letters
Metformin activates a duodenal Ampk–dependent
pathway to lower hepatic glucose production in rats
© 2015 Nature America, Inc. All rights reserved.
Frank A Duca1, Clémence D Côté1,2, Brittany A Rasmussen1,2, Melika Zadeh-Tahmasebi1,2, Guy A Rutter4,
Beatrice M Filippi1 & Tony K T Lam1–3,5
Metformin is a first-line therapeutic option for the treatment
of type 2 diabetes, even though its underlying mechanisms
of action are relatively unclear1–6. Metformin lowers blood
glucose levels by inhibiting hepatic glucose production
(HGP), an effect originally postulated to be due to a
hepatic AMP-activated protein kinase (AMPK)-dependent
mechanism5,6. However, studies have questioned the
contribution of hepatic AMPK to the effects of metformin on
lowering hyperglycemia1,3,4, and a gut–brain–liver axis that
mediates intestinal nutrient- and hormone-induced lowering
of HGP has been identified7. Thus, it is possible that
metformin affects HGP through this inter-organ crosstalk.
Here we show that intraduodenal infusion of metformin for
50 min activated duodenal mucosal Ampk and lowered HGP
in a rat 3 d high fat diet (HFD)-induced model of insulin
resistance. Inhibition of duodenal Ampk negated the HGPlowering effect of intraduodenal metformin, and both duodenal
glucagon-like peptide-1 receptor (Glp-1r)–protein kinase A
(Pka) signaling and a neuronal-mediated gut–brain–liver
pathway were required for metformin to lower HGP.
Preabsorptive metformin also lowered HGP in rat models
of 28 d HFD–induced obesity and insulin resistance and
nicotinamide (NA)–streptozotocin (STZ)–HFD-induced type 2
diabetes. In an unclamped setting, inhibition of duodenal Ampk
reduced the glucose-lowering effects of a bolus metformin
treatment in rat models of diabetes. These findings show
that, in rat models of both obesity and diabetes, metformin
activates a previously unappreciated duodenal Ampk–
dependent pathway to lower HGP and plasma glucose levels.
Diabetes is characterized by disrupted glucose homeostasis due
partly to increased hepatic glucose production (HGP)8. The biguanide metformin lowers plasma glucose levels by reducing HGP9–11,
in addition to improving insulin sensitivity12, in diabetic humans and
rodent models of diabetes. It was originally postulated that metformin
decreases HGP by inhibiting mitochondrial complex I13, which results
in an elevation of AMP levels and activation of AMPK through
liver kinase B1 (LKB1)-dependent phosphorylation of AMPK5,14.
Validation of this mechanism includes data showing that chemical
inhibition of AMPK negates the ability of metformin to inhibit HGP
in hepatocytes6, and hepatic knockout of Stk11 (encoding Lkb1)
abolishes the ability of chronic metformin to activate hepatic Ampk
and lower plasma glucose levels in diabetic rodents5. However, that
the glucose-lowering effect of acute oral metformin is intact in mice
that lack Ampk in the liver1, and that only high doses of metformin
inhibit hepatic complex I respiration3, challenge the originally postulated underlying mechanism. Specifically, recent studies report that
metformin inhibits HGP through a hepatic AMPK–independent
mechanism, either by negating the ability of glucagon to increase
hepatic cAMP levels and stimulate HGP4 or by decreasing hepatic
mitochondrial redox states and lowering the conversion of metabolites to glucose3. Additionally, chronic metformin treatment reduces
hepatic lipogenesis and subsequent lipid accumulation to improve
insulin sensitivity2. These findings indicate that the underlying
mechanisms responsible for the HGP- and glucose-lowering effects
of metformin in type 2 diabetes remain unclear.
The gastrointestinal tract triggers negative feedback systems
to maintain glucose homeostasis7. Specifically, duodenal lipids
activate a duodenal cholecystokinin (CCK)-1 receptor– and gut–
brain–liver-dependent neuronal network to lower HGP in healthy
rodents, and jejunal nutrient- and leptin-sensing lower HGP in
both healthy and diabetic rodents15–18. The pharmacological nature
of such sensory mechanisms, however, is completely unknown. To
this end, some early evidence indicates a potential role of the gut
in mediating the effects of metformin. For example, intraduodenal
(as compared to intraportal and intravenous) administration of
metformin leads to the greatest drop of plasma glucose levels19, and
chronic metformin administration increases Glp-1 secretion and
alters the microbiota profile 20–22. Furthermore, Ampk is expressed
in the intestine, and metformin increases intestinal Ampk activity23.
We hypothesize that preabsorptive metformin activates both a
duodenal Ampk–dependent mechanism and a neuronal relay to
lower HGP and plasma glucose levels in individuals with diabetes
or obesity (Fig. 1a).
1Toronto
General Research Institute and Department of Medicine, University Health Network, Toronto, Ontario, Canada. 2Department of Physiology, University
of Toronto, Toronto, Ontario, Canada. 3Department of Medicine, University of Toronto, Toronto, Ontario, Canada. 4Section of Cell Biology, Division of
Diabetes, Endocrinology and Metabolism, Department of Medicine, Imperial College of London, Imperial Centre for Translational and Experimental Medicine,
Hammersmith Hospital, London, UK. 5Banting and Best Diabetes Centre, University of Toronto, Toronto, Ontario, Canada. Correspondence should be addressed to
T.K.T.L. ([email protected]).
Received 16 August 2014; accepted 12 December 2014; published online 6 April 2015; corrected after print 7 May 2015; doi:10.1038/nm.3787
506
VOLUME 21 | NUMBER 5 | MAY 2015 nature medicine
letters
Ampk
GP
Day 1 2
Duodenum
Clamp
Duodenal
HFD
and
i.v. catheterization
0 min
90
150
200
3
–1
[3- H]Glucose (0.4 µCi min )
Insulin (1.2 mU kg–1 min–1)
SRIF (3 µg kg
–1
–1
min )
Basal
Clamp
d 16
8
7
***
6
5
4
3
2
1
0
14
12
10
8
***
6
4
2
0
Figure 1 Intraduodenal metformin infusion activates
Saline
Metformin
Saline Metformin Portal
Portal
metformin
metformin
duodenal Ampk and lowers HGP in the preabsorptive
Glucose (as needed)
state. (a) Schematic representation of the working
Intraduodenal infusion
hypothesis. (b) Experimental procedure and
pancreatic (basal insulin)–euglycemic clamp protocol.
2.0
SRIF, somatostatin. (c–e) The rates of glucose infusion (c),
16
1.5
2.0
**
HGP (d), and glucose uptake (e) during pancreatic
*
14
clamps of HFD-fed rats infused with intraduodenal
12
1.5
1.0
saline (n = 7), intraduodenal metformin (n = 6), or
10
1.0
0.5
portal vein metformin (n = 5). Basal, the average HGP
8
0.5
over the period from 60 to 90 min during the pancreatic
6
0
Saline Metformin
(basal insulin)–euglycemic clamp; clamp, the average
4
0.2
pAmpk
HGP from 180 to 200 min. (f,g) A representative western
2
0.1
tAmpk
blot (n = 2) and a quantitative analysis of pAmpk protein
0
Saline Metformin Portal
0
expression normalized to tAmpk protein expression (f) and
Saline Metformin
Saline Metformin
metformin
Ampk activity (g) in the duodenal mucosa of HFD-fed rats
infused for 50 min with intraduodenal saline or metformin. *P < 0.05, **P < 0.01; calculated by unpaired t-test; n = 6 for each group. Values are
shown as mean ± s.e.m. Unless noted, ***P < 0.001 versus all other groups as determined by ANOVA with Tukey’s post hoc test.
We first examined the effects of intraduodenal metformin infusion
on HGP in 3 d hyperphagic HFD (lard-oil enriched; Supplementary
Table 1)-induced insulin-resistant rats (the experimental model is
outlined in Fig. 1b, and the insulin-resistance phenotype was confirmed (data not shown)). Given that metformin lowers HGP independently of insulin action1,4, we chose to assess changes in glucose
metabolism using the pancreatic (basal insulin)–euglycemic clamp to
establish our studies in a non–insulin stimulated environment. When
metformin (200 mg kg−1) was infused into the duodenal lumen of rats
for only 50 min, the glucose infusion rate required to maintain eugly­
cemia was higher than in saline-infused control rats (Fig. 1c), and
HGP was reduced from its basal level in metformin-treated but not
saline-infused controls (Fig. 1d and Supplementary Fig. 1a). Glucose
uptake (Fig. 1e) and plasma insulin and glucose levels remained comparable among groups (Supplementary Table 2). A portal vein infusion of the same dose of metformin for 50 min did not raise the glucose
infusion rate (Fig. 1c) or lower HGP compared to saline infusion
(Fig. 1d and Supplementary Fig. 1a), indicating that the HGPlowering effect of intraduodenal metformin infusion is preabsorptive.
It is possible that differences in experimental design and conditions
(such as chronic versus acute metformin treatment, slow constant
infusion versus bolus injection of metformin, pancreatic clamp versus
unclamp studies and various rodent fasting periods) between our study
and previous studies1–5 may have accounted for the observed lack of a
direct hepatic effect of metformin when delivered by portal infusion.
Nonetheless, to strengthen our claim about the preabsorptive effects
of metformin, we repeated the infusion-clamp studies using a lower
and clinically relevant dose of metformin. Intraduodenal infusion
of metformin at 50 mg kg−1 resulted in a higher glucose infusion
rate (Supplementary Fig. 1b) and lower HGP (Supplementary
Fig. 1c,d) with no difference in glucose uptake (Supplementary Fig. 1e),
as compared to saline infusion. Thus, intraduodenal metformin
administration activates preabsorptive mechanisms to lower HGP.
We found that the duodenal mucosa of metformin-treated rats
had a higher ratio of phosphorylated Ampk (pAmpk) to total Ampk
nature medicine VOLUME 21 | NUMBER 5 | MAY 2015
g
Duodenal mucosal
Ampk activity (pmol min–1)
f
Duodenal mucosal
pAmpk:tAmpk
Glucose uptake
(mg kg–1 min–1)
e
© 2015 Nature America, Inc. All rights reserved.
c
5
HGP (mg kg–1 min–1)
Metformin
b
Duodenal
mucosa
Glucose infusion rate
(mg kg–1 min–1)
a
(tAmpk) (Fig. 1f) and higher Ampk activity (Fig. 1g) than that of
saline-treated rats. We then investigated whether duodenal Ampk
activation per se lowers HGP by using an Ampk activator, A769662.
The glucose infusion rate was higher (Supplementary Fig. 1b)
and HGP was lower following intraduodenal infusion of A769662
(3 mg kg−1) compared to those in saline-infused controls
(Supplementary Fig. 1c,d), but glucose uptake was unchanged
between all groups (Supplementary Fig. 1e). Furthermore, we
co-infused metformin with compound C, an Ampk inhibitor.
Intraduodenal compound C alone did not affect glucose metabolism
(compared to intraduodenal saline), but co-infusion with metformin
fully negated the higher glucose infusion rate (Supplementary Fig. 1f)
and lower HGP (Supplementary Fig. 1g,h) observed with metformin
infusion alone, without changing glucose uptake between groups
(Supplementary Fig. 1i).
However, as compound C is a relatively poor specific inhibitor of
AMPK24, we alternatively confirmed the role of duodenal AMPK
by utilizing an adenovirus encoding the dominant negative–acting
Ampk-mutated protein (Ad-dn-Ampk) (Fig. 2a). We first confirmed
the functionality of this virus by using HEK293 human embryonic
kidney cells infected with either Ad-dn-Ampk or adenovirus encoding
green fluorescent protein (Ad-GFP; control), and assessed protein
expression of acetyl-CoA carboxylase (ACC), a direct downstream
target of AMPK2. Among the cells infected with Ad-GFP, those treated
with metformin (10 mM) for 6 h (as originally described in ref. 25)
exhibited a higher ratio of phosphorylated ACC (pACC) to total
ACC than those treated with saline, and this effect was negated in
Ad-dn-Ampk cells treated with metformin (Supplementary Fig. 2a).
Next, using a procedure that maximized duodenal infection without
infecting the jejunum, ileum or liver, we injected either Ad-dn-Ampk
or Ad-GFP into a 5 cm ligated section of the duodenum of HFD-fed
rats. Intraduodenal infusion of metformin resulted in a higher glucose
infusion rate (Fig. 2b) and lower HGP (Fig. 2c and Supplementary
Fig. 2b) without affecting glucose uptake (Supplementary Fig. 2c)
in Ad-GFP–injected control rats as compared to rats given a saline
507
letters
b
MK-801
Tetracaine
Day 1
6
NTS
Duodenal, HFD Clamp
IV &
cannulation
HVAG/sham
surgeries
NTS
Vagal
terminals
Hepatic
vagotomy
0 min
90
150
200
[3-3H]Glucose (0.4 µCi min–1)
Metformin
NTS infusion
+
–
+
–
–
+
+
–
–
+
–
+
+
–
–
+
14
12
10
***
8
6
4
2
0
Gut saline
Gut metformin
NTS saline
NTS MK-801
+
–
+
–
–
+
+
–
–
+
–
+
+
–
–
+
+
–
–
–
+
–
+
–
–
+
+
–
–
–
+
–
+
–
–
+
1.5
*
1.0
0.5
0
rm
in
e
ex tfo
en rm
di in
n9
–
+
–
+
–
–
+
–
+
–
+
+
–
–
+
–
+
–
–
+
–
fo
–
+
+
–
–
–
+
+
–
–
et
+
–
+
–
–
+
–
+
–
–
M
–
+
–
–
–
d
Basal
Clamp
16
7
***
14
6
5
4
3
2
12
10
***
8
6
4
2
Saline
Metformin
Tetracaine
g
Basal
Clamp
–
+
–
–
–
+
–
–
–
–
8
Glucose infusion rate
(mg kg–1 min–1)
HGP (mg kg–1 min–1)
***
4
3
2
1
0
Gut saline
Gut metformin
NTS saline
NTS MK-801
16
***
c
0
lntraduodenal infusion
Saline
Metformin
MK-329
Exendin-9
Rp-CAMPS
duodenal mucosa was lower in Ad-dn-Ampk–injected rats than
in Ad-GFP–injected rats (Supplementary Fig. 2d). These data
collectively indicate that duodenal AMPK activation is required for
preabsorptive metformin to lower HGP.
SRIF (3 µg kg–1 min–1)
Glucose (as needed)
f
8
7
6
5
+
–
–
–
–
1
Duodenum
+
–
f
***
Insulin (1.2 mU kg–1 min–1)
GP
e
10
–
+
Basal
Clamp
16
14
12
10
8
6
4
2
0
Saline
Metformin
MK-329
Exendin-9
Rp-CAMPS
7
–
+
lin
e
HGP (mg kg
–1
–1
min )
e
infusion, but we found no difference in the glucose infusion rate,
HPG or glucose uptake between Ad-dn-Ampk–injected rats infused
with metformin as compared to saline (Fig. 2b,c and Supplementary
Fig. 2b,c). Following metformin treatment, Ampk activity in the
a
+
–
Saline
Metformin
M
–
+
Sa
+
–
***
Phosphorylated/
unphosphorylated A1
peptide (arbitrary intensity)
–1
min )
–
+
***
HGP (mg kg–1 min–1)
+
–
Figure 2 A duodenal AMPK–GLP-1R–PKA signaling pathway is
required for metformin to lower HGP. (a) Schematic representation
of the working hypothesis. (b,c) The glucose infusion rate (b) and rate
of HGP (c) during pancreatic clamps of HFD-fed rats infected with
either duodenal GFP or dn-Ampk and infused with intraduodenal
saline (n = 5 each group) or metformin (n = 7 each group).
(d,e) The glucose infusion rate (d) and rate of HGP (e) during
pancreatic clamps of HFD-fed rats administered intraduodenal
MK-329, exendin-9, or Rp-CAMPS with saline (n = 5 for each)
or in combination with metformin (n = 6 for each). (f) Quantification
of duodenal Pka activity in clamped tissue of HFD-fed rats treated
with saline, metformin, or metformin and exendin-9 (n = 5 for each
group). Values are shown as mean ± s.e.m. *P < 0.05, ***P < 0.001
versus all other groups; calculated by one-way ANOVA with
Tukey’s post hoc test.
Glucose infusion rate
(mg kg–1 min–1)
© 2015 Nature America, Inc. All rights reserved.
Saline
Metformin
***
10
9
8
7
6
5
4
3
2
1
0
+
–
–
–
+
–
–
+
+
6
0
Saline
Metformin
Tetracaine
h
8
7
+
–
+
***
5
4
3
2
1
0
+
–
–
–
+
–
–
+
+
+
–
+
Basal
Clamp
16
14
12
10
8
***
6
4
2
0
Sa
sh line
am
M
et
fo
sh rmin
am
M
et
fo
H rm
VA in
G
GP
Duodenum
14
12
10
8
6
4
2
0
dn-AMPK
HGP (mg kg–1 min–1)
Metformin
d
Basal
Clamp
GFP
16
Glucose infusion rate
(mg kg–1 min–1)
PKA
***
–1
GLP-1R
?
CCK-1R
Ampk
Vagal
terminals
dn-AMPK
HGP (mg kg
dn-AMPK
8
7
6
5
4
3
2
1
0
Glucose infusion rate
(mg kg–1 min–1)
Exendin-9
Rp-CAMPS
MK-329
c
GFP
Sa
sh line
am
M
et
fo
sh rmin
am
M
et
fo
H rm
VA in
G
b
Glucose infusion rate
(mg kg–1 min–1)
a
Figure 3 A gut–brain–liver neuronal axis is required for the HGP-lowering effect of metformin. (a) Schematic representation of the working hypothesis.
Duodenal metformin triggers the afferent nerve terminals in the duodenum and signals via NMDA receptors in the NTS, which in turn lowers HGP via
the hepatic vagus. SRIF, somatostatin. (b) Experimental procedure and pancreatic (basal insulin)–euglycemic clamp protocol with NTS infusion.
(c,d) The glucose infusion rate (c) and rate of HGP (d) in HFD-fed rats infused with intraduodenal tetracaine alone (n = 5) or in combination with
metformin (n = 6). (e,f) The glucose infusion rate (e) and rate of HGP (f) in HFD-fed rats infused with intraduodenal saline or metformin and NTS saline
(n = 4 for saline, 5 for metformin) or MK-801 (n = 5 for saline, 6 for metformin). (g,h) The glucose infusion rate (g) and rate of HGP (h) in HFD-fed rats
infused with intraduodenal metformin after either a sham surgery (n = 5) or HVAG (n = 6). Values are shown as mean ± s.e.m. ***P < 0.001 versus all
other groups; calculated by one-way ANOVA with Tukey’s post hoc test.
508
VOLUME 21 | NUMBER 5 | MAY 2015 nature medicine
HFD
0 min
24
Duodenal & IV
catheterization
90
200
[3-3H]Glucose (0.4 µCi min–1)
–1
SRIF (3 µg kg
–1
min )
–1
min )
c
8
Clamp
150
Insulin (1.2 mU kg
b
28
–1
28 d
7
***
6
5
4
3
2
1
0
Glucose (as needed)
Saline
Metformin
d
28 d
18
16
14
12
10
8
6
4
2
0
Basal
Clamp
***
Saline
Plasma glucose (mg dL–1)
Day 1
HGP (mg kg–1 min–1)
a
demonstrate that metformin requires a duodenal Ampk–Glp1r–Pka
signaling pathway to lower HGP.
We next examined whether a gut–brain–liver axis mediates the
effect of metformin using three different techniques: intraduodenal infusion of tetracaine, infusion of the NMDA receptor blocker
MK-801 into the nucleus of the solitary tract (NTS) and hepatic
vagotomy (Fig. 3a,b). In 3 d HFD–fed rats, we first administered,
with or without metformin, the local anesthetic tetracaine, in order
to inhibit the neurotransmission of afferent fibers innervating the
duodenum. Tetracaine infusion alone did not affect glucose metabo­
lism compared to saline infusion (Fig. 3c,d and Supplementary
Fig. 3a,b), but it reversed the ability of metformin to alter glucose
metabolism (Fig. 3c,d and Supplementary Fig. 3a). Afferent vagal
signaling synapses at the level of the NTS and activation of NMDA
receptors in the NTS mediate gut nutrient sensing to lower HGP15.
Here we saw that direct infusion of MK-801, as compared to saline,
into the NTS did not by itself affect glucose metabolism (Fig. 3e,f
and Supplementary Fig. 3c,d), but it did attenuate the higher glucose
infusion rate (Fig. 3e) and lower HGP (Fig. 3f and Supplementary
Fig. 3c) caused by metformin infusion without affecting glucose
uptake (Supplementary Fig. 3d). We repeated these studies in rats
that underwent hepatic vagal branch vagotomy (HVAG), thereby
eliminating the neurocommunication between the brain and liver.
HVAG abolished the ability of duodenal metformin to alter glucose
metabolism in rats (Fig. 3g,h and Supplementary Fig. 3e,f) as it
did in sham-operated rats, but HVAG alone did not affect glucose
metabolism. Thus, duodenal metformin activates a gut–brain–liver
axis to inhibit HGP.
150
***
100
50
0
Metformin
Saline
5
i.p. injection:
nicotinamide (170 mg/kg)
STZ (65 mg/kg)
Gastric
catheterization
HFD
9–10 10–11
103 kcal Gastric
food
bolus
experiment
0 min
180
Plasma glucose measurements
Gastric bolus metformin (200 mg/kg)
200
175
150
125
100
75
50
25
0
Duo-GFP
Duo-dn-Ampk
*
*,#
***,#
0
60
180
Time after metformin bolus
(min)
NA-STZ/HFD
h
16
***
12
8
4
0
Metformin
Saline
Metformin
NA-STZ/HFD
Duo-GFP
Duo-dn-Ampk
10
% suppression from
baseline plasma glucose
g
Plasma glucose (mg dL–1)
f
Basal
Infusion
20
NA-STZ/HFD
Intraduodenal infusion
Day 1
e
Basal
Infusion
200
HGP (mg kg–1 min–1)
We next chose to specifically examine the roles of Cck and Glp-1,
as duodenal Cck signaling triggers a gut–brain–liver axis to lower
HGP15 and metformin stimulates Glp-1 release20. Thus, we infused
metformin intraduodenally with either the Cck-1 receptor antagonist
MK-329 or the Glp-1 receptor (Glp1r) antagonist exendin-9 (Ex9)
(Fig. 2a). HFD-fed rats co-infused with metformin and MK-329 had
a higher glucose infusion rate (Fig. 2d) and lower HGP (Fig. 2e and
Supplementary Fig. 2e) than saline-infused rats, but co-infusion with
Ex9 reversed the ability of metformin to alter glucose metabolism
(Fig. 2d,e and Supplementary Fig. 2e,f). Ex9 or MK-329 alone had
no effect on glucose metabolism (Fig. 2d,e and Supplementary
Fig. 2e,f). Thus, intestinal Glp-1 signaling, but not Cck-1 signaling,
is required for metformin’s action.
Given that Cck lowers HGP via Cck-1–receptor activation of Pka
signaling on vagal afferents26, Glp1rs are localized on vagal afferents
innervating the small intestine27, and Glp1r activation induces Pka
signaling28, we examined whether the effect of metformin requires
Pka signaling (Fig. 2a). Even though infusion of the Pka chemical
inhibitor Rp-CAMPS by itself did not alter the glucose infusion
rate (Fig. 2d), HGP (Fig. 2e and Supplementary Fig. 2e) or glucose uptake (Supplementary Fig. 2f) compared to saline infusion,
intraduodenal co-infusion of metformin and Rp-CAMPS abolished
the higher glucose infusion rate (Fig. 2d) and lower HGP (Fig. 2e and
Supplementary Fig. 2e) observed with metformin infusion alone.
Duodenal Pka activity was higher following intraduodenal metformin
infusion than following saline infusion, and this activation was abolished by co-infusion with Ex9 (Fig. 2f), indicating that Glp1r activation
is necessary for metformin to increase Pka activity. These results
Glucose infusion rate
(mg kg–1 min–1)
© 2015 Nature America, Inc. All rights reserved.
letters
0
Time after metformin gastric bolus (min)
0
5
15
30
60
90
120
150
180
***
***
***
–10
–20
–30
***
***
***
–40
–50
NA-STZ/HFD
Figure 4 Intraduodenal infusion of metformin lowers HGP in obese and diabetic rats, and the overall acute glucose-lowering effect of a bolus intragastric
treatment of metformin is dependent on duodenal Ampk signaling. (a) Experimental procedure and pancreatic (basal insulin)–euglycemic clamp
protocol for 28 d HFD–fed rats. SRIF, somatostatin. (b,c) Glucose infusion rate (b) and rate of HGP (c) during a pancreatic clamp of 28 d HFD–fed rats
infused with intraduodenal saline or metformin. (d,e) Plasma glucose levels (d) and rate of HGP (e) in NA–STZ–HFD-induced hyperglycemic rats infused
with intraduodenal saline or metformin (n = 8 for each). In b–d, ***P < 0.001 versus saline as compared by unpaired t-test; n = 6 for each group.
(f) Experimental procedure and gastric infusion protocol. (g,h) Plasma glucose levels (g) and percent suppression of plasma glucose from basal levels (h)
in NA–STZ–HFD-induced hyperglycemic rats injected with either duodenal Ad-GFP or Ad-dn-Ampk (Duo-GFP and Duo-dn-Ampk, respectively) 5 d prior
to treatment with a gastric bolus of metformin (n = 8 for each group). In g, *P < 0.05 and ***P < 0.05 compared to baseline; #P < 0.05 compared to
Duo-dn-Ampk within time points as calculated by two-way ANOVA with Tukey’s post hoc test. In h, ***P < 0.01 versus saline as compared by unpaired
t-test within each time point. Values are shown as mean ± s.e.m.
nature medicine VOLUME 21 | NUMBER 5 | MAY 2015
509
© 2015 Nature America, Inc. All rights reserved.
letters
To determine the therapeutic potential of metformin action
in the duodenum, we infused metformin intraduodenally in rat
models of obesity and type 2 diabetes. First we tested the effects of
pre­absorptive metformin in a chronic (28 d) hyperphagic HFD–fed
obese rat model (Fig. 4a), which is characterized by increased fat mass
and insulin resistance (data not shown). Compared to saline, intra­
duodenal metformin still resulted in a higher glucose infusion rate
(Fig. 4b) and lower HGP (Fig. 4c and Supplementary Fig. 4a)
without affecting glucose uptake (Supplementary Fig. 4b) in this
obese rat model. For the type 2 diabetic model, we injected rats
with nicotinamide (NA) and streptozotocin (STZ) and fed them an
HFD for 5–6 d, inducing mild hyperglycemia and increased HGP
(Supplementary Fig. 4c,d), but not insulin deficiency, as described29
(data not shown). In unclamped conditions, NA–STZ-treated,
HFD-fed type 2 diabetic rats given a 50 min duodenal infusion of metformin exhibited lower plasma glucose levels (Fig. 4d) and lower HGP
(Fig. 4e) compared to rats given saline infusion. Thus, pre­absorptive
duodenal metformin is sufficient to lower plasma glucose levels and
HGP in rat models of obesity and diabetes.
Last, to assess the contribution of duodenal Ampk in the acute
glucose-lowering effect of metformin in an unclamped setting, we
administered an intragastric bolus dose of metformin (200 mg/kg)
to NA–STZ-treated, HFD-fed type 2 diabetic rats that had received
an intraduodenal injection of either Ad-dn-Ampk or Ad-GFP (as
described above) (Fig. 4f). After 60 min, metformin decreased plasma
glucose levels, as compared to baseline, in Ad-GFP–treated rats but
not in Ad-dn-Ampk–treated rats (Fig. 4g). By 180 min after treatment, although metformin decreased plasma glucose levels in both
groups, they were still higher in the Ad-dn–Ampk treated rats than
in the Ad-GFP–treated rats (Fig. 4g). In line with this, suppression
of plasma glucose levels was greater in Ad-GFP–treated rats than in
Ad-dn-Ampk–treated rats, and this difference started as early as 30 min
after gastric bolus and was still present 180 min later (Fig. 4h). Thus,
activation of intestinal Ampk contributes to the rapid, overall ability
of metformin to lower plasma glucose levels in diabetes.
Metformin is currently the most widely prescribed treatment for
type 2 diabetes30. Recently, metformin and other biguanides have
been documented to reduce cancer risk31,32 and increase lifespan in
rodents33, which furthers the need and the urgency to elucidate their
modes of action. Adding to and complementing the known direct
hepatic actions of metformin1–6, we demonstrate that preabsorptive metformin remotely (indirectly and acutely) lowers HGP via
activation of duodenal Ampk and a neuronal network.
Our findings do not rule out the possibility that metformin directly
inhibits HGP through the aforementioned mechanisms1,3–6, as metformin still sufficiently lowers plasma glucose levels in rats 90 min
after an intragastric bolus regardless of whether duodenal Ampk
sensing is disrupted. However, we discovered here that metformin
initially (within the first 60 min) reduces plasma glucose levels only
in rats with intact duodenal Ampk signaling. Furthermore, the initial
drop in glucose levels via duodenal Ampk activation has a major and
sustained contribution to the overall suppression of plasma glucose
levels, highlighting the fact that an intestinal metformin–Ampkdependent pathway that lowers blood glucose levels may have been
overlooked by previous studies.
For example, it is possible that intestinal Ampk mediates the ability
of metformin to rapidly (as early as 20 min after administration) lower
blood glucose levels in mice lacking hepatic Ampk, an ability that was
originally hypothesized to be due to Ampk-independent decreases
in the hepatic energy state1. Furthermore, although a rapid reduction
510
(60 min after metformin treatment) in plasma glucose levels
mirrors a change in plasma lactate/pyruvate ratios following intravenous (i.v.) metformin administration3, there is no evidence that
metformin acts directly in the liver to alter hepatocellular redox state
(that is, an experiment that negates hepatic uptake of metformin was
not performed in vivo). In fact, studies have shown that metformin
accumulates mostly in the gastrointestinal tract after only 30 min of
i.v. metformin administration34. In light of these findings, we propose
that the acute glucose-lowering effect of metformin is largely mediated by intestinal mechanisms, whereas the relatively long-term or
chronic effects of metformin treatment are likely mediated by a direct
hepatic mechanism. For example, a recent report documented that the
lipid-lowering and insulin-sensitizing effects of chronic metformin
treatment were dependent upon the ability of metformin to increase
hepatic Ampk activity and inhibit Acc2.
Preabsorptive metformin’s action is dependent on duodenal Glp1r,
consistent with the view that metformin increases Glp-1 levels20,22.
Although this remains to be tested, it is likely that metformin acts
directly on enteroendocrine L cells localized in the duodenum35 to
stimulate the local release of Glp-1. In light of the necessary upstream
role of duodenal Ampk and the fact that AICAR, an Ampk agonist,
also increases circulating Glp-1 levels20,36, future studies are needed to
elucidate the underlying mechanistic links of Ampk to Glp-1 release.
In summary, we demonstrate that metformin activates a previously
unappreciated duodenal Ampk–Glp-1R–Pka-dependent neuronal
pathway to lower HGP and plasma glucose levels in rat models of
obesity and diabetes, and that activation of duodenal Ampk substantially contributes to the overall, acute glucose-lowering effect
of metformin. Our findings lay the groundwork for the potential
development of specific gut-targeted treatments that could activate
intestinal energy sensor proteins, such as Ampk, to lower HGP and
improve glycemia in individuals with diabetes and obesity.
Methods
Methods and any associated references are available in the online
version of the paper.
Note: Any Supplementary Information and Source Data files are available in the
online version of the paper.
Acknowledgments
The authors are grateful to E. Burdett for technical assistance. This work is
supported by a research grant from the Canadian Institute of Health Research
(MOP-82701 to T.T.K.L.). F.A.D. is a Banting Fellow. B.A.R. is supported by a
Canadian Institute of Health Research Doctoral Vanier Canada scholarship. C.D.C.
is supported by a Banting and Best Diabetes Centre graduate studentship. M.Z.-T.
is supported by a Banting and Best Diabetes Centre graduate studentship. G.A.R.
is supported by the Wellcome Trust Senior Investigator (WT098424AIA), the
Medical Research Council Programme (MR/J0003042/1), the Diabetes UK Project
Grant (11/0004210) and Royal Society Wolfson Research Merit awards. T.K.T.L.
holds the John Kitson McIvor (1915–1942) Endowed Chair in Diabetes Research
and the Canada Research Chair in Obesity at the Toronto General Research
Institute and the University of Toronto.
AUTHOR CONTRIBUTIONS
F.A.D. conducted and designed experiments, performed data analyses and wrote
the manuscript. B.A.R., C.D.C., M.Z.-T. and B.M.F. assisted with experiments.
G.A.R. provided the adenovirus expressing dn-Ampk. T.K.T.L. supervised the
project, designed experiments and edited the manuscript.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
VOLUME 21 | NUMBER 5 | MAY 2015 nature medicine
© 2015 Nature America, Inc. All rights reserved.
letters
1. Foretz, M. et al. Metformin inhibits hepatic gluconeogenesis in mice independently
of the LKB1/AMPK pathway via a decrease in hepatic energy state. J. Clin. Invest.
120, 2355–2369 (2010).
2. Fullerton, M.D. et al. Single phosphorylation sites in Acc1 and Acc2 regulate lipid
homeostasis and the insulin-sensitizing effects of metformin. Nat. Med. 19,
1649–1654 (2013).
3. Madiraju, A.K. et al. Metformin suppresses gluconeogenesis by inhibiting
mitochondrial glycerophosphate dehydrogenase. Nature 510, 542–546 (2014).
4. Miller, R.A. et al. Biguanides suppress hepatic glucagon signalling by decreasing
production of cyclic AMP. Nature 494, 256–260 (2013).
5. Shaw, R.J. et al. The kinase LKB1 mediates glucose homeostasis in liver and
therapeutic effects of metformin. Science 310, 1642–1646 (2005).
6. Zhou, G. et al. Role of AMP-activated protein kinase in mechanism of metformin
action. J. Clin. Invest. 108, 1167–1174 (2001).
7. Lam, T.K. Neuronal regulation of homeostasis by nutrient sensing. Nat. Med. 16,
392–395 (2010).
8. Taylor, S.I. Deconstructing type 2 diabetes. Cell 97, 9–12 (1999).
9. Hundal, R.S. et al. Mechanism by which metformin reduces glucose production in
type 2 diabetes. Diabetes 49, 2063–2069 (2000).
10.Radziuk, J., Zhang, Z., Wiernsperger, N. & Pye, S. Effects of metformin on lactate
uptake and gluconeogenesis in the perfused rat liver. Diabetes 46, 1406–1413
(1997).
11.Stumvoll, M., Nurjhan, N., Perriello, G., Dailey, G. & Gerich, J.E. Metabolic effects
of metformin in non-insulin-dependent diabetes mellitus. N. Engl. J. Med. 333,
550–554 (1995).
12.Salpeter, S.R., Buckley, N.S., Kahn, J.A. & Salpeter, E.E. Meta-analysis: metformin
treatment in persons at risk for diabetes mellitus. Am. J. Med. 121, 149 e2–157
e2 (2008).
13.Owen, M.R., Doran, E. & Halestrap, A.P. Evidence that metformin exerts its antidiabetic effects through inhibition of complex 1 of the mitochondrial respiratory
chain. Biochem. J. 348, 607–614 (2000).
14.Hawley, S.A. et al. Use of cells expressing gamma subunit variants to identify
diverse mechanisms of AMPK activation. Cell Metab. 11, 554–565 (2010).
15.Wang, P.Y. et al. Upper intestinal lipids trigger a gut–brain–liver axis to regulate
glucose production. Nature 452, 1012–1016 (2008).
16.Cheung, G.W., Kokorovic, A., Lam, C.K., Chari, M. & Lam, T.K. Intestinal
cholecystokinin controls glucose production through a neuronal network. Cell Metab.
10, 99–109 (2009).
17.Rasmussen, B.A. et al. Jejunal leptin-PI3K signaling lowers glucose production.
Cell Metab. 19, 155–161 (2014).
18.Breen, D.M. et al. Jejunal nutrient sensing is required for duodenal-jejunal bypass
surgery to rapidly lower glucose concentrations in uncontrolled diabetes. Nat. Med.
18, 950–955 (2012).
19.Stepensky, D., Friedman, M., Raz, I. & Hoffman, A. Pharmacokinetic-pharmacodynamic
analysis of the glucose-lowering effect of metformin in diabetic rats reveals first-pass
pharmacodynamic effect. Drug Metab. Dispos. 30, 861–868 (2002).
nature medicine VOLUME 21 | NUMBER 5 | MAY 2015
20.Maida, A., Lamont, B.J., Cao, X. & Drucker, D.J. Metformin regulates the incretin
receptor axis via a pathway dependent on peroxisome proliferator-activated receptoralpha in mice. Diabetologia 54, 339–349 (2011).
21.Shin, N.R. et al. An increase in the Akkermansia spp. population induced by
metformin treatment improves glucose homeostasis in diet-induced obese mice.
Gut 63, 727–735 (2014).
22.Vardarli, I., Arndt, E., Deacon, C.F., Holst, J.J. & Nauck, M.A. Effects of
sitagliptin and metformin treatment on incretin hormone and insulin secretory
responses to oral and ″isoglycemic″ intravenous glucose. Diabetes 63, 663–674
(2014).
23.Harmel, E. et al. AMPK in the small intestine in normal and pathophysiological
conditions. Endocrinology 155, 873–888 (2014).
24.Bain, J. et al. The selectivity of protein kinase inhibitors: a further update. Biochem.
J. 408, 297–315 (2007).
25.He, G. et al. AMP-activated protein kinase induces p53 by phosphorylating MDMX
and inhibiting its activity. Mol. Cell. Biol. 34, 148–157 (2014).
26.Rasmussen, B.A. et al. Duodenal activation of cAMP-dependent protein kinase
induces vagal afferent firing and lowers glucose production in rats. Gastroenterology
142, 834 e3–843.e3 (2012).
27.Richards, P. et al. Identification and characterization of GLP-1 receptorexpressing cells using a new transgenic mouse model. Diabetes 63, 1224–1233
(2014).
28.Yusta, B. et al. GLP-1 receptor activation improves beta cell function and survival
following induction of endoplasmic reticulum stress. Cell Metab. 4, 391–406
(2006).
29.Samuel, V.T. et al. Fasting hyperglycemia is not associated with increased expression
of PEPCK or G6Pc in patients with Type 2 Diabetes. Proc. Natl. Acad. Sci. USA
106, 12121–12126 (2009).
30.Viollet, B. et al. Cellular and molecular mechanisms of metformin: an overview.
Clin. Sci. (Lond.) 122, 253–270 (2012).
31.Shackelford, D.B. et al. LKB1 inactivation dictates therapeutic response of nonsmall cell lung cancer to the metabolism drug phenformin. Cancer Cell 23,
143–158 (2013).
32.Pollak, M. Overcoming drug development bottlenecks with repurposing: repurposing
biguanides to target energy metabolism for cancer treatment. Nat. Med. 20,
591–593 (2014).
33.Martin-Montalvo, A. et al. Metformin improves healthspan and lifespan in mice.
Nat. Commun. 4, 2192 (2013).
34.Wilcock, C. & Bailey, C.J. Accumulation of metformin by tissues of the normal and
diabetic mouse. Xenobiotica 24, 49–57 (1994).
35.Habib, A.M. et al. Overlap of endocrine hormone expression in the mouse intestine
revealed by transcriptional profiling and flow cytometry. Endocrinology 153,
3054–3065 (2012).
36.Mulherin, A.J. et al. Mechanisms underlying metformin-induced secretion of
glucagon-like peptide-1 from the intestinal L cell. Endocrinology 152, 4610–4619
(2011).
511
ONLINE METHODS
Animal preparation. All animal protocols were reviewed and approved by
the Toronto General and Western Animal Care Committee at the University
Health Network. A total of 170 8-week-old male Sprague-Dawley rats
(250–270 g) were obtained from Charles River Laboratories (Montreal, QC,
Canada) and maintained on a 12 h light-dark cycle, with access to chow and
water ad libitum. Except as noted below, no animals were excluded from
analysis, and rats were randomly assigned into the various diet and surgical
groups detailed below.
© 2015 Nature America, Inc. All rights reserved.
3 d HFD–induced model of insulin resistance. We placed rats on a lardoil–enriched diet (HFD; TestDiet #571R, Purina Mills, IN, USA) (Supplementary
Table 1) for 3 d, starting 1 d after duodenal and vascular cannulations. Rats
that overate on this diet for 3 d developed hypothalamic37 and hepatic insulin
resistance38. Rats that did not overeat were excluded from the study.
4 week HFD–induced model of insulin-resistant obesity. We placed rats on
lard-oil–enriched HFD for 28 d. We performed duodenal and vascular cannulations on day 24 after placement on the HFD and performed the pancreatic
(basal insulin)–euglycemic clamp experiment on day 28 after recovery from
surgical procedures.
NA–STZ–HFD-induced model of type 2 diabetes. We treated rats with a
single injection of nicotinamide (170 mg kg−1 i.p.) followed 15 min later by
a single injection of streptozotocin (65 mg kg−1). 4–5 d later, we performed
duodenal and vascular cannulations and placed rats on a lard-oil enriched
HFD (described above) to induce insulin resistance. Rats that were extremely
hyperglycemic (blood glucose level >300 mg dl −1) were excluded from
the study. 5–6 d after surgery, we subjected rats to the basal [3- 3H]glucose
infusion protocol.
Animal surgery. We performed surgeries 4 d prior to the clamp experiments.
We placed a duodenal catheter 2 cm distal to the pyloric sphincter for infusion
purposes, while carotid artery and jugular vein cannulations were performed
for infusion and sampling during the clamp and non-clamp in vivo experiments.
We flushed the duodenal catheter daily with saline to ensure potency. A subgroup of rats received hepatic portal vein cannulations as described18. A separate group of rats received hepatic vagotomy procedures immediately prior to
duodenal cannulations15. A subgroup of rats underwent stereotaxic surgery
6 d prior to duodenal and vascular surgeries. We stereotactically implanted
a 26-gauge bilateral guide cannula made of stainless steel into the NTS
(0.0 mm on occipital crest, 0.4 mm lateral to midline, 7.9 mm below skull
surface) as previously described39. We monitored rats for recovery from surgery by measuring daily food intake and body weight. Rats that did not fully
recover were excluded from the study. Rats were randomly designated into
groups prior to experiment and no blinding was done during the experimental
procedures described below.
Virus injection. A randomized subset of rats also received a duodenal adenovirus injection prior to the insertion of the duodenal cannula and artery and
vein cannulations, as described previously40. Briefly, we tied off a 4 cm section
of the proximal duodenum (1 cm distal to the pyloric sphincter) with two silk
ligatures to prevent inflow of intestinal fluid and outflow of the virus, flushed
the intestinal section with saline, and administered 0.2 ml of the adenovirus
encoding either the dominant-negative Ampk protein (Ad-dn-Ampk (D157A);
3.1 × 10–9 PFU ml−1) or green fluorescent protein (Ad-GFP, 1.4 × 109 PFU
ml−1) described earlier41 via a 23-gauge needle directly into the duodenal
lumen. After 20 min, we inserted a duodenal catheter into the site of the virus
injection and vascular cannulations were performed as described above.
Pancreatic (basal insulin)–euglycemic clamp procedures. The night
prior to the clamp, we restricted rats to 103 kcal of the HFD. The clamp pro­
cedure was 200 min in duration and performed in unrestrained rats in vivo.
At the onset of the experiment (t = 0 min), a primed intravenous infusion of
[3-3H]glucose (PerkinElmer; 40 µCi bolus; 0.4 µCi min−1) was started and continued throughout the experiment (t = 200 min) in order to measure glucose
nature medicine
kinetics using the tracer-dilution methodology. A subset of rats received NTS
infusions, in which saline and MK-801 infusions were started at t = 90 min
and continued throughout the clamp (t = 200 min) experiment. At t = 90 min,
the clamp was started, in which somatostatin (3 µg kg−1 min−1) was infused
intravenously to inhibit endogenous insulin and glucagon secretion and an insulin infusion was simultaneously administered at a dose of 1.2 mU kg−1 min−1
for the pancreatic (basal insulin)–euglycemic clamp. Additionally, a variable 25% glucose infusion was started at t = 90 min and periodically adjusted
to maintain euglycemia (from t = 120 to t = 200 min). At t = 150 min, the
intraduodenal treatment (0.01 ml min−1) of treatments outlined below was
started and continued throughout the experiment (t = 200 min). Plasma
samples were obtained every 10 min to determine the specific activity of
[3-3H]glucose and measure insulin levels. At the end of the experiments, rats
were anesthetized and tissue samples were collected, immediately flash frozen
and stored at −80 °C until use.
Basal [3-3H]glucose infusion protocol (non-clamp conditions). The night
prior to the infusion studies, we restricted rats to 103 kcal of the HFD. These
studies were performed in the diabetic model described above. The infusion
experiment was 140 min in duration, and performed in unrestrained rats
in vivo. A primed continuous infusion of [3-3H]glucose was started at t = 0 min
and continued until t = 140 min. Duodenal infusions of treatments outlined
below were started at t = 90 min and continued throughout the experiment.
Plasma samples were collected every 10 min from t = 60 min to t = 90 min
and from t = 100 min and t = 140 min to determine plasma glucose levels and
[3-3H]glucose specific activity.
Treatments. The following treatments were infused into the lumen of the
duodenum during the in vivo experiments at a rate of 0.01 ml min−1 for
50 min: saline; metformin (200 mg kg−1 or 50 mg kg−1, Sigma-Aldrich,
St. Louis, MO, USA); A769662 (3 mg kg−1, Tocris Bioscience, Bristol, UK);
compound C (100 µM, Millipore); MK-329 (0.08 mg ml−1, Tocris Bioscience);
vi) Exendin-9 (15 µg ml−1, Tocris Bioscience); Rp-Camps (12 µM, Tocris
Bioscience); and tetracaine (0.01 mg min−1, Sigma-Aldrich). The 200 mg kg−1
metformin dose selected was modified from previous studies utilizing gavage of metformin at 250 mg kg−1 (refs. 1,4), accounting for the fact that
after 50 min, not all of that dose would have emptied into the duodenum. The
50 mg kg−1 dose of metformin was chosen based on the fact that a 50 mg kg−1
intragastric gavage of metformin in rats results in plasma metformin
concentrations (< 20 µM)42 similar to those observed following therapeutic oral treatments of metformin in humans (~3–23 µM)43. The following
treatments were infused into the NTS during the pancreatic (basal-insulin)–
euglycemic clamp experiments at a rate of 0.006 µl min−1: saline and MK-801
(0.03 ng min−1, Sigma-Aldrich)15.
Intragastric bolus protocol. A separate group of NA–STZ-treated, HFD-fed
rats were subject to the viral duodenal infection, fitted with a gastric cannula in
the fundic region of the stomach and switched to an HFD for 5 d. The night prior
to the experiment, we restricted rats to 103 kcal of HFD. For the experiment,
we infused a bolus of metformin (200 mg per kg of body weight; t = 0). Plasma
samples were collected at t = −5, 0, 5, 15, 30, 60, 90, 120, 150 and 180 min.
Tissue collection and preparation for western blotting. Immediately following termination of the experiments, the duodenal mucosa was separated from
the duodenal smooth muscle after removal from anesthetized rats, and the liver
was freeze clamped using steel tongs pre-cooled in liquid nitrogen. The tissues
were lysed on ice with a handheld homogenizer in a lysis buffer containing
50 mM Tris-HCl (pH 7.5), 1 mM EGTA, 1 mM EDTA, 1% (w/v) Nonidet P40,
1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM sodium
pyrophosphate, 0.27 M sucrose, 1 µM dithiothreitol (DTT) and 2× protease
inhibitor cocktail (Roche). The protein concentration of homogenized tissues
was determined using the Pierce 660 nm protein assay (Thermo Scientific).
Cell culture. We infected HEK293 cells with either Ad-GFP (50 µl in 10 ml
medium) or Ad-dn-Ampk (25 µL) described above, and the following day
treated the cells with either saline or metformin (10 mM; Sigma-Aldrich) for 6 h.
doi:10.1038/nm.3787
© 2015 Nature America, Inc. All rights reserved.
We then lysed the cells in a buffer containing 50 mM Tris-HCl (pH 7.5),
1 mM EGTA, 1 mM EDTA, 1% (w/v) Nonidet P40, 1 mM sodium orthovanadate,
50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose and
1 mM DTT, and determined the protein concentration using the Pierce
660 nm protein assay (Thermo Scientific). HEK293 cells were regularly tested
for mycoplasma infection. HEK293 cells were originally sourced from ATCC
(CRL-1573) but were not authenticated prior to their use in this study.
Western blotting. 20 µg of cell culture and 40 µg of tissues lysates (prepared as
described above) were subjected to electrophoresis on 10% acrylamide gels and
transferred to nitrocellulose membranes. The membranes were incubated for
1 h with blocking buffer (either TBS-T containing 5% (w/v) BSA or 5% skim
milk). The membranes were then incubated with the indicated primary antibodies (1:1,000): anti-pAmpk-α Thr172 (Cell Signaling, MA, USA, #2535s),
anti-Ampk-α (Santa Cruz Biotechnology, CA, USA, #sc25792), anti-pAcc (Cell
Signaling, #3661) and anti-Acc (Cell Signaling, #3676), diluted in the blocking
buffer for 16 h at 4 °C. The membranes were washed three times with TBS-T
and incubated with the appropriate secondary HRP-conjugated antibodies
(diluted 1:2,000 in 5% skim milk) at room temperature for 1 h. Finally, the
membranes were washed in TBS-T five times for 5 min each, and the signal
was detected using enhanced chemiluminescence reagent (Pierce, IL, USA).
Immunoblots were developed using a film automatic processor (SRX-101;
Konica Minolta Medical), and films were scanned with the GS-800 calibrated
densitometer (Bio-Rad). Protein levels were quantified by densitometry with
the Quantity One 1-D Analysis Software (Bio-Rad).
Ampk immunoprecipitation (IP) and kinase activity assay. We performed
IP and kinase activity assay as previously described44 with some modifications.
Duodenal mucosal protein (2 mg) was incubated with 20 µl of anti-Ampk
antibody (University of Dundee DSTT, Dundee, UK, #S525D) for 2 h at 4 °C,
and afterwards was incubated with 5 µl of A/G sepharose beads for 1 h at
4 °C. Following centrifugation and removal of supernatant, beads were washed
three times with lysis buffer (with the addition of 0.15 M NaCl) and two times
with buffer A (50 mM Tris-HCl (pH 7.5), 0.1 mM EGTA, 1 mM DTT). Next,
25 µl of buffer A with 0.2 µM AMARA peptide (AMARAASAAALARRR)45
was added to the sample, and kinase reaction was started by adding 25 µl
of a solution containing 10 mM magnesium acetate, 0.2 mM cold ATP and
1.66 µCi sample−1 of [γ-32P]ATP. Assays were carried out at 30 °C for 1 h. The
reaction was stopped with 2 µl of EDTA (0.5 M), and after centrifugation at
8,000 rpm for 3 min, 40 µL of the supernatant was spotted on P81 phosphocellulose papers and Laemmli sample buffer was added to the leftover pellet.
Papers were washed five times in orthophosphoric acid (75 mM) followed
by acetone, and the incorporated radioactivity was measured by scintillation
counting. Activity of the sample was expressed as pmol of γ-32P incorporated
into the peptide per minute. The amount of Ampk protein immunoprecipitated activity was normalized to relative amount of total-Ampk quantified by
western blot of Laemmli-treated pellet following the activity assay.
doi:10.1038/nm.3787
Pka activity assay. Pka activity from whole duodenal samples (5 µg of protein;
extracted according to manufacturer’s instructions and taken directly after the
clamp studies) was measured with the PepTag Assay Kit (Promega, Madison,
WI) as per manufacturer’s instructions with minor modifications26. Samples
were run on a 0.8% agarose gel at 120 V for 17 min. Data were analyzed
using ImageJ (National Institutes of Health software). The A1 peptide is
pho­sphorylated by Pka, and thus a higher ratio of pA1/A1 reflects a higher
degree of Pka activation.
Biochemical analysis. We measured plasma glucose concentrations using
the glucose oxidase method (GM9 Glucose Analyzer, Analox Instruments).
We measured plasma insulin levels by radioimmunoassay (Linco Research,
St. Charles, MO).
Statistical analysis. Power calculations were not performed, but the sample
size for each group was chosen based on study feasibility and prior knowledge
of statistical power from previously published experiments. All groups from
data showed normal variance, and thus results were analyzed using unpaired
Student’s t-test, or one-way or two-way ANOVA (followed by Tukey’s post hoc
test) where appropriate, using GraphPad Prism 5.0d software. Differences
were considered significant at P < 0.05. Data are presented as means ± s.e.m.
For the clamp experiments, the time period of 60–90 min was averaged for
the basal condition, and the time period of 180–200 min was averaged for the
clamp condition. For non-clamp experiments, the time period 60–90 min was
averaged for the basal condition, and the time period from 140–150 min was
averaged for treatment conditions.
37.Ono, H. et al. Activation of hypothalamic S6 kinase mediates diet-induced hepatic
insulin resistance in rats. J. Clin. Invest. 118, 2959–2968 (2008).
38.Wang, J. et al. Overfeeding rapidly induces leptin and insulin resistance. Diabetes
50, 2786–2791 (2001).
39.Filippi, B.M., Yang, C.S., Tang, C. & Lam, T.K. Insulin activates Erk1/2 signaling
in the dorsal vagal complex to inhibit glucose production. Cell Metab. 16, 500–510
(2012).
40.Kokorovic, A. et al. Duodenal mucosal protein kinase C-delta regulates glucose
production in rats. Gastroenterology 141, 1720–1727 (2011).
41.da Silva Xavier, G. et al. Role for AMP-activated protein kinase in glucose-stimulated
insulin secretion and preproinsulin gene expression. Biochem. J. 371, 761–774
(2003).
42.Choi, Y.H., Kim, S.G. & Lee, M.G. Dose-independent pharmacokinetics of metformin
in rats: hepatic and gastrointestinal first-pass effects. J. Pharm. Sci. 95,
2543–2552 (2006).
43.Graham, G.G. et al. Clinical pharmacokinetics of metformin. Clin. Pharmacokinet.
50, 81–98 (2011).
44.Sakamoto, K. et al. Deficiency of LKB1 in skeletal muscle prevents AMPK activation
and glucose uptake during contraction. EMBO J. 24, 1810–1820 (2005).
45.Dale, S., Wilson, W.A., Edelman, A.M. & Hardie, D.G. Similar substrate recognition
motifs for mammalian Amp-activated protein kinase, higher-plant HMG-CoA
reductase kinase-A, yeast SNF1, and mammalian calmodulin-dependent protein
kinase I. FEBS Lett. 361, 191–195 (1995).
nature medicine
CO R R I G E N DA
Corrigendum: Metformin activates a duodenal Ampk–dependent pathway to
lower hepatic glucose production in rats
Frank A Duca, Clémence D Côté, Brittany A Rasmussen, Melika Zadeh-Tahmasebi, Guy A Rutter, Beatrice M Filippi & Tony K T Lam
Nat. Med. 21, 506–11 (2015); doi:10.1038/nm.3787; published online 06 April 2015; corrected after print 7 May 2015
In the version of this article initially published, we incorrectly reported the value for the particles per milliliter of Ad-dn-AMPK (D157A) used in
the study. It was 3.1 × 10–9 PFU ml–1 and not 1.1 × 10–13 PFU ml–1 as originally reported. The errors have been corrected in the HTML and PDF
versions of the article.