Summary
The AMP-activated protein kinase cascade is a sensor of
cellular energy charge, and its existence provides strong
support for the energy charge hypothesis first proposed
by Daniel Atkinson in the 1960s. The system is activated
in an ultrasensitive manner by cellular stresses that
deplete ATP (and consequently elevate AMP), either by
inhibiting ATP production (e.g., hypoxia), or by accelerating ATP consumption (e.g., exercise in muscle). Once
activated, it switches on catabolic pathways, both acutely
by phosphorylation of metabolic enzymes and chronically by effects on gene expression, and switches off
many ATP-consuming processes. Recent work suggests
that activation of AMPK is responsible for many of the
effects of physical exercise, both the rapid metabolic
effects and the adaptations that occur during training.
Dominant mutations in regulatory subunit isoforms
(g2 and g3) of AMPK, which appear to increase the basal
activity in the absence of AMP, lead to hypertrophy of
cardiac and skeletal muscle respectively. BioEssays
23:11121119, 2001. 2001 John Wiley & Sons, Inc.
Introduction
One of the most fundamental parameters that any healthy cell
must maintain is a high ratio of ATP to ADP (of the order of
10:1). Almost all energy-requiring processes in the cell are
driven, either directly or indirectly, by hydrolysis of one or other
of the acid anhydride bonds in ATP, yielding ADP or AMP
(reactions 1 and 2, Box 1). Healthy cells maintain the reactants
and products of these two reactions many orders of magnitude
away from their equilibrium ratios. This is why ATP hydrolysis
is able to perform useful work when coupled to processes
requiring an input of energy. A useful analogy can be drawn
between the adenine nucleotides in a living cell and the
chemicals in an electrical cell or battery (strictly, the latter term
refers to a number of cells arranged in series). The ``battery'' of
the cell is charged up by catabolism or photosynthesis,
converting ADP and Pi to ATP (reaction 3, Box 1). Almost all
other cellular processes are coupled to ATP breakdown and
1112
BioEssays 23.12
;ATPAMP K ADP2
Divide both sides of the equation by [ATP]2:
AMP
ADP 2
;
/
ATP
ATP
i.e., the AMP:ATP ratio varies as the square of the
ADP:ATP ratio.
Review articles
ered. Thus the idea remained somewhat in abeyance until the
arrival on the scene of the AMP-activated protein kinase.
At this point, it is worth discussing why AMP, rather than
ADP, should be the key regulatory molecule. Eukaryotic cells
have a very active adenylate kinase that interconverts ATP,
ADP and AMP (reaction 4, Box 1), and maintains this reaction
close to equilibrium. If it is at equilibrium, the AMP:ATP ratio
will vary as the square of the ADP:ATP ratio (see lower part of
Box 1). Healthy cells under ideal conditions maintain an
ADP:ATP ratio of the order of 1:10 due to the operation of ATP
synthases (reaction 3). Under these conditions, adenylate
kinase will operate from right to left, keeping AMP very low
(AMP:ATP 1:100). If a cellular stress causes the rate of
ATPases (reaction 1) to exceed that of the ATP synthases
(reaction 3), however, the ADP:ATP ratio will rise, and the
adenylate kinase reaction will operate from left to right,
generating AMP. If the ADP:ATP ratio rises by 5-fold, the
AMP:ATP ratio will rise 25-fold. Thus, as pointed out by Hans
Krebs in 1963,(5) the cellular concentrations of AMP change
much more dramatically than do those of ATP or ADP. It
therefore makes sense for any system that monitors cellular
energy status to respond to AMP (or the AMP:ATP ratio) rather
than ADP (or the ADP:ATP ratio).
Structure and regulation of AMP-activated
protein kinase
AMP-activated protein kinase was originally discovered by its
ability to inactivate HMG-CoA reductase(6) and acetyl-CoA
carboxylase.(7) In 1980 Kim's group reported that an acetylCoA carboxylase kinase that they were studying was
stimulated by 50 -AMP, and suggested that it might inhibit fatty
acid synthesis in response to falling energy charge.(8) Five
years later Hegardt's group reported that an HMG-CoA
reductase kinase was also stimulated by AMP.(9) Shortly after
this our laboratory provided evidence that a single protein
kinase could account for both of these observations.(10) When
it became clear that the kinase had multiple physiological
substrates we renamed it the AMP-activated protein kinase
(AMPK) after its allosteric activator.(11) AMPK is now known to
exist as heterotrimeric complexes comprising a, b and g
subunits (Fig. 1). In mammals, each subunit is encoded by two
or three genes (a1, a2, b1, b2, g1, g2, g3) and at least 12
heterotrimeric combinations are possible.(1214) The a subunit
(63 kDa) contains the kinase domain at the N terminus, plus a
C-terminal regulatory domain containing an autoinhibitory
region that inhibits the kinase in the absence of AMP.(15) The b
subunits appear to be the scaffold on which a and g assemble,
via binding to their conserved KIS and ASC domains, respectively.(13) The g subunits contain four tandem repeats of a
structural module called a CBS domain, examples of which are
also found in various other proteins.(16) The functions of CBS
domains are not known, although the example in cystathione
b-synthase (after which they are named) appears to be
BioEssays 23.12
1113
Review articles
which phosphorylates a threonine residue (Thr-172) within the
``activation loop'' of the kinase domain.(18,19) AMP promotes
this phosphorylation both by binding to dephospho-AMPK and
making it a better substrate, and by activating AMPKK itself.(20)
In addition, AMP binding to phospho-AMPK almost completely
prevents dephosphorylation of Thr-172.(21) This complex
mechanism of AMP activation is summarized in Figure 2.
The multiple effects of AMP, coupled with a very low Km of
AMPKK for AMPK, make the AMPK cascade an ultrasensitive
system in which, over a critical range of concentrations, there
is a large response to a small increase in AMP. Our laboratory
recently showed that there is a sigmoidal (i.e., ultrasensitive)
response to the concentration of the activating nucleotide in
intact cells.(22)
Through the action of adenylate kinase (Box 1), AMP and
ATP almost always vary in reciprocal directions in the cell.
Although both AMPK and AMPKK require low concentrations
of ATP to function as protein kinases, high (mM) concentrations of ATP inhibit activation of the system by antagonizing
binding of AMP at the allosteric site(s) on AMPK. In a manner
highly analogous to the effects of these nucleotides on muscle
phosphorylase and phosphofructokinase, rising AMP and
falling ATP activate the AMPK system, which therefore acts
as an ``energy charge sensor''. AMPK is also allosterically
inhibited by phosphocreatine at concentrations that lie within
the physiological range.(23) Since phosphocreatine in muscle
and some other cells acts as a phosphagen, i.e., a short-term
reservoir of ATP, this fits in well with the energy sensor
concept. Phosphocreatine appears to have no effect on
phosphorylation of AMPK by AMPKK (DGH and SAH,
unpublished).
1114
BioEssays 23.12
Review articles
associated with very large increases in AMP and decreases in
ATP. However, AMP does not appear to activate the SNF1
kinase in cell-free assays, so it remains unclear whether a rise
in the AMP:ATP ratio is the signal that switches on the yeast
kinase in vivo.(34) What is known is that derepression of many
glucose-repressed genes involves a direct phosphorylation of
the repressor protein Mig1 by the SNF1 kinase, causing the
repressor to become inactive and to be exported from the
nucleus.(35,36) This provides an interesting model to explain
how AMPK might regulate gene expression in mammals
(see below).
Targets of AMP-activated protein kinase
In his paper in 1964(4) Atkinson wrote: ``These indications that
the level of AMP, or the AMP to ATP ratio, may regulate the
metabolic direction (towards energy release or towards energy
storage) ...... suggest that the level of AMP may function more
generally as a mediator of energy metabolism'' (my italics). In
the last few years there has been an explosive growth in our
knowledge of the downstream targets of the AMPK system,
and this has fully vindicated Atkinson's far-sighted proposal.
Much of this progress came from the use of 5-aminoimidazole4-carboxamide (AICA) riboside. This nucleoside is taken up
into cells and phosphorylated by adenosine kinase to the
monophosphorylated nucleotide, usually referred to as ZMP.
ZMP mimics all four effects of AMP on the AMPK cascade(37)
and, although it is much less potent than AMP itself, in most
cells it accumulates to sufficiently high concentrations when
they are incubated with the riboside. This represents a method
for activating AMPK without disturbing the cellular levels of
ATP, ADP or AMP, thus avoiding the many possible sideeffects of the latter. Its absolute specificity remains uncertain
and, like any pharmacological approach, the results should be
interpreted with caution. A novel alternative approach has
been developed recently, in which a constitutively active form
of the AMPK kinase domain is expressed from an adenoviral
vector.(38) This has its own drawbacks: one is overexpressing
a form of the kinase lacking the accessory subunits, which may
therefore not be correctly localized in the cell. However, a role
for AMPK is strengthened when the same results are obtained
using both AICA riboside and the constitutively active mutant,
as has been done in the case of inhibition of transcription of
lipogenic genes in liver.(38)
The many proposed downstream responses to AMPK
activation are summarized in Table 1. A full description of them
is beyond the scope of this article, but I will make a few general
comments and discuss a few specific examples. In most
cases, the effects have been demonstrated using AICA
riboside only, although in some of these the effects are
nevertheless convincing because the target protein and the
site of phosphorylation on the protein has been identified and
found to correspond to the site phosphorylated by AMPK
in vitro. Many of the effects would change the balance between
BioEssays 23.12
1115
Review articles
Approach
(24,37)
HMG-CoA reductase
Acetyl-CoA carboxylase
Glycerol-P acyl transferase?
AR,* others
AR,* others
AR*
Acetyl-CoA carboxylase
AR*
(52,53)
Endothelial NO synthase?
?
6-phosphofructo-2-kinase
AR*
AR*
Ischaemia, oligomycin
(54,55)
AR*
Ischaemia
AR*
AR*
(37,57)
Pathway affected?
Approach
Reference
(24,37)
(51)
(56)
(26)
(58)
(39,59,60)
(61)
Reference
CAM**
CAM**
CAM**
CAM**
arsenite
(38)
AR*
AR*
AR*
AR,* hypoxia
(44)
AR,*
AR,*
AR,*
AR,*
AR,*
(38)
(38)
(38)
(62)
(44)
(63)
(64)
AR AICA riboside.
1116
BioEssays 23.12
Review articles
Figure 3. Alignment of selected CBS domains from cystathionine b-synthase (CBS) and g subunits of AMPK, showing the location of
naturally occurring and engineered mutations. Residues that tend to be conserved across all CBS domains (bold type), and the likely
locations of b strands (bbbb...) and a-helices (hhhh...), are as discussed by Bateman.(16)
that all of these mutations cause forms of AMPK that are more
active under basal conditions in the absence of a rise in AMP.
It is easier to explain the dominant nature of these mutations
if they cause constitutive activation, rather than inhibition, of
the kinase activity. If AMPK activation does indeed cause
muscle hypertrophy as discussed earlier, these findings could
also help to explain the hypertrophy in both cardiac (g2)
and skeletal muscles (g3), as the muscle would respond
as if it sensed a depletion of ATP even when none had
happened.
These are exciting times for those of us working in the
AMPK field, with new papers appearing almost weekly. An
idea that started as a theoretical concept with Daniel Atkinson
has been provided with a firm experimental basis and is now
turning out to have great relevance to important clinical
conditions.
Acknowledgments
Work in the authors' laboratory is supported by the Wellcome
Trust, Diabetes UK and the Medical Research Council. We
thank Lee Witters and Bruce Kemp for permission to cite their
work prior to its publication.
References
1. Hardie DG, Carling D, Carlson M. The AMP-activated/SNF1 protein
kinase subfamily: metabolic sensors of the eukaryotic cell? Ann Rev
Biochem 1998;67:821855.
2. Hardie DG, Carling D. The AMP-activated protein kinase: fuel gauge of
the mammalian cell? Eur J Biochem 1997;246:259273.
3. Kemp BE, Mitchelhill KI, Stapleton D, Michell BJ, Chen ZP, Witters LA.
Dealing with energy demand: the AMP activated protein kinase. Trends
Biochem Sci 1999;24:2225.
4. Ramaiah A, Hathaway JA, Atkinson DE. Adenylate as a metabolic
regulator. Effect on yeast phosphofructokinase kinetics. J Biol Chem
1964;239:36193622.
5. Krebs H. The Croonian lecture gluconeogenesis. Proc Roy Soc Lond B
1963;159:545564.
6. Beg ZH, Allmann DW, Gibson DM. Modulation of 3-hydroxy-3-methylglutaryl coenzyme: A reductase activity with cAMP and with protein
fractions of rat liver cytosol. Biochem Biophys Res Comm 1973;54:1362
1369.
BioEssays 23.12
1117
Review articles
21. Davies SP, Helps NR, Cohen PTW, Hardie DG. 50 -AMP inhibits
dephosphorylation, as well as promoting phosphorylation, of the AMPactivated protein kinase. Studies using bacterially expressed human
protein phosphatase-2Ca and native bovine protein phosphatase-2AC.
FEBS Lett 1995;377:421425.
22. Hardie DG, Salt IP, Hawley SA, Davies SP. AMP-activated protein kinase:
an ultrasensitive system for monitoring cellular energy charge. Biochem
J 1999;338:717722.
23. Ponticos M, Lu QL, Morgan JE, Hardie DG, Partridge TA, Carling D. Dual
regulation of the AMP-activated protein kinase provides a novel
mechanism for the control of creatine kinase in skeletal muscle. EMBO
J 1998;17:16881699.
24. Corton JM, Gillespie JG, Hardie DG. Role of the AMP-activated protein
kinase in the cellular stress response. Current Biol 1994;4:315324.
25. Salt IP, Johnson G, Ashcroft SJH, Hardie DG. AMP-activated protein
kinase is activated by low glucose in cell lines derived from pancreatic b
cells, and may regulate insulin release. Biochem J 1998;335:533
539.
26. Marsin AS, Bertrand L, Rider MH, Deprez J, Beauloye C, Vincent MF,
Van den Berghe G, Carling D, Hue L. Phosphorylation and activation of
heart PFK-2 by AMPK has a role in the stimulation of glycolysis during
ischaemia. Curr Biol 2000;10:12471255.
27. Kudo N, Barr AJ, Barr RL, Desai S, Lopaschuk GD. High rates of fatty
acid oxidation during reperfusion of ischemic hearts are associated with
a decrease in malonyl-CoA levels due to an increase in 50 -AMP-activated
protein kinase inhibition of acetyl-CoA carboxylase. J Biol Chem
1995;270:1751317520.
28. Winder WW, Hardie DG. Inactivation of acetyl-CoA carboxylase and
activation of AMP-activated protein kinase in muscle during exercise. Am
J Physiol 1996;270:E299E304.
29. Fujii N, Hayashi T, Hirshman MF, Smith JT, Habinowski SA, Kaijser L, Mu
J, Ljungqvist O, Birnbaum MJ, Witters LA, Thorell A, Goodyear LJ.
Exercise induces isoform-specific increase in 50 AMP-activated protein
kinase activity in human skeletal muscle. Biochem Biophys Res Commun
2000;273:11501155.
30. Wojtaszewski JF, Nielsen P, Hansen BF, Richter EA, Kiens B. Isoformspecific and exercise intensity-dependent activation of 50 -AMP-activated
protein kinase in human skeletal muscle. J Physiol 2000;528:221
226.
31. Mu J, Brozinick JT, Valladares O, Bucan M, Birnbaum MJ. A role for
AMP-activated protein kinase in contraction- and hypoxia-regulated
glucose transport in skeletal muscle. Mol Cell 2001;7:10851094.
32. Winder WW, Hardie DG. The AMP-activated protein kinase, a metabolic
master switch: possible roles in Type 2 diabetes. Am J Physiol 1999;
277:E1E10.
33. Carlson M. Glucose repression in yeast. Curr Opin Microbiol 1999;
2:202207.
34. Wilson WA, Hawley SA, Hardie DG. The mechanism of glucose
repression/derepression in yeast: SNF1 protein kinase is activated by
phosphorylation under derepressing conditions, and this correlates with
a high AMP:ATP ratio. Curr Biol 1996;6:14261434.
35. Smith FC, Davies SP, Wilson WA, Carling D, Hardie DG. The SNF1 kinase
complex from Saccharomyces cerevisiae phosphorylates the repressor
protein Mig1p in vitro at four sites within or near Regulatory Domain 1.
FEBS Lett 1999;453:219223.
36. DeVit MJ, Johnston M. The nuclear exportin Msn5 is required for nuclear
export of the Mig1 glucose repressor of Saccharomyces cerevisiae. Curr
Biol 1999;9:12311241.
37. Corton JM, Gillespie JG, Hawley SA, Hardie DG. 5-Aminoimidazole-4carboxamide ribonucleoside: a specific method for activating AMPactivated protein kinase in intact cells? Eur J Biochem 1995;229:558
565.
38. Woods A, Azzout-Marniche D, Foretz M, Stein SC, Lemarchand P, Ferre
P, Foufelle F, Carling D. Characterization of the role of AMP-activated
protein kinase in the regulation of glucose-activated gene expression
using constitutively active and dominant negative forms of the kinase.
Mol Cell Biol 2000;20:67046711.
39. Blazquez C, Geelen MJ, Velasco G, Guzman M. The AMP-activated
protein kinase prevents ceramide synthesis de novo and apoptosis in
astrocytes. FEBS Lett 2001;489:149153.
1118
BioEssays 23.12
40. Amos AF, McCarty DJ, Zimmet P. The rising global burden of diabetes
and its complications: estimates and projections to the year 2010. Diabet
Med 1997;14:S185.
41. Hansen BF, Bdvarsdottir TB, Jensen-Holm HB, Rasmussen K, Jensen
P, Kurtzhals P. Improvement of glucose tolerance in Zucker Obese rats
by 5 days once daily treatment with AICAR. Abstract: Keystone Meeting
on Diabetes. Taos, New Mexico.
42. Bergeron R, Previs SF, Cline GW, Perret P, Russell RR 3rd, Young LH,
Shulman GI. Effect of 5-aminoimidazole-4-carboxamide-1-beta-D-ribofuranoside infusion on in vivo glucose and lipid metabolism in lean and
obese Zucker rats. Diabetes 2001;50:10761082.
43. Buhl ES, Jessen N, Schmitz O, Pedersen SB, Pedersen O, Holman GD,
Lund S. Chronic treatment with 5-aminoimidazole-4-carboxamide-1beta-D-ribofuranoside increases insulin-stimulated glucose uptake and
GLUT4 translocation in rat skeletal muscles in a fiber type-specific
manner. Diabetes 2001;50:1217.
44. Holmes BF, Kurth-Kraczek EJ, Winder WW. Chronic activation of 50 -AMPactivated protein kinase increases GLUT-4, hexokinase, and glycogen in
muscle. J Appl Physiol 1999;87:19901995.
45. Kawanaka K, Han DH, Gao J, Nolte LA, Holloszy JO. Development of
glucose-induced insulin resistance in muscle requires protein synthesis.
J Biol Chem 2001;276:2010120107.
46. Goodyear LJ, Kahn BB. Exercise, glucose transport, and insulin
sensitivity. Annu Rev Med 1998;49:235261.
47. Blair E, Redwood C, Ashrafian H, Oliveira M, Broxholme J, Kerr B,
Salmon A, Ostman-Smith I, Watkins H. Mutations in the gamma(2)
subunit of AMP-activated protein kinase cause familial hypertrophic
cardiomyopathy: evidence for the central role of energy compromise in
disease pathogenesis. Hum Mol Genet 2001;10:12151220.
48. Gollob MH, Green MS, Tang ASL, Gollob T, Karibe A, Hassan AS,
Ahmad F, Lozado R, Shah G, Fananapazir L, Bachinski LL, Roberts R.
Idenification of a gene responsible for familial Wolff-Parkinson-White
syndrome. Engl J Med 2001;344:18231831.
49. Milan D, Jeon JT, Looft C, Amarger V, Robic A, Thelander M, RogelGaillard C, Paul S, Iannuccelli N, Rask L, Ronne H, Lundstrom K, Reinsch
N, Gellin J, Kalm E, Roy PL, Chardon P, Andersson L. A mutation in
PRKAG3 associated with excess glycogen content in pig skeletal
muscle. Science 2000;288:12481251.
50. Hamilton SR, Stapleton D, O'Donnell JB, Kung JT, Dalal SR, Kemp BE,
Witters LA. An activating mutation in the g1 subunit of the AMP-activated
protein kinase. FEBS Lett 2001;500:163168.
51. Muoio DM, Seefeld K, Witters LA, Coleman RA. AMP-activated kinase
reciprocally regulates triacylglycerol synthesis and fatty acid oxidation in
liver and muscle: evidence that sn-glycerol- 3-phosphate acyltransferase
is a novel target. Biochem J 1999;338:783791.
52. Merrill GM, Kurth E, Hardie DG, Winder WW. AICAR decreases malonylCoA and increases fatty acid oxidation in skeletal muscle of the rat. Am J
Physiol 1997;273:E1107E1112.
53. Velasco G, Geelen MJH, Guzman M. Control of hepatic fatty acid
oxidation by 50 -AMP-activated protein kinase involves a malonyl-CoAdependent and a malonyl-CoA-independent mechanism. Arch Biochem
Biophys 1997;337:169175.
54. Kurth-Kraczek EJ, Hirshman MF, Goodyear LJ, Winder WW. 50 AMPactivated protein kinase activation causes GLUT4 translocation in
skeletal muscle. Diabetes 1999;48:16671671.
55. Fryer LGD, Hajduch E, Rencurel F, Salt IP, Hundal HS, Hardie DG,
Carling D. Activation of glucose transport by AMP-activated protein
kinase via stimulation of nitric oxide synthase. Diabetes 2000;49:1978
1985.
56. Abbud W, Habinowski S, Zhang JZ, Kendrew J, Elkairi FS, Kemp BE,
Witters LA, Ismail-Beigi F. Stimulation of AMP-Activated Protein
Kinase (AMPK) Is Associated with Enhancement of Glut1Mediated Glucose Transport. Arch Biochem Biophys 2000;380:347
352.
57. Garton AJ, Campbell DG, Carling D, Hardie DG, Colbran RJ, Yeaman
SJ. Phosphorylation of bovine hormone-sensitive lipase by the AMPactivated protein kinase. A possible antilipolytic mechanism. Eur J
Biochem 1989;179:249254.
58. Chen ZP, Mitchelhill KI, Michell BJ, Stapleton D, Rodriguez-Crespo I,
Witters LA, Power DA, Ortiz de Montellano PR, Kemp BE. AMP-activated
Review articles
59.
60.
61.
62.
63.
64.
65.
66.
67.
BioEssays 23.12
1119