Ghrelin stimulation of growth hormone-releasing hormone neurons is direct in the arcuate nucleus.

Abstract

Ghrelin targets the arcuate nucleus, from where growth hormone releasing hormone (GHRH) neurones trigger GH secretion. This hypothalamic nucleus also contains neuropeptide Y (NPY) neurons which play a master role in the effect of ghrelin on feeding. Interestingly, connections between NPY and GHRH neurons have been reported, leading to the hypothesis that the GH axis and the feeding circuits might be co-regulated by ghrelin.
Here, we show that ghrelin stimulates the firing rate of identified GHRH neurons, in transgenic GHRH-GFP mice. This stimulation is prevented by growth hormone secretagogue receptor-1 antagonism as well as by U-73122, a phospholipase C inhibitor and by calcium channels blockers. The effect of ghrelin does not require synaptic transmission, as it is not antagonized by gamma-aminobutyric acid, glutamate and NPY receptor antagonists. In addition, this hypothalamic effect of ghrelin is independent of somatostatin, the inhibitor of the GH axis, since it is also found in somatostatin knockout mice. Indeed, ghrelin does not modify synaptic currents of GHRH neurons. However, ghrelin exerts a strong and direct depolarizing effect on GHRH neurons, which supports their increased firing rate.
Thus, GHRH neurons are a specific target for ghrelin within the brain, and not activated secondary to altered activity in feeding circuits. These results support the view that ghrelin related therapeutic approaches could be directed separately towards GH deficiency or feeding disorders.

Full text

Page 1

Ghrelin Stimulation of Growth Hormone-Releasing
Hormone Neurons Is Direct in the Arcuate Nucleus
Guillaume Osterstock1,2,3, Pauline Escobar1,2,3, Violeta Mitutsova1,2,3, Laurie-Anne Gouty-Colomer1,2,3,
Pierre Fontanaud1,2,3, Franc ¸ois Molino1,2,3, Jean-Alain Fehrentz3,5, Danielle Carmignac4, Jean
Martinez3,5, Nathalie C. Guerineau1,2,3, Iain C. A. F. Robinson4, Patrice Mollard1,2,3, Pierre-Franc ¸ois
Me ´ry1,2,3*
1Inserm U-661, Montpellier, France, 2CNRS UMR 5203, Institut de Ge ´nomique Fonctionnelle, Montpellier, France, 3Universite ´ Montpellier 1, 2, Montpellier, France,
4Division of Molecular Neuroendocrinology, MRC National Institute for Medical Research, The Ridgeway, Mill Hill, London, United Kingdom, 5CNRS UMR 5247, Institut des
Biomole ´cules Max Mousseron, Montpellier, France
Abstract
Background: Ghrelin targets the arcuate nucleus, from where growth hormone releasing hormone (GHRH) neurones trigger
GH secretion. This hypothalamic nucleus also contains neuropeptide Y (NPY) neurons which play a master role in the effect
of ghrelin on feeding. Interestingly, connections between NPY and GHRH neurons have been reported, leading to the
hypothesis that the GH axis and the feeding circuits might be co-regulated by ghrelin.
Principal Findings: Here, we show that ghrelin stimulates the firing rate of identified GHRH neurons, in transgenic GHRH-
GFP mice. This stimulation is prevented by growth hormone secretagogue receptor-1 antagonism as well as by U-73122, a
phospholipase C inhibitor and by calcium channels blockers. The effect of ghrelin does not require synaptic transmission, as
it is not antagonized by c-aminobutyric acid, glutamate and NPY receptor antagonists. In addition, this hypothalamic effect
of ghrelin is independent of somatostatin, the inhibitor of the GH axis, since it is also found in somatostatin knockout mice.
Indeed, ghrelin does not modify synaptic currents of GHRH neurons. However, ghrelin exerts a strong and direct
depolarizing effect on GHRH neurons, which supports their increased firing rate.
Conclusion: Thus, GHRH neurons are a specific target for ghrelin within the brain, and not activated secondary to altered
activity in feeding circuits. These results support the view that ghrelin related therapeutic approaches could be directed
separately towards GH deficiency or feeding disorders.
Citation: Osterstock G, Escobar P, Mitutsova V, Gouty-Colomer L-A, Fontanaud P, et al. (2010) Ghrelin Stimulation of Growth Hormone-Releasing Hormone
Neurons Is Direct in the Arcuate Nucleus. PLoS ONE 5(2): e9159. doi:10.1371/journal.pone.0009159
Editor: Xin-Yun Lu, University of Texas Health Science Center, United States of America
Received May 12, 2009; Accepted January 8, 2010; Published February 11, 2010
Copyright: ? 2010 Osterstock et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by grants from Institut National de la Sante et de la Recherche Medicale, Centre National de la Recherche Scientifique, the
Universities of Montpellier 1 & 2, National Biophotonics and Imaging Platform (Ireland), Reseau National des Genopoles, Institut Federatif de Recherches 3, Region
Languedoc Roussillon and by core funding from the Medical Research Council (United Kingdom). The funders had no role in study design, data collection and
analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: pierre-francois.mery@igf.cnrs.fr
Introduction
The hypothalamic arcuate nucleus is a heterogeneous structure
involved in the regulation of homeostasis. Its functions rely on the
specific actions of its outputs; for example, growth hormone
releasing hormone (GHRH) and somatostatin are involved in
body growth [1], and neuropeptide Y (NPY) and agouti related
peptide (AgRP) are involved in feeding [2]. The distribution of
receptors and afferent nerve terminals within the arcuate nucleus
are generally diffuse, supporting the view that afferent inputs
coordinate combinations of outputs from this structure. Ghrelin,
the endogenous growth hormone secretagogue [3,4], is one such
hypothalamic input. Indeed, ghrelin not only stimulates the
growth hormone (GH) axis [1], but also induces feeding and
modifies body energy consumption [5,6], as well as modulating the
gonadotropic axis [7]. The ghrelin receptor (GHSR, growth
hormone secretagogue receptor-1) is found in several neuronal
subtypes in the arcuate nucleus [8–11], where a diffuse pattern of
ghrelin-containing terminals has been demonstrated [12].
Recent studies have addressed the organisation of this circuitry.
In addition to its direct effects on the pituitary, ghrelin clearly
targets GH release indirectly at the level of the arcuate nucleus
since: 1) anatomical disconnections between the hypothalamus
and the pituitary gland blunt GH secretion induced by GHS in vivo
[13,14]; 2) the GHSR is expressed in GHRH neurons, which
trigger GH release by the pituitary gland [8–11]; 3) in vivo GHS
treatments enhance GHRH secretion in sheep [1,15] and induce
c-fos expression in GHRH neurons in rodents [16]. Furthermore,
ghrelin and GHS enhance the electrical activity of non-identified
neurons in the arcuate nucleus [17–19], and ghrelin enhances
calcium dynamics in isolated hypothalamic neurons, in vitro
[20,21]. While these results do not provide a specific mechanism
of action, collectively they suggest that ghrelin exerts a direct effect
at the level of GHRH neurons.
PLoS ONE | www.plosone.org1February 2010 | Volume 5 | Issue 2 | e9159

Page 2

In contrast to this, other data suggest an indirect modulation
of GHRH neurons by ghrelin. Indeed, the arcuate nucleus is
intimately involved in the effects of ghrelin on the feeding
circuits [5,6], with NPY neurons appearing as central ghrelin
sensors in this role [2,22]. NPY neurons are the main ghrelin
receptor (GHSR)-expressing cells of the arcuate nucleus [9,10],
and they upregulate c-fos expression in response to ghrelin
perfusion [5]. NPY neurons signal through a complex release of
NPY, AgRP, and c-aminobutyric acid (GABA) [2,22]. Accord-
ingly, the orexigenic effect of ghrelin is absent in NPY/AgRP
double knockout mice, despite unaltered growth and feeding
[22,23]. It is also attenuated in mice whose vesicular GABA
transporter is specifically ablated in AgRP-expressing neurons
[24]. In vitro, the stimulatory effect of ghrelin on NPY neurons
orchestrates electrophysiological changes within the feeding
circuits, including a GABAergic modulation of pro-opiomela-
nocortin (POMC) neurons and a dual GABA/NPYergic
modulation of corticotrophin-releasing hormone (CRH) neu-
rons [2,12]. The role of NPY neurons may not be limited to the
feeding circuits, per se, since GHRH neurons express NPY Y2
receptors which mediate the downregulation of GHRH mRNA
induced by long term fasting in rodents [25,26]. In addition, as
NPY neurons often coexpress GABA [2,12,24], part of the
GABAergic inputs to GHRH neurons [27] might originate from
the NPY neurons themselves. Altogether, these findings suggest
that NPY neurons might be the primary ghrelin sensors of the
arcuate nucleus, funnelling information from within the feeding
circuits to the GH axis.
Here, we took advantage of GHRH-GFP transgenic mice [28]
to investigate whether ghrelin modulates GHRH neurons. We
found that ghrelin stimulated the electrical activity of GHRH
neurons in a direct manner, suggesting that parallel and
apparently independent signalling at GHRH neurons and at
NPY neurons can occur within the very restricted area of the
arcuate nucleus. Our data support the view that ghrelin has
multiple entries within the central nervous system. Thus, encoding
of afferent information by the arcuate nucleus is not only
supported by the identity of its outputs, the efferent neuropeptides,
but also by the mechanism of action of its inputs, such as ghrelin,
which can modulate the endocrine axis independently or in
combination.
Results
Ghrelin Modulated the Firing Rate but Not the Firing
Pattern of GHRH Neurons
We examined the effects of ghrelin on the electrical activity of
identified GHRH neurons in brain slices from GHRH-GFP
mice. In the experiment of Fig. 1A, spontaneous action potentials
were first recorded under control conditions. Addition of 10 nM
ghrelin to the external solution increased the firing rate from
,0.2 to 0.9 Hz, and this stimulation disappeared during the
washout of the peptide. The cumulative histograms of Fig. 1B
summarize the results from similar experiments where the
instantaneous frequencies of the spontaneous action potentials
of GHRH neurons were compared under steady-state conditions
in the absence and presence of 10 nM ghrelin (see Methods for
additional information). The mean distribution under control
conditions was shifted to the right (into the 0–18.5 Hz range) in
the presence of ghrelin (grey area, n=28, paired student’s t-test,
p,0.05). This increase in firing rate was also well described as an
increase in the mean frequency at the half maximal values of the
cumulated histograms (Fig. 1C). Lower concentrations of ghrelin
(0.3–3 nM, n=5 to 10) did not significantly change this
parameter (Fig. 1C), and did not significantly shift the cumulative
distribution of GHRH neuron action potentials (data not shown).
However, 0.3–3 nM ghrelin occasionally enhanced the firing rate
of GHRH neurons, and the proportion of responses increased in
a concentration-dependent manner (Fig. 1D). Since 10 nM
ghrelin always enhanced the electrical activity of GHRH
neurons, the other effects of ghrelin were studied at this
concentration.
The traces of Fig. 1A suggested that ghrelin did not change
the firing pattern. Indeed, the mean skewness of the discharge
density histograms was not changed by ghrelin (supporting
Figure S1A). In accordance with the conclusion that ghrelin
increases firing rates without changing the firing patterns of
GHRH neurons, autocorrelogram analysis only showed differ-
ences in a very narrow range of action potential intervals (20.3
to +0.3 s), (supporting Figure S1B–C). Because the GH axis
exhibits several gender differences [1], the hypothalamic effect
of ghrelin was then investigated in female mice. As summarized
in Fig. 1E, ghrelin (10 nM) increased the electrical activity of all
GHRH neurons tested from female GHRH-GFP mice (p,0.05
in the 0.75–6.25 Hz range, paired student’s t-test), and did not
change their firing pattern (data not shown). Thus, the
stimulatory effect of ghrelin on GHRH neurons occurs in both
sexes.
Because GHRH neurons are such a small population [2,29], a
GHRH releasing agent such as ghrelin (or ghrelin mimetics) might
trigger synchronisation between GHRH neurons [15]. This
synchronicity was then studied using the dual patch clamp
technique. In the example of Fig. 2A, 10 nM ghrelin simulta-
neously enhanced the firing rates of two GHRH neurons. The
cumulative distribution of the action potential frequencies of both
neurons were shifted to the right by the peptide, though to
different extents (Fig. 2B). This quantitative analysis was
complemented with a qualitative analysis, where crosscorrelo-
grams were computed (Fig. 2D), as described in the Methods
section, using the stretches of spike trains recorded under steady-
state conditions (Fig. 2C). In brief, the correlation between these
spike trains consisted in counting the spikes of the neuron ‘‘2’’ at
the specific time delay of 100 ms with respect to the spikes of the
neuron ‘‘1’’. The flat shape of the crosscorrelogram obtained
under control conditions indicated that neuron ‘‘2’’ did not fire at
a preferential time before/after neuron ‘‘1’’. Thus, there was no
correlation between the activities of the neurons. Ghrelin induced
an upward shift in the distribution as expected for a stimulatory
agent, but did not induce a distinctive peak in the cross-
correlogram, suggesting independence between the activities of
the two neurons. Both distributions were contained within the
95%-confidence boundaries of random distributions (dotted lines,
computed as stated in Methods). Furthermore, random inter-event
interval distributions (Fig. 2E) were generated using the distribu-
tions of the experimental sets of data (Fig. 2D), as described in
Methods. They were used to model cross-correlograms between
independent series of data (Fig. 2F), which were almost
undistinguishable from the experimental results (Fig. 2D). These
results were typical of six similar experiments, suggesting that
ghrelin induced neither a hierarchy, nor a correlation of activity,
amongst GHRH neurons.
Pharmacological Profile of the Ghrelin Receptor
Prior to the discovery of ghrelin, it was established that GHS,
such as GHRP-6, enhance the electrical activity of unidentified
neurons in the arcuate nucleus [17,18]. Like ghrelin, they exhibit a
nanomolar affinity for GHSR, the canonical ghrelin receptor of
the GH axis found in the arcuate nucleus [3,4]. We therefore
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org2 February 2010 | Volume 5 | Issue 2 | e9159

Page 3

tested the effects of several GHS of differing structures. The
electrical activity of a GHRH neuron (Fig. 3A, adult male) was
enhanced by GHRP-6, slightly at 10 nM (from ,0.9 to 1.4 Hz)
and more strongly at 100 nM (to ,3.7 Hz). This stimulatory effect
on GHRH neurons was also observed when JMV1843 (10 nM), a
potent in vivo GHSR agonist [30,31], was superfused onto GHRH-
GFP brain slices (Fig. 3B). Furthermore, while the GHSR
antagonist, JMV3002 (1 mM) [33], did not change the activity of
a GHRH neuron when applied alone (Fig. 3C), it blunted the
effect of an addition of 10 nM ghrelin. The stimulatory effect of
ghrelin developed upon washout of JMV 3002. The mean effects
of the GHSs on the distribution of the frequencies of spontaneous
action potentials of GHRH neurons were summarized in Fig. 3D
and 3E. All the GHSR agonists, GHRP-6, JMV1843, and
JMV2952 [32] increased the firing rate of GHRH neurons in a
1–100 nM range compatible with their affinities for GHSR (see
mean frequencies at half maximal values of the cumulated
histograms, Fig. 3D). JMV3002, the GHSR antagonist, was
inactive on its own in the 10 nM to 1 mM range but significantly
antagonized the stimulatory effect of 10 nM ghrelin (Fig. 3E).
Hence, it is likely that GHSR activation mediates the enhance-
ment of the electrical activity of GHRH neurons induced by
ghrelin and the GHS tested in this study.
GHSR expression is seen early, at embryonic day 19 in the rat
pituitary gland as well as in the brain [34,35]. Accordingly, we
found that ghrelin (10 nM) enhanced the firing rate of a GHRH
neuron from immature, 6 day-old, male GHRH-GFP mice
(supporting Figure S2A–B). Much later in life, aged individuals
retain ghrelin-induced GH secretion as well as GHSR expression
in the brain [4,36]. The effect of ghrelin on GHRH neurons in
aged (.22 months-old) male GHRH-GFP mice was indeed
present but heterogeneous, being stimulatory in only 8 out of 13
experiments (supporting Figure S2C–D). Thus, the ghrelin
responsiveness observed at different developmental stages in
GHRH neurons, was compatible with the profile of GHSR
expression in the brain [34–36].
Figure 1. Ghrelin enhanced the activity of GHRH neurons. A, time course of an experiment where the superfusion of a sagittal brain slice with
10 nM ghrelin increased, in a reversible manner, the rate of spontaneous action potentials of a GHRH neuron (individual traces shown on the top). B,
summary of the effects of ghrelin (10 nM) on the cumulative distributions of action potential frequencies in GHRH neurons from adult males; C, mean
effects of 0.3 to 10 nM ghrelin on the rate of spontaneous action potentials in GHRH neurons: the action potential frequencies observed at the half
maximal values of the cumulated histograms were collected in each experiment in the absence and presence of ghrelin (see Methods for details). D,
the proportion of stimulatory effects induced by ghrelin increased in a dose-dependent manner in GHRH neurons. E, summary of the effects of
ghrelin (10 nM) on the distributions of action potential frequencies in GHRH neurons from adult females. In B & E, the symbols and lines are the
means and sem. Statistical differences (p,0.05, paired student-t test) between curves are framed by the grey areas. In D, the bars and lines are the
means and sem of the numbers of experiments indicated. ***, statistical difference from control values (p,0.001, paired student-t test).
doi:10.1371/journal.pone.0009159.g001
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org3February 2010 | Volume 5 | Issue 2 | e9159

Page 4

The Stimulation of GHRH Neurones by Ghrelin Requires
Phospholipase C and Calcium Channels
The canonical effector of GHSR is phospholipase C dependent
[4,6], but GHSR activation by ghrelin can elicit the activation of
other pathways depending on the tissue context [37]. The
involvement of phospholipase C in GHRH neurons was
examined first. Superfusion of a GHRH-GFP brain slice with
10 mM U-73122, a phospholipase C inhibitor [38–40], enhanced
the firing rate of GHRH neurons, from ,2 to 3.5 Hz, and this
preincubation prevented the stimulatory effect of 10 nM ghrelin
(Fig. 4A). In similar experiments, ,10 minutes-long perfusion
with U-73122 significantly increased the electrical activity of
GHRH neurons and further addition of ghrelin had no
significant effects in the presence of the phospholipase C inhibitor
(Fig. 4E). In contrast, ghrelin enhanced the activity of GHRH
neurons in the presence of U-73343 (10 mM, n=4, data not
shown), a U-73122 analog which does not inhibit phospholipase
C activity [38–40].
Ion channels are the final effectors of ghrelin-stimulated
pathways in various excitable cell types [4,6,20,37,41]. In addition,
ghrelin tunes mitochondrial homeostasis and cellular energy
supply in neurons [42]. Thus, the effect of ghrelin was first
examined in the presence of a broad range inhibitor, namely
flufenamic acid which inhibits several families of ionic channels
and causes mitochondrial uncoupling [40,43,44]. As summarized
in Fig. 4E, flufenamic acid (150 mM) fully antagonized the
stimulatory effect of ghrelin in GHRH neurons. The role of ionic
channels was further delimited. First, ghrelin did not enhance the
electrical activity of GHRH neurons in the presence of Gd3+
(100 mM, Fig. 4E), a non-selective blocker of cationic channels
including background channels or voltage dependent channels
[41,45]. In addition, Ni2+(150 mM, Fig. 4B&E), a blocker of low
voltage activated calcium channels [46], as well as Cd2+(100 mM,
Fig. 4C&E), a blocker of high voltage activated calcium channels
[46], both prevented the stimulatory effects of ghrelin upon
GHRH neurons. In contrast, extracellular Cs+(5 mM, Fig. 4D&E)
an inhibitor of the hyperpolarisation-activated cyclic nucleotides-
gated cation channels (HCN) channels [44], significantly enhanced
the electrical activity of GHRH neurons but did not antagonise the
stimulatory effect of ghrelin. Therefore, ghrelin stimulates GHRH
neurons in a phospholipase C and calcium dependent mechanism.
The Effect of Ghrelin on GHRH Neurons Did Not Involve
Somatostatin Input
Both GHRH neurons and somatostatinergic neurons express
some GHSR [10]. The effect of ghrelin on the GH axis might
require synaptic signalling between these two neuronal populations
[1]. Accordingly, we took the opportunity to examine the effect of
Figure 2. Ghrelin did not synchronize the activity of GHRH neurons in dual patch-clamp epxeriments. A, stimulatory effects of ghrelin
(10 nM) on the firing rate of two GHRH neurons recorded simultaneously. Action potential rates were calculated every 30 s. B, cumulative
distributions of the frequency of the action potentials of the GHRH neurons from panel A, showing the extent of the rightward shifts induced by
ghrelin. C, intervals between action potentials of the GHRH neurons from panel A, under control conditions and in the presence of ghrelin, were then
used in generating the cross-correlograms shown in D. The correlations of activity were calculated within consecutive bins of 100 ms during 60 s (see
Methods for further details). Dotted lines indicate the 95% confidence boundaries within which the distributions behave as random, in the absence
and presence of ghrelin. E&F, same as C&D, except that random distributions of instantaneous frequencies of action potentials were generated using
the properties of the experimental data, in the absence and in the presence of ghrelin. The shapes of these cross-correlograms characterizing de-
correlated series of events were almost undistinguishable from the experimental curves.
doi:10.1371/journal.pone.0009159.g002
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org4 February 2010 | Volume 5 | Issue 2 | e9159

Page 5

ghrelin on GHRH neurons in the absence of somatostatin, by
breeding GHRH-GFP mice onto a somatostatin knockout mouse
background [47] (a description of GHRH neurons of these animals
is the subject of another submission). Fig. 5A shows that an identified
GHRH neuron in an adult male somatostatin knockout mouse
exhibited a spontaneous firing rate of ,0.9 Hz under control
conditions, increasing to ,3.3 Hz upon addition of 10 nM ghrelin
to the external solution. This stimulation was found in each
experiment performed in GHRH neurons from GHRH-GFP X
somatostatin null mice, and their mean spontaneous activity was
significantly enhanced, as summarized in Fig. 5B (p,0.05 in the
0.5–6.5 Hz range, paired student’s t-test where). A lower concen-
tration of ghrelin (1 nM) had no significant effect (n=3, data not
shown). Thus, the activation of hypothalamic GHRH neurons by
ghrelin occurs in the absence of somatostatin.
The Effect of Ghrelin on GHRH Neurons Did Not Require
NPY Neurotransmission
The NPY neurons are the predominant GHSR positive cells in
the arcuate nucleus [9–11], and it is thought that NPY can
modulate the GH axis, although the mechanisms are unclear [26].
We first tested a simple mechanism, whereby the NPY Y2
receptors, expressed by GHRH neurons [25], would mediate the
effects of ghrelin. Interestingly, NPY [13–36] (100 nM), a selective
NPY Y2 receptors agonist [48], increased the discharge rate of a
GHRH neuron from an adult male (from ,2.5 to 3.5 Hz, Fig. 6A).
Like ghrelin, NPY [13–36] (100 nM) shifted the cumulated
distribution of action potentials frequencies of GHRH neurons
(p,0.05, in the 0.5–25 Hz range, Fig. 6B). The effect of a lower
concentration (30 nM) of the Y2 receptor agonist was not
significant.
The stimulatory effect of ghrelin was also examined in the
presence of BIIE 0246, a selective NPY Y2 receptor antagonist
[48]. On average, 1 mM BIIE 0246 did not change the activity of
GHRH neurons in adult male mice (Fig. 6C), although it
significantly blunted the stimulatory effect of 100 nM NPY [13–
36] (n=4, data not shown). Ghrelin induced significant rightward
shifts of the distribution of the action potential frequencies
(p,0.05, 0.25–4.75 Hz range for ghrelin + BIIE 0246, and
0.25–8 Hz for ghrelin, Fig. 6D), in the absence or presence of BIIE
0246 (p.0.05, ghrelin alone vs ghrelin + BIIE 0246). Therefore,
Figure 3. Effects of ghrelin receptor ligands on the activity of GHRH neurons. A to C, typical experiments where superfusions with the
agonists GHRP-6 (10 & 100 nM, A) and JMV1843 (100 nM, B) increased action potentials rates in adult male GHRH neurons; the antagonist JMV3002
(1 mM) prevented from the stimulatory effect of ghrelin (10 nM, C). D–E, summaries of the effects of GHSR agonists (D: ghrelin, GHRP-6, JMV1843,
JMV2952) or antagonist (E: JMV3002) on the cumulative distributions of the spontaneous action potentials of GHRH neurons. The action potential
frequencies observed at the half maximal values of the cumulated histograms were averaged according to the absence (white bars) and presence of
agonists and/or antagonist (coloured bars, see Methods for details). In D&E, bars and lines are the means and the sem of the numbers of experiments
indicated. Statistical differences *, p,0.05; **, p,0.01; ***, p,.005 are shown (paired student-t test).
doi:10.1371/journal.pone.0009159.g003
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org5 February 2010 | Volume 5 | Issue 2 | e9159

Page 6

Y2 receptor activation was not required for the stimulatory effect
of ghrelin.
The Stimulatory Effect of Ghrelin Did Not Require Fast
Synaptic Transmission
GABAergic neurotransmission by NPY neurons is intimately
involved in the effects of ghrelin on CRH and POMC neurons
[12]. GABA also modulates GHRH neurons [27], so its potential
involvement in the effect of ghrelin on GHRH neurons was
studied. Fig. 7A shows that 10 nM ghrelin strongly increased the
firing rate of a GHRH neuron, in the continuing presence of an
antagonist of ionotropic GABAA receptors, GABAzine (4-[6-
imino-3-(4-methoxyphenyl)pyridazin-1-yl]
average, 3 mM GABAzine did not significantly modify the firing
rates of GHRH neurons, because its effects were heterogeneous
(Fig. 7B). Nevertheless, ghrelin shifted the distribution of action
potentials frequencies in the presence of the GABAA receptor
antagonist (p,0.05, in the 2–17.5 Hz range). Thus, GABAergic
butanoic acid). On
neurotransmission was not necessary for the stimulatory effect of
ghrelin on GHRH neurons.
Similarly, the involvement of glutamatergic neurotransmission
in the effect of ghrelin was investigated, since this excitatory
transmitter was strongly involved in the muscarinic modulation of
GHRH neurons [27]. CNQX (6-cyano-7-nitroquinoxaline-2,3-
dione), an antagonist at AMPA (a-amino-3-hydroxy-5-me ´thyli-
soazol-4-propionic acid) and kainate receptors was used in
combination with GABAzine. In the experiment shown in
Fig. 7C, the inhibitors slightly diminished the firing rate of the
GHRH neuron (,4.5 and 3.9 Hz in the absence and presence of
CNQX + GABAzine, respectively), but the addition of ghrelin
10 nM, in the continuing presence of inhibitors, induced a robust
increase in the firing rate of the neuron to ,8.3 Hz. On average
(Fig. 7D), the combination of GABAzine + CNQX tended to
weaken the activity of GHRH neurons, though in a non significant
manner, and ghrelin shifted the distribution of action potentials
frequencies in the presence of the inhibitors (p,0.05, 1–12.5 Hz
Figure 4. The effect of ghrelin on GHRH neurons requires phospholipase C and calcium channels. A–D, typical recordings from GHRH
neurons in the absence and presence of the phospholipase C inhibitor U-73122 (10 mM, A); the high voltage-activated calcium channel blocker Cd2+
(100 mM, B); the low voltage-activated calcium channel blocker Ni2+(150 mM, C); the HCN channel blocker Cs+(5 mM, D); and 10 nM ghrelin (A–D), as
indicated by the lines. E, summary of the effects of cellular signalling inhibitors on the cumulated distributions of spontaneous action potentials in
GHRH neurons. The action potential frequencies observed at the half maximal values of the cumulated histograms were averaged according to the
absence (white bars) and presence of inhibitor (blue bars), and in the presence of inhibitor plus 10 nM ghrelin (orange bars, see Methods for details).
Bars and lines are the means and the sem of the numbers of experiments indicated. Statistical differences (vs control values *: p,0.05; **: p,0.01;
***, p,0.005; and vs inhibitor level $, p,0.05, paired student-t test) are shown.
doi:10.1371/journal.pone.0009159.g004
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org6 February 2010 | Volume 5 | Issue 2 | e9159

Page 7

range). Thus the stimulatory effect of ghrelin did not require
AMPA/kainate neurotransmissions.
Ghrelin Did Not Modify Synaptic Currents in GHRH-GFP
Neurons
A modulation of GHRH neuron synaptic currents might play a
subtle role in the effect of ghrelin. The spontaneous glutamatergic
currents and GABAergic currents of GHRH neurons [27] were
recorded as shown in Fig. 8A and 8C. Glutamatergic (recorded at
270 mV, Fig. 8A) and GABAergic (recorded at 230 mV, Fig. 8C)
currents seemed unchanged by the superfusion with ghrelin
(10 nM). It was found that ghrelin did not shift the cumulative
distribution of the amplitudes and of the inter-event intervals of
either the glutamatergic currents (n=11, Fig. 8B), or the
GABAergic currents (n=6, Fig. 8D) in GHRH neurons. In these
experiments, ghrelin did not modify the kinetics of the synaptic
currents (data not shown). A synthetic GHS, JMV1843 100 nM,
did not modify the spontaneous GABAergic and glutamatergic
currents of GHRH-GFP neurons (n=6, data not shown). Thus,
Figure 5. Ghrelin enhances the firing rates of GHRH neurons in
the absence of somatostatin. A, typical experiment where 10 nM
ghrelin increased the firing rate of a GHRH neuron from an adult male
somatostatin 2/2, GHRH-GFP mouse (raw traces are shown on the
top). B, summary of the effects of ghrelin (10 nM) on the distributions of
action potential frequencies in GHRH neurons from adult male
somatostatin 2/2, GHRH-GFP mice. Symbols and lines are the means
and the sem of the numbers of experiments indicated. Statistical
significances (p,0.05, paired student-t test, see methods) between
curves are framed by the grey area.
doi:10.1371/journal.pone.0009159.g005
Figure 6. The effect of ghrelin on the firing rates of GHRH
neurons did not require Y-2 receptors. A, typical experiment
where the Y-2 receptors agonist, NPY [13–36] 100 nM, increased, in a
reversible manner, the spontaneous firing rate in a male GHRH neuron.
Raw traces are shown on top of the panel. B, summaries of the effects
of NPY [13–36] (30 & 100 nM) on the distributions of action potential
frequencies in GHRH neurons from adult male GHRH-GFP mice. C–D,
summaries of the effects of the Y-2 antagonist BIIE0246 alone (C) and of
ghrelin in the absence or presence of BIIE0246 (D) on the distributions
of action potential frequencies in GHRH neurons from adult male
GHRH-GFP mice. Symbols and lines are the means and the sem of the
numbers of experiments indicated. Statistical significance (p,0.05,
paired student-t test) between curves (effect of ghrelin in the absence
or presence of BIIE0246, D) are framed by the grey area.
doi:10.1371/journal.pone.0009159.g006
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org7 February 2010 | Volume 5 | Issue 2 | e9159

Page 8

fast synaptic transmission at GHRH neurons is unaffected by
ghrelin.
Ghrelin Had a Direct Depolarizing Effect on GHRH
Neurons
The signature of the neuromodulatory effect of ghrelin on
GHRH neurons was further investigated with the perforated
patch-clamp technique [49], where amplitudes and kinetics of
action potentials can be quantified (as shown by the individual
traces of Fig. 9A). In the recording of Fig. 9A–B, the spontaneous
action potentials of a GHRH neuron were collected under control
conditions in the current-clamp mode (0 pA). Superfusion of the
slice with ghrelin 10 nM increased the firing rate of the neuron
(Fig. 9B, top panel) and this stimulation was mirrored by a
decrease in the resting membrane potential (Fig. 9B, bottom
panel). In similar experiments, ghrelin consistently decreased the
mean action potentials intervals (from 4.3162.0 s to 1.4060.77 s,
n=8, p,0.05, paired student t-test: Fig. 9C), without changing the
skewness of the interval distribution (data not shown), consistent
with the results from extracellular recordings. Ghrelin consistently
depolarizedGHRHneurons
255.3162.15 mV, n=8, p,0.005, paired student t-test: Fig. 9D)
and did not alter the parameters of the action potentials (Table 1).
Similar results were found when ghrelin was applied in the
presence of the AMPA/kainate antagonist DNQX (6,7-dinitro-
quinoxaline-2,3-dione, 15 mM) plus the GABAA antagonist
GABAzine 3 mM, which eliminated spontaneous synaptic depo-
larisations and hyperpolarizations (data not shown). These
experiments showed that ghrelin modified an intrinsic ionic
current of GHRH neurons. This was not studied further, however,
because of space-clamp limitations [12].
(from
261.8862.81 mV to
Discussion
The GH axis is a well-known target for GHS and there is
evidence that GHS can stimulate GHRH secretion [1,4,15,22].
Our direct recordings of identified GHRH neurons in GHRH-
GFP mice have confirmed that ghrelin enhances their spontaneous
firing rate, providing a direct explanation for the hypothalamic
effect of GHS on the GH axis. This stimulation was direct,
required GHSR, phospholipase C and voltage-dependent calcium
channels, and paralleled other effects related to the modulation of
NPY neuronal activity in the arcuate nucleus [2]. Thus, the
growth axis and the appetite network have independent
hypothalamic sensors for ghrelin, despite the fact that they overlap
within the arcuate nucleus.
Ghrelin exerted a direct stimulation on GHRH neurons and,
importantly, did not modify spontaneous synaptic currents. This is
unlike the muscarinic M1-mediated modulations of GHRH neurons
[27], and consistent with the observation that a muscarinic
antagonist, atropine, does not blunt the effect of ghrelin (unpub-
lished data). The stimulatory effect of ghrelin was mimicked by
GHSR agonists and fully antagonized by a GHSR antagonist [30–
33]. It was interesting that the GHSR antagonist JMV3002 did not
modify the spontaneous activity of GHRH neurons, suggesting that
ghrelin responsiveness may normally require acute activation, and
arguing against a constitutive activity of unliganded GHSR [50].
The effector of the GHSR in GHRH neurons was likely to be
phospholipase C, since the stimulation of the firing rate induced by
ghrelin was prevented by U-73122, a pharmacological blocker of the
hydrolysis of phosphatidylinositol 4,5-bisphosphate to inositol
phosphates [38–40]. Activation of the phospholipase C pathway
generally enhances intracellular calcium dynamics, and indeed,
ghrelin elicits calcium transients in isolated hypothalamic neurons
Figure 7. Ghrelin enhanced the firing rate of GHRH neurons
during GABAAreceptor inhibition. A&B, typical experiments where
spontaneous action potentials of GHRH neurons were recorded as
ghrelin 10 nM was applied in the continuing presence of 3 mM
GABAzine, a GABAAreceptor antagonist, alone (A) or together with
20 mM CNQX, a AMPA/kainate receptors antagonist (B). C&D, summa-
ries of the stimulatory effect of ghrelin (10 nM) on the distributions of
action potential frequencies in GHRH neurons in the presence of 3 mM
GABAzine (C) or in the presence of 3 mM GABAzine + 20 mM CNQX (D).
Symbols and lines are the means and the sem of the numbers of
experiments indicated. Statistical significance (p,0.05, paired student-t
test) between mean values recorded in the presence of inhibitors alone
and in the presence of inhibitors plus ghrelin is framed by the grey area.
Note that the mean control distributions are shown as lines and sem
omitted, for clarity.
doi:10.1371/journal.pone.0009159.g007
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org8 February 2010 | Volume 5 | Issue 2 | e9159

Page 9

from immature animals, including some GHRH-positive neurons
[20,21]. In the present study, the effect of ghrelin was antagonised
by voltage-dependent calcium channels blockers (with either Ni2+or
Cd2+), but not by neurotransmission disruption (with the combina-
tion of CNQX + GABAzine). Therefore, it is likely that ghrelin
required and/or targeted high and low voltage-activated calcium
channels in GHRH neurons. In comparison, N-type channels were
involved in the generation of the calcium transients by ghrelin in
cultures of NPY neurons [51]. A requirement for calcium channels
might not be ubiquitous because ghrelin enhanced the firing rate of
unidentified neurons of the arcuate nucleus in calcium-depleted
medium [12,52,53]. This treatment not only slows down neuro-
transmission, but eliminates voltage-dependent calcium influx as
well. For a comparison, a calcium-deprived medium profoundly
altered the action potentials kinetics in GHRH neurons, which
became silent within minutes, precluding further studies (unpub-
lished data). Perforated patch clamp results showed that ghrelin
depolarized GHRH neurons in a tonic manner, and did not
significantly modify the kinetics of the spontaneous action potentials.
A stimulation of low voltage-activated calcium channels might
account for this depolarization, although other mechanisms might
be involved. Indeed, calcium influx controls a variety of background
conductances, including some Gd3+-sensitive transient receptor
potential channels [41,44]. Furthermore, it was interesting to notice
that narrow range blockers (of calcium channels) were as efficient in
eliminating the ghrelin stimulation than the broader range
compounds Gd3+and flufenamic acid [40,43,44]. Future work will
dissect out the molecular events linking membrane and internal
targets of ghrelin, notably mitochondria, in GHRH neurons [42].
The important role of calcium ions might explain why ghrelin less
consistently enhanced the firing rate of GHRH neurons of aged
mice. Indeed, calcium buffering is impaired in aged neurons [54],
and some of them might not tolerate the elevation of the firing rate
(the present study) and the elevation in intracellular calcium [21]
induced by ghrelin.
Ghrelin increased the spontaneous firing rate, but did not
modify either the firing pattern or the synchronisation amongst
GHRH neurons. This characterizes a simple mechanism for the
Figure 8. Ghrelin did not modify spontaneous synaptic currents of GHRH neurons. A&D, raw traces of spontaneous glutamatergic (230 mV)
and GABAergic (270 mV) synaptic currents recorded in the absence and presence of 10 nM ghrelin in GHRH neurons from adult male GHRH-GFP mice.
The effects of ghrelin (10 nM) were summarized, on the amplitude (B&E) and intervals (C&F) of glutamatergic (B&C) and GABAergic (E&F) synaptic
currents. The cumulative distributions are represented as symbols and lines, i.e. the means and the sem of the numbers of experiments indicated.
doi:10.1371/journal.pone.0009159.g008
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org9 February 2010 | Volume 5 | Issue 2 | e9159

Page 10

hypothalamic stimulation of the GH axis. Electrical activation of
the arcuate nucleus, which recruits GHRH neurons, is a relevant
trigger of GH secretion [55], and in vivo GH secretion is
potentiated with increasing duration, but not increasing frequency,
of electrical-field stimulations of the arcuate nucleus [55]. The
effect of ghrelin is similar, since it promotes a sustained electrical
activity of GHRH neurons. Synchronisation of the GHRH
network might be facultative for factors, like ghrelin, which
enhance the amplitude and not the frequency of GH pulses [1].
Some mathematical models have incorporated an antagonistic
effect of ghrelin on the hypothalamic effects of somatostatin [1],
but this is not essential, since in the mouse, the stimulatory effect of
ghrelin on GHRH neurons was observed in the absence of
somatostatin.
There is debate as to the origin of ghrelin that exerts a
hypothalamic effect. Peripheral ghrelin crosses the blood brain
barrier [56,57], as seen in the arcuate nucleus where GHRH cells
bodies are located [28]. Peripheral ghrelin or GHS induce rapid c-
fos expression in GHRH neurons [16,17,58]. Thus, ghrelin is
clearly capable of acting as a hormone to activate the GH axis at
the hypothalamic level. Peripheral ghrelin can also stimulate GH
cells directly, and therefore promote a synergy of effects at the
pituitary gland level [1,4]. In addition, the ghrelin-containing
synapses, found within the arcuate nucleus [12], might have a
specific effect at the hypothalamic level of the GH axis. Their
origin remains unclear, however, and they might represent a small,
or a very specialized, population as there is very little measurable
ghrelin in the hypothalamus. Moreover, whereas ghrelin-positive
synapses connect NPY neurons and GABAergic synapses [12], it is
unknown if ghrelin neurons synapse onto GHRH neurons. If so,
their basal tonic activity would be expected to be low in acute
brain slices since JMV3002, the GHSR antagonist [33], did not
change the firing rate of GHRH neurons. Importantly, while
GHRH might have some properties of a GHSR agonist [59–61],
mouse GHRH did not mimic the stimulatory effect of ghrelin on
GHRH neurons (unpublished data).
This direct modulation of GHRH neurons by ghrelin parallels
the direct effect of ghrelin on NPY neurons, which orchestrates the
activity of the appetite network [2]. GABA and NPY, two products
of NPY neurons, were not involved in the effect of ghrelin on
GHRH neurons (present study), and AgRP is thought to be
ineffective on the GH axis, notably at the hypothalamic level [62].
In addition, GHRH neurons are unlikely to regulate NPY neurons
directly because they project towards the median eminence, but
not to the arcuate nucleus [22]. Although an attractive concept,
our results do not support the notion of a co-ordination of
functions by ghrelin at multiple targets in the arcuate nucleus.
Instead, ghrelin can modulate independent targets and regulate
body functions in an independent manner. These findings do not
exclude the fact that other products might synchronize the activity
of the GH axis and the feeding circuits [25,26]. Our data
demonstrating that Y2 receptor activation stimulated the electrical
activity of identified GHRH neurons is quite provocative in this
respect, since these same Y2 receptors are mandatory for an
adaptation of GHRH neurons to prolonged fasting [25,26]. We
speculate that the growth axis and the appetite network do overlap
under some circumstances, perhaps following afferent rewiring [1].
Alternately, the different hypothalamic effects of ghrelin might
evolve independently during development, since ghrelin and
GHSR are expressed at early stages in life [34,35,63]. GHRH
neurons were responsive to ghrelin in 6 day-old mice and this
mechanism might participate in the GH secretion elicited upon
ghrelin injection in immature rats, aged 1–3 weeks [64,65].
Figure 9. Ghrelin changed the excitability of GHRH neurons. A,
recordings from a GHRH neuron in the absence and presence of 10 nM
ghrelin, in the perforated patch-clamp configuration. B, time course of
the effect of ghrelin 10 nM on the firing rate (upper graph) and on the
resting potential (lower graph) of the GHRH neuron shown in A. C,
summary of the effects of ghrelin 10 nM on the mean action potential
intervals in GHRH neurons recorded in the perforated patch-clamp
configuration. D, mean amplitude of the resting potential in GHRH
neurons in the absence and presence of 10 nM ghrelin (same
experiments as in C). Bars and lines are the means and the sem of
the numbers of experiments indicated. Statistical difference (p,0.05,
paired student-t test) with the control level is indicated.
doi:10.1371/journal.pone.0009159.g009
Table 1. Effects of ghrelin on the properties of action
potentials in GHRH neurons (n=8).
ParametersControl Ghrelin 10 nM
Amplitude (mV)Threshold
245.763.4
243.962.2
Peak4.663.05.362.0
After hyperpolarisation
254.063.8
251.663.1
Peak - threshold49.863.549.762.9
AHP - threshold
27.761.1
28.861.0
Duration (ms) Time-to-peak 5.9260.46 6.1760.34
Time-to-AHP12.2060.9412.6960.88
Half-width 1.6060.13 1.6360.19
doi:10.1371/journal.pone.0009159.t001
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org 10February 2010 | Volume 5 | Issue 2 | e9159

Page 11

Nitric oxide (NO), another product of NPY neurons [66,67]
might orchestrate the activity of the arcuate nucleus, without the
need for synaptic rewiring. Indeed, a NO synthase inhibitor
antagonizes the effect of ghrelin on food intake [68], and NO
synthase mediate the effects of ghrelin in the pituitary gland [69].
However, ghrelin and NO have opposite effects on excitability in
the arcuate nucleus (our unpublished data) [70], suggesting that
NO release cannot account for the major molecular effects of
ghrelin.
Ghrelin directly activated GHRH neurons, and this modulation
obviously concerns the GH axis, and does not require NPY
neurons involved in feeding. Although the same receptors, GHSR,
are involved in both regulatory effects, there might be differences
in the subsequent transduction pathways underlying the effects of
ghrelin in GHRH neurons and NPY neurons. For a comparison,
GHSR is expressed in both GH cells and GHRH neurons [1,4],
but the effects of ghrelin are not identical in both these cell types.
Insights into these mechanisms could assist in the development of
pharmacological agents in the treatment of feeding disorders or
GH deficiencies [4]. Ghrelin receptors can be found at the NPY
nerve terminals, accounting for presynaptic modulation of
POMC- and CRH-neurons [12]. It is not known if GHSR also
localize to GHRH nerve terminals. Future studies will be needed
to address the mechanisms of action of ghrelin at the median
eminence, characterized by its abundance of fenestrated blood
vessels. At this location, ghrelin might modulate the activity of
nerve terminals relevant to the GH axis.
Materials and Methods
All animal studies complied with the animal welfare guidelines
of the European Community, and/or UK Home Office
guidelines, as appropriate. They were approved by the Direction
of Veterinary departments of Herault, France (agreement number
34.251) and the Languedoc Roussillon Institutional Animal Care
and Use Committee (#CE-LR-0818).
Slice Preparation for Electrophysiological Recordings
Adult 12–16 week-old, GHRH-GFP mice [27,28] were
anesthetized by isoflurane inhalation, killed by decapitation, and
brains quickly removed into cold (0–2uC) solution-1 [in mM; 92
NMDG-Cl, 2.3 KCl, 1 CaCl2, 6 MgCl2, 26 NaHCO3, 1.2
KH2PO4, 25 glucose, 0.2 ascorbic acid, 0.2 thiourea; pH 7.4
gassed with 95% CO2, 5% O2]. Sagittal sections (300 mm-thick)
were cut with a microtome (Integraslice 7550, Campden Inst.,
UK) and stored at 34uC in solution-2 [in mM; 115 NaCl, 2.5 KCl,
1 CaCl2, 4 MgCl2, 26 NaHCO3, 1.25 NaH2PO4, 25 glucose, 0.2
ascorbic acid, 0.2 thiourea; pH 7.4, gassed with 95% CO2, 5%
O2] for at least 45 min. When indicated, young, 6 days-old
GHRH-GFP male mice (Gouty-Colomer et al. submitted), or
aged, 22–30 months-old GHRH-GFP male mice [29] were
investigated without modifications of the method.
Patch-Clamp Recordings
Slices were immobilized with a nylon grid in a submersion
chamber on the stage of an upright microscope (Axioskop FS2,
Carl Zeiss) and superfused with solution-3 [in mM; 125 NaCl, 2.5
KCl, 2 CaCl2, 1 MgCl2, 26 NaHCO3, 1.25 NaH2PO4, 12 glucose;
pH 7.4, gassed with 95% CO2, 5% O2] at a rate of ,1.5 ml/min
for at least 15 min at 30–32uC. A variant was used when NiCl2,
CdCl2and GdCl3were included [in mM; 138 NaCl, 2.5 KCl, 2
CaCl2, 1 MgCl2, 3 NaHCO3, 1.25 NaH2PO4, 10 HEPES, 12
glucose; pH 7.4, adjusted with NaOH, saturated with 100% O2].
Slices were viewed with a x63 immersion objective and Nomarski
differential interference contrast optics. Infrared differential
interference contrast illumination was used to visualize neurons
deeper in the slices and the images captured with an infrared
camera (C2400, Hamamatsu Photonics, Massy, France). Borosil-
icate glass pipettes were connected to the head stage of an EPC-9/
2 amplifier (HEKA, Lambrecht, Germany) to acquire and store
data using Pulse 8.09 software. Agonists were bath-applied, and
solutions were changed by switching the supply of the perfusion
system from one to another. Typically, the effect of ghrelin 10 nM
reached steady-state within 6–8 minutes, and the mean recovery
time from this effect was ,25 minutes. Activity was recorded for at
least 4 min at steady state under each condition.
For extracellular recordings of spontaneous action potentials,
pipettes (5–7 MV) were filled with (in mM), 130 NaCl, 2.5 KCl,
10 HEPES, 10 Glucose, 2 CaCl2, 1 MgCl2, pH 7.4 with NaOH
(295 mOsm adjusted with NaCl). Neuronal activity was recorded
in the voltage clamp mode (0 mV) of the loose-patch configuration
[27]. For whole cell recordings, pipettes (6–8 MV) were filled with (in
mM), 2.25 KCl, 125.3 KMeSO3, 10 HEPES, 0.1 EGTA acid, 1
MgCl2, 2 MgATP, 0.5 Na-GTP, 5 Na2-phospocreatine, 2 Na-
pyruvate, 2 malate, pH 7.2 with KOH (295 mOsm adjusted with
KMeSO3). Voltage- or current-clamp recordings were then
performed as described [27]. For perforated patch-clamp recordings,
gramicidin-D (50 mg/ml in dimethylsulfoxide) was dissolved at
50 mg/ml in the internal medium. The tips of the recording
electrodes (4–6 MV) were filled with the protein-free solution, and
backfilled with the antibiotic-containing medium [49]. Perforation
of the membrane patch was evaluated in the cell-attached
configuration under current-clamp at 0 pA, and recordings were
started when resting membrane potential was ,250 mV and
action potential amplitude was .50 mV.
Chemicals
Chemicals were from Sigma-Aldrich (L’isle d’Abeau, France)
except d-Glucose (Euromedex, France); tetrodotoxin (Latoxan,
France); BIIE 0246 (Tocris bioscience, Bristol, UK). U-73122 and
U-73343 were prepared as 10 mM stock solutions in DMSO and
kept frozen at 220uC until use; flufenamic acid was prepared daily
as 0.5 M stock solutions in DMSO.
Data Analysis
Standard off-line detection of spontaneous events (action
potentials or synaptic currents) were performed with Axograph
4.0 (Axon Instruments Inc., Foster City, CA). In brief, a template
was generated and used to scan the raw trace for similar
waveforms. All matching events were stored and, when present,
false positive events were discarded, either manually or automat-
ically on the basis of their amplitude or kinetics. Other calculations
and analysis were performed with IgorPro (Wavemetrics, Lake
Oswego, OR). The cumulative distributions were generated from
stretches of .4 minutes-long series of data (such as amplitude or
frequency of either action potentials or synaptic currents) recorded
at steady state. The distribution histogram of this stretch was
calculated using the appropriate binning interval (common to all
the experiments) and normalized to the number of events.
Cumulated distributions of the normalized data were then
generated using the same binning intervals. This presentation
allowed the statistical analysis (using the Kolmogorov-Smirnoff
test, see below) and permitted inspection of the distributions. The
modulation of GHRH neurons essentially shifted the position of
the cumulated distributions in either direction, and did not modify
the mean slope of the distributions. Accordingly, the frequency at
the half maximum of the cumulated distributions was used as an
index of the position of the cumulated distribution.
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org11 February 2010 | Volume 5 | Issue 2 | e9159

Page 12

Auto-correlograms were generated as follows: we constructed a
counting variable N(t,ds) corresponding to the number of events
falling at distance t from an other event of the signal, within bin ds
[71]. The histogram of this counting variable, once suitably
normalised for bin size ds and total measurement time T,
constitutes the auto-correlogram. To compute the corresponding
confidence limits, we relied on Brillinger results [72], according to
whom the square root of the cross-correlation distribution can be
approximated to a normal distribution of mean P0, the mean
density of the process, and of sem 1/(4 ds T)1/2. A 95% confidence
interval was thus computed as P061.96/(4 ds T)1/2. Note that
boundary effects inherent for finite data were corrected for, by
sub-weighting extreme values appropriately. Crosscorrelograms
were computed in a similar way. The approximate distribution
used for confidence intervals being now, mean (P0P1)1/2, with
P0and P1the mean density of the two processes, and of sem 1/
(4 ds T)1/2. A 95% confidence interval was also computed as
(P0P1)1/261.96/(4 ds T)1/2. The temporal organisation of stretches
of action potentials was also evaluated with a statistical test, which
required a randomisation of the neuronal activity, based on the
statistics of the activity itself. The procedure was to use the inter-
event intervals of the spontaneous action potentials and draw,
from this empirical distribution, a shuffled sequence of random
intervals. Thus, this artificial signal was totally decorrelated and
had the same histogram signature than the empirical series of data.
Comparisons between cross-correlograms generated with the
artificial and the empirical data were then performed.
Statistics
In each experiment, the Kolmogorov-Smirnoff (KS) test was
used to test the statistical difference between two distributions
obtained at steady-state (typically in the absence and in the
presence of an agonist). Data were then expressed as mean 6
standard-error-of-the-mean (sem) and the averaged distributions
were compared at each abscissa value with a paired student-t test,
to delineate the ranges of differences between untreated and
treated distributions. p,0.05 was taken as significant (ns, not
significant). Mean distributions are represented as lines connecting
the mean values (symbols) and error bars represent the sem. For
clarity, only part of the mean 6 sem values are shown in the
graphs.
Supporting Information
Figure S1
pattern of GHRH neurons. A, on average, ghrelin (10 nM)
strongly diminished the mean intervals of action currents in
GHRH neurons from adult males, and had no significant effect on
the skewness of the density histograms of these intervals, suggesting
that it did not shift the range of firing rates of GHRH neurons. B–
C, auto-correlogram analysis of the action currents intervals in the
absence and presence of ghrelin 10 nM. B, analysis of a typical
Ghrelin changed the firing rate but not the firing
individual experiment where the autocorrelograms of the action
current interval distributions are shown. Superimposed are the
95%-confidence boundaries of random distributions computed
from the data sets. The firing rate of the GHRH neuron was
enhanced by ghrelin (as evidenced by the upward shift of the
distribution), without a change in the bursting behaviour (similar
monotonous distributions), and the distributions were framed
within the boundaries of random distributions. C, mean
autocorrelogram distributions where solid lines are the means of
24 experiments. Statistical significance (paired student-t test)
between curves was found in a very narrow range of action
current intervals (20.3 to +0.3 s, shaded grey area), in accordance
with the conclusion that ghrelin increased firing rates without
changing its firing patterns. These findings agree with previous
observations that GH secretion evoqued by electric stimulation of
the arcuate nucleus is potentiated with increasing burst durations,
but not with increasing stimuli frequency [43].
Found at: doi:10.1371/journal.pone.0009159.s001 (0.14 MB TIF)
Figure S2
changed during development. A, time course of an experiment
where a single GHRH neuron was recorded from an immature
GHRH male mouse (PN6). C, simultaneous recordings of GHRH
neurons from an aged (24 months-old, C) male GHRH-GFP
mouse, and where 10 nM ghrelin enhanced the activity of one
neuron, but induced a transient inhibitory effect in the other
GHRH neuron. B&D, summaries of the effects of ghrelin (10 nM)
on the distributions of action current frequencies in GHRH
neurons from immature PN6 (B) and aged 22–30 months-old (D)
male GHRH-GFP mice. Note that the effects of ghrelin on
GHRH neurons were heterogeneous in aged animals. Symbols
and lines are the means and the sem of the numbers of
experiments indicated. Statistical significances (paired student-t
test) between curves are shown by the grey areas.
Found at: doi:10.1371/journal.pone.0009159.s002 (0.25 MB TIF)
The stimulatory effect of ghrelin on GHRH neurons
Acknowledgments
We thank the staff of the animal room facilities at the IFR3, for taking care
of the transgenic mouse line; MN Mathieu and S Debiesse for mice
genotyping. We thank Professor MJ Low for the generous gift of
somatostatin 2/2 mice; Drs. Elodie Fino, Fre ´de ´rique Scamps, Philippe
Lory and Didier Gagne for sharing expertise and reagents; Drs. Ge ´rard
Alonso and Norbert Chauvet for help with the brain anatomy; Drs. Michel
Desarmenien, Emmanuel Bourrinet and Professor Jean-Yves Le Guennec
for discussions; Dr. David Odson for careful reading of the manuscript.
Author Contributions
Conceived and designed the experiments: IR PM PFM. Performed the
experiments: GO PE VM LAGC PFM. Analyzed the data: GO PE PFM.
Contributed reagents/materials/analysis tools: PF FM JAF DC JM IR PM
PFM. Wrote the paper: FM NG IR PM PFM.
References
1. Farhy LS, Bowers CY, Veldhuis JD (2007) Model-projected mechanistic bases
for sex differences in growth hormone regulation in humans. Am J Physiol Regul
Integr Comp Physiol 292: R1577–93.
2. Abizaid A, Horvath TL (2008) Brain circuits regulating energy homeostasis.
Regul Pept 149: 3–10.
3. Bowers CY, Momany FA, Reynolds GA, Hong A (1984) On the in vitro and in
vivo activity of a new synthetic hexapeptide that acts on the pituitary to
specifically release growth hormone. Endocrinology 114: 1537–45.
4. Smith RG (2005) Development of growth hormone secretagogues. Endocr Rev
26: 346–360.
5. Nakazato M, Murakami N, Date Y, Kojima M, Matsuo M, et al. (2001) A role
for ghrelin in the central regulation of feeding. Nature London 409: 194–
198.
6. Hosoda H, Kojima M, Kangawa K (2006) Biological, physiological, and
pharmacological aspects of ghrelin. J Pharmacol Sci 100: 398–410.
7. Tena-Sempere M (2008) Ghrelin and reproduction: ghrelin as novel regulator of
the gonadotropic axis. Vitam Horm 77: 285–300.
8. Bennett PA, Thomas GB, Howard AD, Feighner SD, van der Ploeg LH, et al.
(1997) Hypothalamic growth hormone secretagogue-receptor (GHS-R) expres-
sion is regulated by growth hormone in the rat. Endocrinology 138: 4552–4557.
9. Tannenbaum G, Lapointe M, Beaudet A (1998) Expression of Growth
Hormone Secretagogue- -Receptor in the Mediobasal Hypothalamus. Endocri-
nology 139: 4420–4423.
10. Willesen MG, Kristensen P, Rømer J (1999) Co-localization of growth hormone
secretagogue receptor and NPY mRNA in the arcuate nucleus of the rat.
Neuroendocrinology 70: 306–316.
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org 12 February 2010 | Volume 5 | Issue 2 | e9159

Page 13

11. Lein ES, Hawrylycz MJ, Ao N, Ayres M, Bensinger A, et al. (2007) Genome-
wide atlas of gene expression in the adult mouse brain. Nature London 445:
168–176.
12. Cowley MA, Smith RG, Diano S, Tscho ¨p M, Pronchuk N, et al. (2003) The
distribution and mechanism of action of ghrelin in the CNS demonstrates a
novel hypothalamic circuit regulating energy homeostasis. Neuron 37: 649–661.
13. Hickey GJ, Drisko J, Faidley T, Chang C, Anderson LL, et al. (1996) Mediation
by the central nervous system is critical to the in vivo activity of the GH
secretagogue L-692,585. J Endocrinol 148: 371–380.
14. Popovic V, Miljic D, Micic D, Damjanovic S, Arvat E, et al. (2003) Ghrelin
Main Action of Growth Hormone Release is Exerted at hypothalamic Level.
J Clin Endocrinol Metab 88: 3450–3453.
15. Guillaume V, Magnan E, Cataldi M, Dutour A, Sauze N, et al. (1994) Growth
hormone (GH)-releasing hormone secretion is stimulated by a new GH-releasing
hexapeptide in sheep. Endocrinology 135: 1073–1076.
16. Dickson S, Luckman S (1997) Induction of c-fos Messenger Ribonucleic Acid in
Neuropeptide Y and Growth Hormone (GH)-Releasing Factor Neurons in the
Rat Arcuate Nucleus Following Systemic Injection of the GH secretagogue, GH-
releasing peptide-6. Endocrinology 138: 771–777.
17. Bailey AR, Smith RG, Leng G (1998) The nonpeptide growth hormone
secretagogue, MK-0677, activates hypothalamic arcuate nucleus neurons in
vivo. J Neuroendocrinol 10: 111–118.
18. Hewson AK, Viltart O, McKenzie DN, Dyball RE, Dickson SL (1999) GHRP-
6-induced changes in electrical activity of single cells in the arcuate,
ventromedial and periventricular nucleus neurons of a hypothalamic slice
preparation in vitro. J Neuroendocrinol 11: 919–923.
19. Traebert M, Riediger T, Whitebread S, Scharrer E, Schmid HA (2002) Ghrelin
acts on leptin-responsive neurons in the rat arcuate nucleus. J Neuroendocrinol
14: 580–586.
20. Kohno D, Gao HZ, Muroya S, Kikuyama S, Yada T (2003) Ghrelin directly
interacts with neuropeptide-Y-containing neurons in the rat arcuate nucleus:
Ca2+ signaling via protein kinase A and N-type channel-dependent mechanisms
and cross-talk with leptin and orexin. Diabetes 52: 948–956.
21. Kuramochi M, Kohno D, Onaka T, Kato S, Yada T (2005) Galanin-like peptide
and ghrelin increase cytosolic Ca2+ in neurons containing growth hormone-
releasing hormone in the arcuate nucleus. Regul Pept 126: 85–89.
22. Dimaraki EV, Jaffe CA (2006) Role of endogenous ghrelin in growth hormone
secretion, appetite regulation and metabolism. Rev Endocr Metab Disord 7:
237–29.
23. Chen HY, Trumbauer ME, Chen AS, Weingarth DT, Adams JR, et al. (2004)
Orexigenic action of peripheral ghrelin is mediated by neuropeptide Y and
agouti-related protein. Endocrinology 145: 2607–2612.
24. Tong Q, Ye CP, Jones JE, Elmquist JK, Lowell BB (2008) Synaptic release of
GABA by AgRP neurons is required for normal regulation of energy balance.
Nat Neurosci 11: 998–1000.
25. Lin S, Lin EJ, Boey D, Lee NJ, Slack K, et al. (2007) Fasting inhibits the growth
and reproductive axes via distinct Y2 and Y4 receptor-mediated pathways.
Endocrinology 148: 2056–2065.
26. Luque RM, Park S, Kineman RD (2007) Severity of the catabolic condition
differentially modulates hypothalamic expression of growth hormone-releasing
hormone in the fasted mouse: potential role of neuropeptide Y and
corticotropin-releasing hormone. Endocrinology 148: 300–309.
27. Baccam N, Alonso G, Costecalde T, Fontanaud P, Molino F, et al. (2007) Dual-
level afferent control of growth hormone-releasing hormone (GHRH) neurons in
GHRH-green fluorescent protein transgenic mice. J Neuroscience 27:
1631–1641.
28. Balthasar N, Me ´ry PF, Magoulas CB, Mathers KE, Martin A, et al. (2003)
Growth hormone-releasing hormone (GHRH) neurons in GHRH-enhanced
green fluorescent protein transgenic mice: a ventral hypothalamic network.
Endocrinology 144: 2728–2740.
29. Alonso G, Sanchez-Hormigo A, Loudes C, El Yandouzi T, Carmignac D, et al.
(2007) Selective alteration at the growth-hormone- releasing-hormone nerve
terminals during aging in GHRH-green fluorescent protein mice. Aging Cell 6:
197–207.
30. Broglio F, Boutignon F, Benso A, Gottero C, Prodam F, et al. (2002) EP1572: a
novel peptido-mimetic GH secretagogue with potent and selective GH-releasing
activity in man. J Endocrinol Invest 25: RC26–RC28.
31. Moulin A, Demange L, Berge ´ G, Gagne D, Ryan J, et al. (2007) Toward potent
ghrelin receptor ligands based on trisubstituted 1,2,4-triazole structure. 2.
Synthesis and pharmacological in vitro and in vivo evaluations. J Med Chem 50:
5790–5806.
32. Guerlavais V, Boeglin D, Mousseaux D, Oiry C, Heitz A, et al. (2003) New
active series of growth hormone secretagogues. J Med Chem 46: 1191–1203.
33. Moulin A, Demange L, Ryan J, Mousseaux D, Sanchez P, et al. (2008) New
trisubstituted 1,2,4-triazole derivatives as potent ghrelin receptor antagonists. 3.
Synthesis and pharmacological in vitro and in vivo evaluations. J Med Chem 51:
689–693.
34. Kamegai J, Wakabayashi I, Kineman RD, Frohman LA (1999) Growth
hormone-releasing hormone receptor (GHRH-R) and growth hormone
secretagogue receptor (GHS-R) mRNA levels during postnatal development in
male and female rats. J Neuroendocrinol 11: 299–306.
35. Katayama M, Nogami H, Nishiyama J, Kawase T, Kawamura K (2000)
Developmentally and regionally regulated expression of growth hormone
secretagogue receptor mRNA in rat brain and pituitary gland. Neuroendocri-
nology 72: 333–340.
36. Sun Y, Garcia JM, Smith RG (2007) Ghrelin and growth hormone secretagogue
receptor expression in mice during aging. Endocrinology 148: 1323–1329.
37. Dezaki K, Kakei M, Yada T (2007) Ghrelin uses Galphai2 and activates voltage-
dependent K+ channels to attenuate glucose-induced Ca2+ signaling and insulin
release in islet beta-cells: novel signal transduction of ghrelin. Diabetes 56:
2319–27.
38. Zheng L, Paik WY, Cesnjaj M, Balla T, Tomic ´ M, et al. (1995) Effects of the
phospholipase-C inhibitor, U73122, on signaling and secretion in pituitary
gonadotrophs. Endocrinology 136: 1079–88.
39. Mousseaux D, Le Gallic L, Ryan J, Oiry C, Gagne D, et al. (2006) Regulation of
ERK1/2 activity by ghrelin-activated growth hormone secretagogue receptor
1A involves a PLC/PKC epsilon pathway. Br J Pharmacol 148: 350–65.
40. Liu X, Lee K, Herbison AE (2008) Kisspeptin excites gonadotropin-releasing
hormone neurons through a phospholipase C/calcium-dependent pathway
regulating multiple ion channels. Endocrinology 149: 4605–14.
41. Kato M, Sakuma Y (1999) The effect of GHRP-6 on the intracellular Na+
concentration of rat pituitary cells in primary culture. J Neuroendocrinol 11:
795–800.
42. Andrews ZB, Liu ZW, Walllingford N, Erion DM, Borok E, et al. (2008) UCP2
mediates ghrelin’s action on NPY/AgRP neurons by lowering free radicals.
Nature 454: 846–51.
43. Jordani MC, Santos AC, Prado IM, Uyemura SA, Curti C (2000) Flufenamic
acid as an inducer of mitochondrial permeability transition. Mol Cell Biochem
210: 153–8.
44. Teruyama R, Armstrong WE (2007) Calcium-dependent fast depolarizing
afterpotentials in vasopressin neurons in the rat supraoptic nucleus.
J Neurophysiol 98: 2612–21.
45. Beedle AM, Hamid J, Zamponi GW (2002) Inhibition of transiently expressed
low- and high-voltage-activated calcium channels by trivalent metal cations.
J Membr Biol 187: 225–38.
46. Fox AP, Nowycky MC, Tsien RW (1987) Kinetic and pharmacological
properties distinguishing three types of calcium currents in chick sensory
neurones. J Physiol 394: 149–72.
47. Low MJ, Otero-Corchon V, Parlow AF, Ramirez JL, Kumar U, et al. (2001)
Somatostatin is required for masculinization of growth hormone-regulated
hepatic gene expression but not of somatic growth. J Clin Invest 107:
1571–1580.
48. King PJ, Williams G, Doods H, Widdowson PS (2000) Effect of a selective
neuropeptide Y Y(2) receptor antagonist, BIIE0246 on neuropeptide Y release.
Eur J Pharmacol 396: R1–R3.
49. Pieraut S, Laurent-Matha V, Sar C, Hubert T, Me ´chaly I, et al. (2007) NKCC1
phosphorylation stimulates neurite growth of injured adult sensory neurons.
J Neuroscience 27: 6751–6759.
50. Lall S, Balthasar N, Carmignac D, Magoulas C, Sesay A, et al. (2004)
Physiological studies of transgenic mice overexpressing growth hormone (GH)
secretagogue receptor 1A in GH-releasing hormone neurons. Endocrinology
145: 1602–1611.
51. Kohno D, Gao HZ, Muroya S, Kikuyama S, Yada T (2003) Ghrelin directly
interacts withneuropeptide-Y-containing neurons in the rat arcuate nucleus:
Ca2+ signaling via protein kinase A and N-type channel-dependent mechanisms
and cross-talk with leptin and orexin. Diabetes 52: 948–56.
52. Riediger T, Traebert M, Schmid HA, Scheel C, Lutz TA, et al. (2003) Site-
specific effects of ghrelin on the neuronal activity in the hypothalamic arcuate
nucleus. Neurosci Lett 341: 151–155.
53. Yanagida H, Morita T, Kim J, Yoshida K, Nakajima K, et al. (2008) Effects of
ghrelin on neuronal activity in the ventromedial nucleus of the hypothalamus in
infantile rats: an in vitro study. Peptides 29: 912–918.
54. Foster TC (2007) Calcium homeostasis and modulation of synaptic plasticity in
the aged brain. Aging Cell 6: 319–25.
55. Dickson SL, Leng G, Robinson ICAF (1993) Growth hormone release evoked
by electrical stimulation of the arcuate nucleus in anesthetized male rats. Brain
Res 623: 95–100.
56. Diano S, Farr SA, Benoit SC, McNay EC, da Silva I, et al. (2006) Ghrelin
controls hippocampal spine synapse density and memory performance. Nat
Neurosci 9: 381–388.
57. Banks WA, Burney BO, Robinson SM (2008) Effects of triglycerides, obesity,
and starvation on ghrelin transport across the blood-brain barrier. Peptides 29:
2061–2065.
58. Hewson AK, Dickson SL (2000) Systemic administration of ghrelin induces Fos
and Egr-1 proteins in the hypothalamic arcuate nucleus of fasted and fed rats.
J Neuroendocrinol 12: 1047–1049.
59. Howard AD, Feighner SD, Cully DF, Arena JP, Liberator PA, et al. (1996) A
receptor in pituitary and hypothalamus that functions in growth hormone
release. Science 273: 974–977.
60. Kojima M, Hosoda H, Date Y, Nakazato M, Matsuo H, et al. (1999) Ghrelin is a
growth-hormone-releasing acylated peptide from stomach. Nature London 402:
656–660.
61. Casanueva FF, Camin ˜a JP, Carreira MC, Pazos Y, Varga JL, et al. (2008)
Growth hormone-releasing hormone as an agonist of the ghrelin receptor GHS-
R1a. Proc Natl Acad Sci U S A 105: 20452–20457.
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org 13 February 2010 | Volume 5 | Issue 2 | e9159

Page 14

62. Tamura H, Kamegai J, Shimizu T, Ishii S, Sugihara H, et al. (2005) The effect
of agouti-related protein on growth hormone secretion in adult male rats. Regul
Pept 125: 145–149.
63. Liu YL, Yakar S, Otero-Corchon V, Low MJ, Liu JL (2002) Ghrelin gene
expression is age-dependent and influenced by gender and the level of
circulating IGF-I. Mol Cell Endocrinol 189: 97–103.
64. Hayashida T, Nakahara K, Mondal MS, Date Y, Nakazato M, et al. (2002)
Ghrelin in neonatal rats: distribution in stomach and its possible role.
J Endocrinol 173: 239–245.
65. Pinilla L, Barreiro ML, Tena-Sempere M, Aguilar E (2003) Role of ghrelin in
the control of growth hormone secretion in prepubertal rats: interactions with
excitatory amino acids. Neuroendocrinology 77: 83–90.
66. Bitran M, Tapia W, Eugenı ´n E, Orio P, Boric MP (1999) Neuropeptide Y
induced inhibition of noradrenaline release in rat hypothalamus: role of receptor
subtype and nitric oxide. Brain Res 851: 87–93.
67. Morley JE, Alshaher MM, Farr SA, Flood JF, Kumar VB (1999) Leptin and
neuropeptide Y (NPY) modulate nitric oxide synthase: further evidence for a role
of nitric oxide in feeding. Peptides 20: 595–600.
68. Gaskin FS, Farr SA, Banks WA, Kumar VB, et al. (2003) Ghrelin-induced
feeding is dependent on nitric oxide. Peptides 24: 913–918.
69. Rodrı ´guez-Pacheco F, Luque RM, Tena-Sempere M, Malago ´n MM, Castan ˜o JP
(2008) Ghrelin induces growth hormone secretion via a nitric oxide/cGMP
signalling pathway. J Neuroendocrinol 20: 406–412.
70. Riediger T, Giannini P, Erguven E, Lutz T (2006) Nitric oxide directly inhibits
ghrelin-activated neurons of the arcuate nucleus. Brain Res 1125: 37–45.
71. Cox DR (1965) On the Estimation of the Intensity Function of a Stationary
Point Process. J Royal Statistical Soc Series B 27: 332–337.
72. Brillinger D (1976) Estimation of the Second-Order Intensities of a Bivariate
Stationary Point Process. Journal of the Royal Statistical Society Series B 38:
60–66.
Ghrelin and GHRH Neurons
PLoS ONE | www.plosone.org14 February 2010 | Volume 5 | Issue 2 | e9159