations returned to near basal levels by 120 min. The insulin AUC during the period of meal absorption was significantly greater in plasma than lymph (18,837 ± 3,166 vs. 11,282 ± 2,318 pM/min, P < 0.05; Fig. 2B).
Portal vein concentrations of GLP-1 were approximately half the lymph values in the fasting state, 7.1 ± 1.7 and 14.2 ± 1.5 pM, respectively (P = 0.02). Following intragastric Ensure, portal vein GLP-1 levels increased six- to sevenfold to a peak of ∼50 pM at 30 min (Fig. 3A). The plasma peak of GLP-1 was transient, and levels returned to near fasting within 60–90 min. Postload GLP-1 concentrations were dramatically different in intestinal lymph, with a peak level at 30 min of 308 ± 73 pM that was ∼20-fold greater than basal values. Concentrations of GLP-1 in lymph remained elevated for 120 min after the meal. Lymph GLP-1 AUC was significantly greater than portal plasma levels (15,564 ± 1,293 vs. 3,137 ± 803 pM/min, respectively, P < 0.01; Fig. 3B).
Fasting GLP-1 levels in peripheral plasma were also lower than concentrations measured in lymph sampled over a similar interval, 7.1 ± 0.7 vs. 16.8 ± 5.1 (P = 0.08). Following intragastric nutrients, GLP-1 levels in blood rose quickly to a peak level of 13.2 ± 3.1 pM at 15 min and remained elevated over the next 2 h (Fig. 4). In contrast, the peak in lymph GLP-1 following intragastric feeding was more delayed, occurring at 45 min but considerably higher, 391.7 ± 180.3 pM. Concentrations of lymph GLP-1 in these rats remained elevated above baseline levels for the entire 4 h of sampling. Lymph GLP-1 AUC was significantly greater than plasma GLP-1 AUC (44,443 ± 4,748 vs. 315 ± 131 pM/min, respectively, P < 0.001).
Plasma concentrations of PYY were measured in the samples taken from the jugular vein and compared with lymph levels. Fasting PYY was at the limit of detection of the assay in both lymph and plasma, 15.6 pg/ml. In response to the meal, levels of PYY increased to peak concentrations in plasma and lymph at 30 min (Fig. 5). Lymph concentrations were greater than plasma levels throughout the course of sampling, but generally less than twice as high. The lymph AUC for PYY was significantly greater than the plasma AUC (51,348 ± 10,206 vs. 19,734 ± 6,367 pg·ml−1·min−1, respectively, P = 0.03).
The GLP-1 response in plasma and lymph was also measured in a group of six rats given carbohydrate or lipid directly in the duodenum by continuous infusion. Jugular plasma levels of GLP-1 increased from basal levels of 10 ± 2 pM to average values of 44 ± 7 pM during the glucose infusion. Basal lymph GLP-1 values were 54 ± 4 pM and increased to stimulated levels of 129 ± 34 pM (P < 0.05 vs. plasma comparison of both basal and stimulated levels). In four other rats, lymph GLP-1 was 23.6 ± 9.6 in the fasting state and rose to mean levels of 140 ± 80 pM with intraduodenal lipid.
Plasma and lymph DPP-IV activity.
Fasting DPP-IV activity was ∼20-fold higher in fasting plasma than lymph, 42.1 ± 4.2 and 1.8 ± 0.3 nmol dipeptide release·min−1·ml−1, respectively (P < 0.001). Plasma levels did not differ during saline or liposyn infusion and did not change from the fasting value (Fig. 6). Lymph DPP-IV activity increased significantly following lipid infusion, with mean values of 6.1 ± 1.0 compared with 1.9 ± 0.3 nmol·min−1·ml−1 with saline (P < 0.001).
These studies demonstrate that intestinal lymph contains measurable quantities of insulin, glucose, GLP-1, and PYY and that the profile of these factors during nutrient absorption generally mirrors plasma levels. However, there are major differences among insulin, PYY, and GLP-1 in their relative concentrations in lymph and plasma. Our results suggest that insulin is secreted primarily in the bloodstream and gets into the lymph compartment by filtration from capillaries. On the other hand, the significantly higher lymph-to-plasma ratio of GLP-1 indicates that it is concentrated in the lymph compartment. The profile of PYY is intermediate between insulin and GLP-1, with values slightly greater in lymph than plasma, but not dramatically different. These findings indicate that a relatively greater portion of secreted GLP-1 bypasses the capillary drainage of the intestinal villi and ends up in the lymph drainage pool. This is different from what has been previously reported for other gastrointestinal hormones. In addition, intestinal lymph appears to be a protected environment for GLP-1 in that the concentrations of DPP-IV, an enzyme that leads to rapid inactivation of GLP-1, are considerably lower in lymph than plasma. These results indicate that sampling of intestinal lymph is a potentially valuable experimental approach to studying GLP-1 secretion in vivo. Moreover, our findings suggest that there may be differential partitioning of gastrointestinal hormones between the lymph and plasma, and raise the possibility that this differential transport plays a role in the biological activity of these factors.
Based on standard analytic techniques, intestinal lymph is a reliable medium for the analysis of insulin and GLP-1 by RIA. The linear response to serial dilutions of lymph and the virtual absence of interfering substances in the assays suggest that the peptide measurements were specific. Because recovery of added peptide was quantitative, it seems that there is little degradation of insulin or GLP-1 in lymph. Bioactive GLP-1-(7-36) amide is rapidly metabolized in plasma by DPP-IV, a ubiquitous protease in the vasculature carrying blood, to the closely related but inactive peptide GLP-1-(9-36) amide (14, 20). The significantly lower levels of DPP-IV in lymph than plasma suggest that a higher percentage of the GLP-1 in this pool is bioactive compared with plasma. However, because the antibody used in our RIA recognizes intact and metabolized GLP-1 equally, we cannot comment on the relative percentages of intact and metabolized GLP-1.
The glucose peak in portal plasma was transient, and levels were only minimally elevated throughout most of the absorptive phase of the meal. This pattern reflects the typical postprandial regulatory responses that maintain glucose homeostasis, primarily insulin-dependent clearance of glucose and suppression of glucose production. In contrast, the glycemic excursion in intestinal lymph was greater and more prolonged. Lymph glucose concentrations equilibrate rapidly with plasma levels during intravenous glucose administration (35). However, following meals, concentrations of lymph glucose are predominantly a function of glucose absorption from the gut (4) and are only secondarily affected by the regulatory influences maintaining glucose tolerance in plasma. In the postprandial setting, disappearance of glucose from the intestinal lymph compartment occurs when it empties in the venous circulation without significant clearance along the course of the lymphatics.
The relative concentrations of insulin in the portal plasma and intestinal lymph provide some insight into the likely derivation of this peptide in the two fluid compartments. The sampling site for intestinal lymph at the level of the superior mesenteric lymph duct is distal to the drainage from the pancreas (2). However, in the only previous study of lymph drainage from the pancreas, flow was noted to be surprisingly low (2), raising doubt as to whether significant amounts of insulin could arise primarily from drainage of the islets. An alternative possibility is that intestinal lymph insulin originates in the blood compartment. Our results are compatible with the latter possibility, since insulin concentrations in portal plasma are higher at every time point than levels in lymph. In addition, although insulin concentrations peak at 30 min in portal blood, the peak levels in lymph are not until 60 min, a delay that is consistent with movement from plasma to lymph. Yang and colleagues (35) previously reported a similar lag between plasma and lymph insulin concentrations during intravenous insulin administration; in this condition, transport from plasma is the only source for lymph insulin. These investigators reported a ratio of peripheral plasma/lymph insulin of 3:2 during intravenous insulin infusions, relative concentrations comparable to our findings before and after a meal. Therefore, the results of our Ensure feeding studies support the conclusion that insulin secreted after meals reaches intestinal lymph secondarily due to filtration from blood capillaries supplying the gut interstitium.
The relative concentrations of GLP-1 in the plasma and intestinal lymph differ dramatically from those of insulin. GLP-1 concentrations were significantly greater in lymph than plasma, with postprandial values 5- to 10-fold higher in the intestinal lymph compartment than in portal blood. These results are most consistent with GLP-1 release directly in the lymphatic system draining the intestinal mucosa rather than transport from the blood plasma to lymph. Based on immunocytochemical studies, most gastrointestinal endocrine cells, including L cells, are polar with an apical end in contact with the gut lumen and a basal end in close proximity to submucosal capillaries and lacteals (17, 28). The fenestrated capillary network of the intestinal villus is much nearer the basal aspect of the mucosal cells than the initial lacteals (2), and so is the first space exposed to substances absorbed or secreted by the mucosa. In the only previous report of gastrointestinal hormones from intestinal lymphatics, concentrations of gastrin, neurotensin, vasoactive intestinal peptide, substance P, and bombesin were ∼50% those of plasma in a dog model (19), consistent with primary clearance of peptides released from mucosal cells or enteric neurons by blood capillaries. The case for GLP-1 seems to be much different in that the greatly elevated lymph, compared with portal plasma, levels suggest a different means of distribution in the vasculature draining the villus.
PYY is a particularly apt comparator for GLP-1 in these studies. Similar to GLP-1, PYY is also a product of intestinal L cells that is released after eating (1, 3). Although lymph levels of PYY were generally higher than those in peripheral plasma, this was not true at the 15-min time point and was not invariable among all of the rats at other time points. The difference in the relative lymph/plasma levels of GLP-1 and PYY, ostensibly released by the same gastrointestinal endocrine cells, indicates that there can be differential partitioning of gut hormones between the lymph and plasma compartments after their release from intestinal endocrine cells.
How significant quantities of a small molecule, like GLP-1, would bypass the usual collection in submucosal capillaries is unclear but consistent with the findings reported here. The significantly different lymph/plasma hormone levels of PYY and GLP-1 suggest that the extraordinarily high lymph concentrations of GLP-1 are not due simply to a concentration effect caused by the smaller pool size and slower turnover of lymph (10, 26). This conclusion is consistent with the previous report of lower gut hormone concentrations in canine lymph compared with plasma (19). Rather, it seems likely that the higher lymph concentrations of GLP-1 indicate targeted secretion of peptide in the intestinal lymphatic system. Given that intestinal lymph flow rates after meals are ∼4 ml/h (23), and portal blood flow rates in rats are nearly 100 times that (10, 18), even with the higher peak levels of GLP-1 in lymph the majority of this hormone is delivered in the blood stream. Nonetheless, the substantial elevation of GLP-1 levels in intestinal lymph raises the possibility that the relatively high levels in this compartment are related to some regulatory effect. Lymphatic function is regulated by autonomic and peptidergic neurotransmitters and a variety of humoral factors (4, 22), so it is conceivable that GLP-1 contributes to this process. Thus, although the lymphatic system may not be the major route for delivery of GLP-1 in the circulation, our findings raise the possibility that GLP-1 has specific effects mediated in this compartment.
The GLP-1 response to the Ensure meal was greater than the response to glucose or lipid given individually, despite the greater total load of nutrients in the intraduodenal infusions. Previous reports have demonstrated that both lipid- and carbohydrate-containing meals are potent stimuli for GLP-1 release (4). The findings from this study raise the possibility that a mixed meal augments the L cell response beyond that of lipid or glucose given in isolation.
These are the first data we are aware of measuring DPP-IV activity in lymph and assessing the changes in activity over time after feeding. Our results demonstrate that intestinal lymph is a compartment where peptides normally metabolized by DPP-IV may be protected. Surprisingly, the levels of DPP-IV activity increase in lymph but not plasma during nutrient absorption. Although it is conceivable that the DPP-IV could arrive in intestinal lymph by filtration through the capillary system, this seems improbable because capillaries are restrictive of large macromolecules such as DPP-IV (mol wt 175,000; see Ref. 9). This raises the possibility that lymph DPP-IV is released from the gut. The significance of increasing DPP-IV activity in lymph after meals is not clear but is interesting to consider in light of the concurrent and disproportionate increase of GLP-1.
Perspectives and significance.
We have demonstrated that intestinal lymph is enriched with the gastrointestinal hormone GLP-1 before and after meal intake. This is the first report of a gastrointestinal hormone being more concentrated in lymph compared with plasma. These findings suggest that lymph draining the gut is a medium through which the function of L cells, and potentially other types of intestinal endocrine cells, can be examined. It will be important to determine how gastrointestinal hormones, besides GLP-1 and PYY, are distributed between lymph and plasma in the lymph fistula model to gain insights into the possible differential function of gut endocrine cells. In addition, the potential of gastrointestinal hormones to mediate physiological actions through their concentration in intestinal lymph merits further consideration.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-56863, DK-57900, and DK-17844 and by funds from the Procter and Gamble.
We thank Kay Ellis for precise measurements of hormones in plasma and lymph and Joe Kluener for other technical support.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked ““advertisement”” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society