0013-7227/05/$15.00/0
Printed in U.S.A.
Endocrinology 146(1):3–10
Copyright © 2005 by The Endocrine Society
doi: 10.1210/en.2004-0968
Role of Neuronal Energy Status in the Regulation
of Adenosine 5ⴕ-Monophosphate-Activated Protein
Kinase, Orexigenic Neuropeptides Expression, and
Feeding Behavior
Kichoon Lee, Bing Li, Xiaochun Xi, Yeunsu Suh, and Roy J. Martin
Department of Animal Sciences (K.L.), The Ohio State University, Columbus, Ohio 43210; Neurobehavioral Laboratory
(B.L., X.X., Y.S., R.J.M.) Louisiana State University Pennington Biomedical Research Center, Baton Rouge, Louisiana
70808; and School of Human Ecology (B.L., R.J.M.), Louisiana State University AgCenter and College of Agriculture, Baton
Rouge, Louisiana 70803
decreased phosphorylation of AMPK and decreased AgRP expression. Overexpression of a dominant-inhibitory mutant of
AMPK significantly decreased low-glucose- or 2-DG-induced
AgRP expression. Furthermore, ex vivo hypothalamus culture
in high glucose concentrations decreased both expression and
phosphorylation of AMPK and expression of both AgRP and
neuropeptide Y, whereas pyruvate supplementation suppressed a 2-DG-induced AgRP expression. Finally, our in vivo
studies clearly show that central administration of pyruvate
dramatically delayed 2-DG-induced food intake. These data
indicate that modulation of ATP levels in neuronal cells triggers a cascade of events via AMPK that modulate feeding
behavior to restore energy status of cells. (Endocrinology 146:
3–10, 2005)
Nutrient sensing in the hypothalamus is tightly related to food
intake regulation. However, the mechanisms by which the
nutrient-sensing cells of the brain translate this signal of energy need into feeding behavior via regulation of neuropeptide expression are not known. To address this issue, we investigated two neuronal cell lines expressing agouti-related
protein (AgRP), ex vivo hypothalamic tissues, and in vivo
whole animals. Maintaining cells in a low cellular ATP concentration generated by low glucose, 2-deoxyglucose (2-DG),
ATP synthesis inhibitor, and 5-aminoimidazole-4-carboxamide 1-␤-D-ribofuranoside increased phosphorylation of AMPactivated protein kinase (AMPK) and increased AgRP expression, whereas maintaining cells in high ATP status by high
glucose and pyruvate supplementation in 2-DG-treated cells
A
LARGE BODY of evidence supports the roles of the
central nervous system in regulating feeding behavior
and body fat content (1, 2). The brain receives signals of
whole-body energy status via hormones and metabolites
such as leptin, ghrelin, insulin, glucagon-like peptide 1, glucose, amino acids, and fatty acids. It is proposed that these
diverse and complex signals are integrated by neurons in the
arcuate nucleus, including neuropeptide Y (NPY)/agoutirelated protein (AgRP) neurons and proopiomelanocortin
(POMC)/cocaine-amphetamine-related transcript neurons.
Growing experimental evidence supports the classical glucostatic (3, 4) and ischymetric hypotheses (5). Specifically,
there is increasing evidence for existence of cells in the arcuate nucleus and hindbrain that have the correct gene expression (6 – 8), electrophysiology (9), and feeding behavioral
response to nutrients that supply energy to the cell. For
example, central administration of 2-deoxy-d-glucose (2-DG), a
nonmetabolizable glucose analog that inhibits glucose use,
increases AgRP and NPY gene expression and elicits gluco-
privic food intake (10, 11). However, the mechanisms by
which signals from metabolites are translated into neurochemical signals still remain unsolved.
To address this issue, we developed in vitro and ex vivo
systems for cellular level studies. In the N1E-115 and
GT1-7 neuroblastoma cell lines, glucose suppressed expression of AgRP, whereas 2-DG induced AgRP expression. Next, we provided evidence that the modulation of
cellular ATP concentration by glucose, 2-DG, pyruvate,
ATP synthesis inhibitor, and 5-aminoimidazole-4-carboxamide 1-␤-d-ribofuranoside (AICAR) regulates AgRP expression, probably through the AMP-activated protein
kinase (AMPK) pathway. Our findings in vitro were further
confirmed by ex vivo and in vivo studies where changes
in neuronal energy status affect AMPK phosphorylation
and neuropeptide expression, leading to changes in food
intake. Our present data suggest that energy sensing through
ATP status is a main switch to regulate neuropeptide expression and food intake regulation.
First Published Online September 16, 2004
Abbreviations: AgRP, Agouti-related protein; AICAR, 5-aminoimidazole-4-carboxamide 1-␤-d-ribofuranoside; AMPK, AMP-activated
protein kinase; 2-DG, 2-deoxy-d-glucose; FBS, fetal bovine serum;
ICV, intracerebroventricular; NPY, neuropeptide Y; POMC, proopiomelanocortin.
Endocrinology is published monthly by The Endocrine Society (http://
www.endo-society.org), the foremost professional society serving the
endocrine community.
Materials and Methods
Cell culture and treatment
N1E-115 neuroblastoma was obtained from ATCC (Manassas, VA),
and GT1-7 cell line was a generous gift from Dr. Pamela Mellon (University of California, San Diego, San Diego, CA). The cells were cultured
in DMEM containing 25 mm glucose, 10% fetal bovine serum, 50 U of
penicillin, and 50 ␮g/ml streptomycin at 37 C in 5% CO2. For analysis
of the effects of metabolites on AgRP expression, the confluent cells were
3
4
Endocrinology, January 2005, 146(1):3–10
further cultured in DMEM containing a low percentage (1%) of serum
because serum contains glucose and factors (for example insulin and
leptin) that can affect AMPK phosphorylation and neuropeptide expression. To test glucose or 2-DG levels on AgRP expression, cells were
cultured in various levels (0.5, 1, 2.5, 5, 10, and 25 mm) of glucose or 2.5
mm glucose plus various concentrations of 2-DG (0, 1, 2.5, 5, 10, and 25
mm) for 16 h. All of the experiments with cell lines were done using
media containing 1% fetal bovine serum (FBS) and 2.5 mm glucose.
Plasmids and transfection
pcDNA3-Myc-tagged wild-type and dominant-inhibitory mouse
AMPK␣ (AMPK␣2WT and AMPK␣2KD) (12) were subcloned into pTREpuro plasmid (Invitrogen Corp., Carlsbad, CA). GT1-7 neuroblastoma
cells (⬃70% confluent) were transfected with pTRE-puro-AMPK␣2WT
or pTRE-puro-AMPK␣2KD plasmids using the LipofectAMINE reagent
as described by the manufacturer’s protocol (Invitrogen). Transfected
cells were grown 5 d in DMEM containing 1 ␮g/ml puromycin, and the
pool of puromycin-resistant clones was amplified for subsequent treatments. Transfected cells were incubated with medium containing 1%
FBS plus 1, 2.5, or 10 mm glucose, or 2.5 mm glucose plus 10 mm 2-DG
for 16 h.
Ex vivo experiments
Fresh hypothalamuses with the dimension of 3.2 mm ⫻ 3.5 mm ⫻ 2
mm and the average weight of approximately 7 mg were excised from
mice and immediately cultured in the media containing DMEM, 1% FBS,
and various concentrations of glucose (1, 2.5, 5, and 10 mm) for 2 h. The
hypothalamuses were also incubated with media containing DMEM in
2.5 mm glucose with either 10 mm 2-DG or 2-DG (10 mm) plus pyruvate
(10 mm) for 2 h. The protein was extracted from the hypothalamus and
subjected to Western blot for phospho-AMPK, AMPK, and ␤-actin. The
total RNA was also extracted from the hypothalamuses and subjected
to quantitative real-time RT-PCR for AgRP, NPY, POMC, and cyclophilin. The results were reported as percentages of values obtained from
expression of target genes divided by cyclophilin expression with six
samples per group.
Quantitative real-time RT-PCR
Real time RT-PCR was performed in a 25-␮l final reaction volume
using TaqMan 1000 Rxn PCR core reagent kit (Applied Biosystems,
Branchburg, NJ). Cyclophilin mRNA levels from each sample were used
as internal controls to normalize the mRNA levels (13). Detailed methods and PCR condition, using ABI PRISM 7700 sequence detector (Applied Biosystems), are described in our previous report (6). The data
within the linear region of the amplification curve are analyzed according to ABI’s user bulletin No. 2. The TaqMan probe is dual labeled with
5⬘-FAM and 3⬘-BHQ. Primer and probe sequences are available on
request.
ATP concentration
Total cellular ATP concentration was measured using the bioluminescent somatic cell assay kit (Sigma Chemical Co., St. Louis, MO)
according to the manufacturer’s instructions.
Western blot
The cells and tissues were homogenized in ice-cold lysis buffer as
described in a previous report (14). The membranes were incubated with
polyclonal primary antibodies, phospho-AMPK antibody (Cell Signaling Technology, Inc., Beverly, MA; 1:300 dilution), and AMPK antibody
(Cell Signaling; 1:1000 dilution) in blocking solution for 1 h at room
temperature and washed five times for 10 min each with TBS plus 0.05%
Tween 20. The membranes were further incubated with horseradish
peroxidase-conjugated secondary antibody for 1 h at room temperature.
Experimental animals, cannulation, and
intracerebroventricular (ICV) injection
Male Sprague Dawley rats, 230 –270 g, were maintained on a 12-h
light, 12-h dark cycle with lights on at 0700 h. All rats were singly housed
Lee et al. • Regulation of AMPK
and fed ad libitum. Drinking water was available at all times. Animal
experiments were approved by the Pennington Biomedical Research
Center Animal Care and Use Committee. For cannulation, rats were
anesthetized with Ketamine/acepromazine/xylazine, and 24-gauge
guide cannulas (Plastics One, Roanoke, VA) were placed in the lateral
cerebroventricle (0.8 mm posterior to bregma, 1.4 mm lateral to midline,
and 3.5 mm below the skull). After a 10-d recovery, rats were injected
with 10 ␮l of saline, 2-DG (3 mg), or 2-DG (3 mg) plus pyruvate (4 mg)
via the cannula during 0900 –1030 h. Food intake was recorded at 30 min,
1 h, and 2 h after injection.
Data presentation
Data from in vitro studies are presented as mean ⫾ sem of at least three
independent experiments performed in duplicate for in vitro experiments. Comparisons involving more than two groups were analyzed
using ANOVA. Student’s t test was used when determining the significant difference between two groups. The minimum level of significance
was set at P ⬍ 0.05.
Results
We first selected two independent neuronal cell lines,
N1E-115 and GT1-7 neuroblastoma cell lines expressing
AgRP (15, 16), to examine whether changes in energy status,
such as glucose deprivation or high levels of glucose, can
affect expression of AgRP, a potent hunger neuropeptide. As
shown in Fig. 1A, increasing glucose levels produced a dosedependent decrease in AgRP expression in both cell lines.
Because glucose is a major energy source, we next tested
whether AgRP expression is related to changes in cellular
energy availability. Around 2.5–5 mm glucose, cellular ATP
concentrations were sharply increased up to 2-fold, compared with 1 mm glucose. Above 5 mm glucose, ATP levels
were not further increased (Fig. 1B).
AMPK functions as a fuel sensor in many tissues and
organs where it inhibits anabolic pathways when cellular
ATP levels are depleted and when AMP rises in response to
limited energy availability (17). AMPK has been known to be
expressed in the brain and to be activated in response to
glucose deprivation (18). In our study, it was of interest to
investigate whether various energy statuses at different glucose concentrations could modulate the phosphorylation
state of AMPK. Our Western blot, using an antibody specific
for phosphorylated threonine at the 172nd amino acid of
AMPK, showed that glucose lower than that of physiological
levels increased AMPK phosphorylation, whereas elevation
of glucose concentration over the physiological range dramatically decreased phosphorylation of AMPK (Fig. 1C).
2-DG has been known as an inhibitor of glucose use and a
low-glucose mimetic in a wide variety of physiological situations (19). We further investigated whether 2-DG could reverse
the down-regulation of AgRP expression as previously shown
with high levels of glucose in Fig. 1A. In both cell lines, 2-DG
linearly increased expression of AgRP in a dose-dependent
manner (Fig. 2A). Because 2-DG inhibits glycolysis, we tested
whether 2-DG-induced AgRP expression is related to changes
in cellular energy availability. Treatment of neuronal cell cultures with various 2-DG concentrations showed a significant
dose-dependent decrease in cellular ATP levels in both cell lines
(Fig. 2B). In addition, the reduction of cellular ATP concentration by 2-DG treatment increased phosphorylation of AMPK in
a dose-dependent manner (Fig. 2C). Taken together, a decrease
in cellular ATP levels by either low glucose or increased 2-DG
Lee et al. • Regulation of AMPK
FIG. 1. Glucose suppresses expression of AgRP expression in neuronal cell lines. A, Two cell lines were cultured at various concentrations
of glucose for 16 h. Total RNAs were subjected to quantitative realtime RT-PCR for AgRP gene expression and normalization with cyclophilin (CYC) gene expression. Data are presented as percentile
mean ⫾ SEM value of five independent experiments. B, Relative ATP
concentrations were determined using the luciferase bioassay as described in Materials and Methods. Data are presented as percentage
changes of mean ⫾ SEM values of four independent experiments.
Statistical differences vs. 0.5 mM (glucose): *, P ⬍ 0.05; **, P ⬍ 0.01;
***, P ⬍ 0.001. C, Cell extracts were prepared and subjected to Western blot using anti-phospho-AMPK (p-AMPK) and anti-AMPK.
increased AMPK phosphorylation, which is presumably associated with increased AgRP expression in these low-energy
statuses.
Because 2-DG blocks glycolysis and also inhibits mitochondrial respiration caused by depletion of pyruvate, we
tested whether pyruvate can bypass the blockage of glycolysis by 2-DG treatment and inhibit 2-DG-induced AgRP
expression. As shown in Fig. 3A, supplementation of pyruvate resulted in a reduction of AgRP expression compared
with the basal condition. Furthermore, pyruvate attenuated
or reversed a 2-DG-induced increase in AgRP expression in
both cell lines. We then studied whether pyruvate could
increase cellular ATP concentration in 2-DG-treated cells, in
which ATP concentration might be a downstream effector
molecule to change AgRP expression. Cellular levels of ATP
were measured in these cells treated with 2-DG and pyruvate. Incubation of N1E-115 cell lines with 5 mm pyruvate
significantly increased cellular ATP levels by 18%, whereas
2-DG decreased ATP levels by 12% (Fig. 3B). In addition,
pyruvate prevented the decrease in cellular ATP levels in
Endocrinology, January 2005, 146(1):3–10
5
FIG. 2. 2-DG increases expression of AgRP expression in neuronal
cell lines. A, Two cell lines were cultured at various concentrations of
2-deoxyglucose for 16 h. Total RNAs were subjected to quantitative
real-time RT-PCR for AgRP gene expression. Data are presented as
percentile mean ⫾ SEM value of four independent experiments. Statistical differences vs. 0 mM (2-DG): * and #, P ⬍ 0.05; **, P ⬍ 0.01;
***, P ⬍ 0.001. B, Relative ATP concentrations were determined using
the luciferase bioassay as described in Materials and Methods. C, Cell
extracts were prepared and subjected to Western blot using antiphospho-AMPK (p-AMPK) and anti-AMPK.
2-DG-treated N1E-115 cells. These results indicate that supplementation of pyruvate rescues the low-energy status of
2-DG-treated cells, which is associated with induction of
AgRP expression.
The decreased ATP content of the neuronal cell lines after
2-DG treatment and its rescue by pyruvate suggests that
cellular ATP concentration may be negatively correlated
with the expression patterns of AgRP expression as shown
in Fig. 3. Therefore, we next asked whether direct modulation
of intracellular ATP concentration could affect AgRP expression. To study the effect of ATP depletion on AgRP expression, we used sodium azide (NaN3), which affects mitochondrial oxidative phosphorylation. As shown in Fig. 4B, NaN3
decreased ATP concentration and increased the amount of
phosphorylated AMPK in a dose-dependent manner. Furthermore, depletion of ATP by NaN3 treatment increased
expression of AgRP in a dose-dependent manner (Fig. 4A).
These data clearly show that changes in ATP concentration
directly regulate AgRP expression, suggesting a possibility
that ATP concentration can regulate activation of AMPK that
serves as a downstream effector molecule in the signaling
pathway of glucose- or 2-DG-regulated AgRP expression.
Because low cellular ATP levels are associated with increased AgRP expression, the most probable mechanism is
that the decreased levels of ATP together with the concomitant increase in AMP levels become a signal that is respon-
6
Endocrinology, January 2005, 146(1):3–10
FIG. 3. Pyruvate suppresses 2-DG-induced AgRP expression in N1E115 neuronal cell line. Cells were cultured in media containing 2.5 mM
glucose (control) plus pyruvate (5 mM), 2-DG (10 mM), or 2-DG (10 mM)
plus pyruvate (5 mM) for 16 h. A, Total RNA was extracted for quantitative real-time RT-PCR to measure AgRP expression. B, Relative
ATP concentrations were measured. Data are presented as mean ⫾
SEM value of three to five independent experiments. *, P ⬍ 0.05; **,
P ⬍ 0.01. Cell extracts were prepared and subjected to Western blot
using anti-phospho-AMPK (P-AMPK) and anti-AMPK.
sible for the induction of AgRP expression. To test this possibility, N1E-115 cells were treated with AICAR, a cellpermeable AMP analog. As shown in Fig. 4C, 2 mm AICAR
could increase AgRP expression linearly up to 2-fold at 4 h
in a time-course study. AICAR induced phosphorylation of
AMPK and AgRP expression in a dose-dependent manner
(Fig. 4D). These results show that direct modulation of either
ATP or AMP concentrations could affect AgRP expression,
indicating again that cellular energy status can be an important signal for regulation of AgRP expression.
The experiments presented so far suggested that changes
in cellular energy status regulate AgRP expression possibly
via AMPK. Next, we show direct evidence that AMPK is involved in regulation of AgRP expression. We overexpressed
wild-type and dominant-negative AMPK␣2 and tested whether
expression of the AMPK␣2-DN construct could suppress induction of AgRP expression in response to low glucose and
2-DG. Low glucose induced expression of AgRP in cells
expressing AMPK␣2WT, whereas AMPK␣2-DN expression
blocked low-glucose-induced AgRP expression (Fig. 5B).
In addition, AMPK␣2-DN overexpression attenuated 2-DGinduced AgRP expression. These results clearly show the
contribution of AMPK in the regulation of AgRP expression.
Lee et al. • Regulation of AMPK
FIG. 4. Activators of AMPK, sodium azide, and AICAR induce expression of AgRP in neuronal cell lines. A, N1E-115 neuronal cells
were treated with indicated dosage of sodium azide for 4 h, and total
RNA extracted from those cells was used for quantitative real-time
RT-PCR for AgRP expression. A–D, Data are presented as percentile
mean ⫾ SEM value of three independent experiments. Statistical differences vs. control (no treatment): *, P ⬍ 0.05; **, P ⬍ 0.01. B,
Relative ATP concentrations in the cells treated with various dosages
of sodium azide were measured. Cell extracts were prepared and
subjected to Western blot using anti-phospho-AMPK (P-AMPK), and
anti-AMPK. C, Time-course study of AICAR (2 mM) on AgRP expression. Three independent experiments were performed in duplicate. D,
Dose-dependent increases in AgRP expression in response to AICAR.
N1E-115 cells were treated with indicated dosages of AICAR for 4 h,
and protein was extracted from those cells and subjected to Western
blot for phosphorylated AMPK (P-AMPK) and total AMPK.
Our in vitro model allows more direct testing of the effect
of glucose on AgRP expression. However, this system cannot
address the indirect effects of interactions between cells
within the hypothalamus because astrocytes and glial cells
have been known to provide nutrients to the neighboring
neuronal cells (20). Therefore, we further investigated
whether changes in cellular energy status could affect neuropeptide gene expression ex vivo. The fresh hypothalamuses
were cultured ex vivo in 1, 2.5, 5, or 10 mm glucose concentrations for 2 h. AgRP expression was sharply decreased from
1–2.5 mm glucose but not further decreased at 5 and 10 mm
glucose concentrations (Fig. 6A), which indicates saturation
of the responsiveness above a 2.5-mm glucose concentration.
NPY expression was decreased with increased glucose concentration in a dose-dependent manner (Fig. 6B.), which
agrees well with the down-regulation of NPY gene expression in the hypothalamus by high-energy statuses, such as
feeding or ICV glucose. Expression of POMC was not significantly different but had a tendency to be up-regulated
with increased glucose concentrations (Fig. 6C). These stud-
Lee et al. • Regulation of AMPK
FIG. 5. Overexpression of AMPK attenuated low-glucose- or 2-DGinduced AgRP expression. A, The cell lysates from stably transfected
GT1-7 cells with vectors containing either AMPK␣2-WT or AMPK␣2-DN
were subjected to Western blot for total AMPK and ␤-actin (b-actin).
B, Stably transfected cells were incubated with medium containing 1,
2.5, or 10 mM glucose or 2-DG (10 mM) overnight. Then, the AgRP
expression in these cells was measured by quantitative real-time PCR.
Results are expressed as percentile mean ⫾ SEM of four experiments.
Statistical differences between WT-AMPK vs. DN-AMPK: *, P ⬍ 0.05.
ies, using ex vivo hypothalamic tissue culture, are the first
reports to show that modulation of nutrients can regulate
neuropeptide expression ex vivo. In addition, Western blot for
phospho-AMPK and total AMPK showed that, compared
with 10 mm glucose, hypothalamus cultured in 1 mm glucose
increased AMPK phosphorylation, suggesting that low glucose activated AMPK more highly (Fig. 6D). Furthermore,
the AMPK protein levels were elevated in the hypothalamus
treated with low glucose.
We next confirmed our in vitro finding that pyruvate supplementation suppressed 2-DG-induced AgRP expression in
the ex vivo condition. Ex vivo hypothalamus cultured with
2-DG significantly increased AgRP expression, and supplementation of pyruvate suppressed 2-DG-induced AgRP expression (Fig 7A). To relate to food intake, we then tested
whether pyruvate coadministration with 2-DG would attenuate 2-DG-induced food intake. Consistent with our previous reports and others (21, 22), in these studies, ICV 2-DG
robustly stimulated food intake with the greatest intake
(2.73 ⫾ 0.19 g) occurring within the first 30 min after 2-DG
administration, whereas the saline-injected rats ate 0.88 ⫾
0.23 g within the first hour. Furthermore, pyruvate supplementation dramatically inhibited 2-DG-induced food intake
for the first 30 min (Fig. 7B). For the next 30 min, suppression
of 2-DG-induced food intake by pyruvate was not observed,
and food intake was increased dramatically, suggesting
pyruvate supplementation delays 2-DG-stimulated feeding.
Discussion
It is clear that neuronal cells sense nutrient availability and
signal relevant neuronal pathways that lead to feeding behavior. Until now, it was not known how specific neuronal
cells convert nutrient status into feeding signals. To under-
Endocrinology, January 2005, 146(1):3–10
7
FIG. 6. Glucose regulates neuropeptide expression in hypothalamus
in ex vivo culture. A–C, Fresh hypothalamuses were cultured in various concentrations of glucose for 2 h. Total RNA was extracted from
hypothalamuses for quantitative real-time RT-PCR for AgRP (A),
NPY (B), and POMC (C). Data are presented as percentile mean ⫾ SEM
value of six mouse hypothalamuses. Statistical differences vs. 1 mM
(glucose): *, P ⬍ 0.05. D, Hypothalamuses were cultured ex vivo at 1
or 10 mM glucose concentrations for 2 h, and total protein was extracted from hypothalamuses and subjected to Western blot for phosphorylated AMPK (p-AMPK) and total AMPK. ␤-Actin serves as a
validation for loading equal amounts of samples.
stand the cellular and molecular mechanisms by which nutrients regulate the expression of feeding relevant neuropeptides, it is necessary to develop an in vitro system for direct
cellular level studies. It has been shown that the N1E-115
mouse neuroblastoma cell line expresses the gene product
for neuropeptides, including AGRP, that play a role in feeding behavior (15). N1E-115 neuroblastoma cells have been
used as a neuronal model system. These cells also have been
used to study polyunsaturated fatty acid modulation of
cAMP formation (23) and insulin-stimulated myoinositol uptake (24). Raising the glucose concentration from 5.6 to 25
mmol/liter was associated with decreased myoinositol uptake, with an inhibitory constant (Ki) of 20.4 mmol/liter for
N1E-115 cells (25). These characteristics indicate the potential
utility of the N1E-115 neuroblastoma cell line as a model for
studies of nutrients regulation of expression of feeding relevant neuropeptides. In addition, Gt1-7, a mouse hypothalamic neuronal cell line, has been known to express AgRP
mRNA (16).
Using two mouse neuroblastoma cell lines, we demonstrated for the first time that glucose directly regulates the
expression of AgRP. In addition, two cell lines sharply increased cellular ATP concentrations in a range of 1–5 mm
glucose concentration, and thereafter the cellular ATP concentrations were saturated. It is noteworthy to mention that
extracellular glucose concentration of the brain is approximately 2.4 mm in normoglycemia and that glucose concentration is maintained in a range of 0.2– 4.5 mm even in hypoand hyperglycemic states, whereas blood glucose concen-
8
Endocrinology, January 2005, 146(1):3–10
FIG. 7. Pyruvate supplementation suppressed 2-DG-induced AgRP
expression, and central administration of pyruvate delays 2-DGinduced hyperphagia. A, Hypothalamuses were cultured ex vivo in
media containing 2.5 mM glucose, 2.5 mM glucose plus 2-DG (10 mM),
or 2.5 mM glucose plus 2-DG plus pyruvate (pyr) (10 mM) for 2 h. Data
are presented as percentile mean ⫾ SEM value of at least six mouse
hypothalamuses. *, P ⬍ 0.05; **, P ⬍ 0.01. B, 2-DG (3 mg) or 2-DG (3
mg) plus pyruvate (4 mg) in a 10-␮l volume were injected into the
lateral ventricle of rats. Food intake was measured at indicated time
intervals. Data are presented as mean ⫾ SEM value of four rats per
group. Statistical differences between 2-DG vs. 2-DG plus pyruvate:
***, P ⬍ 0.001.
trations vary in a much wider range of approximately 2.8 –
15.2 mm in hypoglycemia and hyperglycemia (26). In addition, degrees of AMPK phosphorylation are linearly
decreased with increased glucose concentration at physiological levels. Therefore, minute changes in glucose concentration or energy status can be monitored and sensed by a
delicate sensing system in the brain to generate hunger or
satiety signals. Lastly, magnitudes of change in AgRP expression in response to glucose concentrations are similar to
the degree of changes in AgRP expression in the arcuate
during fasting and refeeding conditions (27). Our data on
AgRP expression in two independent neuronal cell lines
clearly demonstrated that these cells could respond to glucose deprivation or high levels of glucose, providing a reliable in vitro system to study the cellular mechanisms of
nutrient regulation of AgRP expression. In addition, in vitro
studies allow more direct testing of the effect of glucose on
AgRP expression without the disadvantage of indirect effects
brought about by interactions of cells with other areas of the
brain and changes in blood hormones or substrates.
Inhibition of glucose use by 2-DG has been known to cause
glucoprivic food intake and to increase AgRP and NPY expression in the arcuate (10, 11). However, it is not clear how
glucoprivic conditions are converted to specific signals to
AgRP neurons, which stimulate food intake. Our study
showed that modulation of glucose availability in two cell
Lee et al. • Regulation of AMPK
FIG. 8. Schematic diagram. Glucose down-regulates AgRP expression through AMPK. 2-DG inhibits glycolysis, decreases cellular ATP
levels, increases phosphorylation of AMPK, and increases AgRP expression. Pyruvate bypasses glycolysis and suppresses 2-DG-induced
AgRP expression. Direct modulation AMPK by ATP synthesis inhibitor and AICAR (a cell-permeable adenosine analog) regulates AgRP
expression. TCA, tricarboxylic acid cycle.
lines could affect AgRP expression, indicating a direct effect
of glucose and 2-DG on these neuronal cell lines. Because
inhibition of glycolysis by 2-DG increased AgRP expression,
it was hypothesized that glucose metabolites in the glycolysis
pathway could be an effector molecule that is responsible for
regulation of AgRP expression. To test this hypothesis, pyruvate was supplemented and suppressed a 2-DG-induced
AgRP expression both in cell line cultures and hypothalamus
culture ex vivo. These findings suggested that downstream
metabolites of the glycolysis pathway beyond pyruvate or
energy status might be responsible for regulation of AgRP
expression.
Our studies clearly show that glucoprivic conditions elicit
low cellular ATP concentration and that direct modulation of
ATP synthesis regulates AgRP expression. In addition, expression of dominant negative AMPK in the neuronal cell
line blocked low-glucose-induced AgRP expression and attenuated 2-DG-induced AgRP expression. These findings
indicate that increased expression of AgRP in glucoprivic
conditions is, at least partially, mediated through the regulation of AMPK.
It is possible that depletion of ATP in the hypothalamus of
2-DG-treated rats can be rescued by pyruvate supplementation, which readily generates enough ATP through the
tricarboxylic acid cycle to normalize the energy state, thereby
resulting in the suppression of 2-DG-induced food intake for
the first 30 min. However, the lack of supplemented pyruvate
that may occur during the 30- to 60-min period probably
resumed 2-DG-induced food intake. Therefore, it will be
interesting to investigate whether a continuous supply of
pyruvate in the brain can delay glucoprivic food intake over
Lee et al. • Regulation of AMPK
a long-term period. Taken together, our in vitro, ex vivo, and
in vivo studies provide strong evidence that restoring neuronal energy status by pyruvate supplementation decreased
AgRP expression and food intake through the phosphorylation states of AMPK.
Because AgRP expression is colocalized with NPY expression in the same neurons of the arcuate, NPY expression can
be also regulated in the same manner as AgRP within the
same cells. Acute ip administration of 2-DG has been known
to increase food intake and both AgRP and NPY expression
in the arcuate (10). Most recent studies also show that adenoviral expression of AMPK in the arcuate increased food
intake and expression of these orexigenic peptides (28).
Therefore, it is likely that in response to glucose fluctuation,
the metabolic and nutritional signals in the NPY/AgRP neurons can be generated and converted into the secondary
signals that possibly share the same initiation steps. But the
common signals, for example AMPK activities, need to be
diverged to regulate two independent expressions of genes
unless these promoters contain the same response elements.
It will be interesting to investigate the downstream signaling
pathway of AMPK connecting AMPK to the regulation of
AgRP and NPY gene expression.
In agreement with our studies, several reports show that
peripheral administration of lactate and pyruvate decreased
food intake in rats (29). In fact, lactate is produced by astrocytes and glial cells in the brain and converted to pyruvate
by lactate dehydrogenase (20), providing an energy source to
neighboring neuronal cells. In addition, blood lactate and
pyruvate can be transported across the blood-brain barrier
(30) and used as alternative energy sources during periods
of decreased glucose availability. Pyruvate supplementation
has been known to increase ATP content and enhance energetic status during brain hemorrhagic shock (31). Based
upon our findings and previous reports, we hypothesized
that modulation of nutrient availability and use would affect
food intake through changes in neuronal energy status. Consistent with our hypothesis, ICV infusion of brain fuels, including glucose, glycerol, and ␤-hydroxybutyrate, reduced
food intake (32), and mercaptoacetate, an inhibitor of fatty
acid oxidation, caused lipoprivic feeding (10). These studies
further support our conceptual hypothesis. However, additional studies revisiting these experiments from the view of
energy status will reconcile new concepts and help to better
understand mechanisms of energy-privic feeding.
As a conclusion, our studies provide direct evidence of the
critical role played by neuronal cell energy status in the
activation and expression of AMPK, modulation of NPY and
AGRP expression, and subsequent food intake. The data
support our overall hypothesis that energy status of the
nutrient-sensing cells of the brain is the major switch for the
control of neuropeptide expression and feeding behavior.
Manipulation of the energy status of the cells included modification of the rate of neuronal glucose, pyruvate oxidation,
and direct manipulation of ATP concentration.
Fairly recent reports show that inhibition of ␤-oxidation by
C75 administration generates low energy status, which stimulates food intake through AMPK (33). In addition, most
recent studies show that both adenoviral expression of
AMPK in the arcuate and ICV injection of AICAR increased
Endocrinology, January 2005, 146(1):3–10
9
food intake, suggesting involvement of AMPK in food intake
regulation (28, 34). Our studies provide strong evidence that
glucose, a major cellular energy source, regulates food intake
through the cellular ATP status-AMPK axis. Therefore, the
final common pathway for sensing glucose, fatty acids, and
other nutrients can be merged into the energy status of the
cell. These key findings should provide novel intervention
opportunities for manipulation of feeding behavior.
Acknowledgments
We are grateful to Dr. Morris Birnbaum (University of Pennsylvania)
for generously providing vectors containing AMPK␣2WT and
AMPK␣2KD and to Dr. Pamela Mellon (University of California, San
Diego) for providing the GT1-7 cell line.
Received July 27, 2004. Accepted September 9, 2004.
Address all correspondence and requests for reprints to: Dr. Kichoon
Lee, Department of Animal Sciences, The Ohio State University, Columbus, Ohio 43210. E-mail: [email protected]; or Roy J. Martin, Pennington Biomedical Research Center, Baton Rouge, Louisiana 70808.
E-mail: [email protected].
This study was supported in part by Biotechnology for Students and
Teachers program of LSU AgCenter and the LSU Pennington Biomedical
Research Center.
References
1. Schwartz MW, Woods SC, Porte Jr D, Seeley RJ, Baskin DG 2000 Central
nervous system control of food intake. Nature 404:661– 671
2. Schwartz MW 2002 Neuronal pathways regulating food intake and body
adiposity. Ann Endocrinol (Paris) 63:117–120
3. Mayer J 1953 Glucostatic mechanism of regulation of food intake. N Engl J Med
249:13–16
4. Mayer J 1955 Regulation of energy intake and the body weight: the glucostatic
theory and the lipostatic hypothesis. Ann NY Acad Sci 63:15– 43
5. Nicolaidis S, Even PC 1990 The ischymetric control of feeding. Int J Obes
14(Suppl 3):35– 49; discussion 50 –52
6. Li B, Xi X, Roane DS, Ryan DH, Martin RJ 2003 Distribution of glucokinase,
glucose transporter GLUT2, sulfonylurea receptor-1, glucagon-like peptide-1
receptor and neuropeptide Y messenger RNAs in rat brain by quantitative real
time RT-PCR. Brain Res Mol Brain Res 113:139 –142
7. Kang L, Routh VH, Kuzhikandathil EV, Gaspers LD, Levin BE 2004 Physiological and molecular characteristics of rat hypothalamic ventromedial nucleus glucosensing neurons. Diabetes 53:549 –559
8. Zhou J, Roane DS, Xi X, Bogacka I, Li B, Ryan DH, Martin RJ 2003 Short-term
food restriction and refeeding alter expression of genes likely involved in brain
glucosensing. Exp Biol Med (Maywood) 228:943–950
9. Silver IA, Erecinska M 1998 Glucose-induced intracellular ion changes in
sugar-sensitive hypothalamic neurons. J Neurophysiol 79:1733–1745
10. Sergeyev V, Broberger C, Gorbatyuk O, Hokfelt T 2000 Effect of 2-mercaptoacetate and 2-deoxy-d-glucose administration on the expression of NPY,
AGRP, POMC, MCH and hypocretin/orexin in the rat hypothalamus. Neuroreport 11:117–121
11. Fraley GS, Dinh TT, Ritter S 2002 Immunotoxic catecholamine lesions attenuate 2-DG-induced increase of AGRP mRNA. Peptides 23:1093–1099
12. Yin W, Mu J, Birnbaum MJ 2003 Role of AMP-activated protein kinase in cyclic
AMP-dependent lipolysis in 3T3–L1 adipocytes. J Biol Chem 278:43074 – 43080
13. Medhurst AD, Harrison DC, Read SJ, Campbell CA, Robbins MJ, Pangalos
MN 2000 The use of TaqMan RT-PCR assays for semiquantitative analysis of
gene expression in CNS tissues and disease models. J Neurosci Methods
98:9 –20
14. Lee K, Villena JA, Moon YS, Kim KH, Lee S, Kang C, Sul HS 2003 Inhibition
of adipogenesis and development of glucose intolerance by soluble preadipocyte factor-1 (Pref-1). J Clin Invest 111:453– 461
15. Roth JD, Yee DK, Kisley LR, Fluharty SJ 2002 Modeling the pathways of
energy balance using the N1E-115 murine neuroblastoma cell line. Brain Res
Mol Brain Res 103:146 –150
16. Khong K, Kurtz SE, Sykes RL, Cone RD 2001 Expression of functional melanocortin-4 receptor in the hypothalamic GT1–1 cell line. Neuroendocrinology
74:193–201
17. Hardie DG, Carling D 1997 The AMP-activated protein kinase–fuel gauge of
the mammalian cell? Eur J Biochem 246:259 –273
18. Culmsee C, Monnig J, Kemp BE, Mattson MP 2001 AMP-activated protein
kinase is highly expressed in neurons in the developing rat brain and promotes
neuronal survival following glucose deprivation. J Mol Neurosci 17:45–58
10
Endocrinology, January 2005, 146(1):3–10
19. Brown J 1962 Effects of 2-deoxyglucose on carbohydrate metabolism: review
of the literature and studies in the rat. Metabolism 11:1098 –1112
20. Tsacopoulos M, Magistretti PJ 1996 Metabolic coupling between glia and
neurons. J Neurosci 16:877– 885
21. Miller CC, Martin RJ, Whitney ML, Edwards GL 2002 Intracerebroventricular
injection of fructose stimulates feeding in rats. Nutr Neurosci 5:359 –362
22. Miselis RR, Epstein AN 1975 Feeding induced by intracerebroventricular
2-deoxy-d-glucose in the rat. Am J Physiol 229:1438 –1447
23. Murphy MG, Byczko Z 1992 Further studies of the mechanism(s) of polyunsaturated-fatty-acid-mediated increases in intracellular cAMP formation in
N1E-115 neuroblastoma cells. Neurochem Res 17:1069 –1077
24. Dunlop ME, Hill MA, Larkins RG 1989 Acute changes in myo-inositol uptake
and 22Na⫹ flux in murine neuroblastoma cells (N1E-115) following insulin.
FEBS Lett 220:84 – 88
25. Dunlop ME, Hill MA, Larkins RG 1988 The influence of insulin and sorbinil
on myoinositol uptake in peripheral nerve from normal and diabetic rats and
a neuroblastoma cell line (N1E-115). Diabetes Res 8:51–57
26. Silver IA, Erecinska M 1994 Extracellular glucose concentration in mammalian brain: continuous monitoring of changes during increased neuronal activity and upon limitation in oxygen supply in normo-, hypo-, and hyperglycemic animals. J Neurosci 14:5068 –5076
27. Bertile F, Oudart H, Criscuolo F, Maho YL, Raclot T 2003 Hypothalamic gene
expression in long-term fasted rats: relationship with body fat. Biochem Biophys Res Commun 303:1106 –1113
Lee et al. • Regulation of AMPK
28. Minokoshi Y, Alquier T, Furukawa N, Kim YB, Lee A, Xue B, Mu J, Foufelle
F, Ferre P, Birnbaum MJ, Stuck BJ, Kahn BB 2004 AMP-kinase regulates food
intake by responding to hormonal and nutrient signals in the hypothalamus.
Nature 428:569 –574
29. Langhans W, Damaske U, Scharrer E 1985 Different metabolites might reduce
food intake by the mitochondrial generation of reducing equivalents. Appetite
6:143–152
30. Cremer JE, Cunningham VJ, Pardridge WM, Braun LD, Oldendorf WH 1979
Kinetics of blood-brain barrier transport of pyruvate, lactate and glucose in
suckling, weanling and adult rats. J Neurochem 33:439 – 445
31. Mongan PD, Capacchione J, Fontana JL, West S, Bunger R 2001 Pyruvate
improves cerebral metabolism during hemorrhagic shock. Am J Physiol Heart
Circ Physiol 281:H854 –H864
32. Davis JD, Wirtshafter D, Asin KE, Brief D 1981 Sustained intracerebroventricular infusion of brain fuels reduces body weight and food intake in rats.
Science 212:81– 83
33. Kim EK, Miller I, Aja S, Landree LE, Pinn M, McFadden J, Kuhajda FP,
Moran TH, Ronnett GV 2004 C75, a fatty acid synthase inhibitor, reduces food
intake via hypothalamic AMP-activated protein kinase. J Biol Chem 279:
19970 –19976
34. Andersson U, Filipsson K, Abbott CR, Woods A, Smith K, Bloom SR, Carling
D, Small CJ 2004 AMP-activated protein kinase plays a role in the control of
food intake. J Biol Chem 279:12005–12008
Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving the
endocrine community.