Present pharmacological treatments of type 2 diabetes aim mainly at increasing insulin secretion by pancreatic β-cells, at improving insulin action on liver, fat, or muscle, and, more recently, at favoring renal glucose elimination. Although efficacious, these treatments do not have the rapid and long-term efficacy of bariatric surgery on correcting diabetic hyperglycemia, suggesting that alternate therapeutic targets can still be identified. Neuronal circuits in the brain, in particular in the hypothalamus and brain stem, can exert strong control on glucose homeostasis. The activity of these circuits is controlled by a variety of cell types—neurons, glial cells, tanycytes—that are sensitive to interoceptive signals such as glucose and various hormones, including insulin, leptin, or ghrelin. When activated, these neuronal circuits can regulate, by controlling autonomic nervous system activity, the secretion of insulin and glucagon by pancreatic endocrine cells as well as glucose metabolic pathways in liver, fat, and muscle. Targeting these neuronal circuits could lead to innovative therapies for diabetes. In the report by Scarlett et al. (1) in this issue of Diabetes and in a previous publication (2), the Schwartz group demonstrated that a single intracerebroventricular (i.c.v.) injection of fibroblast growth factor 1 (FGF1) in diabetic mice and rats leads to a long-term (up to 18 weeks) correction of diabetic hyperglycemia.
FGF1 belongs to the fibroblast growth factor (FGF) family, which contains 18 members (3). Most of the FGFs act as paracrine regulators because they bind to heparan sulfate proteoglycans present in the extracellular space, preventing their diffusion beyond the local environment where they are secreted. The exceptions to this rule are FGF15 (or the human ortholog FGF19) and FGF21, which lack the heparan sulfate proteoglycan binding site and, thus, act as classic endocrine hormones. FGF1 binds to all FGF receptors (FGFRs) (FGFR1 to FGFR4) and their splice variants; high-affinity binding for FGFs requires receptor dimerization and association with heparan sulfates (3). In contrast, FGF15/19 and FGF21 bind to FGFRs in obligatory association with the coreceptor β-klotho. Intracerebroventricular injections of FGF19 induce a short-term normalization of glycemia in diabetic rats and mice; this effect is probably through activation of arcuate nucleus neurons and the regulation of insulin and counterregulatory hormones secretion (4–8).
In their previous study, Scarlett et al. (2) showed that i.c.v. FGF1 normalized glycemia in ob/ob, db/db, and high-fat diet–fed mice by a mechanism that could not be ascribed to increased insulin sensitivity, insulin secretion, or insulin-independent glucose disposal, referred to as glucose effectiveness. In the current study (1), the basis for the hypoglycemic effect of i.c.v. FGF1 was investigated in Zucker diabetic fatty (ZDF) rats. Diabetes development in these rats follows a well-described pattern, with an increased in insulinemia preceding the appearance of hyperglycemia. The onset of diabetes is triggered by a failure of the β-cells to further increase insulin secretion and by glucolipotoxicity-induced apoptosis, which reduces β-cell mass (9,10). Here, Scarlett et al. (1) showed that a single i.c.v. injection of FGF1 in ZDF rats reversibly normalized their glycemia for approximately 30 days. In addition, when injected in prediabetic rats, development of hyperglycemia was delayed by 3–4 weeks, an effect associated with a transient increase in β-cell mass at 3 weeks posttreatment, which was nevertheless reduced within 4 weeks to the low level found in vehicle-treated rats. Analysis of the components of glucoregulation showed, similarly to the previous mouse study, no difference in insulin secretion, insulin sensitivity, or glucose effectiveness. In contrast, the authors found an increase in liver Gck mRNA expression, Gck activity, and increased index of glycolytic flux as determined by increased plasma lactate following glucose injection. They conclude (1) that the main effect of i.c.v. FGF1 is to increase hepatic glucose uptake, probably through a sympathetic regulation.
The results of this and the preceding study (1,2) are quite spectacular, further supporting the central nervous system as an important target for diabetes treatment. Of importance, i.c.v. FGF1 does not induce hypoglycemia in diabetic animals and has no impact on glycemia in regular chow–fed mice. In addition, the mechanisms involved do not involve changes in food intake or body weight. Then what are the mechanisms responsible for these long-term effects? A number of specific questions could be asked: what is the primary site of action of FGF1 in the brain? How is the signal generated and why does it last so long? How is it transmitted to the periphery and what is the primary peripheral target? Why is it observed only in diabetic animals and does not induce hypoglycemia?
When FGF1 is injected i.c.v., a strong induction of c-Fos immunostaining is observed in tanycytes lining the third ventricle (2). These astroglial cells, which line the basolateral part of the third ventricle, are exposed to the cerebrospinal fluid and form intricate contacts with neurons from the dorsomedial, ventromedial, or arcuate nuclei of the hypothalamus; some also extend projections to the blood capillaries of the median eminence (11) (Fig. 1). Their role is not fully understood, but they influence the activity of the arcuate neurons that control feeding and glucose homeostasis, possibly in response to variations in cerebrospinal fluid and/or blood nutrients or hormones (12,13). The anatomical organization of the tanycytes–arcuate neurons barrier and the fact that FGF1 is a poorly diffusible protein suggests that its sustained action may be due to its retention in the immediate vicinity of tanycytes, thereby providing constant stimulation of FGFRs. Alternatively, or in addition, FGF1 may induce the reorganization of neuronal circuits or tanycyte-neuron connections that may have been hampered by the diabetic milieu. The stability of these repaired structures could then explain the long-lasting effect of FGF1 and the fact that it has no hypoglycemic effect in regular chow-fed mice. Using FGF1 variants that lack the heparan sulfate proteoglycan binding site may help resolve these questions.
Scarlett et al. (1) speculate that the action of i.c.v. FGF1 on hepatic glucose uptake could be directed by an increase in sympathetic activity. However, the firing activities of the sympathetic or vagal nerves have not been directly recorded. It would be important to determine which branch of the autonomic nervous system is activated. Indeed, instead of invoking an increase in sympathetic activity in the control of hepatic glucose uptake, an activation of the vagal nerve could explain the transient increase in β-cell mass and insulinemia in the prediabetic ZDF rats. Indeed, increasing vagal activity, for instance by lesion of the ventromedial hypothalamus, strongly increases β-cell proliferation (14), and the observed hyperinsulinemia can also explain the increased expression of hepatic Gck mRNA and Gck activity. Thus, whether glycemia normalization is primarily due to a β-cell protective effect or to an increase in hepatic glucose uptake is difficult to determine based on the presented data. More generally, because rats are studied several weeks after FGF1 treatment, distinguishing between the primary events that trigger glycemia normalization versus secondary events is not straightforward.
Nevertheless, this study is a new illustration of the importance of the central nervous system in glucoregulation. It also indicates that understanding the pathogenesis of type 2 diabetes requires understanding precisely what these brain control mechanisms are, how they become dysfunctional, and whether they precede and cause deregulated insulin secretion and insulin action. Also, earlier studies showed that peripheral injection of FGF1 normalizes glycemia in diabetic mice (15), but the effect was short-lived and dependent on the adipocyte FGFR1. Thus, the beneficial effect of FGF1 as presented in this study depends on its delivery to the central nervous system. This still poses a major challenge because of the need for these drugs to cross the blood-brain barrier. Targeting the basal hypothalamus may be more feasible because of the leakiness of the blood-brain barrier, but if delivery needs to specifically target tanycytes, this may represent an additional difficulty.
See accompanying article, p. 654.
Funding. The work in B.T.’s laboratory is supported by a grant from the Swiss National Science Foundation (310030B_163458), a European Research Council Advanced Grant (694798-INTEGRATE), and a grant from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement no. 115881 (RHAPSODY). This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation program, the European Federation of Pharmaceutical Industries and Associations, and the Swiss State Secretariat for Education, Research and Innovation under contract number 16.0097.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.