While glucokinase was shown to be rate-limiting for overall glucose metabolism [15], glucose transport in rat islet cells was found to be an extremely rapid process, equilibrating intracellular and extracellular glucose concentrations within seconds [21]. The molecular basis for this rapid transport in rodent islets was clarified with the cloning of the rat liver facilitative glucose transporter isoform GLUT2 which is selectively expressed in liver, islet beta cells and the basolateral membranes of kidney tubular epithelial cells and intestinal mucosa cells [1]. Because the transporter has a high-Km for glucose (around 15 mmol/l) GLUT2 was proposed to be a glucose sensor protein in beta cells in concert with glucokinase [1]. Indirect evidence for this concept was found by observing a parallel reduction of islet GLUT2 expression and loss of glucose-induced insulin release in Zucker diabetic rats (ZDR), a model in which the male animals become obese, insulin resistant and overtly diabetic between week 7 and 9 after birth, while the obese and insulin resistant female animals remain non-diabetic [22]. In the male animals, the severity of diabetes was strongly correlated to the number of GLUT2-immunoreactive beta cells, while liver GLUT2 was not affected to the same extent [22]. In the female ZDR animals a 24-day treatment with 0.2–0.4 mg⋅ kg–1⋅ day–1 dexamethazone induced diabetes, which was accompanied by a loss of glucose-induced (but not arginine-induced) insulin release [23]. The animals had down regulated high-Km glucose uptake in islet cells to about 50% of the normal value, which correlated well with the observed decrease [23] in the number of GLUT2 immunoreactive beta cells (from 100 to 24%). Since total islet GLUT2 protein decreased by only 30%, while pancreatic GLUT2 mRNA even increased in the Zucker diabetic animals [23], it was suggested that:(i) hyperglycaemia increases beta-cell GLUT2 transcription;(ii) down regulation of GLUT2 expression is mediated at the (post) translational level and (iii) not all beta cells are affected to the same degree by loss of GLUT2 expression [24]. The first hypothesis is further supported by observations that glucose increases beta-cell GLUT2 transcription both in cultured islets [25] and in glucose-infused rats [26]. The idea of heterogeneity in beta-cell GLUT2 expression is in agreement with the observation that rat beta cells are heterogeneous in glucose-induced proinsulin biosynthesis [27]. However, when beta-cell subsets were flow sorted on the basis of low and high NADPH autofluorescence after incubation in 7.5 mmol/l glucose [28], no differences were observed in terms of glucose uptake, GLUT2 mRNA level and protein abundance [29], suggesting that this issue needs further investigation in the Zucker model.

Transplantation of non-diabetic GLUT2-positive islets into diabetic recipients has suggested that loss of GLUT2 expression in islets can be experimentally induced by the diabetic environment [30]. In the db mouse, an animal model of diabetes and obesity with a mutation in the leptin receptor gene [31], transplantation of GLUT2-positive db/+ islets under the kidney capsule of db/db animals led to the disappearance of GLUT2 protein from islet beta cells, while kidney GLUT2 remained unaltered [30]. Conversely, transplantation of db/db GLUT2-negative islets into db/+ recipients resulted in the reappearance of the trans-