Corresponding author D. Burdakov: Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1PD, UK. Email: email@example.com
Central orexin/hypocretin neurones are critical for sustaining consciousness: their firing stimulates wakefulness and their destruction causes narcolepsy. We explored whether the activity of orexin cells is modulated by thyrotropin-releasing hormone (TRH), an endo-genous stimulant of wakefulness and locomotor activity whose mechanism of action is not fully understood. Living orexin neurones were identified by targeted expression of green fluorescent protein (GFP) in acute brain slices of transgenic mice. Using whole-cell patch-clamp recordings, we found that TRH robustly increased the action potential firing rate of these neurones. TRH-induced excitation persisted under conditions of synaptic isolation, and involved a Na+-dependent depolarization and activation of a mixed cation current in the orexin cell membrane. By double-label immunohistochemistry, we found close appositions between TRH-immunoreactive nerve terminals and orexin-A-immunoreactive cell bodies. These results identify a new physiological modulator of orexin cell firing, and suggest that orexin cell excitation may contribute to the arousal-enhancing actions of TRH.
Hypothalamic neurones containing the neuropeptide transmitters hypocretins/orexins (orexin neurones) are critical for normal sleep/wake transitions and vital adaptive behaviours (de Lecea et al. 1998, 2006; Sakurai et al. 1998). Their firing promotes wakefulness (Adamantidis et al. 2007), while their loss leads to narcolepsy (Chemelli et al. 1999; Lin et al. 1999). The activity of orexin neurones is also thought to stimulate feeding and reward-seeking (Sakurai et al. 1998; de Lecea et al. 2006). Orexin neurones are located in the lateral hypothalamic area (LHA) but project to all brain areas except the cerebellum, with especially dense innervation of regions regulating cognitive arousal (Peyron et al. 1998; Sakurai et al. 1998). Despite their importance for brain state control, the regulation of electrical activity of orexin neurones is not fully understood. Recent reports indicate that they are innervated by fibres originating in the dorsomedial hypothalamus (DMH), a centre critical for circadian organization of diverse behaviours. About 35% of DMH neurones that project to the LHA contain thyrotropin-releasing hormone (TRH) (Chou et al. 2003). Although first characterized as a regulator of the pituitary–thyroid axis, TRH also functions as a neurotransmitter within the brain, acting as a potent CNS stimulant independently of its actions on pituitary secretion (Lechan & Fekete, 2006). The neural mechanisms underlying these effects are incompletely understood.
Considering that TRH and orexin neurones both regulate cognitive arousal, we hypothesized that these two neural systems may interact. Here, we investigate this idea using transgenic GFP-tagging of orexin neurons, patch-clamp electrophysiology in living brain slices, and immunofluorescence staining in the mouse brain.
Animal procedures were carried out in accordance with the Animals (Scientific Procedures) Act, 1986 (UK). Coronal slices (250 μm thick) were obtained from 13- to 22-day-old transgenic mice selectively expressing enhanced green fluorescent protein (eGFP) in orexin cells, as previously described (González et al. 2008). Briefly, orexin-eGFP cells were identified in brain slices by epifluorescence, whole-cell recordings were made at 36°C using an EPC-10 amplifier (Heka, Lambrecht, Germany), and data were sampled using Patchmaster software (Heka, Lambrecht, Germany). Most recordings consisted of alternating 2 min-long current-clamp traces and ∼30 s-long voltage-clamp protocols (see below). Because of the latter, breaks can be seen in current-clamp traces in Figs 1 and 2, but the shown duration of these ∼30 s-long breaks is compressed for presentation clarity.
Pipettes were pulled from borosilicate glass and had tip resistances of 3–5 MΩ when filled with intracellular solution containing (in mm): KCl 130, Hepes 10, EGTA 0.1, MgCl2 2, K2ATP 5, and NaCl 2 (pH 7.25 with KOH). This solution was used in most recordings, but in certain experiments (when stated in Results), we instead used ‘low-Cl−’ (potassium gluconate) or ‘low-K+’ pipette solutions. The potassium gluconate solution contained (in mm): potassium gluconate 120, KCl 10, EGTA 0.1, Hepes 10, K2ATP 4, Na2ATP 1, and MgCl2 2 (pH 7.3 with KOH). The ‘low K+’ solution contained (in mm): CsCl 112, TEA-Cl 20, MgCl2 2, Hepes 10, Na2ATP 5, and Cs-EGTA 0.2 (pH 7.25 with CsOH).
‘Control’ extracellular solution contained (in mm): NaCl 125, KCl 2.5, MgCl2 2, NaH2PO4 1.2, NaHCO3 21, CaCl2 2, and glucose 1. ‘Low Ca2+’ extracellular solution instead contained 9 mm MgCl2 and 0.3 mm CaCl2. ‘Low Na+’ extracellular solution contained (in mm): NMDG (N-methyl-d-glucamine)-Cl 125, KCl 2.5, MgCl2 2, NaH2PO4 1.2, NaHCO3 21, CaCl2 2, and glucose 1. Extracellular solutions were bubbled with 95% O2–5% CO2 during the experiments. To calculate membrane conductance, whole-cell current was recorded during voltage steps (Fig. 3A), and conductance was determined as the slope of the line of best fit to the current–voltage relationship between −120 and −70 mV (i.e. where the relationship was the most linear). The dose–response curve in Fig. 1G was obtained by fitting a modified Hill equation to the data:
where Vmax is the maximal change in membrane potential, EC50 is the concentration that gives half-maximal response, and h is the Hill coefficient. The fit shown in Fig. 1G was obtained using EC50= 6.2 nm, Vmax= 10.9 mV, and h= 1. Data were analysed using SciPy (http://www.scipy.org/) and plotted with Matplotlib (http://matplotlib.sourceforge.net/). Student's t test was used to determine statistical significance. Values are shown as mean ±s.e.m. Tetrodotoxin, KB-R7943 and ZD7288 were from Tocris, TRH and TRH free acid were from Phoenix Pharmaceuticals, and all other chemicals were from Sigma.
Mice were anaesthetized with sodium pentobarbital and perfused via the ascending aorta with 10 ml of Ca2+-free Tyrode solution (37°C) followed by 10 ml of fixative containing 4% paraformaldehyde, 0.5% glutaraldehyde and 0.2% picric acid in 0.16 m phosphate buffer, pH 6.9, 37°C, followed by 50 ml of the same, but ice-cold, fixative. Brains were dissected, immersed in fixative for 90 min, and rinsed for 24 h in 0.1 m phosphate buffer (pH 7.4) containing 10% sucrose. Brains were then cut in 1 mm slabs and rinsed in 0.1% sodium borohydride for 30 min prior to freezing. Coronal sections were cut on a cryostat (Microm, Heidelberg, Germany) at 14 μm thickness and thaw-mounted onto gelatine-coated glass slides. Conventional immunofluorescence was employed for orexin using monoclonal anti-orexin antibodies (1 : 400) raised in mouse; these antibodies were a gift from Drs K. Eriksson and E. Mignot and their specificity was confirmed by cell body staining restricted to the LHA, and by parallel immunofluorescence performed with three other different monocolonal antibodies raised against orexin, which produced identical staining patterns of LHA cell bodies and terminals throughout the brain. The Tyramide Signal Amplification (TSA) protocol (Perkin Elmer, Waltham, MA, USA) was used to visualize TRH by polyclonal anti-TRH antiserum (1 : 2000; raised in rabbit; gift of Dr T. Visser, see Klootwijk et al. 1995), as previously described (Broberger et al. 1999). For quantification, five sections at regularly spaced intervals were sampled from the LHA of four brains, and the total number of orexin-immunopositive cell bodies on each side were counted, as well as those in close apposition with TRH-immunopositive terminals. By ‘close apposition’ we mean that (a) there is no observable gap between terminal and cell body/dendrite, and (b) the density of terminals on cell body/dendrite is not lower than that in the surrounding neuropil. Images were captured by a Zeiss LSM 510 META confocal microscope.
TRH induces depolarization and increases spontaneous firing of orexin cells
To examine the effects of TRH on the membrane potential of orexin neurones, we performed whole-cell current-clamp recordings from orexin cells identified by specific expression of eGFP in mouse brain slices. Bath application of TRH (10–500 nm) induced reversible membrane depolarization in 27 of 28 cells tested (control −54.3 ± 1.1 mV, TRH −43.3 ± 1.3 mV, P < 0.001, n= 28), and increased action potential frequency (Fig. 1). We recently described two distinct types of orexin neurons in the LHA (Williams et al. 2008); here we observed that TRH had excitatory effects on both of these cell types. TRH-induced depolarization was dose dependent (Fig. 1G), and persisted in the presence of tetrodotoxin (Fig. 1D; mean depolarization caused by 250 nm TRH in TTX was 12.8 ± 2.0 mV, n= 6 cells), indicating that it does not require synaptic transmission mediated by action potentials. Two types of control experiments indicated that this interaction between TRH and LHA orexin cells is not an artefact. First, a biologically inactive TRH analogue, ‘TRH free acid’, failed to alter the membrane potential of orexin cells (n= 4 cells, Fig. 1E). Second, TRH did not change the membrane potential or firing of LHA cells that did not express eGFP and did not possess the electrophysiological properties of orexin cells (n= 4, data not shown).
TRH-mediated depolarization is direct and requires extracellular Na+
TRH-induced depolarization and stimulation of firing persisted under conditions of synaptic isolation (using a low Ca2+/high Mg2+ extracellular solution, see Methods), suggesting that a direct (postsynaptic) mechanism is involved (Fig. 2A; mean depolarization in low Ca2+ was 32.4 ± 2.3 mV, n= 4 cells). In other types of neurones, TRH-induced membrane depolarization can result from a reduction in membrane K+ conductance and/or an increase in excitatory Na+-containing currents (e.g. Kolaj et al. 1997). Under our experimental conditions, TRH-induced depolarization could, in theory, also be caused by a Cl− current, since we used a ‘high-chloride’ pipette solution. However, we found that TRH-induced depolarization was not affected by changing the reversal potential for Cl− from 0 mV to −60 mV (Fig. 1C) when we switched to a ‘low-Cl−’ pipette solution (see Methods), arguing against any major involvement of Cl− channels (Fig. 1C, n= 4). In contrast, reducing extracellular Na+ concentration (‘low Na+’ solution, see Methods), which would diminish the depolarization caused by an increase in Na+ currents but not that caused by a decrease in K+ currents, reduced TRH-evoked depolarization by about 80% (Fig. 2B and D, n= 4 cells), without major effects on pre-stimulation membrane potentials (which were: control, −54.3 ± 1.1 mV; low Na+, −48.3 ± 4 mV, P > 0.05). Because this suggests that the TRH effect requires Na+ influx, we tested whether it could be mediated by the electrogenic Na+/Ca2+ exchanger (Burdakov et al. 2003) or by the H-current (a hyperpolarization-activated mixed Na+/K+ current, Hille, 2001). However, the Na+/Ca2+ exchange blocker KB−R7943 did not suppress TRH-induced depolarization (Fig. 2C, n= 4 cells) when used at a concentration (70 μm) expected to block the Na+/Ca2+ exchanger and previously shown to reverse depolarization induced by other neuropeptides (Burdakov et al. 2003). Similarly, the depolarizing responses to TRH persisted when the H-current was blocked with ZD7288 (100 μm) (Fig. 2D, n= 3 cells). This suggest that TRH-induced depolarization involves an influx of Na+ via pathways other than H-currents or Na+/Ca2+ exchangers.
Effect of TRH on the membrane current–voltage relationship of orexin cells
To study postsynaptic currents triggered by TRH, we looked at membrane current–voltage (I–V) relationships using a low Ca2+/high Mg2+ extracellular solution to isolate postsynaptic effects (Fig. 3). To obtain the net TRH-modulated current, we subtracted the steady-state whole-cell I–V relationships obtained without TRH from those in the presence of TRH (Fig. 3B). This revealed that TRH activated a current with a reversal potential of about −25 mV (Fig. 3C, n= 4 cells). Because this reversal potential is positive to the resting membrane potential of orexin cells (∼−50 to −55 mV), the TRH current is expected to depolarize the cell and increase firing probability, as observed in Fig. 1. In hypothalamic neurones, the current–voltage relationship that we observed for the TRH-activated current (Fig. 3C) is considered suggestive of a non-selective cation current (e.g. Cowley et al. 2001; Fioramonti et al. 2004). Our current-clamp data imply that this current contains a substantial Na+ component (Fig. 2D). An equivalent analysis of K+ contribution cannot be performed using current clamp, because K+ substitution would disrupt the ability of neuronal membrane to maintain physiological membrane potentials, and so instead we analysed the effect of K+ substitution using voltage clamp. Using a ‘low K+’ pipette solution (see Methods) reduced the net TRH-induced membrane conductance (calculated from the slope of the I–V relationship, see Methods) by ∼50% (from 0.9 ± 0.13 nS to 0.43 ± 0.12 nS, n= 4, P < 0.05), supporting the idea that the non-selective current activated by TRH contains both Na+ and K+ components.
Anatomical interrelationship of TRH terminals and orexin neurones
The histochemical distribution of peptides was studied in glutaraldehyde-perfused mouse brains. In comparison with non-glutaraldehyde-containing fixative (C. Broberger & E. Horjales-Araujo, unpublished observations), this method of fixation caused some attenuation of the orexin-immunoreactive signal and less intensely stained cell bodies, but a marked increase in the detection of TRH-immunoreactive terminals (Hökfelt et al. 1989). Orexin-like immunoreactivity was observed in cell bodies in the LHA largely co-extensive with the rostrocaudal span of the arcuate nucleus, through a field delimited dorsoventrally by the mammillothalamic tract and the fornix (Fig. 4A and C), as previously described in the mouse (Wagner et al. 2000). TRH-immunoreactive terminals were observed in several hypothalamic nuclei. A prominent terminal plexus could be seen concentrated over the DMH, extending laterally into the LHA (Fig. 4B). The TRH-immunoreactive terminal field was located largely medial to the cluster of orexin-immunoreactive cell bodies. However, the two regions were not mutually exclusive but overlapped (Fig. 4C), and TRH-like immunoreactive terminals could be seen in the LHA. At high magnification, TRH-immunoreactive terminals could be observed forming close appositions on orexin-immunoreactive cell somata and proximal dendrites (Fig. 4D–F). In a semiquantitative analysis of four brains, 28.5 ± 0.9% of all counted orexin-immunoreactive neurones (n= 1386) were in apposition with one or more TRH-immunoreactive terminals.
Our study identifies a new excitatory stimulus for central orexin neurones, and suggests a previously unanticipated interaction between the orexin/hypocretin and TRH systems in the regulation of cognitive arousal and adaptive behaviour.
Electrophysiological actions of TRH
After the discovery of TRH as the hormone that controls the hypothalamus–pituitary–thyroid axis, non-neuroendocrine roles for the peptide soon began to emerge, consistent with its wide distribution in the brain. TRH is now well known as a neurotransmitter with excitatory effects on diverse types of neurones (Lechan & Fekete, 2006). Our results indicate that LHA orexin neurones are part of the network of TRH-activated cells. In our experiments, TRH action involved Na+-dependent membrane depolarization and an activation of a mixed cationic current. A similar mechanism was suggested to explain the action of TRH on spinal motoneurons of the frog (Nicoll, 1977). It would be desirable to establish the molecular identity of TRH-activated current(s); however, as in other recent studies of hypothalamic neurones (e.g. Cowley et al. 2001; Fioramonti et al. 2004), we were unfortunately not able to do this here due to the lack of specific drugs for manipulating candidate channels. Our pharmacological experiments performed so far argue against any major involvement of the H-current or the Na+/Ca2+ exchanger (Fig. 2D). However, we would like to stress that, while the results shown in Fig. 3 are consistent with a mixed cation channel, they do not rule out that TRH may in parallel modulate other channel(s), for example voltage-gated K+ and Ca2+ currents; these additional aspects of TRH action remain to be investigated.
Kolaj et al. (1997) reported that, in mammalian spinal neurones, TRH activates a non-selective cation current, while simultaneously blocking a resting K+ conductance. In contrast, TRH-induced inhibition of a leak-like K+ conductance, most likely composed of acid-sensitive TASK channels, appears to be the predominant mechanism underlying TRH-induced excitation in other neurones (e.g. Talley et al. 2000; Broberger & McCormick, 2005). Although leak-like K+ currents are present in orexin cells and can be modulated by acid (Williams et al. 2007), in this study TRH did not significantly modulate these currents when investigated using voltage ramps (n > 10, not shown). These results support the view that the electrophysiological effects of TRH are cell-type dependent. Two G-protein-coupled TRH receptors are known, but their similar affinity for TRH and the notorious absence of reliable antagonists and immunostaining reagents precluded us from identifying the receptor subtype in this study. A challenge for the future is to correlate the electrical effects of TRH with a specific type of receptor in situ.
In addition to the electrophysiological effects, we found TRH-containing terminals in close apposition to orexin cells in the LHA. While most orexin neurones responded electrically to TRH irrespective of their anatomical location, we identified TRH-immunoreactivity in close proximity to only about 30% of orexin-immunoreactive neurones. However, given the documented difficulties in visualizing the full extent of this peptide in the brain (Hökfelt et al. 1989), this may well be an underestimate of the actual percentage of orexin neurones directly contacted by TRHergic terminals. Furthermore, in our staining conditions, only proximal dendrites of orexin-immunoreactive cell bodies could be observed, leaving open the possibility of further TRH contacts on distal dendrites. Finally, because TRH is a neuropeptide, it is likely that direct synaptic contact is not required for orexin neurones to be excited, since neuropeptides are considered able to diffuse some distance from their site of release (‘volume transmission’, reviewed in Agnati et al. 1995).
Orexin neurones play a major role in the control of arousal and feeding behaviour, and their loss causes narcolepsy, metabolic abnormalities, and inability to increase foraging during hunger (Lin et al. 1999; de Lecea et al. 2006). Their excitation by TRH is thus likely to contribute to the potent arousal-promoting actions of this neuropeptide, in concert with its recently described actions in the thalamus (Broberger & McCormick, 2005). Of the brain regions that express TRH mRNA (Ebling et al. 2006), four project heavily to orexin cells: the anterior hypothalamus, the medial preoptic area, the bed nucleus of the stria terminalis, and the DMH (Sakurai et al. 2005). The first two regions are important for thermoregulation. Central or peripheral administration of TRH increases body temperature (Schuhler et al. 2007), but this effect is probably not directly linked to orexin cells because TRH microinjections into the LHA do not affect body temperature (Shintani et al. 2005). However, TRH has been linked to arousal from hibernation in hamsters (Tamura et al. 2005), and thermosensitive cells in the anterior/preoptic hypothalamus are capable of modulating the activity of arousal-related cells in the LHA (Krilowicz et al. 1994). Thus, it is possible that TRH may be involved in co-ordination of thermoregulatory and arousal systems through the projection to orexin cells from the anterior hypothalamus/preoptic area. The third area containing TRH mRNA, the bed nucleus of the stria terminalis, participates in the regulation of emotional states. Its links to the orexin system have not yet been studied in full, but emotions are capable of triggering cataplexy in narcoleptic humans (Dauvilliers et al. 2007), suggesting a role for orexin cells in the physiological responses to emotions. It is possible that TRHergic cells in the bed nucleus of the stria terminalis may be part of the circuitry modulating such responses. Finally, the DMH is fundamental for coordinating behavioural timing, in particular food entrainment of circadian rhythms (Fuller et al. 2008). Of the DMH neurones that project to the LHA, about a third express TRH mRNA (Chou et al. 2003), suggesting an important role for TRH in the circadian control of sleep, wakefulness and feeding. The stimulating properties of TRH on orexin cells may contribute to this circadian control of alertness.
In conclusion, we propose that the effects of TRH on orexin neurones are likely to be involved in a variety of vital behaviours orchestrated by the orexin system, and may be relevant for the design of drug therapies involving TRH analogues.
This study was primarily supported by the European Research Council (grant to D.B.), and also by the Swedish Research Council, Wenner-Gren Foundations, Rut and Arvid Wolff's Foundation, and Petrus and Augusta Hedlund's Foundation (grants to C.B.).