Glucose-dependent and Glucose-sensitizing Insulinotropic Effect of Nateglinide: Comparison to Sulfonylureas and Repaglinide

Nateglinide, a novel D-phenylalanine derivative, stimulates insulin release via closure of KATP channels in pancreatic β-cell, a primary mechanism of action it shares with sulfonylureas (SUs) and repaglinide. This study investigated (1) the influence of ambient glucose levels on the insulinotropic effects of nateglinide, glyburide and repaglinide, and (2) the influence of the antidiabetic agents on glucose-stimulated insulin secretion (GSIS) in vitro from isolated rat islets. The EC50 of nateglinide to stimulate insulin secretion was 14 μM in the presence of 3mM glucose and was reduced by 6-fold in 8mM glucose and by 16-fold in 16mM glucose, indicating a glucose-dependent insulinotropic effect. The actions of glyburide and repaglinide failed to demonstrate such a glucose concentration-dependent sensitization. When tested at fixed and equipotent concentrations (~2x EC50 in the presence of 8mM glucose) nateglinide and repaglinide shifted the EC50s for GSIS to the left by 1.7mM suggesting an enhancement of islet glucose sensitivity, while glimepiride and glyburide caused, respectively, no change and a right shift of the EC50. These data demonstrate that despite a common basic mechanism of action, the insulinotropic effects of different agents can be influenced differentially by ambient glucose and can differentially influence the islet responsiveness to glucose. Further, the present findings suggest that nateglinide may exert a more physiologic effect on insulin secretion than comparator agents and thereby have less propensity to elicit hypoglycemia in vivo.


INTRODUCTION
The homeostatic maintenance of blood glucose concentration is an integrated process predominantly regulated by the anti-hyperglycemic hormone insulin. When blood glucose rises, uptake of glucose into the/-cells leads to an elevation of ATP/ADP ratio and a sequence of ionic events. The resultant increase in intracellular Ca 2 +triggers exocytosis and insulin release. [1] The generation of insulin occurs through a precursor, proinsulin, whose biosynthesis is stimulated by nutrient secretagogue like glucose. [2] Aside from nutrient secretagogues, there are a number of non-nutrient *Address for correspondence: Metabolic and Cardiovascular Diseases, Novartis Institute for Biomedical Research, 556 Morris Avenue Summit, NJ 07901, USA. Tel.: 908-277-5703, Fax: 908-277-4756, e-mail: shiling.hu@pharma.novartis .com agents, which exert insulinotropic action via mechanisms other than stimulating the biosynthesis of insulin. As representatives of a class of non-nutrient insulin secretagogues, sulfonylureas (SUs) like glyburide (GLY) and glimepiride (GLI) act on pancreatic ]B-cells by blocking KATP channels. [3][4][5] Agents that share the primary mechanism of action with SUs include repaglinide (REP), a non-SUs benzoic acid derivative, [6][7][8] and nateglinide (NAT), a novel oral hypoglycemic agent recently marketed. [9,1] The structures of these hypoglycemic agents are shown in Figure 1.
In treating type 2 diabetes, SUs and REP can cause long-lasting hypoglycemia under both normoglycemic and hyperglycemic conditions in animal models. [11,2] NAT, on the other hand, demonstrates an enhanced activity under hyperglycemic conditions due to glucosesensitive action. [9,3,141 In line with the in vivo data, our earlier study characterizing the KATP channel-blocking effect by hypoglycemic drugs showed that NAT but not GLY and REP had an increased potency at elevated glucose concen-tration. [151 The aim of the present study was to obtain further evidence for a glucose-sensitive insulinotropic action by NAT. We investigated the interaction between glucose and NAT with regard to the stimulation of insulin release in vitro from rat pancreatic islets by determining the influence of glucose concentration on NATinduced insulin secretion as well as the influence of NAT on glucose-stimulated insulin secretion (GSIS). Such an interaction was also studied with the comparator insulinotropic agents like glyburide (GLY) and repaglinide (REP). Our results indicated that the islet secretory response to glucose stimulation was sensitized by NAT and REP, but not by the SUs, GLY and GLI. In addition, stimulation of insulin secretion in vitro by NAT was glucose-dependent while the effects of REP and GLY showed little or no glucose-sensitivity.
After islets were settled, 500 1 2x treatment of the acute glucose/drug concentrations was added to each tube (1 ml final volume). In the first study, the concentration-dependence of in vitro insulin secretion induced by hypoglycemic agents during lh static incubation in the presence of low (3 mM, G3), moderately elevated (8mM, G8) and severely elevated (16mM, G16) glucose was investigated. Each drug at 6-7 concentrations (4 tubes/concentration) including drug free control in the presence of G3, G8 and G16 were orderly set in a rack. Tubes were incubated at 37C with intermittent hand shaking for 1 hour. In the second study, insulin secretion during one hour incubation at eight glucose concentrations (0, 3, 5, 6.5, 8, 9.5, 11 and 16mM, 4 tubes/each glucose concentration) was measured in the presence or absence of one test drug (NAT, GLY, GLI or REP) at comparably effective concentrations approximately 2x respective EC50s at G8. At the end of lh static incubation, islet media (500bl/tube) were transferred to 96 deep-well plates and stored at -20C for subsequent insulin analysis.

Insulin Scintillation Proximity
Assay (SPA) The incubation media were diluted by factors ranging from lx to 20x depending on the concentrations of glucose/drugs/inhibitors. The diluted media were assayed for insulin content with SPA. [17] The assay employed commercially available products including a guinea pig anti-rat insulin specific antibody (Linco Research Inc) and scintillation proximity Type I reagent coupled to protein A (Amersham Life Science), and was performed as a single step assay. All samples were assayed in duplicate. The preparation of 96 well sample plates was made by sequentially pipetting standard/unknown samples, anti-insulin serum, 125I-insulin tracer, and SPA reagent, and the final volume equaled 175b1/well. The plates were incubated and vortexed on a titer plate shaker for approximately [18][19][20] hours overnight at room temperature before being placed into a Wallac Microbeta 1450 Liquid Scintillation Counter to be read under a normalization protocol. The output was in counts per minute (CPM).

Data Analysis
The sample insulin concentration was calculated by utilizing a template set up in Excel spreadsheet that possessed statistical analysis functions. The calculated concentration was eventually adjusted to reflect the degree of dilution. The intra-and inter-assay coefficients of variation were generally between 5% and 8%. EC50 s were calculated from 5-points dose-response curves fit with 4-parameter Hill sigmoidal equation in Sigmaplot (version 4.01).
However, in cases where high concentrations of glucose/drugs caused a decrease of insulin release, the values were excluded from the curve fitting analysis. Statistical significance was determined with t-test (single-tailed). P < 0.05 was considered significantly different.

GSIS in Nutrient-free and Nutrient-rich Incubation Media
To validate the islet static incubation assay and insulin SPA assay, we investigated GSIS in physiologically relevant nutrient-rich medium DMEM or in nutrient-free buffers such as Phosphate Buffered Saline (PBS) and Krebs Ringer Bicarbonate (KRB). While DMEM was rich in amino acids, the salines had glucose as the sole exogenous substrate. In all three types of media, insulin secretion from freshly isolated islets during 1 hour static incubation was stimulated by glucose in a concentration-dependent fashion, as shown in Figure 2. The characteristics of GSIS, however, differed considerably from medium to medium in several aspects: (1) the basal level of insulin release (at GO) was significantly lower in DMEM and KRB than in PBS; (2) the stimulation factor was 3.5-fold in PBS, 16.2-fold in KRB, and 26.7fold in DMEM as glucose concentration increased from 0 to 16 mM; (3) the EC50s (glucose concentration at which a half-maximal insulin release was achieved) were 5.8 mM and 6.0mM, respectively, in PBS and DMEM; (4) in KRB, there was no significant increment in GSIS at glucose concentration up to 10mM. Our results reinforce the importance of selection of incubation media and the presence of exogenous nutrients (e.g., amino acids) for islet function, as have been repeatedly discussed by others.  Thus, DMEM seems to be an optimal choice for the study of insulin secretory response in isolated islets. [glucose] (mM)   13.5 U/islet at G3, GS, and at G16 (n =96), demonstrating a glucose-dependence of insulin secretion. The ability of NAT, GLY and REP to stimulate insulin secretion was evaluated when the glucose concentration was maintained at 3 (G3), 8 (GS), or 16 (G16) raM. Representative results with REP at six concentrations are shown in Figure 3, in which the amount of insulin secretion at each glucose level in the absence (basal) and presence of REP are displayed. The basal insulin secretion in the absence of REP (shown with empty symbols) increased as glucose concentration is elevated. REP stimulated insulin secretion in a concentration-dependent manner at all glucose concentrations tested.
Parallel studies were carried out with NAT and GLY. The concentration-response curves for all drugs tested at G3, G8 and G16 were pooled and shown, respectively, in Figures 4A, B,   NAT. This concentration of NAT was approximately 2 x of the ECs0 of insulinotropic effect by NAT at glucose concentration of 8 mM (2.3 Figure 5A illustrates the data of glucoseinsulin response pooled from six independent experiments (n =4 in each experiment). The EC50s were direct readouts of the parameters in curve fitting of the data in Figures 4A  interpreted as an increase in islet sensitivity to glucose. Moreover, the insulinotropic effect of NAT was additive to that of glucose, since the presence of NAT substantially increased the maximal value of GSIS.
Parallel studies on GSIS was performed in the absence and presence of hypoglycemic drugs GLY (100nM), GLI (100nM) and REP (50nM). The concentrations of the drugs were so chosen that they were about equally effective in stimulating insulin secretion (approximately 2x respective EC50 s at G8 obtained from study 1).
While GLI had not been tested in study 1   of GSIS with and without secretagogues are tabulated (Tab. II). The EC50 of REP was reduced by 2.3mM, a magnitude slightly greater than that with NAT, suggesting an increased sensitivity of islets to glucose. However, the SUs, GLY and GLI, caused, respectively, a pronounced increase (by 3.4

mM) and no change in
EC50s. In fact, GSIS in the presence of GLY hardly reached plateau at the highest glucose concentration tested (16mM). The actual ECs0 might therefore be even greater than the value obtained from Hill sigmoidal fitting.

DISCUSSION
The islet incubation assay in conjunction with insulin SPA assay adopted in this work allowed in vitro study of cumulative insulin secretion from pancreatic islets during static incubations. The method was validated through investigation of GSIS in nutrient-free or nutrient-rich incubation media. While in all media insulin was released in a glucose concentration-dependent manner, the composition of media profoundly altered the basic sigmoidal relationship.
The lower insulin secretory capacity in nutrientfree salines may be attributable to multiple factors such as (1) impairment of glucose-sensing resulting from a reduced islet content of glucokinase; (2) decrease in rate of glycolysis resulting from insufficient activation of phosphofructokinase in response to a rise in hexose concentration; (3) reduction of insulin content in the islet cells. [21] It is therefore conceivable that nutrient-free salines are inappropriate for in vitro insulin study. The present study assessed the influence of glucose on in vitro insulin secretion stimulated by hypoglycemic drugs. Our data showed that the augmentation by NAT of insulin release was glucose-sensitive, as evidenced by a respective 6-and 16-fold increase in potency with an elevation of glucose from 3mM to 8 and 16mM. These changes, albeit not drastic, indicated an ability of NAT to "self-correct" for the maintenance of glucose homeostasis. These data are in qualitative agreement with findings from studies utilizing the buffer-perfused pancreas [22] as well as from in vivo studies in rat or dog. [11,12] The glucose sensitization of the NAT's insulinotropic action in vitro and in vivo and the glucose desensitization of GLY's action are also consistent with the observations that NAT allows whereas SUs prevent nutrient-stimulated protein biosynthesis in f3 cells. [23,24] The glucose-sensitive insulinotropic effect of NAT predicts that it would be a more effective drug in hyperglycemic patients than in normal individuals, depending, of course on the "health" of the pancreas or the residual /3-cell mass. In addition, the reduced insulinogenic potency of NAT at low glucose level may be translated to an increased safety margin due to reduced propensity of serious hypoglycemia. Conversely, GLY showed a greater potency at low glucose concentrations, which may contribute to relatively high risk of hypoglycemia known to be associated with GLY therapy. [25][26][27] The insulinotropic effect of REP was more potent by 5-fold in the presence of moderately high glucose (8 mM) than in the presence of low glucose (3mM). The sensitivity to glucose, however, diminished at high glucose level of 16mM. We found that the insulinotropic drugs tested differentially altered the in vitro GSIS. NAT and REP appeared to sensitize the secretory response of islets to glucose in two ways: (1) their effect was additive to the effect of glucose, as evidenced by an increase in maximal insulin release in the presence of drugs (Figs. 5A and 5C); (2) both drugs left-shifted the ECs0s for GSIS, suggesting an increased islet sensitivity to glucose. The interpretation for such an interaction between drugs and glucose is lacking at present. Some recent studies indicated that GSIS cannot be attributed solely either to the closing of KATP channels following an increase in intracellular ATP/ADP, or the role of glucose as a nutrient to cover the energy expenditure in the islet cells [28]. On the other hand, the insulinotropic action of NAT is likely to be mediated by KATP channel-independent as well as KATP channel-dependent pathways, while that of REP appears to be solely due to its closing of KATP channels in /-cells. [7,29] Taken together, the closure of KATP channels may not be sufficient to account fully for all the effects of hypoglycemic agents upon glucose sensitivity as well as other biophysical and biochemical variables in the islet cells.
The reported results on the effect of SUs on GSIS are rather controversial and lack consen-sus. The data in this study with SUs showed that the insulinotropic effect of GLY and GLI was additive to glucose stimulated insulin release (Figs. 5B and 5D), since the maximal insulin release was markedly increased in the presence of SUs. These agents, however, failed to increase sensitivity of islets to glucose as indicated by an increase (with GLY) or no change (GLI) of their EC50s. The data are ir agreement with the in vivo results of Groop et al. [ The mechanism(s) by which the non-SU insulinotropic drugs (NAT and REP) and the SUs (GLY and GLI) exerted differential effect on GSIS are yet to be established. It is speculated that the non-SU drugs bind to the SU receptor at a molecular site distinct from that for SUs and hence interact with glucose in a distinct pattern. In this context, the existence of a common SU receptor with distinct sites for GLY and REP [81 and the presence of a specific binding site for NAT in addition to a common SU receptor [33] have been proposed to be responsible for the common and differential insulin-stimulating processes by these drugs. Alternatively, the sensitizing or desensitizing efficacy of hypoglycemic agents on GSIS may be linked to differences in their capacity to be inserted into the phospholipid domain of the plasma membrane. SUs, GLY and GLI, have been known to be internalized into -cells to exert their action [21,341 while NAT appears to act extracellularly. [35] This distinction may lead to different modification of responsiveness to Ca 2+ of the effector system for insulin release and in turn, different glucose-sensitivity of islets.
In conclusion, nateglinide demonstrated a glucose-dependent and glucose-sensitizing insulinotropic action on isolated rat islets. These properties further distinguish nateglinide from other SU receptor ligands, raise the question of whether KATp-channel closure is the sole mechanism of action of this agent and predict a low hypoglycemic potential during therapeutic use of nateglinide.