5-HT inhibition of rat insulin 2 promoter Cre recombinase transgene and proopiomelanocortin neuron excitability in the mouse arcuate nucleus

A number of anti-obesity agents have been developed that enhance hypothalamic 5-HT transmission. Various studies have demonstrated that arcuate neurons, which express proopiomelanocortin peptides (POMC neurons), and neuropeptide Y with agouti-related protein (NPY/AgRP) neurons, are components of the hypothalamic circuits responsible for energy homeostasis. An additional arcuate neuron population, rat insulin 2 promoter Cre recombinase transgene (RIPCre) neurons, has recently been implicated in hypothalamic melanocortin circuits involved in energy balance. It is currently unclear how 5-HT modifies neuron excitability in these local arcuate neuronal circuits. We show that 5-HT alters the excitability of the majority of mouse arcuate RIPCre neurons, by either hyperpolarization and inhibition or depolarization and excitation. RIPCre neurons sensitive to 5-HT, predominantly exhibit hyperpolarization and pharmacological studies indicate that inhibition of neuronal firing is likely to be through 5-HT1F receptors increasing current through a voltage-dependent potassium conductance. Indeed, 5-HT1F receptor immunoreactivity co-localizes with RIPCre green fluorescent protein expression. A minority population of POMC neurons also respond to 5-HT by hyperpolarization, and this appears to be mediated by the same receptor-channel mechanism. As neither POMC nor RIPCre neuronal populations display a common electrical response to 5-HT, this may indicate that sub-divisions of POMC and RIPCre neurons exist, perhaps serving different outputs.

The CNS control of food intake involves complex interactions between circulating hormones, nutrients, neuropeptides, monoamines and other neurotransmitters. These act at a variety of hypothalamic areas (including the paraventricular nucleus (PVN) and the lateral (LHA) and medial hypothalamic areas) to modulate orexigenic and anorexigenic neural pathways (Broberger, 2005). At least two populations of neurons within the arcuate nucleus (ARC) of the hypothalamus contribute to the central circuitry that controls energy homeostasis. These neurons make up part of the melanocortin pathway, which consists of cells containing neuropeptide Y (NPY) along with the endogenous melanocortin antagonist, agouti-related protein (AgRP) and cells containing alpha-melanocyte stimulating hormone (␣-MSH) and other proopiomelanocortin (POMC) derived peptides (Ellacott and Cone, 2004;Cone, 2005). These neurons are key targets for the hormones leptin and insulin, the actions of which effect an anorexigenic output (Niswender et al., 2004).
Food intake is accompanied by changes in the release of monoamines in the hypothalamus (Schwartz et al., 1990), and sympathomimetic drugs (e.g. d-fenfluramine) have long been used to reduce food intake and appetite (Ioannides-Demos et al., 2005). Consequently, pharmacological manipulation that results in enhancement or inhibition of 5-HT synaptic transmission reduces and increases food intake, respectively, in animals and humans (Halford et al., 2005). Electrophysiological recordings from neurons of transgenic mice expressing green fluorescent protein (GFP) under the control of the POMC promoter demonstrate that 5-HT depolarizes arcuate POMC neurons (Heisler et al., 2002), an action also observed for leptin (Cowley et al., 2001;Choudhury et al., 2005). There are at least 14 different 5-HT receptor subtypes and many are present at significant levels in the hypothalamus, notably 5-HT 1B , 5-HT 1F , 5-HT 2A , 5-HT 2B , 5-HT 2C and 5-HT 7 (Hoyer et al., 2002). However, it is still unclear exactly which 5-HT receptor subtypes contribute to modulation of activity in the hypothalamic circuits that sub-serve long-term control of food intake and energy expenditure. In addition, the underlying mechanisms by which 5-HT receptor activation alters the electrical activity of these ARC neurons are unknown. Studies, using selective 5-HT receptor subtype agonists and antagonists, have demonstrated both hyperphagic and hypophagic responses in animal studies. Unfortunately, many of these ligands lose their receptor selectivity at higher concentrations, resulting in some uncertainty over receptor subtype identity in relation to changes in food intake. Nevertheless, there are two main subtypes proposed as key mediators of the anorexigenic action of 5-HT, the 5-HT 1B and 5-HT 2C receptors (Ramos et al., 2005). Neurons expressing 5-HT 1B (Makarenko et al., 2002) and 5-HT 2C receptors (Clemett et al., 2000) are present in hypothalamic feeding centers (i.e. PVN, LHA and ARC), but are also found in brain areas not implicated in energy homeostasis (Hoyer et al., 2002). The 5-HT 2C receptor knockout mouse demonstrates increased food intake and obesity (Tecott et al., 1995) and the hypophagic action of d-fenfluramine is attenuated in the 5-HT 1B knockout mouse (Lucas et al., 1998). However, global 5-HT 1B and 5-HT 2C receptor knockout mice also develop physiological abnormalities such as seizures, anxiety and aggression (Tecott et al., 1995;Ramboz et al., 1996). These observations highlight the multi-functional role of the serotonergic system, but also raise the question whether the feeding behaviors observed are due to the lack of a given receptor in the energy regulatory centers and/or in areas associated with other physiological or pathological outcomes. In addition, although histochemical and in situ hybridization studies demonstrate that many 5-HT receptor subtypes are located in energy regulatory centers, these data provide no information about receptor-mediated alteration of neuronal function, most importantly in relation to changes in neuronal excitability.
Recent studies have indicated that another population of arcuate neurons, defined by GFP expression driven by the rat insulin 2 promoter Cre recombinase transgene (RIPCre), which are distinct from NPY/AgRP and POMC neurons, are involved in the regulation of body weight and energy homeostasis (Cui et al., 2004;Choudhury et al., 2005). Thus, we have examined the actions of 5-HT on the electrical activity of this population of arcuate neurons and show that they respond to 5-HT in a heterogeneous manner with the majority of responding neurons displaying hyperpolarization and reduced excitability.

Electrophysiology
Individual arcuate neurons were identified by epifluorescence and differential interference contrast optics using an upright Zeiss Axioskop-2 FS plus microscope. Slices were continually perfused with a modified external solution (0.5 mM CaCl 2 and 2.5 mM MgCl 2 , no ascorbate and pyruvate) at a flow rate of 5-10 ml/min and a bath temperature of 33°C. For high potassium experiments, the normal external solution was replaced with a solution containing (in mM) NaCl 130, KCl 20, CaCl 2 0.5, MgCl 2 2.5, D-glucose 10, D-mannitol 15, Hepes 10, pH 7.4. Patch-clamp recordings were performed using borosilicate patch pipettes (4 -8 M⍀) filled with an internal solution containing (in mM) K-gluconate 130, KCl 10, EGTA 0.5, Hepes 10, NaCl 1, CaCl 2 0.28, MgCl 2 3, Na 2 ATP 3, tris-GTP 0.3, phosphocreatine 14 (pH 7.2). Whole-cell series resistance (Rs) was compensated using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA) in current (I fast ) and voltage-clamp modes (Rs: 30 -60 and 10 -30 M⍀ respectively). Voltage and current commands were manually or externally driven using PClamp 9.2 software and injected into neurons via the patchclamp amplifier. Under current clamp, hyperpolarizing current pulses (between Ϫ5 and Ϫ20 pA, at a frequency of 0.05 Hz) were used to monitor input and series resistance at resting membrane potentials. In addition, input resistance was calculated from I-V relationships evoked from a holding potential of Ϫ70 mV (Ϯ5-30 pA, 0.5 s pulse duration). Voltage clamp recordings of transient voltage-dependent potassium (I A ) conductance and the delayed and inward rectifying potassium conductances were performed as described in Smith et al. (2007). Whole cell currents and voltages were filtered at 5 and 2 kHz respectively, and digitized at 50 kHz using Pclamp 9.2 software. All data were stored un-sampled on digital audiotape for off-line analysis using Clampex 9.2 or Igor pro. Membrane potentials were either replayed un-sampled on an EasyGraph TA240 chart recorder (Gould, Ballainvilliers, France), or digitized and imported into Abode illustrator for illustration purposes.
Drugs were added to the external solution and applied to slices via the perfusion system or locally applied using a broken tipped pipette (ϳ4 m diameter) positioned above the recording neurons, as previously described (Choudhury et al., 2005). At least 10 min of stable control data were recorded before the application of any drug, and antagonists were applied for at least 10 min prior to challenge with agonist. Neuronal integrity was determined by biophysical and gross anatomical assessment, as described previously Claret et al., 2007).

Immunocytochemistry
Hypothalamic sections (30 m) from paraformaldehyde perfused brains were processed as previously described (Choudhury et al., 2005). Primary antibodies used were rabbit polyclonal antibodies raised to the C-or N-terminal domains of the 5-HT 1F receptor and were obtained from MBL International (MA, USA; cat No. LS-3344 and LS-3338, respectively). Slices were incubated with primary antibody (1:300 dilution) for 48 h at 4°C, following which they were incubated with anti-rabbit secondary antibody conjugated to Alexa Fluor 549 (1:800 dilution) for 1 h. RIPCre-GFP expression and 5-HT 1F receptor localization were detected using a confocal microscope (BioRad MRC 100).

PCR
A 475 base pair fragment encoding a region of the 5-HT 1F receptor was detected by PCR from mRNA extracted from mouse hypothalamus using the following primers: forward GGAAGCTGAGTT-GAGATGATGGC, reverse CACGTACAACAGATGATGTCG.

Statistical analysis
Responsive neurons were distinguished from non-responding neurons based on the criterion that the change in membrane potential (⌬Vm) induced by the drug challenge was Ϯ3 times the standard deviation of the mean membrane potential prior to addition of the drug Claret et al., 2007). Consequently, a neuron was considered hyperpolarized or depolarized if the membrane potential was altered by Ն3 mV. Results are expressed as the meanϮS.E.M. of the defined responses, with the number of cells studied. Statistical significance was determined on all neurons examined within a data set using a Student's twotailed paired t-test or ANOVA, followed by Bonferroni's post hoc test, where appropriate. Comparisons between 5-HT Ϯ agonist or 5-HT Ϯ antagonist on neuron responses were made by paired two-tailed Student's t-test. A P value of less than 0.05 was considered statistically significant.

5-HT modulates the excitability of RIPCre neurons
Whole-cell current-clamp recordings were made from RIPCre hypothalamic arcuate neurons identified by the expression of GFP. Under control recording conditions, and consistent with previous reports (Choudhury et al., 2005;Smith et al., 2007) RIPCre neurons were characterized (nϭ111) by a high mean input resistance, 1.5Ϯ0.1 G⍀, and spontaneously fired sodium mediated action potentials from a mean resting membrane potential of Ϫ57.0Ϯ0.5 mV at a frequency of 4.0Ϯ0.2 Hz. Changes in the electrical excitability of RIPCre neurons (as assessed by changes in membrane potential and firing frequency) were elicited by bath (1 M) or locally (2-5 M) applied 5-HT with no discernable difference in the output in relation to the method of application. Application of 5-HT resulted in clear heterogeneous responses in this population of neurons. Overall, 5-HT inhibited 50% (55/111), excited 25% (28/111) or did not alter the excitability of 25% (28/ 111) of RIPCre neurons. In neurons, which displayed an inhibitory response to 5-HT there was a rapid hyperpolarization ( Fig. 1A) from a mean membrane potential of Ϫ50.6Ϯ0.8 mV to Ϫ63.3Ϯ1.1 mV (nϭ55, PϽ0.05), which resulted in a decrease in neuronal firing rate, from a mean value of 4.1Ϯ0.3 Hz to 0.6Ϯ0.2 Hz (PϽ0.05). This hyperpolarization and reduction in excitability was accompanied The predominant response to bath applied 5-HT (1 M) was hyperpolarization and inhibition of firing. This action of 5-HT was readily reversible and a second application of 5-HT produced the same effect. (B) A smaller proportion of RIPCre neurons responded to 5-HT by depolarization and increased excitability. A diary plot of firing frequency for this neuron is shown, with bath-applied 5-HT demonstrating a clear excitation. (C) The remaining proportion of RIPCre neurons tested was insensitive to 5-HT, with no evidence of a ⌬Vm or of firing frequency. Expanded regions of this recording are displayed to show more clearly that bath-applied 5-HT had no effect on membrane potential. The diary plot of firing frequency is shown for this neuron. (D) Increasing doses of 5-HT were locally-applied to a RIPCre neuron, shown previously to respond to 5-HT by hyperpolarization. Increasing the duration of pressure ejection of 5-HT (5 s-25 s) increased the magnitude and duration of the 5-HT response. Note that increasing the dose of 5-HT did not induce receptor desensitization at the time intervals used for successive 5-HT challenges, and there was no evidence for heterogeneity of response to 5-HT. by a decrease in whole-cell input resistance from a mean value of 1.50Ϯ0.09 G⍀ to 1.24Ϯ0.23 G⍀ (PϽ0.05), indicating that an increase in conductance underlies this response. 5-HT-mediated hyperpolarizing or depolarizing neuronal responses (nϭ8) were also obtained (data not shown) from slices exposed to 0.5 M TTX. RIPCre neurons that were excited by 5-HT, were rapidly depolarized (Fig. 1B) from a mean membrane potential of Ϫ52.4Ϯ1.1 mV to Ϫ47.2Ϯ1.2 mV (nϭ28, PϽ0.05), and this was accompanied by an increase in firing rate from a mean value of 3.1Ϯ0.5 Hz to 5.2Ϯ0.8 Hz (PϽ0.05). The depolarization and increased firing were not accompanied by any significant change in whole-cell input resistance (from 1.60Ϯ0.11 G⍀ to 1.68Ϯ0.13 G⍀). In RIPCre neurons where electrical activity was unaffected by application of 5-HT (Fig. 1C), there was no significant change in either resting membrane potential (Ϫ50.0Ϯ0.9 mV to Ϫ50.4Ϯ0.9 mV) or firing rate (4.6Ϯ0.5 Hz to 4.9Ϯ0.6 Hz).
We found no evidence for any dose-dependent effect of 5-HT with respect to these alterations in excitability; RIPCre neurons responded to 5-HT by hyperpolarization, depolarization or no response regardless of concentration (up to 10 M tested) or duration of 5-HT application. An example of such a single reproducible outcome is shown in Fig. 1D, where increasing amounts of 5-HT were pressure ejected (by altering the duration of pressure application) onto a RIPCre neuron, previously shown to respond to 5-HT by hyperpolarization. A dose-dependent increase in the hyperpolarizing response amplitude and duration was observed (nϭ3), with no evidence for the direction of the response being dependent upon the duration (dose) of 5-HT ejection. The changes in membrane potential and spike firing induced by 5-HT (and other agonists, see below) at the concentrations utilized in this study on RIPCre neurons were reversible on washout within 15-30 min. Moreover, no desensitization of response to 5-HT or Fig. 2. 5-HT 2 receptors are not responsible for 5-HT-induced RIPCre neuron depolarization. (A) Bath-applied ␣-me 5-HT (1 M) did not affect the membrane potential or firing frequency of a RIPCre neuron, which had been shown previously to respond to 5-HT with depolarization (upper trace). Expanded traces and the corresponding diary plot of firing frequency for this neuron are shown. (B) Increasing the concentration of ␣-me 5-HT to 10 M mimicked the depolarizing effect of 5-HT on the same RIPCre neuron. The expanded traces (lower) show the depolarization more clearly. Note that the depolarization of this neuron by either agonist was sufficient to cause severe action potential truncation. The presence of (C) the selective 5-HT 2C antagonist, SB242084 (100 nM) (D) ketanserin (10 nM) or (E) the 5-HT 2B antagonist SB204741 (100 nM) did not prevent locally-applied 5-HT from depolarizing and increasing the firing frequency of RIPCre neurons. (F) Application of 5 M BW72C86, a selective 5-HT 2B agonist, had no effect on the excitability of RIPCre neurons that responded to 5-HT by depolarization.
agonists was observed when re-applied following several minutes of washout (e.g. Fig. 1A, D). In subsequent experiments, detailed below, selective agonists and antagonists were applied to RIPCre neurons following an initial 5-HT challenge, which was used to ascertain the category of the response (depolarizing, hyperpolarizing or non-responsive).

BRL54443 inhibits RIPCre neurons by increasing a voltage-dependent K ؉ conductance
In an attempt to identify the conductance(s) modulated by BRL54443 that give rise to the hyperpolarizing response, RIPCre neurons were voltage-clamped in an external solution containing 10 M bicuculline, 2 mM kynurenic acid and 1 M TTX to block synaptic transmission and regenerative Na ϩ spikes. Neurons were held at Ϫ70 mV and voltage pulses (500 ms duration) were stepped from Ϫ90 to Ϫ10 mV in 5 mV increments, with a 5 ms pre-pulse stepped to Ϫ170 mV to deactivate voltage dependent potassium conductances (Fig. 6A). As described for POMC (Roseberry et al., 2004) and RIPCre  neurons, the inward rectifier potassium (K IR ) conductance in arcuate neurons is extremely small under our control recording conditions. To increase the magnitude and induce a shift in the reversal potential (to approxi-mately Ϫ50 mV) of the K IR conductance, the external potassium concentration was raised from 2.5-20 mM . We have previously shown , for RIPCre neurons, that the K IR conductance is blocked by 100 M Ba 2ϩ , the transient voltage-dependent potassium (I A ) conductance by 4 mM 4-aminopyridine, and the delayed rectifier type voltage-dependent potassium conductance partly blocked by 40 mM external TEA (see also Fig. 6B). Use of these blockers allowed identification and relative isolation of the main potassium currents observed in these neurons under our recording conditions. Local application of 20 nM BRL54443 to RIPCre neurons had no effect on K IR , with a linear slope conductance (measured between Ϫ90 and Ϫ50 mV) in 20 mM [K o ] of 1.6Ϯ0.5 nS in control and 1.6Ϯ0.5 nS (nϭ8) in the presence of BRL54443 (data not shown). Similarly, 20 nM BRL54443 did not significantly change the peak amplitude of I A (measured at Ϫ10 mV; 112Ϯ6% of control, PϾ0.05, nϭ8). In contrast, 20 nM BRL54443 reversibly increased the steady state depolarization-activated potassium current (measured at the end of the Ϫ10 mV command pulse) by 36Ϯ8% (PϽ0.05, nϭ8, Fig. 6C, D).

DISCUSSION
The primary aim of this study was to examine the effect of 5-HT on RIPCre neuron electrical excitability. It had previously been reported that 5-HT 2C receptor agonists could depolarize and excite POMC neurons (Heisler et al., 2002). Our results showed that although RIPCre neurons could be depolarized by 5-HT this was not the majority response, with only 25% demonstrating an increase in excitability. In fact, half of all neurons tested displayed a hyperpolarizing, inhibitory, response to 5-HT application. The 5-HT induced changes in excitability of RIPCre neurons are likely to be predominantly directly mediated rather than synaptically driven as similar responses are observed in the presence of TTX in current clamp recordings. The increased voltage-dependent potassium currents observed in voltage-clamped RIPCre and POMC neurons in the presence of TTX and GABA and glutamate receptor antagonists are also consistent with a direct effect of 5-HT on these neurons. Previous studies of POMC and RIPCre neurons have demonstrated that peptide agonists and hormones do not elicit responses from every identified neuron (Cowley et al., 2001;Choudhury et al., 2005;Plum et al., 2006;Claret et al., 2007;Könner et al., 2007). It was expected that 5-HT would alter the excitability of RIPCre neurons by one or more of the following receptor subtypes; 5-HT 1A,B , 5-HT 2A,B,C or 5-HT 7 , as these have been shown to be expressed in the medial hypothalamus or arcuate, and implicated in 5-HT modulation of feeding (Hoyer et al., 2002;Ramos et al., 2005). Indeed, we reasoned that RIPCre neu- ron depolarization could occur through the same receptor mechanism reported for POMC neurons. However, surprisingly we were unable to demonstrate, using selective 5-HT 2A,2B,2C antagonists and a selective 5-HT 2B agonist that the 5-HT-induced depolarization of RIPCre neurons was due to any of these subtypes. Clearly the receptor subtype and identity of the channel mechanisms underlying the 5-HT depolarization of RIPCre neurons will require further analysis.
5-HT-induced hyperpolarization of RIPCre neurons was unlikely to be mediated by a member of the 5-HT 2 family as ␣-me 5-HT was ineffective, the 5-HT responses were insensitive to 5-HT 2A,B,C antagonists and a selective 5-HT 2B agonist. In contrast, the observation that 5-CT hyperpolarized RIPCre neurons suggested that the receptor underlying the 5-HT hyperpolarization is a member of the 5HT 1 family, or the 5-HT 5 or 5-HT 7 receptor. However, the lack of effect of 8-OH-DPAT, CGS12066B and L694247 as agonists and SB 224289 and sub-micromolar methiothepin as antagonists to 5-HT-mediated hyperpolarization, effectively ruled out 5-HT 1A,B,D and 5-HT 5,7 receptors as responsible. The observation that low concentrations (10 or 20 nM) of the mixed 5-HT 1E,F receptor agonist, BRL54443 hyperpolarized and that micromolar concentrations of methiothepin prevented 5-HT induced hyperpolarization of RIPCre neurons indicates that one of these subtypes is likely responsible. Unfortunately, no selective 5-HT 1F receptor agonist or antagonist was available to allow us to distinguish between these 5-HT receptor isoforms, but as mice do not possess the gene for 5-HT 1E receptors (Bai et al., 2004) we suggest that the 5-HTinduced hyperpolarization of RIPCre neurons is mediated by activation of the 5-HT 1F receptor. The human cloned 5-HT 1F receptor has previously been demonstrated to require a methiothepin concentration of ϳ1 M to inhibit a functional response (Adham et al., 1993). Previous studies have indicated (20 nM) hyperpolarization of a POMC neuron. Expanded traces are shown below for each agonist application ((i)-(iii)). Note that a second application of BRL54443 (20 nM) ϳ3 min after the first was unable to elicit a response, and even after ϳ9 min an increased dose of BRL54443 was required to produce significant hyperpolarization and reduction in firing. (E, F) BRL54443 reversibly increased the amplitude of the steady-state potassium current in POMC neurons. Representative current families in the absence and presence of BRL54443 (20 nM) are shown (E) with expanded single test pulses to Ϫ10 mV (F) to demonstrate this action of BRL54443 is reversible and limited to the steady-state, not the peak, potassium current amplitude. that mRNA for the 5-HT 1F receptor is present in human brain (Hoyer et al., 1994(Hoyer et al., , 2002, and in situ hybridization studies of mouse brain indicate its presence in numerous regions, including the hypothalamus (Bonnin et al., 2006). In support of this contention, mRNA for the 5-HT 1F receptor was detected by PCR in mouse hypothalamus in agreement with a previous study in rat (Lovenberg et al., 1993). Furthermore, immunohistochemical analysis of mouse ARC slices, using two separate antibodies to the 5-HT 1F receptor demonstrated its expression in most RIPCre neurons as well as other, undefined, neurons in the ARC. BRL54443 increased current through a channel that has the basic characteristics of a steady-state, delayed rectifier-like potassium channel with no effect on K IR or I A , the other major potassium currents present in these neuron types . Previous studies have demonstrated that a number of 5-HT receptor subtypes are capable of coupling to K ϩ channels although this appears primarily mediated via Ca 2ϩ -activated K ϩ channels or G-protein-gated inwardly rectifying K ϩ channels (Raymond et al., 2001), rather than the delayed rectifier family of K ϩ channels. Although we demonstrate that BRL54443 increases the current through a voltage-gated potassium conductance, it is unlikely that the typical delayed rectifier or I A potassium channels (i.e. Kv1-4 families) underlie the 5-HTmediated hyperpolarization. 5-HT and BRL54443 hyperpolarize RIPCre and POMC neurons from a membrane potential of Ϫ50 to Ϫ55 mV, a range where most delayed rectifier channels are closed. One plausible candidate group of voltage-gated potassium channels is the KCNQ gene family. These encode Kv7 channel subunits, which are expressed widely in the brain, exhibit voltage dependent activation, are functionally active at the resting membrane potential of many neurons and can be modulated by neurotransmitters (Hansen et al., 2008). Additionally, the coupling mechanism by which this 5-HT receptor increases potassium channel current is presently unknown. Studies of the human 5-HT 1F receptor, expressed in 3T3 or fibroblast cell lines, indicate that this receptor subtype is capable of engaging with multiple cell signal transduction pathways (i.e. cAMP, inositol phosphates and intracellular calcium), in a cell-specific manner (Adham et al., 1993).
Our results do support the contention that at least part of the anorexigenic actions of 5-HT in the hypothalamus may be via the central melanocortin system (Heisler et al., 2002(Heisler et al., , 2006. Indeed, the main action of 5-HT on identified POMC neurons, albeit on a minority of the population (25%) we sampled in the arcuate, was depolarization and increased excitation in agreement with the previous report (Heisler et al., 2002). This result is consistent with the report that low concentrations of 5-HT directly stimulate ␣-MSH release from POMC neurons in the hypothalamus (Tiligada and Wilson, 1989). However, POMC neurons can also respond to 5-HT by hyperpolarization and inhibition of firing, and our current clamp and voltage clamp data using BRL54443 as an agonist, suggest this outcome may also be driven by the same 5-HT receptor-voltage-activated potassium channel mechanism described for ARC RIPCre neurons. Thus, POMC neurons also appear capable of divergent electrical responses to 5-HT and this may be linked to differ-ential outputs, such as altered food intake and energy expenditure or modulation of peripheral glucose levels or fat storage, sub-served by different groups of POMC neurons (Cone, 2005). Indeed, there is additional evidence that multiple 5-HT-dependent mechanisms contribute to modulation of POMC neuron excitability as Heisler et al. (2006) have shown that 5-HT, by activating 5-HT 1B receptors on NPY/ AgRP neurons, can reduce the inhibitory drive onto POMC neurons (i.e. indirect excitation).
The responses of the RIPCre neurons investigated here indicate that the predominant action of 5-HT is to hyperpolarize, and inhibit, these neurons. This occurs through activation of a voltage-gated K ϩ conductance and is likely mediated by the 5-HT 1F receptor subtype. Unfortunately, it is not yet possible to judge the physiological significance of this inhibition of RIPCre neuron function, or its potential relation to 5-HT-mediated physiological outcomes such as reduced food consumption or mood alteration. This is because we do not presently know the identity of the transmitter(s)/peptide(s) released from this neuronal population, nor how these neurons relate to the NPY/AgRP and POMC neuron ARC circuitry, although there is evidence to suggest that RIPCre neurons may be part of the melanocortin system (Choudhury et al., 2005;Smith et al., 2007).