Original Research Article
Endurance exercise training increases APPL1 expression and
improves insulin signaling in the hepatic tissue of diet-induced obese
mice, independently of weight loss
R Marinho1, ER Ropelle2, DE Cintra2, CT De Souza3, ASR Da Silva4, FC Bertoli1,
E Colantonio1, V D´Almeida1, JR Pauli1,2
1. Departamento de Biociências, Universidade Federal de São Paulo, Santos, SP,
Brasil. Curso de Educação Física – Modalidade Saúde.
2. Faculdade de Ciências Aplicadas, Universidade Estadual de Campinas, UNICAMP,
Limeira, SP, Brasil. Curso de Ciências do Esporte e de Nutrição.
3. Universidade do Extremo Sul Catarinense, Laboratório de Bioquímica e Fisiologia,
Santa Catarina, Criciúma, SC, Brasil.
4. Universidade de São Paulo, USP, Ribeirão Preto, SP, Brasil. Escola de Educação
Física e Esporte.
Please address correspondence to: José Rodrigo Pauli, PhD, Curso de Ciências do
Esporte, FCA-UNICAMP, Rua Pedro Zaccaria, 1300, Jardim Santa Luzia, Limeira, SP,
Brasil, 13484-350, Fax: +55 19 3701-6680, e-mail: firstname.lastname@example.org
RUNNING TITLE: Training and hepatic insulin signaling
KEYWORDS: training, high fat diet, diabetes, insulin signaling and liver
Abbreviations used are:
Akt, protein kinase B/Akt; APPL1, endosomal adaptor protein; Foxo1, nuclear forkhead
box O1; G6Pase, glucose 6-phosphatase; GSK3, glycogen synthase kinase 3; ITT,
insulin tolerance test; Kitt, rate constant for plasma glucose disappearance; PTT,
pyruvate tolerance test; PEPCK, phosphoenolpyruvate carboxykinase; PGC-1?,
peroxisome proliferator-activated receptor-? coactivator 1?; PI3K, phosphoinositide 3-
kinase; TRB3, a mammalian homolog of Drosophila tribbles.
Received 8 February 2011; Revised 9 September 2011; Accepted 15 September 2011
Journal of Cellular Physiology
© 2011 Wiley-Liss, Inc.
Hepatic insulin resistance is the major contributor to fasting hyperglycemia in type 2
diabetes. The protein kinase Akt plays a central role in the suppression of
gluconeogenesis involving forkhead box O1 (Foxo1) and peroxisome proliferator-
activated receptor gamma co-activator 1 alpha (PGC-1?), and in the control of glycogen
synthesis involving the glycogen synthase kinase beta (GSK3?) in the liver. It has been
demonstrated that endosomal adaptor protein APPL1 interacts with Akt and blocks the
association of Akt with its endogenous inhibitor, tribbles-related protein 3 (TRB3),
improving the action of insulin in the liver. Here, we demonstrated that chronic exercise
increased the basal levels and insulin-induced Akt serine phosphorylation in the liver of
diet-induced obese mice. Endurance training was able to increase APPL1 expression
and the interaction between APPL1 and Akt. Conversely, training reduced both TRB3
expression and TRB3 and Akt association. The positive effects of exercise on insulin
action are reinforced by our findings that showed that trained mice presented an
increase in Foxo1 phosphorylation and Foxo1/PGC-1? association, which was
accompanied by a reduction in gluconeogenic gene expressions (PEPCK and G6Pase).
Finally, exercised animals demonstrated increased at basal and insulin-induced GSK3?
phosphorylation levels and glycogen content at 24 hours after the last session of
exercise. Our findings demonstrate that exercise increases insulin action, at least in
part, through the enhancement of APPL1 and the reduction of TRB3 expression in the
liver of obese mice, independently of weight loss.
Hepatic insulin resistance is the major contributor to fasting hyperglycemia in
type 2 diabetes. During feeding, the increase in circulating pancreatic insulin inhibits
hepatic glucose output through the activation of the Ser/Thr kinase Akt (Nakae et al.
1999; Cho et al. 2001; Taniguchi et al. 2006). Schenck and colleagues showed that the
endosomal adaptor protein APPL1 regulates the activity of Akt (Schenk et al. 2008). It
has recently been reported that APPL1 increases hepatic insulin sensitivity by
potentiating insulin-mediated suppression of the gluconeogenic program, through Akt
phosphorylation (Cheng et al. 2009). In primary hepatocytes, the phosphorylation of Akt
by insulin and its downstream targets and the inhibition of the gluconeogenic program
are markedly attenuated by knockdown of APPL1 expression, but enhanced by its
overexpression. Moreover, the potentiation of APPL1 mediated by Akt activation by
insulin results in a marked enhancement of GSK3? phosphorylation and increases
glycogen accumulation in the mouse liver (Cheng et al. 2009).
The TRB3 (a mammalian homolog of Drosophila) is emerging as an important
player in the regulation of insulin signaling (Matsushima et al. 2006; Hegedus et al.
2007). TRB3 specifically blocks the actions of insulin in the liver by binding to the
enzyme Akt, reducing its phosphorylation (Du et al. 2003). Due to stimulation of Akt
inhibits glucose production by the liver, excess TRB3 may cause hepatic insulin
resistance. Interestingly, APPL1 interacts with Akt and blocks the association of Akt with
its endogenous inhibitor, TRB3, through direct competition, thereby promoting Akt
translocation to the plasma membrane and the endosomes for further activation (Cheng
et al. 2009).
Once activated, Akt increases the phosphorylation of different substrates in the
liver, including forkhead box O1 (Foxo1) and glycogen synthase kinase-3 (GSK3). In the
absence of insulin, Foxo1 interacts with the peroxisome proliferator-activated receptor
gamma coactivator 1 ? (PGC-1?), transactivating the two gluconeogenic genes,
phosphoenolpyruvate carboxykinase (PEPCK) and glucose 6-phosphatase (G6Pase)
(Dentin et al. 2007; Li et al. 2007; Puigserver et al. 2003). Upon insulin stimulation,
activated Akt phosphorylates Foxo1 at three conserved sites, inducing Foxo1
translocation to the cytoplasm and thereby reducing its transcriptional activity (Barthel et
al. 2001). Furthermore, insulin-induced Akt phosphorylation increases the glycogen
content by activating glycogen synthase through the inhibition (phosphorylation) of the
glycogen synthase kinase 3 beta (GSK3?). Thus, the control of the Akt activity in the
liver is critical to the control of glucose homeostasis.
Physical exercise is known to be essential in the treatment of type 2 diabetes.
Both the acute and persistent effects of exercise on glucose uptake and disposal have
important implications for individuals with diabetes in terms of chronic metabolic control
and the acute regulation of glucose homeostasis (Perseghin et al. 1996; Houmard et al.
1999; O’Gorman et al. 2006). The molecular mechanisms associated with insulin
sensitivity that are enhanced in response to exercise training may be related to
increased expression and/or the activation of key proteins that regulate glucose
metabolism (Röcki et al. 2008; Da Silva et al. 2010). Several studies have demonstrated
that exercise improves insulin signaling in the hepatic tissue (Heled et al. 2004; Hoene
et al. 2009; Ropelle et al. 2009); however, the mechanisms responsible for increase
liver insulin sensitivity, mediated by exercise, are poorly investigated. Therefore, the aim
of this study was to investigate the expressions of APPL1 and TRB3 in the liver of diet-
induced obese mice. In parallel, we investigated insulin signaling by monitoring Akt
serine phosphorylation and its substrates.
MATERIALS AND METHODS
Four-week-old male Swiss mice (15–20 g) from Federal University of São Paulo
Breeding Center were employed. All the animals were handled according to the
University guidelines for the use of animals in experimental studies and conforming to
the Guide for the Care and Use of Laboratory Animals, published by the US National
Institutes of Health (NIH publication No. 85-23 revised 1996). All experimental protocols
were approved by the Federal University of São Paulo (UNIFESP) Ethics Committee.
The mice were always housed in individual cages, subjected to a standard light-dark
cycle (6:00 a.m. to 6:00 p.m./6:00 p.m. to 6:00 a.m.) and room temperature was
maintained stable (23 ± 2°C). After random selection, animals were submitted to a
control or high-fat diet (HFD), as presented in Table 1, and distributed in to three
groups: control mice fed on standard rodent chow (Lean; n=6) for sixteen weeks;
sedentary mice fed with a HFD (HFD; n=6) also for sixteen weeks; mice fed with a HFD
for sixteen weeks and submitted to 8-week endurance exercise training (HFD exe; n=6).
In some experiments, we evaluated control mice submitted to exercise (Lean exe; n=6).
Determination of cumulative energy intake
The food intake of the animals (lean, HFD and HFD exe) was evaluated every
week. Thereafter standard chow or high-fat diet was given and food intake was
determined by measuring the difference between the weight of chow given and the
weight of chow at the end of a 24-h period. Thus, cumulative energy intake during 8
weeks of experimental period was determined in kcal.
Endurance exercise training protocol
The training sessions were performed during the light cycle and consisted of 60-
min swimming sessions, five days/week for eight weeks in an apparatus adapted for
mice containing warm water (32-33ºC). Volume and intensity of training were gradually
increased until the mice swan for 60 min wearing caudal dumbbells weighing 5% of their
body weight. Thereafter, volume and intensity were constant. The HFD group were
placed in the swimming apparatus for 10 min twice a week to simulate the water stress
associated with the experimental protocol.
Insulin tolerance test (ITT)
Sixteen, 24, 36 and 48h after the last session training, as indicated in the time
course experiments, the mice were submitted to an insulin tolerance test (ITT; 1.5 U/kg
body weight of insulin), after six hours fasting. Briefly, 1.5 IU/kg of human recombinant
insulin (Humulin R) from Eli Lilly (Indianapolis, IN, USA) was injected intraperitoneally in
mice, the blood samples were collected at 0, 5, 10, 15, 20, 25 and 30 minutes from the
tail for serum glucose determination. The rate constant for plasma glucose
disappearance (Kitt) was calculated using the formula 0.693/biological half life (t1/2).
The plasma glucose t1/2 was calculated from the slope of the last square analysis of the
plasma glucose concentration during the linear phase of decline (Bonora et al. 1989).
Fasting glucose and serum insulin quantification
Plasma glucose levels were determined by a colorimetric method using a glucose
meter (Advantage. Boehringer Mannheim, USA). Plasma was separated by
centrifugation (1100 g) for 15 min at 4?C and stored at -80?C until assay.
Radioimmunoassay (RIA) was employed to measure serum insulin, according to a
previous description (Scott et al. 1990).
Glycogen content and Pyruvate tolerance test (PTT)
Glycogen content in liver fragments was measured according to a previously
described method (Pimenta et al. 1989). The pyruvate tolerance test was performed to
estimate gluconeogenesis as follows. Mice were starved for 16 hr, and then injected
intraperitoneally with pyruvate (2 g/kg) dissolved in saline. Blood glucose levels were
determined in the tail blood every 30 min for 2 hours using a glucose meter (Advantage.
Boehringer Mannheim, USA).
Tissue extraction, immunopreciptation and immunoblotting
Twenty four hours after the last session of the endurance exercise training
protocol, exercised mice and the other groups were anesthetized with intraperitoneal
(i.p.) injection of sodium thiopental (40 mg/kg body weight). Following the experimental
procedures, the mice were euthanized under anesthesia (thiopental 200 mg/kg)
following the recommendations of the NIH (publication n° 85- 23). All animals were fed
ad libitum and food was withdrawn six hours before experimental procedures.
As soon as anesthesia was assured by the loss of pedal and corneal reflexes,
the abdominal cavity was opened, the cava vein exposed, and 0.2 ml of normal saline
or insulin (10-6 mol/L) were injected. After insulin injection, hepatic tissue was excised.
The liver was pooled, minced coarsely and homogenized immediately in extraction
buffer (mM) (1% Triton-X 100, 100 Tris, pH 7.4, containing 100 sodium pyrophosphate,
100 sodium fluoride, 10 EDTA, 10 sodium vanadate, 2 PMSF and 0.1 mg of
aprotinin/ml) at 4 ºC with a Polytron PTA 20S generator (Brinkmann Instruments model
PT 10/35) operated at maximum speed for 30 sec. The extracts were centrifuged at 11
000 rpm and 4°C in a Beckman 70.1 Ti rotor (Palo Alto, CA) for 40 min to remove
insoluble material, and the supernatants of these tissues were used for protein
quantification, using the Bradford method (Bradford, 1976). Aliquots of the resulting
supernatants containing 2.0 mg total protein were used for immunoprecipitation with
antibodies against Akt and FoxO1 at 4°C overnight, followed by SDS-PAGE, transferred
to nitrocellulose membranes, and blotted with APPL1, TRB3 and PGC-1? antibodies, as
described previously (De Souza et al., 2005).
In direct immunoblotting experiments, 150 mg protein extracts, obtained from the
liver were separated by SDS-PAGE, transferred to nitrocellulose membranes, and
blotted with antiphospho [Ser473] Akt, anti-GSK3?, anti-phospho-GSK3? (Ser-9), anti-
PEPCK and anti-phospho-Foxo1 (Cell Signaling Technology, Beverly, MA), anti-APPL1
(Abcam, Inc., Cambridge, UK), anti-Akt, anti-G6Pase, anti-Foxo1, anti-TRB3, anti-PGC-
1? and anti-?-tubulin (Santa Cruz Biotecnology, Inc., Santa Cruz, CA).
In both immunoprecipitation and immunoblotting experiments, proteins were
denaturated by boiling in Laemmli buffer (Laemmli 1970). For immunoblotting, the
sample buffer contained 100 mM dithiothreitol (DTT), whereas for immunoprecipitation,
50 mM DTT was used. Specific bands were labeled with a chemioluminescence kit
(Sigma) and visualization was performed by exposure of the membranes to X-ray films.
Band intensities were quantified by optical densitometry (Scion Image software,
ScionCorp, Frederick, MD) of the developed autoradiographs
Liver obtained from six mice were examined to determine the expression and
tissue distribution of the APPL1 protein. Hydrated, 4-µm sections of paraformaldehyde-
fixed, paraffin-embedded tissue were transferred and coated with poly-L-lysine and
fixed in methanol for 10 min. The sections were washed twice in 0.01 M phosphate-
buffered saline (PBS), pH 7.4 and incubated in PBS with 1% bovine serum albumin
(BSA, w/v) for 30 min to block nonspecific background. Upon BSA removal, the sections
were treated with primary antibody anti APPL1 (Santa Cruz Biotechnology, Inc., Santa
Cruz, CA) at a 1:100 dilution in PBS/BSA overnight at 4°C. After washing in PBS, slides
were covered with a 1:200 dilution of FITC-conjugated secondary donkey anti-rabbit IgG
(Santa Cruz Biotechnology, Inc., Santa Cruz, CA) in PBS, for 120 min. The sections
were subsequently washed in PBS and incubated for 10 min with DAPI (Vector
Laboratories, Burlingame, Calif., USA), prepared according to the manufacturer’s
instructions. Analysis and documentation of results were performed using a Leica FW
4500 B microscope. The frequency of positive cells was assessed using the Imagelab®
Analysis software (version 2.4). The frequency of positive cells was determined in 100
All numerical results are expressed as the means ± S.E.M. of the indicated
number of experiments. The results of blots are presented as direct comparisons of
bands in autoradiographs and quantified by densitometry using the Scion Image
software. Data were analyzed by ANOVA followed by post hoc analysis of significance
(Bonferroni test) when appropriate, comparing experimental and control groups. The
level of significance was set at p<0.05.
Physiological and metabolic parameters
Figure 1 shows results corresponding to the insulin tolerance test (ITT), body and
epididymal fat weight, fasting serum glucose and insulin levels of the Lean, obese
(HFD) and exercised obese mice (HFD exe). According to the ITT results (time-
dependent), the obese group demonstrated statistically reduced insulin sensitivity,
compared to lean mice. Conversely, endurance training increased insulin sensitivity at
16 and 24h after last session of the training protocol, compared to the HFD group. After
48 h, we also observed that exercised mice showed lower values compared to those
evaluated after 16h, compared to the lean group (Fig. 1A). Based on these results,
subsequent experiments were performed 24h the after last session of the training
As expected, we observed that the HFD group presented higher body weight,
epididymal fat, cumulative energy intake, fasting glucose and serum insulin, compared
to the control animals (Fig. 1B, C, D, E and F, respectively). In comparison with the
sedentary obese mice, exercised obese mice presented slight reduction in the total
body weight or epididymal fat mass, but without significantly difference (Fig. 1B and C,
respectively); however, when compared to the HFD group, exercised mice presented
lower fasting glucose (Fig. 1E) and fasting insulin (Fig. 1F). In relation to food intake, we
observed an increase in trained animals at the beginning of training (first three weeks)
and after this period the food intake was normalized (data not shown). However in the
analysis of total cumulative food intake, HFD exe consumed the similar amount of
energy (kcal) compared to HFD mice (Fig. 1D). We also performed the same exercise
protocol in lean mice and observed that exercise only increased insulin sensitivity and
did not alter the other metabolic parameters (data not shown).
Endurance exercise training increases insulin signaling by Akt phosphorylation
and increases APPL1 and decreases TRB3 expression in the liver of obese mice
We also determined the effect of exercise on insulin sensitivity in the liver of
obese mice. At 24h after the last session of the training protocol, mice received an i.v.
injection of insulin and the hepatic tissue was obtained to evaluate Akt serine
phosphorylation. We observed that insulin-induced Akt phosphorylation was reduced in
the liver of obese mice compared to that of the lean group (Fig. 2A upper panel).
Endurance exercise training was able to restore insulin sensitivity in the liver of obese
animals by increasing insulin-induced Akt phosphorylation to the levels of control mice.
Interestingly, endurance training increased Akt phosphorylation by about 35% in the
liver of obese mice compared to the obese group under resting conditions (Fig. 2A
upper panel). The Akt protein levels did not differ between groups (Fig. 2A lower panel).
To explore the mechanisms responsible for increased liver insulin sensitivity,
mediated by exercise in obese mice, we determined the liver expression of APPL1 and
TRB3 in lean, obese and exercised animals. Western blot analysis revealed that APPL1
expression was reduced in the liver of obese animals by about 64%, compared to the
control group. In contrast, the exercise program significantly increased levels of the
APPL1 protein in the liver of obese mice (Fig. 2B).
Subsequently, we performed immunoprecipitation experiments to evaluate the
association between APPL1 and Akt. Under fasting conditions, the APPL1/Akt
interaction was reduced in obese mice, compared to the lean group. On the other hand,
endurance exercise training increased this interaction by about 46%, compared to
obese animals (Fig. 2C).
Next, we evaluated the TRB3 expression and its association with Akt in lean,
obese and exercised obese mice. The high fat diet treatment increased TRB3
expression and TRB3/Akt association by 52 and 55%, compared to value in the lean
group; conversely, training reduced TRB3 expression and TRB3/Akt association by 55
and 48%, compared to obese mice at rest (Fig. 2D and 2E, respectively). Endurance
training did not alter the APPL1 and TRB3 expressions in the livers of lean mice (Fig.
To evaluate whether the localization of APPL1 was affected by the high-fat diet or
endurance exercise, we performed immunohistochemistry in liver samples of mice fed
on a standard diet (Lean), high-fat diet (HFD) or high-fat plus exercise (HFD exe) (Fig.
3A). Immunohistochemistry with an anti-APPL1 (APPL1)-specific antibody revealed that
obesity reduced the APPL1 content, but exercise was able to increase the APPL1
expression (frequency of positive cells to APPL1) (Fig. 3B). In addition, we found that
APPL1 immunoreactivity was highly localized in the cytoplasm, and a few positive cells
to APPL1 were found in the perinuclear and nuclear compartment in the liver of control
animals. Interestingly, in obese and exercised obese mice, this profile was unchanged
Endurance exercise training increases Foxo1 activity and reduces gluconeogenic
enzymes in the hepatic tissue of obese mice
To confirm whether the improvement in insulin signaling and Akt activation,
induced by exercise, was able to suppress the gluconeogenesis program, we
determined the effect of exercise on insulin-induced Foxo1 phosphorylation. Twenty-
four after the last session of the training protocol, mice received an i.v. injection of
insulin and a liver sample was obtained to evaluate Foxo1 phosphorylation. We verified
that insulin-induced Foxo1 phosphorylation was reduced in the liver of obese mice and
was restored after endurance exercise training (Fig. 4A upper panel). The Foxo1 protein
levels did not differ between the groups (Fig. 4A lower panel).
PGC-1? expression and the interaction between Foxo1 and PGC-1? in mice
were then analyzed. Endurance exercise training normalized the PGC1? protein levels
in the liver of obese mice (Fig. 4B). In addition, the immunoprecipitation assay revealed
that the high fat diet increased Foxo1/PGC1? interaction (32%), but that training
attenuated this interaction (Fig. 4C upper panel). The Foxo1 protein levels did not differ
between the groups (Fig. 4C lower panel).
We also determined the expressions of PEPCK and G6Pase in the liver of
animals under fasting conditions. The hepatic tissue extracts were blotted with anti-
PEPCK and anti-G6Pase antibodies. In the hepatic tissue of obese mice, PEPCK and
G6Pase expressions were increased by 52 and 61%, respectively, compared to the
lean group (Fig. 4D and E, respectively). Interestingly, after training, PEPCK and
G6Pase protein levels were decreased by 38 and 44%, compared to obese mice at rest
(Fig. 4D and E, respectively).
Endurance exercise training increases GSK3? phosphorylation and glycogen
content and decreases gluconeogesis in the liver of obese mice
Consistent with Akt activation and with the reduction in gluconeogenic gene
expressions, we evaluated the GSK3??phosphorylation and glycogen synthesis after
endurance training in obese animals. Insulin-induced GSK3? phosphorylation was
reduced in the liver of obese mice, and was restored (40%) at 24 hours after the last
session of endurance exercise training (Fig. 5A upper panel). The GSK3? protein levels
did not differ between the groups (Fig. 5A lower panel).
We also evaluated the hepatic glycogen content. In obese mice, the glycogen
content was reduced by about 50%, compared to the lean group (Fig. 5B). In the liver of
the exercised group, the glycogen content increased compared to lean (35%) and
obese mice (55%) (Fig. 5B). Finally, to determine the role of endurance exercise in
modulating gluconeogenesis, we also examined blood glucose levels in mice following
intraperitoneal injection of pyruvate in a separate set of experiments (Fig. 5C). In lean
mice, blood glucose increased to a peak level at 60 min after pyruvate administration,
and decreased thereafter. However, in obese mice with reduced APPL1 expression,
blood glucose continued to rise at 90 min, and did not significantly decrease until 120
min. Compared to the exercised mice, the glucose area after pyruvate challenge was
much greater in obese mice.