Beta-Cell response to nonglucose secretagogues
Proteins and amino acids
The insulinotropic effect of oral proteins was first described almost 50 years ago [51–53] and recently confirmed [54–56]. After the ingestion of a small amount of proteins (30–50 g) [51,55] or a larger amount of proteins (2 g kg−1) , plasma insulin was raised two- to threefold over baseline and remained persistently elevated for 90 or 240 minutes, respectively. In either case, blood glucose did not change, whereas both GLP-1 and GIP levels were raised threefold over fasting values [55,56]. Because the effect of the incretins is glucose-dependent , the insulinotropic effect of oral proteins was most likely elicited by a direct stimulation of β cells by amino acids. In support of this hypothesis, it has been shown that insulin response to different protein solutions was closely related to the increase of plasma amino acids [57–59] with large differences among the individual amino acids unexplained by the functional group : phenylalanine and glycine shown to be the most potent (+30 pmol L−1), histidine, tyrosine, and, surprisingly, arginine as neutral, and the others as intermediate (+10.5 pmol L−1).
In vitro studies  have reported that the insulinotropic effect of amino acids is mostly dependent on the amino acid type, duration of exposure and concentration; furthermore only combination of amino acids stimulates insulin secretion when added at physiologic concentrations, whereas higher concentrations of individual amino acids are required to activate insulin secretion . Accordingly, the intravenous administration of a relatively small number of amino acids individually promotes insulin release while the maximum stimulus is elicited by mixed amino acids . In contrast to other amino acids, homocysteine in vitro showed a dose-dependent negative effect on insulin secretion in pancreatic β cells .
When either amino acids or proteins were ingested with glucose, plasma insulin levels were not significant or only slightly different from those after oral glucose alone but at lower glucose concentration [52–55,59,60,64,65] suggesting a direct effect on β-cell glucose sensitivity. Interestingly, the insulinotropic effect of proteins was neither blunted by insulin resistance  nor by diabetes  and was largely explained by the enhanced GLP-1 and GIP responses.
Lipids and free fatty acids
The first evidence of the insulinotropic effect of nonesterified fatty acids (NEFA) dates back five decades [68–70]. Several subsequent studies, described later, led to heterogeneous results which may be explained by differences in study protocols, dose (pharmacologic/physiologic), chemical structure (level of saturation, length of the carbon chain), type of administration (acute/chronic, oral/intravenous) and in subjects’ individual features (body weight, fasting glycemia, family history of diabetes) [61,71].
In lean subjects, an acute elevation of NEFA has a minor effect upon basal insulin concentration, but significantly increased glucose-stimulated insulin secretion [72–75]; this effect is blunted in obese subjects [73,74], and it is not confirmed when NEFA are infused for 10 h  or ingested simultaneously with glucose [66,77]. The acute enhancement of glucose-stimulated insulin secretion seems to be dependent on the type of NEFA ingested, being more relevant after monounsaturated than after polyunsaturated and saturated NEFA .
Studies investigating the chronic effects of NEFA elevation yielded more conflicting results. A 24–48-h lipid infusion has been reported to reduce , increase  or not significantly change  glucose-induced insulin secretion. A 72-h increase in plasma NEFA concentration produced an enhanced insulin response to the hyperglycemic clamp, but no effect on basal insulin concentration and, interestingly, the effect was opposite in the subjects with family history of type 2 diabetes (T2DM) . In overweight and obese subjects 24-h repeated ingestion  of polysaturated NEFA resulted in a greater increase of plasma insulin levels than in mono- and fully saturated, although the analysis of C-peptide profiles revealed a reduction in insulin clearance. The evidence that NEFA may affect insulin clearance has also been reported in healthy lean subjects [80,83,84] and represents another confounding factor when investigating the insulinotropic effects of NEFA.
Animal [85–87] and in vitro [70,88–90] studies show that an acute elevation of NEFA increases both basal and glucose stimulated insulin secretion while reducing insulin clearance, whereas chronic elevation of NEFA may increase basal insulin secretion but inhibits the secretion stimulated by glucose. Different effects were elicited by different NEFA, depending on their chain length and degree of saturation [91,92].
Slow beta-cell response modes and adaptation mechanisms
The secretion mechanisms discussed above are relatively rapid and typically suited to coping with the insulin needs of a meal. The healthy β cell can also respond with slower modes, adapting to insulin demand if required by the metabolic conditions.
Glucose-induced potentiation of insulin secretion
Slow response modes are unveiled by prolonged exposure to hyperglycemia or repeated glycemic stimuli. During a hyperglycemic clamp, particularly at high glucose levels (>10mmol L−1), the second-phase response exhibits a slow progressive rise over time . When two consecutive hyperglycemic episodes are brought about, the insulin response after the second one is higher compared to the first .With a prolonged (three day) infusion of glucose at a low constant rate, the insulin response assessed with a hyperglycemic clamp before and after the infusion is increased more than twofold . Similarly, a prolonged glucose infusion makes the β-cell dose-response, assessed with the graded glucose infusion test, steeper . When glucose is infused intravenously to mimic the response to an oral glucose test, increasing glucose doses and concentrations produces an upward shift of the β-cell dose-response, that is, insulin secretion becomes greater for the same glucose level . The common denominator of these phenomena is that sustained hyperglycemia potentiates insulin secretion; this mechanism provides an additional resource to control glucose levels.
Another classical experiment showing the potentiating effect of exposure to hyperglycemia involves the use of arginine as a secretagogue. An arginine bolus elicits a burst of insulin secretion similar to that of the IVGTT. When the arginine bolus is administered in a hyperglycemic state created by a hyperglycemic clamp, the insulin secretion response is potentiated compared to the basal state . In normal subjects, the magnitude of this potentiation increases almost in proportion to the glucose levels until it reaches a plateau above ∼30mmol L−1 glucose, where the response to arginine is more than fivefold the response at basal glucose. The initial slope of this curve, denoted as the glucose potentiation slope, is an index of the ability of glucose to potentiate insulin secretion .
Adaptation to insulin sensitivity
A widely studied adaptive mechanism of the β cell is the modulation of its response by prevalent insulin sensitivity. It is a longstanding observation that obesity is accompanied by insulin resistance and hypersecretion ; the widespread use of the IVGTT for the simultaneous assessment of insulin sensitivity and secretion has consolidated the notion that insulin secretion, and in particular fasting insulin secretion and the first-phase response (AIR), is inversely related to insulin sensitivity [27,29,96]. Thus, insulin-resistant subjects with normal glucose tolerance secrete more insulin to cope with the increased insulin demand.
Based on observations largely derived from the IVGTT, it has been proposed that the relationship between insulin secretion and insulin sensitivity is hyperbolic . According to this paradigm, a more appropriate index of β-cell function is the product of the insulin secretion and sensitivity indices, as on the hyperbola representing normal compensation this index is constant. The product of the insulin secretion and sensitivity indices has been historically denoted as the “disposition index” . The disposition index, rather than the absolute insulin secretion response, should reflect intrinsic β-cell function.
Although the experimental evidence supporting the disposition index concept has been mostly derived for AIR and the IVGTT assessment of insulin sensitivity, the paradigm has been often assumed to apply to other indices of insulin secretion. This has been criticized because it has been shown that the assumption of a hyperbolic relationship is not valid for all indices [97,98]. In the presence of inverse relationships of other kinds, the disposition index provides a biased assessment of β-cell function .
Physiologically more important is that β-cell glucose sensitivity obtained from an OGTT is not related to insulin sensitivity [31,99]. Therefore, while insulin resistance does upregulate fasting insulin secretion and the first-phase response, the most important β-cell function index of the OGTT is not affected. This implies that compensation for insulin resistance is of particular importance for the tuning of fasting insulin secretion and therefore fasting glucose, while the secretion mechanisms that regulate postprandial glucose are less influenced by this form of adaptation.