Address correspondence and reprint requests to Jenni Harvey, Neurosciences Institute, Ninewells Hospital and Medical School, University of Dundee, Dundee DD1 9SY, UK. E-mail: email@example.com
It is well documented that leptin is a circulating hormone that plays a key role in regulating food intake and body weight via its actions on specific hypothalamic nuclei. However, leptin receptors are widely expressed in the CNS, in regions not generally associated with energy homeostasis, such as the hippocampus, cortex and cerebellum. Moreover, evidence is accumulating that leptin has widespread actions in the brain. In particular, recent studies have demonstrated that leptin markedly influences the excitability of hippocampal neurons via its ability to activate large conductance Ca2+-activated K+ (BK) channels, and also to promote long-term depression of excitatory synaptic transmission. Here, we review the evidence supporting a role for this hormone in regulating hippocampal excitability.
The hormone leptin is a highly conserved 167-amino acid protein that is encoded by the obese (ob) gene (Zhang et al. 1994). It is predominantly, although not exclusively, synthesized by adipocytes and it circulates in the plasma in amounts proportional to body fat content (Maffei et al. 1995; Considine et al. 1996). Leptin displays a high degree of homology amongst different species and it is also analogous in structure to other cytokines (Madej et al. 1995). It was first identified by its ability to regulate food intake and body weight via its actions in the hypothalamus (Jacob et al. 1997; Spiegelman and Flier 2001). However, recent studies have shown that the neuronal actions of leptin are not confined to the hypothalamus. Indeed, evidence is accumulating that this hormone has widespread biological actions in the CNS.
Expression cloning techniques were initially performed by Tartaglia et al. (1995) to isolate the leptin receptor (Ob-R) from mouse choroid plexus. Ob-R was found to be encoded by the diabetes (db) locus located within the 5.1-cm interval of mouse chromosome 4 (Tartaglia et al. 1995). The leptin receptor shows greatest homology to the class I cytokine superfamily of receptors which are characterized by extracellular motifs of four cysteine residues and a number of fibronectin type III domains (Heim 1996). The leptin receptor is known to exist as a homodimer and is activated by conformational changes that occur following ligand binding to the receptor (Devos et al. 1997).
Six leptin receptor isoforms, generated by alternate slicing of the db gene, have been identified so far (Lee et al. 1996; Wang et al. 1998). These isoforms, termed Ob-Ra to Ob-Rf, have identical extracellular N-terminal domains comprising of over 800 amino acids, but have distinct intracellular C-terminal regions. All the leptin receptor isoforms, except Ob-Re (Lee et al. 1997), are membrane spanning receptors that contain a 34-amino acid trans-membrane region. Ob-Re is distinct from the other isoforms and is thought to be a soluble form of the receptor as it is the predominant leptin binding site in the plasma. The remaining isoforms can be classed as either short isoforms (Ob-Ra, c, d and f) with a C-terminal domain of 30–40 residues, or the long isoform (Ob-Rb) with an intracellular domain comprising 302 amino acids in length.
Neuronal leptin receptor expression
High levels of leptin receptor mRNA and protein are expressed in both rodent and human hypothalamus (Schwartz et al. 1996a; Savioz et al. 1997; Elmquist et al. 1998; Hakansson et al. 1998). In particular, specific hypothalamic nuclei (ventromedial hypothalamus, arcuate nucleus and dorsomedial hypothalamus) that are involved in regulating energy homeostasis are highly enriched with leptin receptors. Leptin receptor mRNA and immunoreactivity are also highly expressed in many extra-hypothalamic brain regions including hippocampus, brainstem, cerebellum, amygdala and substantia nigra (Mercer et al. 1996; Elmquist et al. 1998; Hakansson et al. 1998; Grill et al. 2002; Figlewicz et al. 2003). In the hippocampus, the distribution of leptin receptor immunoreactivity has been well characterized, and the CA1/CA3 and dentate gyrus regions exhibit high levels of leptin receptor mRNA and immunolabelling (Mercer et al. 1996; Shanley et al. 2002a). Moreover, in primary hippocampal cultures, dual-labelling approaches have shown that leptin receptor immunoreactivity is found on axonal processes and somato-dendritic regions (Shanley et al. 2002a). It is also highly expressed at hippocampal synapses (Shanley et al. 2002a), suggesting a possible role for this hormone in modulating synaptic function in this brain region.
Transport of leptin into the brain
Leptin is thought to enter the brain via two distinct mechanisms. A saturable transport system is thought to enable leptin to cross the blood–brain barrier via receptor-mediated transcytosis (Banks et al. 1996). Indeed, the short leptin receptor isoforms, which are capable of binding and internalizing leptin, have been detected on brain microvessels (Golden et al. 1997; Bjorbaek et al. 1998). In mice, impairments in leptin transport across the blood–brain barrier develop in tandem with obesity; a process that can be reversed by modest weight reduction (Banks and Farrell 2003). Moreover, recent studies have also shown that leptin transport is regulated by epinephrine (Banks 2001) and triglycerides (Banks et al. 2004). Leptin is also likely to be transported to the brain via the CSF (Schwartz et al. 1996b), as the choroid plexus, the key site for production of CSF, expresses high levels of Ob-Ra and could mediate transport of leptin from the blood to the CSF (Bjorbaek et al. 1998). In addition, leptin has the potential to be made and released locally in the CNS. In support of this possibility, leptin mRNA and immunoreactivity are widely expressed throughout the brain (Morash et al. 1999; Ur et al. 2002). Thus, like other neuropeptides such as oxytocin and vasopressin (Ludwig and Pittman 2003), leptin may be released from neuronal dendrites and signal in a retrograde manner to modulate neuronal function. However, in the absence of evidence supporting dendritic release of leptin, it is likely that peripherally derived leptin has the ability to modulate hippocampal function. Indeed, Banks et al. (2000) have shown that leptin can be transported across the blood–brain barrier to all brain regions. Furthermore, the expression levels of glucocorticoids in the hippocampus are markedly altered following intraperitoneal administration of leptin (Proulx et al. 2001).
Leptin receptor-dependent signalling pathways
The leptin receptor is a class I cytokine receptor (Tartaglia et al. 1995) that activates analagous signalling cascades to other members of this receptor superfamily, such as interleukin 6 and leukemia-inhibitory factor receptors (Ihle 1995). Thus, following leptin binding and subsequent receptor activation, janus tyrosine kinases (JAKs), and in particular JAK2 (Baumann et al. 1996; Bjorbaek et al. 1997), are activated. JAK2 then associates with specific C-terminal domains of the leptin receptor which results in trans-phosphorylation of JAK2 and subsequent phosphorylation of specific tyrosine residues located within the C-terminal domain. This chain of events in turn acts as a catalyst to enable the recruitment and activation of various downstream signalling molecules including signal transducers and activators of transcription (STAT) transcription factors, insulin receptor substrate (IRS) proteins, phosphoinositide 3-kinase (PI 3-kinase) and the Ras-Raf-mitogen-activated protein kinase (MAPK) signalling pathway (for reviews, see Harvey 2003; Hegyi et al. 2004).
The long form of the leptin receptor (Ob-Rb) is thought to be the predominant signalling competent isoform as a result of the expression of various signalling motifs within its C-terminal domain, and the inability of the short leptin receptor isoforms to undergo tyrosine phosphorylation (Bjorbaek et al. 1997). However, the short isoforms are capable of signalling in some cell types. For instance, the MAPK signalling cascade is stimulated following activation of recombinant Ob-Ra expressed in either CHO or human embryonic kidney (HEK) 293 cells (Bjorbaek et al. 1997; Yamashita et al. 1998). Furthermore, in hepatocytes, which fail to express the signalling competent Ob-Rb, leptin still has the ability to inhibit the effects of glucagon (Zhao et al. 2000).
Modulation of hippocampal function by leptin
Regulation of hippocampal excitability
Previous studies have demonstrated that leptin inhibits peripheral insulin-secreting cells (Harvey et al. 1997), glucose-responsive hypothalamic neurons (Spanswick et al. 1997) and nucleus tractus solitarius neurons (Williams and Smith 2006) via activation of ATP-sensitive potassium (KATP) channels. Similarly, leptin inhibits rat hippocampal neurons by increasing a K+ conductance (Shanley et al. 2002a), but, in contrast to these other cell types, KATP channels are not the cellular target for leptin in hippocampal neurons. Thus, the leptin-induced hyperpolarization and increased K+ conductance were inhibited by the Ca2+ and voltage-dependent K+-channel blocker, TEA, but not the sulfonylurea, tolbutamide (Shanley et al. 2002a). Moreover, in single channel recordings, leptin increased the activity of a charybdotoxin-sensitive K+ channel, consistent with the activation of large conductance Ca2+-activated K+ (BK) channels (Shanley et al. 2002a). It is well documented that BK channels consist of a pore-forming α subunit (Slo) with or without a regulatory β subunit (Toro et al. 1998). In HEK293 cells expressing either hSlo or hSlo + hSloβ1, together with Ob-Rb, application of leptin via the patch pipette evoked a rapid increase in BK-channel activity. As leptin was capable of altering BK-channel activity in HEK293 cells expressing only the α subunit, it is likely that β subunits are not a prerequisite for this effect of leptin. In support of this, the effects of leptin were blocked by low nanomolar concentration of charybdotoxin (Shanley et al. 2002a,b), even although expression of the BK-channel β4 subunit, the predominant β subunit in the CNS (Behrens et al. 2000), reduces the sensitivity of BK channels to iberiotoxin and charybdotoxin (Behrens et al. 2000; Meera et al. 2000).
The ability of leptin to modulate BK-channel activity involved a PI 3-kinase-driven mechanism (Fig. 1), as the effects of leptin were inhibited or reversed by the PI 3-kinase inhibitor wortmannin (Shanley et al. 2002a,b). More recent studies have shown that a complex series of events downstream of PI 3-kinase couple leptin receptors to rapid alterations in the actin cytoskeleton and subsequent stimulation of BK channels (O'Malley et al. 2005). This process shows parallels to the effects of leptin on hypothalamic neurons and insulinoma cells, as its ability to modulate KATP-channel function also depends on PI 3-kinase-dependent reorganization of actin filaments (Harvey et al. 2000; Mirshamsi et al. 2004). However, in hypothalamic neurons and insulinoma cells, the precise identity of the intermediate signalling molecules linking PI 3-kinase activity to alterations in actin dynamics is unknown. In hippocampal neurons, leptin receptor-driven activation of PI 3-kinase has been shown to result in a rapid and highly localized increase in the levels of phosphatidylinositol-3,4,5-trisphosphate (PtdIns(3,4,5)P3) at synapses, which promotes the depolymerization and reorganisation of actin filaments. This in turn results in the activation and clustering of BK channels at the hippocampal synapses (O'Malley et al. 2005). However, it is unclear how the elevations in PtdIns(3,4,5)P3 levels influence actin dynamics in hippocampal neurons. As PtdIns(3,4,5)P3 can activate Rho GTPases, which play a key role in regulating actin dynamics, it is feasible that leptin influences the actin cytoskeleton by modifying Rho GTPase activity (Attoub et al. 2000). Alternatively, PtdIns(3,4,5)P3 may directly bind to and alter the activity of actin-binding proteins, thereby influencing actin dynamics (Janmey et al. 1999). The decrease in the levels of the PI 3-kinase substrate, phosphatidylinositol-4,5-bisphosphate (PtdIns(4,5)P2), that occurs following PI 3-kinase activation, may also trigger changes in actin dynamics as PtdIns(4,5)P2 is known to associate with and modulate various cytoskeletal proteins (Janmey et al. 1999).
In hippocampal neurons, BK-channel activation results in generation of the fast after hyperpolarization, which is in turn responsible for repolarization of action potentials. Thus, BK channels are likely to play a key role in determining action potential firing rates and burst firing patterns. Thus, it is conceivable that BK-channel activation by leptin regulates the level of hippocampal excitability. Indeed in a Mg2+-free culture model of epileptiform-like activity, application of leptin induced a rapid and reversible attenuation of the enhanced global levels of intracellular Ca2+ (Shanley et al. 2002b). In contrast, leptin failed to alter the basal levels of intracellular Ca2+ in control conditions. The effects of leptin on this model of epileptiform-like activity were mimicked by a selective BK-channel opener, NS 1619, but were occluded by either iberiotoxin or charybdotoxin, indicating that leptin-induced activation of BK channels underlies this process. Moreover, in parallel with the signalling pathways coupling leptin receptors to stimulation of single BK channels, a PI 3-kinase-, but not MAPK-, dependent signalling cascade underlies this process.
In another model of epileptiform-like activity, acute hippocampal slices were bathed in Mg2+-free medium and subsequent application of leptin reduced the frequency of interictal discharges (Shanley et al. 2002a). Leptin receptor activation underlies this process as leptin inhibited the interictal discharge frequency in slices from Zucker lean, but not obese fa/fa rats (Shanley et al. 2002a). It is interesting to note that in Mg2+-free conditions the frequency of interictal events was significantly higher in slices from Zucker fa/fa rats than in age-matched lean controls, suggesting that rodents that are insensitive to leptin also have an increased level of neuronal excitability. Moreover, the ability of leptin to modulate this excitability is not confined to the hippocampus, as leptin can also markedly influence the firing frequency of hypothalamic neuropeptide Y/agouti-related protein neurons (Takahashi and Cone 2005). Thus, fasting which reduces the circulating levels of leptin resulted in an increase in the spike frequency of neuropeptide Y neurons, whereas direct administration of leptin into the hypothalamus reduced the spike frequency in fasted animals (Takahashi and Cone 2005). In contrast, leptin is reported to increase the frequency of penicillin-evoked epileptic discharges in the somatomoter cortex, suggesting that leptin may have pro-convulsant activity in this brain region (Ayyildiz et al. 2006).
It is well established that unregulated hyperexcitability in the hippocampus is associated with the onset of temporal lobe epilepsy. It is also known that, despite intensive anti-epileptic drug research and development, many individuals with this and other forms of epilepsy display resistance to standard drug therapies. However, several lines of evidence indicate that the incidence and manageability of seizures can be markedly improved by moderate changes in energy homeostasis with diets such as fasting, the ketogenic diet and calorie restriction (Greene et al. 2003). This has led to the hypothesis that metabolic disturbances may influence the severity and frequency of epileptic seizures. Thus, it is tempting to speculate that alterations in the circulating levels of the metabolic hormone leptin may be one of many factors contributing to these disturbances in energy balance in epilepsy and which may in turn influence neuronal excitability.
Leptin induces a novel form of NMDA receptor-dependent long-term depression (LTD)
In addition to its effects on hippocampal neuronal excitability, under conditions of enhanced excitability leptin also markedly alters the strength of excitatory synaptic transmission. Thus, following removal of Mg2+ or blockade of GABAA receptors, leptin induced a novel form of hippocampal LTD (Durakoglugil et al. 2005). This contrasts with the actions of leptin under physiological conditions (1 mm Mg2+), as it promotes the induction of hippocampal long-term potentiation (LTP; Shanley et al. 2001; Wayner et al. 2004), via facilitating NMDA receptor function (Shanley et al. 2001). Recent studies have suggested that activation of NR2A-containing NMDA receptors is required for the induction of LTP, whereas NR2B subunit activation promotes LTD induction (Liu et al. 2004; Massey et al. 2004). Thus, the ability of leptin to bi-directionally modulate the strength of hippocampal excitatory synaptic transmission may reflect differential effects of this hormone on NMDA receptor subunits under different levels of excitability. In support of this possibility, leptin is reported to preferentially enhance NR2B-mediated NMDA responses in cerebellar granule cells (Irving et al. 2006).
The LTD induced by leptin in the CA1 region of the hippocampus was inhibited by the competitive NMDA receptor antagonist D-APV, but not by group 1a and group 5 metabotropic glutamate receptor antagonists, indicating that this form of LTD is NMDA dependent, but independent of metabotropic glutamate receptors. Moreover, leptin-induced LTD shares at least some similar expression mechanisms to LTD induced by low-frequency-induced stimulation (LFS), as leptin did not reduce synaptic responses further following saturation of LFS-induced LTD, whereas LFS still depressed synaptic transmission after leptin-induced LTD (Durakoglugil et al. 2005). Leptin-induced LTD is likely to be expressed postsynaptically, as the synaptic depression induced by leptin was not accompanied by a change in the corresponding paired-pulse facilitation ratio. In contrast, the depression evoked by adenosine, which is known to act presynaptically, was paralleled by a significant change in the paired-pulse facilitation ratio.
Durakoglugil et al. (2005) also evaluated the signalling pathways underlying leptin-induced LTD and demonstrated that inhibition of PI 3-kinase failed to attenuate, but rather markedly enhanced, the level of depression induced by leptin. This suggests that leptin-induced LTD is negatively regulated by PI 3-kinase (Fig. 1). Similarly, inhibition of serine/threonine protein phosphatases 1/2A, but not protein phosphatase 2B, enhanced the depressant effects of leptin. The negative regulation of leptin-induced LTD by PI 3-kinase is in marked contrast to the role of this enzyme in hippocampal LTP, as the ability of leptin to facilitate NMDA receptor-dependent LTP is PI 3-kinase dependent (Shanley et al. 2001). Thus, these data indicate that the ability of leptin to influence different forms of hippocampal synaptic plasticity not only occurs under differential conditions, but also that divergent leptin receptor-driven signalling cascades mediate these processes.
There is growing evidence that, in addition to its role in regulating energy balance, the hormone leptin has widespread actions in the CNS (Fig. 2). In the hippocampus, leptin is a potent regulator of neuronal excitability as it has the ability to inhibit epileptiform-like activity via a process involving PI 3-kinase-driven activation of BK channels (Shanley et al. 2002a,b). Under conditions of enhanced excitability, leptin also promotes a long-lasting inhibition (LTD) of excitatory synaptic strength; a process that is negatively regulated by PI 3-kinase (Durakoglugil et al. 2005). The ability of leptin to markedly alter the excitability of hippocampal neurons via both synaptic and non-synaptic mechanisms may have important implications for the role of this hormone in regulating hippocampal hyperexcitability.