Leptin Grows Up And Gets A Neural Network

From linear thinking to networks

The hypothalamus integrates sensory information with hormonal signals to regulate the activity of the autonomic nervous system and control hormone secretion from the pituitary gland. The organization and activity of hypothalamic neural circuits are critical for the integration of these sensory and hormonal signals. The adipocyte-derived hormone leptin acts directly on the hypothalamus, but attempts to find dominant sites of leptin action in the regulation of energy balance have failed. Body adiposity is a complex phenotype that is the integration of linked functions of energy intake, expenditure, and partitioning. Although the neurotransmitters GABA and glutamate tend to dominate the regulation of forebrain neural circuits, the role of neuropeptides in mediating the neural control of energy balance has received the majority of experimental attention, with a focus on neuropeptide Y (NPY) and melanocortin (POMC) containing neurons of the arcuate nucleus (ARC). The notion that these two populations of neurons represent a direct and critical site for bidirectional regulation of energy homeostasis by leptin has been enormously influential and has provided the conceptual framework for much of the work on how leptin functions to regulate body adiposity. In this issue of Neuron, Vong et al. provide the simple, yet paradigm-shifting observation that leptin controls this circuit primarily through a distributed network of GABAergic neurons (Vong et al., 2011).

In previous studies, the leptin receptor has been specifically deleted from a number of neuronal cell types defined by common neuropeptidergic expression or developmental origin, including POMC neurons (Balthasar et al., 2004), AgRP neurons (van de Wall et al., 2008), SF1-positive VMH neurons (Dhillon et al., 2006), and others. In every case, these experiments yielded mild obesity syndromes with only a small percentage of the adiposity seen in the global leptin receptor knockout. Taking a different approach assessing the role of GABAergic and glutamatergic neurotransmission in mediating the biological effects of leptin, Vong et al. constructed Vgat-ires-Cre and Vglut2-ires-Cre transgenic mice, and crossed these with previously characterized Lepr^flox/flox mice (Balthasar et al., 2004) to specifically delete the leptin receptor from GABAergic and glutamatergic neurons. The weight gain in the Vgat-ires-Cre, Lepr^flox/flox mice was 83% (females) to 86% (males) of that seen in the global leptin receptor knockout mice. By comparison, the increase in weight in Vglut2-ires-Cre, Lepr^flox/flox mice was minimal. While mice lacking leptin receptor in GABAergic neurons exhibited up to a 10-fold increase in adipose mass, those lacking leptin receptor exclusively in glutamatergic neurons exhibited a modest, but significant, 2-fold increase in adipose mass. Significant hyperphagia and an increase in lean mass were seen in the Vgat-ires-Cre, Lepr^flox/flox mice, while neither of these changes were observed in the Vglut2-ires-Cre, Lepr^flox/flox mice. Not surprisingly, the former were diabetic, while the latter were euglycemic, and exhibited normal fasting insulin levels. The striking conclusion is that only very modest effects result from the cumulative action of leptin on glutamatergic, neuropeptidergic, and other non-GABAergic neuronal cell types.

The GABAergic NPY/AgRP neuron is the only characterized GABAergic neuron known to express leptin receptors, and since deletion of leptin receptor from these cells has little effect (van de Wall et al., 2008), a critical question involves defining the GABAergic leptin responsive neurons responsible for the bulk of leptin action. Lowell and colleagues collected data to address this by injecting fasted Vgat-ires-Cre, Lox-GFP reporter mice with leptin and identifying leptin-activated GABAergic neurons by costaining cells for GFP and pSTAT3, a marker of leptin receptor signaling. Positive cells were only found in ARC, DMH, and LH. Of course, this experimental paradigm would only identify neurons activated by an acute increase in leptin; neurons regulated by a decrease in leptin may not be identified by this method.

The networks of leptin-receptor expressing GABAergic neurons described here undoubtedly control multiple CNS circuits, but the most well characterized is the NPY/AgRP and POMC neurons that project to over 100 different brain regions to coordinately regulate food intake and energy expenditure (Cone, 2005). Lowell and colleagues next sought to address the relative contributions of NPY/AgRP versus distributed GABAergic interneurons in leptin-induced inhibition of POMC neurons, as measured electrophysiologically. Deletion of leptin receptors globally or selectively in GABAergic neurons enhanced inhibitory tone onto POMC neurons, as reflected by increased frequency and amplitude of IPSCs in POMC neurons. In contrast, deletion of leptin receptors in POMC or NPY/AgRP neurons alone had no measurable effect on inhibitory inputs to POMC neurons. Again, the conclusions are inescapable whether one examines whole animal energy and glucose homeostasis or individual hypothalamic circuits: leptin acts via a network of GABAergic neurons to reduce inhibitory tone to POMC neurons (Figure 1).

Figure 1. Leptin Regulates the ARC to PVN Circuit Via a Distributed Network of Predominantly GABAergic Neurons.

Diagram shows one example of the hundreds of POMC and NPY/AgRP projections throughout the CNS, and the network of leptin receptor expressing GABAergic neurons controlling this circuit. Major sites of leptin-responsive GABAergic neurons identified in Vong et al. include ARC, LH, and DMH. Note that most of the GABAergic and glutamatergic inputs to the POMC and NPY/AgRP neurons remain undefined. Target neurons in the PVN are highly heterogeneous, and are also regulated by a complex network of GABAergic and glutamatergic inputs.

From neuropeptides to neurotransmitters and circuits

The desire to functionally associate the role of a single neuropeptide system with an hypothalamic function stems in large part from the study of releasing hormone containing neurons, such as the corticotropin releasing hormone (CRH) neurons. Within this framework, neuropeptides are primary effectors that control hormone release from the anterior pituitary. However, hypothalamic circuits regulating energy expenditure are far more complex. The leptin receptor is expressed in dozens of sites in the forebrain and brainstem. While previous research had largely focused on the control of neuropeptide synthesis and release by direct leptin action on POMC and NPY/AgRP neurons, these data focus research on the control of neurotransmitter release by leptin with broad implications for hormonal control of information processing by hypothalamic circuits.

One of the most well characterized subcircuits involved in energy homeostasis involves neurons in the paraventricular nucleus (Figure 1). Many of these neurons are hypophysiotropic neuropeptidergic neurons that project to the median eminence where they release peptides that control the release of pituitary hormones, while others project to the brainstem regions controlling autonomic outflow. Both classes of cells can express melanocortin receptors and NPY receptor subtypes, and receive dense projections from POMC and NPY/AgRP neurons from the ARC. Analysis of this subcircuit reinforces the findings presented, and raises some further questions. First, just as the electrical activity of POMC and NPY/AgRP neurons are controlled by leptin and metabolic state (Takahashi and Cone, 2005; Vong et al., 2011), the same properties have been found in melanocortin-4 receptor expressing PVN motoneurons (Ghamari-Langroudi et al., 2011). In short, the firing frequency of these cells increases in fasted mice, and this increase can be inhibited if animals are fasted but given leptin peripherally. This finding reinforces the concept that leptin responsive neurons controlling the activity of a neural circuit are distributed, and the effects on the circuit are distributed across multiple cells in the circuit, rather than residing in a single neuronal cell type like the arcuate POMC neuron. Surprisingly, over 90% of MC4R neurons in the PVN are potently inhibited in response to leptin (Ghamari-Langroudi et al., 2011; Ghamari-Langroudi et al., 2010); thus minor sites of leptin action outside the GABAergic network exist throughout the circuit.

Refining the phenotype

It is important not to lose sight of the fact that energy homeostasis is a complex phenomena involving multiple integrated circuits that are not only tonically regulated by leptin, but regulated as well by a host of other inputs, such as satiety and hunger signals, the energy demands of lean mass, and ambient temperature, to name a few. The animal models created here will serve as important tools for determining if leptin regulation of these individual circuits works through alterations in glutamatergic or GABAergic inputs. There may be important leptin-mediated subfunctions that are not predominantly controlled by inhibitory GABAergic cells. Indeed, earlier data show that leptin receptor expression non-GABAergic POMC cells are disproportionately responsible for the cumulative effect of leptin in the brain (Balthasar et al., 2004). Deletion of leptin receptor from the POMC neuron recapitulates around 20% of the obesity from global deletion, yet these neurons comprise a much smaller fraction of all leptin-receptor expressing cells (Balthasar et al., 2004). The data in Vong et al. may also point to a specific role of glutamatergic cells in mediating the effects of leptin on the responsiveness to neuronal and hormonal gut-derived satiety signals.

From the synapse to synaptic plasticity

In all neural circuits the point of communication between neurons is the synapse and defects in the function of presynaptic components of synapses have been implicated in a variety of neurological disorders (Waites and Garner, 2011). Such defects may include errors in axonal targeting of GABAergic neurons to these circuits, alterations in synaptic density, reduced synapse stability, and degradation of synaptic vesicle release. Leptin clearly can cause dynamic changes in presynaptic organization in the ARC (Pinto et al., 2004) and is required for normal targeting of ARC axons in the hypothalamus (Bouret et al., 2004). Genetic or environmental insults that affect any of these processes may have a significant impact on the regulatory actions of leptin. The lack of major alterations in body weight in mice that lack leptin receptors in glutamatergic neurons is surprising, but glutamatergic synapses are remarkably plastic throughout life. The findings of Vong et al are indeed important for they demonstrate the importance of presynaptic regulation by leptin and define the outcome in terms of the activity of POMC neurons. Together, these results may imply that leptin functions at multiple cellular levels, in much the same way that glucocorticoids and estrogen regulate multiple aspects of neuronal function in other homeostatic forebrain pathways. While it is clear that interest in the organization and regulation of hypothalamic neural circuitry originates in a desire to understand physiological mechanisms underlying homeostatic systems, it is equally clear that such understanding will only be gained through experimental dissection of the neurobiological events responsible for the function of the circuitry. The landmark paper by Vong et al. creates a new conceptual framework for the study of the hypothalamic neural circuitry mediating energy homeostasis.