Introduction

Diabetes is increasing steadily all over the world. The International Diabetes Federation estimates that more than 382 million people have diabetes globally and this number is expected to exceed 590 million by 2035 [1]. This is a serious health concern since people with diabetes are vulnerable to both short- and long-term complications, including cardiovascular and renal disease, blindness and amputation [2]. Given the continued prevalence and incidence of diabetes and its considerable health burden, a better understanding of its pathogenesis is required to facilitate the development of new strategies for its treatment.

Ever since its discovery [3], insulin has been the mainstay for the treatment of uncontrolled insulin-deficient diabetes mellitus (uDM). In fact, because of severe beta cell deficiency, the administration of exogenous insulin is necessary for survival. However, despite progress in understanding the causes, consequences and progression of diabetes, and the advancements in the development of insulin and its analogues, maintaining tight glycaemic control without adverse consequences such as hypoglycaemia and weight gain still remains a challenge [4–6]. This further highlights the need for alternative approaches or adjuncts to insulin.

Leptin regulation of glucose metabolism

Growing evidence suggests that, in addition to the regulation of energy homeostasis, the adiposity hormone leptin also plays a key role in glucose metabolism [7]. In support of this, rodent models of leptin deficiency are characterised by insulin resistance and diabetes [8, 9], and leptin treatment lowers blood glucose and insulin levels [10] independent of changes in food intake and body weight [11]. Moreover, leptin administration in both rodents [12, 13] and humans [14] ameliorates the severe insulin resistance and diabetes phenotype characteristic of other models of leptin deficiency that are not associated with obesity, such as lipodystrophy, a condition characterised by loss of adipose tissue owing to mutations that impair adipogenesis [15]. Taken together, these data suggest that leptin regulates glycaemic control in addition to energy balance in both rodent models and clinical settings.

Recent evidence suggests that leptin also has glucose-lowering effects in uDM [16, 17], another model of acquired leptin deficiency [18]. Because insulin is required for the synthesis and storage of triacylglycerol into adipose tissue, in the absence of insulin therapy in uDM there is a depletion of body fat stores. This rapid, progressive loss of body fat stores is accompanied by a pronounced decrease in plasma leptin levels [18]. Thus, uDM is characterised by hyperglycaemia and hyperphagia, and both insulin- and leptin-deficiency have been implicated in these responses.

Since the central nervous system (CNS) plays a critical role in mediating the majority of the actions of leptin on energy homeostasis, we and others hypothesised that the glucose-lowering effects of leptin are mediated by the brain. In support of this, leptin administration directly into the brain, at doses that have no effect when administered peripherally, normalises blood glucose levels in rodent models of uDM [19–22]. This glucose-lowering effect of leptin occurs via a mechanism that is independent of reduced food intake, increased urinary glucose loss or recovery of pancreatic beta cells. Moreover, this glucose-lowering effect of leptin occurs via an insulin-independent mechanism characterised by a normalisation of hepatic glucose production (HGP) and increased rates of glucose uptake in peripheral tissues, including brown adipose tissue, muscle and heart [20]. Collectively, these data establish that leptin signalling in the brain in uDM has the capacity to normalise diabetic hyperglycaemia.

Mechanisms mediating the glucose-lowering effects of leptin

The mechanism(s) that mediates the glucose-lowering effects of leptin in uDM remains to be fully elucidated and is an area of active investigation. uDM is characterised by hyperphagia, hyperglucagonaemia and hypercorticosteronaemia, and each of these is thought to increase glucose production and promote hyperglycaemia [23]. At the same time, the hypothalamic–pituitary–thyroid (HPT) axis is inhibited and, similar to other conditions of leptin deficiency (ob/ob mice and fasting), each of these abnormalities is corrected by leptin treatment [24].

Several lines of evidence implicate hyperglucagonaemia in the hyperglycaemia and ketosis characteristic of uDM [25], and indicate that the glucose-lowering action of leptin is mediated via a normalisation of plasma glucagon levels. For example, hyperglucagonaemia is a hallmark of diabetes in both humans and rodent models [26, 27], and the suppression of glucagon by somatostatin in a pharmacologically induced dog model of diabetes or in humans with type 1 diabetes lowers blood glucose and HGP [28, 29]. Recent findings suggest that hyperglucagonaemia is required for diabetic hyperglycaemia, since glucagon receptor null (Gcgr −/−)-mice fail to develop streptozotocin (STZ; a beta cell toxin)-induced diabetes mellitus (STZ-DM) [30], an effect reversed by restoration of glucagon receptor signalling to the liver [31]. Furthermore, evidence suggests that the glucose-lowering effects of leptin in uDM are accompanied by a normalisation of plasma glucagon levels [17, 19, 20].

However, a similar argument has been made regarding the role of the hypothalamic–pituitary–adrenal (HPA) axis, as uDM is characterised by increased glucocorticoid secretion [32] and this response is also reversed by intracerebroventricular (i.c.v.) infusion of leptin [20]. Recent work further indicates that when fasted rats are studied within 24 h after induction of STZ-DM, the normalisation of elevated corticosterone levels by leptin is sufficient to account for the normalisation of hyperglycaemia, while plasma glucagon levels remain elevated [33]. Thus, reversal of hyperglucagonaemia and HPA axis activation have each been separately advanced to explain the glucose-lowering effects of leptin [17, 33], yet data suggest that once uDM is established, neither mechanism is sufficient to explain leptin’s glucose-lowering actions.

For example, the hypothesis that leptin-mediated reversal of HPA axis activation explains its glucose-lowering effects [33] predicts that circulating corticosterone levels must be elevated for diabetic hyperglycaemia to occur, but previous studies demonstrate that adrenalectomy (ADX) does not prevent hyperglycaemia from developing in rats with uDM fed ad libitum [32, 34, 35], and that the ability of leptin to lower plasma glucocorticoid levels is not required for its glucose-lowering actions [34, 36]. Thus, normalisation of corticosterone levels cannot by itself fully explain leptin’s glucose-lowering effects [33, 34]. While the discrepancy between these findings is not readily apparent, differences in the severity and duration of uDM, or other environmental factors may play a role.

Similarly, the observation that Gcgr −/− mice are protected from hyperglycaemia following STZ is confounded by the marked increase in circulating levels of both glucagon like peptide-1 (GLP-1) and fibroblast growth factor-21 [37, 38], which appear to protect against diabetes irrespective of deficient glucagon signalling. In addition, neither systemic administration of the GLP-1 receptor agonist liraglutide, which reverses hyperglucagonaemia in STZ-DM rats, nor administration of a glucagon-neutralising antibody, when used at a dose that ameliorates diabetic ketosis (and blocks glucagon-induced hyperglycaemia in non-diabetic rats), has any impact on either diabetic hyperglycaemia [39] or elevated rates of HGP. Moreover, physiological leptin replacement in STZ-DM rats is sufficient to normalise both plasma glucagon and corticosterone levels [36], but has little effect on diabetic hyperglycaemia, suggesting that the normalisation of these two hormones is insufficient to explain the potent glucose-lowering effect of i.c.v. leptin.

We therefore propose that in the setting of established uDM, hyperphagia, hyperglucagonaemia and hypercorticosteronaemia, are among several redundant mechanisms that contribute to diabetic hyperglycaemia (Fig. 1). Thus, in order to reverse diabetic hyperglycaemia, leptin must simultaneously normalise several neuroendocrine and autonomic responses, and reversal of one alone is insufficient. This is analogous to the counter-regulatory response (CRR) to hypoglycaemia, an ‘emergency state’ in which the brain perceives a fall in energy availability (i.e. blood glucose). In response to hypoglycaemia, several redundant mechanisms are engaged to raise blood glucose levels, including stimulation of food intake, activation of the HPA axis, stimulation of glucagon secretion and inhibition of insulin secretion [40]. Interestingly, many of the same metabolic and neuroendocrine responses engaged during hypoglycaemia are also elicited during conditions characterised by leptin deficiency, including uDM. However, given the redundant nature of the CRR system, the ability to recover from hypoglycaemia remains largely intact when a single component of the response is blocked [40] but is dramatically impaired if two or more (e.g. both adrenal and glucagon) responses are blocked [28, 41, 42]. Thus, although both adrenal and glucagon responses contribute, neither is required because of the extensive redundancy built into the CRR.

Fig. 1
figure 1

Normalisation of blood glucose levels by leptin in uDM. uDM is characterised by hyperglycaemia and both insulin- and leptin-deficiency because of the loss of pancreatic beta cells and subsequent depletion of adipose tissue stores, respectively. In response to reduced insulin and leptin signalling, the brain initiates a set of behavioural, autonomic and neuroendocrine responses designed to raise blood glucose levels. This effect is mediated by both a reduction in peripheral tissue glucose uptake and increased rates of HGP, attributable in part, to increased glucagon secretion from pancreatic alpha cells and activation of the HPA axis. CNS leptin administration normalises hyperglycaemia in uDM by both potently suppressing HGP and increasing glucose uptake despite persistent severe insulin deficiency, an effect associated with the normalisation of both elevated plasma glucagon and corticosterone (Cort.) levels

Full size image

Neurocircuitry controlling leptin’s glucose-lowering effects

While leptin receptors are expressed throughout the brain, including several hypothalamic and extrahypothalamic areas involved in the control of energy balance and autonomic function [43], several observations implicate a role for the hypothalamic ventromedial nucleus (VMN) in the glucose-lowering effects of leptin in uDM. In support of this, neurons in the VMN express the leptin receptor [43] and are activated by leptin [44], injection of leptin into this brain area increases glucose uptake in peripheral tissues [45, 46] and deletion of leptin receptors from neurons in the VMN causes mild obesity and an insulin-resistant phenotype in mice [47, 48]. In addition, this brain area is also involved in the control of glucagon secretion [49].

Based on these observations, we hypothesised a role for the VMN in mediating the glucose-lowering actions of leptin. Our findings demonstrated that leptin administration to the VMN (at a tenfold lower dose than given i.c.v.) was sufficient to normalise blood glucose levels and other metabolic disturbances associated with uDM [50]. This glucose-lowering effect was also accompanied by the normalisation of both increased HGP and hyperglucagonaemia characteristic of uDM [50]. However, in mice in which leptin receptors were deleted from VMN neurons, the glucose-lowering effects of CNS leptin remained intact [50]. Thus, leptin signalling in the VMN is sufficient, but not required to lower blood glucose levels in uDM and suggests that the VMN may be part of a distributed network through which leptin can act to lower glucose independently of insulin.

Additional studies have further examined the CNS mechanism(s) whereby leptin normalises blood glucose levels in uDM. The available evidence suggests that the glucose-lowering effects of leptin require melanocortin signalling [51, 52]; however, increased CNS melanocortin signalling is not sufficient to lower blood glucose levels in uDM [51, 52]. Consistent with this, i.c.v. leptin administration failed to normalise blood glucose levels in leptin receptor-deficient mice in which leptin receptors were re-expressed in pro-opiomelanocortin (POMC) neurons [53]. By contrast, re-expression of leptin signalling to both γ-aminobutyric acid (GABA) and POMC neurons in otherwise leptin receptor-deficient mice is sufficient to mediate the majority of the glucose-lowering effects of leptin [53]. Future studies are further warranted to better characterise and understand the neurons and neurocircuits that mediate the glucose-lowering and other neuroendocrine effects of leptin in uDM. To facilitate this, new advancements in neuroscience technologies, such as optogenetics and designer receptors exclusively activated by designer drugs (DREADDs), hold promise in the functional mapping and manipulation of discrete populations of cells involved in glycaemic control [54, 55], similar to that which has been applied to increase our knowledge of feeding behaviour [56–58].

Conclusions

The research following the discovery of the hormone leptin more than two decades ago [59] has fundamentally increased our understanding of the CNS mechanisms regulating energy homeostasis. However, while the therapeutic potential of leptin for the treatment of obesity remains limited [60], as leptin plays a key role in glycaemic control, independent of its effects on energy balance, it might be useful as a therapy for other conditions. Leptin therapy has already been approved for the treatment of lipodystrophy and has glucose-lowering effects in rodent models of type 1 and type 2 diabetes [61]. This raises the possibility that leptin may be a potentially useful adjunct to insulin treatment in the management of diabetes.