Ion Channel Regulation by AMPK

HYPOXIA AND CONSEQUENCES

Ion Channel Regulation by AMPK

The Route of Hypoxia-Response Coupling in theCarotid Body and Pulmonary Artery

A. Mark Evans,a D. Grahame Hardie,b Chris Peers,c

Christopher N. Wyatt,d Benoit Viollet,f ,g Prem Kumar,e

Mark L. Dallas,c Fiona Ross,b Naoko Ikematsu,b

Heidi L. Jordan,d Barbara L. Barr,d J. Nicole Rafferty,a

and Oluseye Ogunbayoa

aCentre for Integrative Physiology, University of Edinburgh, Edinburgh, United KingdombUniversity of Dundee, Dundee, United Kingdom

cUniversity of Leeds, Leeds, United KingdomdWright State University, Dayton, Ohio, USA

eUniversity of Birmingham, Birmingham, United Kingdomf Universite Paris Descartes and gInserm, Paris, France

Vital homeostatic mechanisms monitor O2 supply and adjust respiratory and cir-culatory function to meet demand. The pulmonary arteries and carotid bodiesare key systems in this respect. Hypoxic pulmonary vasoconstriction (HPV) aidsventilation−perfusion matching in the lung by diverting blood flow from areas withan O2 deficit to those rich in O2, while a fall in arterial pO2 increases sensory afferentdischarge from the carotid body to elicit corrective changes in breathing patterns. Wediscuss here the new concept that hypoxia, by inhibiting oxidative phosphorylation,activates AMP-activated protein kinase (AMPK) leading to consequent phosphorylationof target proteins, such as ion channels, which initiate pulmonary artery constrictionand carotid body activation. Consistent with this view, AMPK knockout mice exhibit animpaired ventilatory response to hypoxia. Thus, AMPK may be sufficient and necessaryfor hypoxia-response coupling and may regulate O2 and thereby energy (ATP) supply atthe whole body as well as the cellular level.

Key words: AMP-activated protein kinase (AMPK); hypoxia; calcium; carotid body;pulmonary artery; hypoxic pulmonary vasoconstriction (HPV)

Introduction

AMP-activated protein kinase (AMPK) is aheterotrimer comprising catalytic α and regu-latory β and γ subunits, the genes for which arehighly conserved in all eukaryotic species, in-cluding vertebrates, invertebrates, plants, fungi,

Address for correspondence: A. Mark Evans, Centre for IntegrativePhysiology, College of Medicine and Veterinary Medicine, Universityof Edinburgh, Hugh Robson Building, George Square, Edinburgh EH89XD, United Kingdom. Voice: +44 131 6511501; fax: +44 131 [email protected]

and protozoa.1 Furthermore, AMPK is ubiq-uitously expressed and mediates cellular re-sponses that maintain metabolic homeostasisin response to a variety of metabolic stressesthat either accelerate ATP consumption (e.g.,muscle contraction) or that inhibit ATP pro-duction (e.g., glucose deprivation or ischemia).In addition to this cell-autonomous role in theregulation of energy balance, recent findingspoint to AMPK as a key player in mediatingresponses to cytokines that determine whole-body energy balance. For example, anorexo-genic agents such as leptin reduce food intake

Hypoxia and Consequences: Ann. N.Y. Acad. Sci. 1177: 89–100 (2009).doi: 10.1111/j.1749-6632.2009.05041.x c© 2009 New York Academy of Sciences.

89

90 Annals of the New York Academy of Sciences

in rodents by inhibiting AMPK activity in thehypothalamus, while low glucose and orexi-genic agents such as ghrelin increase food in-take by activating AMPK in the hypothala-mus.2 The adipokines leptin and adiponectinalso increase whole body energy expenditure byactivating AMPK and thus increase fatty acidoxidation in muscle and liver.2 Furthermore,we have suggested that AMPK is also involvedin the response to hypoxia in those cells in or-gans such as the carotid body and pulmonaryarteries, which serve to monitor O2 supply andmodulate cardiorespiratory function in orderto maintain arterial pO2 within physiologicallimits.3

Regulation of AMPK in Responseto Metabolic Stress

Metabolic stresses that either inhibit ATPproduction or accelerate ATP consumptionelevate the cellular ADP/ATP ratio. This isamplified by the adenylate kinase reaction(2ADP ↔ ATP + AMP) into a much largerrise in AMP/ATP ratio, which is monitoredby AMPK as an index of metabolic stress.4

Once activated, AMPK serves to restore energyhomeostasis by stimulating catabolic pathwaysthat generate ATP, and by suppressing ATP-consuming processes not essential to short-termsurvival.

AMPK is activated >100-fold by phospho-rylation at Thr-172 within the α subunit by up-stream kinases, of which the most important isthe tumor suppressor LKB1.5 LKB1 appears tophosphorylate Thr-172 constitutively, but bind-ing of AMP to the two exchangeable sites on theγ subunit of AMPK6,7 activates the kinase byinhibiting dephosphorylation of Thr-172,8 thuscausing a switch to the active, phosphorylatedform. Binding of AMP to the γ subunit causesa further allosteric activation of the phosphory-lated kinase by up to 10-fold—the combinationof the two effects causing as much as 1000-fold activation.9 This dual mechanism of acti-vation ensures great sensitivity. In the absence

of a metabolic stress, mitochondrial oxidativephosphorylation maintains cellular ATP : ADPin most cells at 10 : 1, and adenylate kinasethen maintains the ATP : AMP ratio at about100 : 1.4 Because the affinities of the γ subunitsites for AMP, ADP, and ATP are quite similar,in unstressed cells these sites are likely to be al-most entirely occupied by ATP; both ATP andADP compete with AMP for binding at the γ

subunit sites but, unlike AMP, do not cause theactivating effects.6,7

Even in cells undergoing a metabolic stress,the concentration of AMP is still likely to bemuch lower than that of ATP or ADP, so thatthe proportion of AMPK in the active, AMP-bound form may also be rather small. However,this mechanism also means that any further in-creases in AMP would cause additional activa-tion, and it is unlikely that the system wouldever become saturated. As discussed above, themajor source of AMP in the cell is likely to beadenylate kinase, which exists as multiple iso-forms in distinct subcellular locations.10 Thus,it is possible that there may be localized pools ofhigh AMP concentration produced at specificlocations during particular types and degrees ofmetabolic stress, and this may allow for differ-ential regulation of different AMPK isoforms(see below).

In many cells, AMPK can also be acti-vated by a Ca2+-mediated pathway. Stim-uli that increase cytoplasmic Ca2+ activatethe calmodulin-dependent protein kinaseCaMKKβ (CaMKK2), which also activatesAMPK by phosphorylating Thr-172 in anAMP-independent manner.11 Because in-creases in cytoplasmic Ca2+ often trigger ATP-requiring processes such as muscle contractionor membrane trafficking, this mechanism mayanticipate the consequent demand for ATP. Inaddition, AMPK appears to be regulated byglycogen via the β subunit, which contains acarbohydrate-binding domain that causes thecomplex to bind to glycogen particles. This mayallow AMPK to act as a “glycogen sensor” thatensures rapid resynthesis of glycogen when ithas become depleted.12

Evans et al.: AMPK and Hypoxia-Response Coupling 91

Figure 1. AMP-activated protein kinase (AMPK) subunits and their expression in pul-monary arterial smooth muscle. (A) The domain structure of AMPK subunit isoforms, indicatingfunctional regions where known: NTD, N terminal domain; CBS motif x 2 = AMP-bindingBateman domain; GBD, glycogen-binding domain. (B) Increase in AMPK α1 and α2 catalyticactivities in pulmonary arterial smooth muscle and consequent acetyl CoA carboxylase phos-phorylation in response to hypoxia and 5-aminoimidazole-4-carboxamide riboside (AICAR)in the presence and absence of the AMPK antagonist compound C (comp. C). (C) Westernblot showing the AMPK subunits present in pulmonary arterial smooth muscle (i and ii, pul-monary; iii, liver) and the relative activity conferred by each subunit isoform as determinedby immunoprecipitate kinase assay.

In light of the above, AMPK has been re-ferred to as the “guardian of cellular energy.”1

Yet, while many of its downstream targets areinvolved in energy metabolism, it is becom-ing clear that AMPK may also phosphory-late targets such as ion channels and trans-porters, and thereby contribute to other aspects

of cell function.3 At least 12 different AMPKheterotrimers can be formed from the multi-ple isoforms of the catalytic subunits (α1 andα2) and regulatory subunits (β1, β2, γ1, γ2,γ3), with splice variants adding further diversity(Fig. 1A). Variations in heterotrimeric subunitcombination may target AMPK to different


subcellular locations13,14 and may thereby con-fer target specificity. Thus, selective expressionof a particular AMPK isoform and/or targetprotein(s) could determine functional outcomesconsequent to AMPK activation in a cell- andsystem-specific manner.3 This is already appar-ent with respect to metabolic control: In skele-tal muscle, AMPK activation promotes glucoseuptake via plasma membrane translocation ofthe glucose transporter GLUT4 in an additivemanner with insulin,15 while in adipocytes itinhibits insulin-stimulated glucose uptake viaGLUT4.16

Oxygen-Sensing Cells

With these considerations in mind we beganinvestigations into the possibility that AMPKmight mediate the responses to hypoxia of cellsor organs that have evolved to monitor O2 sup-ply to the body. These include carotid bodytype I cells,17 pulmonary arterial smooth mus-cle18 and endothelial cells,18,19 neuroepithelialbody cells,20 and neonatal adrenal chromaf-fin cells.21 Each plays a discrete role in deter-mining the whole-body response to deficits inO2 supply. Importantly, they all respond to lev-els of hypoxia that lie within the physiologicalrange (20–60 mm Hg) and which elicit little orno response from cells that do not function tomonitor O2 supply.22,23 Irrespective of the typeof O2-sensing cell under study, an importantand consistent finding is that relatively mild lev-els of hypoxia inhibit mitochondrial oxidativephosphorylation as indicated by depolarizationof the mitochondrial membrane potential23

and/or an increase in cellular β-NAD(P)Hlevels.20,24,25 Inhibition of oxidative phospho-rylation has therefore been proposed to under-pin, at least in part, chemotransduction by hy-poxia, and this view has been reinforced byrecent studies showing that functional mito-chondria are required for O2-sensing in im-mortalized neonatal adrenal chromaffin cells.26

The sole argument against this proposal hasbeen that the affinity of cytochrome c oxi-

dase for O2 is too high to allow for the inhi-bition of mitochondrial oxidative phosphory-lation by physiological levels of hypoxia (60–20 mm Hg). However, this parameter mayvary dramatically in a cell- and tissue-specificmanner that is dependent on both metabolicstate and rate.27 Indeed, the carotid body isknown to have a very high rate of O2 con-sumption.28,29 That mitochondrial inhibition iskey to chemotransduction by hypoxia is furtheremphasized by the fact that inhibitors of themitochondrial electron transport chain mimicand/or occlude chemotransduction by hypoxiain all O2-sensing cells.23,26,30 It is notable, there-fore, that mitochondrial inhibitors have longbeen recognized as providing metabolic stresssufficient for activation of AMPK.31 In fact,one such inhibitor, metformin, is used in thetreatment of type 2 diabetes and is knownto mediate its therapeutic effects via AMPKactivation.31,32

AMPK Activation Underpins HPV

Investigations on pulmonary arterial smoothmuscle have suggested that cellular ATP lev-els remain remarkably stable in the presenceof hypoxia (although a drop in phosphocrea-tine levels was noted33), despite the fact thatinhibition of mitochondrial oxidative phospho-rylation by hypoxia would be expected to pre-cipitate a drop in the cellular energy status.However, we found that exposure of pulmonaryarterial smooth muscle to physiological hypoxiaprecipitated an increase in the AMP/ATP ra-tio, which varied as the square of the ADP/ATPratio, in line with expectation, if the adeny-late kinase reaction was at equilibrium.34 Thiswas associated with a concomitant, twofold in-crease in AMPK activity, as well as increasedphosphorylation of acetyl CoA carboxylase(ACC), an established marker of AMPK ac-tion, within the smooth muscle (Fig. 1B). Theseeffects were mimicked by 5-aminoimidazole-4-carboxamide riboside (AICAR), which is takenup and metabolized to yield the AMP mimetic


ZMP (AICAR monophosphate) and therebyactivates AMPK without affecting the cellularAMP/ATP ratio. Furthermore, in each caseAMPK activation and ACC phosphorylationwas attenuated by the nonselective AMPK an-tagonist compound C.35

There are, however, multiple isoforms of theα, β, and γ subunits that comprise AMPK.Each heterotrimer may be differentially reg-ulated by a given stimulus and may phosphory-late discrete target proteins and thereby mod-ulate different aspects of cell function. Thus,it may be significant to note that analysis ofAMPK activities in pulmonary arterial smoothmuscle suggests that the α1, α2, β2, γ1, andγ2 subunit isoforms are present (Fig. 1C), withonly limited γ3 subunit expression and lit-tle or no expression of the β1 subunit iso-form apparent.34 These data suggest that asfew as four heterotrimeric subunit combina-tions may contribute to the regulation of pul-monary arterial smooth muscle in response tometabolic stresses, i.e., the α1β2γ1, α1β2γ2,α2β2γ1, and α2β2γ2 hetrotrimers. Of these ac-tivities, that conferred by the α1 catalytic sub-unit isoform contributed ∼80% of total activity,and the α1β2γ1 represented the predominantheterotrimer.34

Also noteworthy was the finding thatAMPK-α1-associated catalytic activity wasmuch lower (≤ 50%) in smooth muscle fromthe main pulmonary artery when comparedto second and third order branches, i.e., it isinversely related to pulmonary artery diame-ter, as is the magnitude of pulmonary arteryconstriction by hypoxia.36 Most importantly,perhaps, AMPK-α1 activity was at least four-fold higher in second and third order branchesof the pulmonary arterial tree when comparedwith systemic (mesenteric) arteries,34 which di-

late in response to hypoxia. In marked contrast,the smaller AMPK-α2-associated activity re-mained virtually unchanged. Thus, this selec-tive expression of AMPK-α1 activity may insome way offer the pulmonary selectivity re-quired for HPV. However, it should be notedthat hypoxia also increased catalytic activ-

ity associated with AMPK-α2-containing het-erotrimers, and the targets that these phospho-rylate may also be of significance to functionaloutcomes.

In isolated pulmonary artery rings, hy-poxia (16–21 mm Hg) induces a biphasic con-striction, with an initial transient constriction(Phase 1) followed by a slow tonic constric-tion (Phase 2). Each represents a discrete event(Fig. 2A), and they are both initiated imme-diately on exposure to hypoxia.37 The initialtransient constriction peaks within 5–10 minof the hypoxic challenge, whilst the underly-ing, tonic constriction peaks after 30–40 min.When the endothelium is removed the gradualamplification of Phase 2, which is driven by therelease of an endothelium-derived vasocon-strictor,19 is not observed, and the Phase 1constriction now declines to a maintainedplateau.18 A Phase 1-like constriction hasbeen induced in response to AMPK acti-vation by AICAR, but this has been infre-quently observed (unpublished observation).However, AICAR consistently induced a slow,sustained, and reversible constriction of pul-monary artery rings (Fig. 2B) that exhibitedall the primary characteristics of Phase 2 ofHPV (Fig. 2C).34 Thus, with respect to constric-tion by either AICAR or hypoxia: (1) removalof the pulmonary artery endothelium attenu-ates the tonic constriction; (2) the endothelium-dependent component of constriction was abol-ished upon removal of extracellular Ca2+ andtherefore requires Ca2+ influx into the en-dothelium; (3) constriction mediated by mecha-nisms intrinsic to the smooth muscle remainedin the absence of extracellular Ca2+, but wasabolished by blocking the mobilization of sar-coplasmic reticulum (SR) Ca2+ stores withryanodine and caffeine, and by blocking Ca2+

mobilization dependent upon cyclic adenosinediphosphate-ribose (cADPR), an endogenousregulator of ryanodine receptors (RyRs),38 with8-bromo-cADPR; (4) the AMPK antagonistcompound C inhibited HPV and constrictionin response to AICAR.35 Therefore, the pro-file of pulmonary artery constriction versus


Figure 2. Constriction of pulmonary artery ringsin response to hypoxia and AMP-activated proteinkinase (AMPK) activation with 5-aminoimidazole-4-carboxamide riboside (AICAR). (A) Schematic dia-gram shows the identified components of hypoxicpulmonary vasoconstriction. (B) Constriction of pul-monary artery rings (200–500 μm i.d.), with andwithout endothelium, by hypoxia (16–21 Torr) andAICAR (1 mM). (C) Profile of the constriction of pul-monary artery rings in response to hypoxia andAICAR (1 mM) with and without endothelium andin the presence and absence of: extracellular Ca2+;ryanodine (10 μM) and caffeine (10 mM); 8-bromo-cADPR (8Br, 300 μM).

experimental intervention remains astonish-ingly consistent whether hypoxia or AICARis used as the stimulus (Fig. 2C). In eachcase, maintained smooth muscle constrictionis driven by cADPR-dependent mobilizationof SR Ca2+ stores via RyRs, which is sup-ported by consequent store−release-activatedCa2+ influx that is not itself directly regulatedby hypoxia34; this is indicated by the fact that

removal of extracellular Ca2+ attenuates con-striction of pulmonary arteries without the en-dothelium (Fig. 2C).

However, the mechanisms involved in theprocess of smooth muscle constriction by hy-poxia appear far more complex than suggestedby the above. Our preliminary investigationsindicate that hypoxia may promote constric-tion by mobilizing Ca2+ from lysosome-relatedstores, via two pore channels,39 and/or fromcentral SR compartments via RyRs.37,40 In ad-dition, data suggest concomitant inhibition byhypoxia of smooth muscle dilation via mecha-nisms that: (1) limit Ca2+ uptake/release from afunctionally segregated SR compartment situ-ated proximal to the plasma membrane,37,40,41

and (2) attenuate subsequent Ca2+ removal(Fig. 3A). As discussed in detail elsewhere,42

a variety of functionally compartmentalizedion channels and transporters have been impli-cated in these two processes, and it is quite pos-sible that they may be differentially regulatedupon exposure to hypoxia by compartmen-talized AMPK heterotrimers. As mentionedpreviously, however, the degree and type ofmetabolic stress experienced in one cellularcompartment may differ from that experiencedby another. Moreover, due to the targeting ofdifferent isoforms, the capacity of adenylate ki-nase to compensate for a localized fall in ATPavailability may vary from one compartment toanother. Consequently it is possible that a fallin local ATP levels, rather than AMPK, maymodulate certain processes.

We are, therefore, in the process of determin-ing which of these protein targets are directlyphosphorylated and regulated by AMPK. Inthis respect we first focused on one classicaland consistent observation, namely, that hy-poxia modulates the delayed rectifier K+ cur-rent (likely carried by homo/heteromultimersthat incorporate Kv2.1 or Kv1.5 subunits) butnot the large conductance Ca2+-activated K+

current (BKCa).43,44 Our preliminary findings(unpublished) have been striking in that theysuggest that AMPK phosphorylates recombi-nant Kv2.1 in an AMP-activated manner and


Figure 3. Possible targets for AMP-activated protein kinase (AMPK)-dependent phospho-rylation in pulmonary arterial smooth muscle. (A) Schematic diagram shows putative proteintargets for AMPK-dependent phosphorylation within pulmonary arterial smooth muscle, andwhich may either be inhibited or activated by AMPK in response to hypoxia in order topromote smooth muscle contraction ( , confirmed; , under investigation): voltage-dependentK+ current (Kv); sarco/endoplasmic reticulum Ca2+ ATPase (SERCA) subtype 2a and 2b;ryanodine receptor (RyR) subtypes 1, 2, and 3; two-pore channel subtype 2 (TPC2); vacuolarproton pump (V-H+-ATPase); ADP-ribosyl cyclase (ARC). (B) Schematic diagram depicts theeffect of AMPK on the voltage-dependent activation and inactivation curve of recombinantKv2.1 expressed in HEK293 cells.

thereby confers a leftward shift in both theactivation and inactivation curves of Kv2.1stably expressed in HEK293 cells (Fig. 3B),and in a manner not dissimilar from the ef-fects of mitochondrial inhibitors45 and hypoxia(S.V. Smirnov, personal communication) on the

endogenous delayed rectifier in this cell type. Itshould be noted, however, that the functionalconsequence of Kv regulation by hypoxia is de-batable, as it is not yet clear whether this servesto modulate voltage-gated Ca2+ influx or Ca2+

removal.


Figure 4. Modulation of K+ currents by AMP-activated protein kinase (AMPK) in carotidbody type I cells. Inhibition by AMPK activation of (A) the large conductance Ca2+-activatedK+ current (IBTX = iberiotoxin) and (B) TASK-like leak K+ current in acutely isolated carotidbody type I cells. (C) Inhibition by AMPK activation of the large conductance Ca2+-activatedK+ current conferred by recombinant KCa1.1 expression in HEK293 cells and the AMP-dependent phosphorylation by AMPK of purified KCa1.1.

AMPK and Carotid BodyExcitation by Hypoxia

Irrespective of any cell-specific variation insignaling mechanism, if AMPK were to playa central role in hypoxia-response coupling inall O2-sensing cells we would expect its activa-tion to mimic precisely the effects of hypoxiain a given O2-sensing cell type. To determinewhether or not this was the case we next inves-tigated the role of AMPK in mediating carotidbody excitation in response to a drop in sys-temic arterial pO2. Upon exposure to hypoxia,carotid body type I cells depolarize, precipitat-ing voltage-gated Ca2+ influx rather than en-doplasmic reticulum Ca2+ release. This causesneurosecretion, which ultimately leads toincreased sensory afferent discharge to the

brain stem and corrective changes in breath-ing patterns.28 It is well-established that thisprocess is driven, in the main, by the inhibi-tion of discrete populations of O2-sensitive K+

channels.Like hypoxia, AMPK activation by AICAR

inhibited the O2-sensitive K+ currents carriedby TASK-like leak K+ channels46,47 and BKCa

channels47,48 in rat carotid body type I cells(Fig. 4A). Importantly, our investigations onthe regulation of recombinant K+ channels byAMPK provide further support for these find-ings. Briefly, we have now shown that AMPKphosphorylates recombinant BKCa channels(KCa1.1) in an AMP-activated manner andthereby inhibits voltage-dependent activationof macroscopic currents carried by KCa1.1stably expressed in HEK293 cells47 (Fig. 4B).


Further preliminary data suggest that AMPKactivation also inhibits the whole-cell cur-rent carried by recombinant TASK3 channels,but not TASK1 channels stably expressed inHEK293 cells (unpublished observations).

In marked contrast, but again consistent withthe effects of hypoxia, no inhibition of the O2-insensitive, voltage-gated Kv channel currentwas observed in rat carotid body type I cells.This observation has greater implications, how-ever, due to the fact that Kv2 channel subunitshave been proposed to underpin the delayedrectifier in rat type I cells,49 and our studies onrecombinant Kv2.1 show that channels formedby this subunit are phosphorylated and therebyregulated by AMPK. Thus, either an AMPK-insensitive Kv2 subunit other than Kv2.1 isexpressed in rat type I cells, or these cells donot express (or target to the plasma membrane)the AMPK heterotrimer required for Kv2regulation.

Such considerations may also address cell-,species-, and even strain-dependent variationsthat have been noted with respect to ion chan-nel regulation by hypoxia. For example andconsistent with observations on rat type I cells,BKCa currents are inhibited by hypoxia in typeI cells from the DBA/2J strain of mouse,50

while BKCa currents remain unaffected by hy-poxia in type I cells derived from either theA/J strain of mouse50 or from rabbit51; nor,as mentioned previously, are they inhibited byhypoxia in rat pulmonary arterial smooth mus-cle cells. Furthermore, fast inactivating Kv cur-rents appear to be inhibited by hypoxia inrabbit (Kv4.1/Kv4.3)52 and mouse (Kv3.1-/Kv3.2-/Kv3.3)53 carotid body type I cells, inmarked contrast to the hypoxia-insensitive na-ture of Kv currents in rat type I cells.

Consequent to K+ channel inhibition byAMPK activation, as is the case for hypoxia,54

a reversible depolarization of rat carotid bodytype I cell membrane potential was induced,which precipitated an increase in the intracel-lular Ca2+ concentration via Ni2+- and Cd2+-sensitive voltage-gated Ca2+ influx pathways.This effect was reversed by the AMPK antag-

Figure 5. AMP-activated protein kinase (AMPK)mediates carotid body activation in response to hy-poxia. (A) The AMPK antagonist compound C at-tenuates the increase in carotid body afferent fiberdischarge induced by hypoxia (< 150 mm Hg). (B)The increase in frequency of breathing in responseto hypoxia is attenuated in AMPK α2 knockout (KO)mice.

onist compound C. Furthermore, studies onthe isolated carotid body demonstrated thatAICAR and hypoxia increased sensory affer-ent fiber discharge in a manner that was atten-uated by the AMPK antagonist compound C(Fig. 5A).47 Taken together, these findings sug-gest that AMPK likely mediates the excitatoryeffects of hypoxia on isolated carotid body typeI cells and on the carotid body in vitro. This pos-sibility now receives strong support from ourpreliminary findings in vivo, which show that theincrease in ventilation frequency triggered byhypoxia is attenuated in AMPK-α2 knockoutmice (Fig. 5B; unpublished observation), as it isgenerally recognized that the acute ventilatoryresponse to a fall in arterial pO2 is mediated bythe carotid body.


Figure 6. AMPK may regulate O2 and thereby energy (ATP) supply at the whole-bodylevel. Schematic diagram shows inhibition of mitochondrial oxidative phosphorylation byhypoxia, AMPK activation consequent to an increase in the AMP/ATP ratio and the induction,via cell-specific Ca2+ signaling processes, of (1) hypoxic pulmonary vasoconstriction, whichaids ventilation perfusion matching in the lung, and (2) carotid body activation, which leadsto corrective changes in breathing patterns.

Summary

Our findings are consistent with the viewthat inhibition of mitochondrial oxidative phos-phorylation by hypoxia leads to a rise in thecellular AMP/ATP ratio, and consequent ac-tivation of the catalytic activity of AMPK inpulmonary arterial smooth muscle and carotidbody type I cells. The reliance of O2-sensingcells on mitochondria for ATP supply and ahigher level of AMPK expression/activity maydetermine, at least in part, the relative sensitiv-ity of such cells to physiological hypoxia. There-after, the differential response of discrete cellu-lar compartments to metabolic stress combinedwith subcellular compartmentalization of dif-ferent AMPK heterotrimeric subunit combi-nations and/or protein targets may elicit thecharacteristic response of each tissue type to hy-poxia. Thus, AMPK may be sufficient and nec-essary for hypoxia-response coupling and may

regulate O2 and thereby energy (ATP) supplyat the whole-body level (Fig. 6).

Acknowledgments

This work was supported by the WellcomeTrust (A.M.E., D.G.H., C.P.), the AmericanHeart Association (C.N.W.), and the EuropeanUnion FP6 Program (B.V.).

Conflicts of Interest

The authors declare no conflicts of interest.

References

1. Hardie, D.G. 2007. AMP-activated/SNF1 protein ki-nases: conserved guardians of cellular energy. Nat.

Rev. Mol. Cell. Biol. 8: 774–785.2. Kahn, B.B. et al. 2005. AMP-activated protein ki-

nase: ancient energy gauge provides clues to modernunderstanding of metabolism. Cell Metab. 1: 15–25.


3. Evans, A.M. 2006. AMP-activated protein kinaseand the regulation of Ca2+ signalling in O2-sensingcells. J. Physiol. 574: 113–123.

4. Hardie, D.G. & S.A. Hawley. 2001. AMP-activatedprotein kinase: the energy charge hypothesis revis-ited. Bioessays 23: 1112–1119.

5. Hawley, S.A. et al. 2003. Complexes between theLKB1 tumor suppressor, STRAD alpha/beta andMO25 alpha/beta are upstream kinases in the AMP-activated protein kinase cascade. J. Biol. 2: 28.

6. Scott, J.W. et al. 2004. CBS domains form energy-sensing modules whose binding of adenosine ligandsis disrupted by disease mutations. J. Clin. Invest. 113:274–284.

7. Xiao, B. et al. 2007. Structural basis for AMP bindingto mammalian AMP-activated protein kinase. Nature

449: 496–500.8. Davies, S.P. et al. 1995. 5′-AMP inhibits dephospho-

rylation, as well as promoting phosphorylation, of theAMP-activated protein kinase. Studies using bacteri-ally expressed human protein phosphatase-2C alphaand native bovine protein phosphatase-2AC. FEBS

Lett. 377: 421–425.9. Suter, M. et al. 2006. Dissecting the role of 5′-AMP

for allosteric stimulation, activation, and deactivationof AMP-activated protein kinase. J. Biol. Chem. 281:32207–32216.

10. Dzeja, P. & A. Terzic. 2009. Adenylate kinase andAMP signaling networks: metabolic monitoring, sig-nal communication and body energy sensing. Int. J.

Mol. Sci. 10: 1729–1772.11. Hawley, S.A. et al. 2005. Calmodulin-dependent pro-

tein kinase, kinase-beta is an alternative upstream ki-nase for AMP-activated protein kinase. Cell Metab. 2:9–19.

12. McBride, A. et al. 2009. The glycogen-binding do-main on the AMPK beta subunit allows the kinase toact as a glycogen sensor. Cell Metab. 9: 23–34.

13. Salt, I. et al. 1998. AMP-activated protein kinase:greater AMP dependence, and preferential nuclearlocalization, of complexes containing the alpha2 iso-form. Biochem. J. 334(Pt 1): 177–187.

14. Gregor, M. et al. 2006. Plectin scaffolds recruitenergy-controlling AMP-activated protein kinase(AMPK) in differentiated myofibres. J. Cell. Sci. 119:1864–1875.

15. Hayashi, T. et al. 1998. Evidence for 5′ AMP-activated protein kinase mediation of the effect ofmuscle contraction on glucose transport. Diabetes 47:1369–1373.

16. Salt, I.P., J.M. Connell & G.W. Gould. 2000.5-aminoimidazole-4-carboxamide ribonucleoside(AICAR) inhibits insulin-stimulated glucose trans-port in 3T3-L1 adipocytes. Diabetes 49: 1649–1656.

17. Verna, A., M. Roumy & L.M. Leitner. 1975. Loss of

chemoreceptive properties of the rabbit carotid bodyafter destruction of the glomus cells. Brain Res. 100:13–23.

18. Dipp, M., P.C. Nye & A.M. Evans. 2001. Hypoxicrelease of calcium from the sarcoplasmic reticulum ofpulmonary artery smooth muscle. Am. J. Physiol. Lung

Cell Mol. Physiol. 281: L318–L325.19. Gaine, S.P., M.A. Hales & N.A. Flavahan. 1998. Hy-

poxic pulmonary endothelial cells release a diffusiblecontractile factor distinct from endothelin. Am. J.

Physiol. 274: L657–L664.20. Youngson, C. et al. 1993. Oxygen sensing in airway

chemoreceptors. Nature 365: 153–155.21. Thompson, R.J., A. Jackson & C.A. Nurse. 1997.

Developmental loss of hypoxic chemosensitivity inrat adrenomedullary chromaffin cells. J. Physiol.

498(Pt 2): 503–510.22. Stea, A. & C.A. Nurse. 1991. Whole-cell and

perforated-patch recordings from O2-sensitive ratcarotid body cells grown in short- and long-term cul-ture. Pflugers Arch. 418: 93–101.

23. Duchen, M.R. & T.J. Biscoe. 1992. Relative mito-chondrial membrane potential and [Ca2+]i in type Icells isolated from the rabbit carotid body. J. Physiol.

450: 33–61.24. Duchen, M.R. & T.J. Biscoe. 1992. Mitochondrial

function in type I cells isolated from rabbit arterialchemoreceptors. J. Physiol. 450: 13–31.

25. Shigemori, K. et al. 1996. Adenine nucleotides via ac-tivation of ATP-sensitive K+ channels modulate hy-poxic response in rat pulmonary artery. Am. J. Physiol.

270: L803–L809.26. Buttigieg, J. et al. 2008. Functional mitochondria are

required for O2 but not CO2 sensing in immortalizedadrenomedullary chromaffin cells. Am. J. Physiol. Cell

Physiol. 294: C945–C956.27. Gnaiger, E. et al. 1998. Mitochondrial oxygen affin-

ity, respiratory flux control and excess capacity ofcytochrome c oxidase. J. Exp. Biol. 201: 1129–1139.

28. Gonzalez, C. et al. 1994. Carotid body chemore-ceptors: from natural stimuli to sensory discharges.Physiol. Rev. 74: 829–898.

29. Whalen, W.J. et al. 1981. Cat carotid body: oxygenconsumption and other parameters. J. Appl. Physiol.

50: 129–133.30. Weissmann, N. et al. 2003. Effects of mitochondrial

inhibitors and uncouplers on hypoxic vasoconstric-tion in rabbit lungs. Am. J. Respir. Cell. Mol. Biol. 29:721–732.

31. Hardie, D.G. 2008. Role of AMP-activated proteinkinase in the metabolic syndrome and in heart dis-ease. FEBS Lett. 582: 81–89.

32. Shaw, R.J. et al. 2005. The kinase LKB1 mediatesglucose homeostasis in liver and therapeutic effectsof metformin. Science 310: 1642–1646.


33. Buescher, P.C. et al. 1991. Energy state and vaso-motor tone in hypoxic pig lungs. J. Appl. Physiol. 70:1874–1881.

34. Evans, A.M. et al. 2005. Does AMP-activated proteinkinase couple inhibition of mitochondrial oxidativephosphorylation by hypoxia to calcium signaling inO2-sensing cells? J. Biol. Chem. 280: 41504–41511.

35. Robertson, T.P. et al. 2008. AMP-activated proteinkinase and hypoxic pulmonary vasoconstriction. Eur.

J. Pharmacol. 595: 39–43.36. Kato, M. & N.C. Staub. 1966. Response of small

pulmonary arteries to unilobar hypoxia and hyper-capnia. Circ. Res. 19: 426–440.

37. Dipp, M. & A.M. Evans. 2001. Cyclic ADP-ribose isthe primary trigger for hypoxic pulmonary vasocon-striction in the rat lung in situ. Circ. Res. 89: 77–83.

38. Galione, A., H.C. Lee & W.B. Busa. 1991.Ca(2+)−induced Ca2+release in sea urchin egg ho-mogenates: modulation by cyclic ADP-ribose. Science

253: 1143–1146.39. Calcraft, P.J. et al. 2009. NAADP mobilizes calcium

from acidic organelles through two-pore channels.Nature 459: 596–600.

40. Kinnear, N.P. et al. 2008. Lysosomes co-localize withryanodine receptor subtype 3 to form a trigger zonefor calcium signalling by NAADP in rat pulmonaryarterial smooth muscle. Cell Calcium 44: 190–201.

41. Boittin, F.X. et al. 2003. Vasodilation by the calcium-mobilizing messenger cyclic ADP-ribose. J. Biol.

Chem. 278: 9602–9608.42. Evans, A.M. 2009. The role of intracellular ion chan-

nels in regulating cytoplasmic calcium in pulmonaryarterial smooth muscle: which store and where? Adv.

Exp. Med. Biol. In Press43. Smirnov, S.V. et al. 2002. Electrophysiologically dis-

tinct smooth muscle cell subtypes in rat conduit andresistance pulmonary arteries. J. Physiol. 538: 867–878.

44. Platoshyn, O. et al. 2006. Acute hypoxia selectively in-hibits KCNA5 channels in pulmonary artery smoothmuscle cells. Am. J. Physiol. Cell Physiol. 290: C907–C916.

45. Firth, A.L., K.H. Yuill & S.V. Smirnov. 2008.Mitochondria-dependent regulation of Kv currentsin rat pulmonary artery smooth muscle cells. Am. J.

Physiol. Lung Cell Mol. Physiol. 295: L61–L70.46. Buckler, K.J., B.A. Williams & E. Honore. 2000.

An oxygen-, acid- and anaesthetic-sensitive TASK-like background potassium channel in rat arte-rial chemoreceptor cells. J. Physiol. 525(Pt 1): 135–142.

47. Wyatt, C.N. et al. 2007. AMP-activated protein kinasemediates carotid body excitation by hypoxia. J. Biol.

Chem. 282: 8092–8098.48. Peers, C. 1990. Hypoxic suppression of K+ currents

in type I carotid body cells: selective effect on theCa2(+)-activated K+ current. Neurosci. Lett. 119: 253–256.

49. Lopez-Lopez, J.R., C. Gonzalez & M.T. Perez-Garcia. 1997. Properties of ionic currents fromisolated adult rat carotid body chemoreceptorcells: effect of hypoxia. J. Physiol. 499(Pt 2): 429–441.

50. Otsubo, T. et al. 2006. Differential expression of oxy-gen sensitivity in voltage-dependent K channels ininbred strains of mice. Adv. Exp. Med. Biol. 580: 209–214; discussion 351–359.

51. Lopez-Lopez, J.R., D.A. De Luıs & C. Gonzalez.1993. Properties of a transient K+ current inchemoreceptor cells of rabbit carotid body. J. Physiol.

460: 15–32.52. Sanchez, D. et al. 2002. Molecular identification

of Kvalpha subunits that contribute to the oxygen-sensitive K+ current of chemoreceptor cells ofthe rabbit carotid body. J. Physiol. 542: 369–382.

53. Perez-Garcıa, M.T. et al. 2004. Characterization ofthe Kv channels of mouse carotid body chemorecep-tor cells and their role in oxygen sensing. J. Physiol.

557: 457–471.54. Buckler, K.J. & R.D. Vaughan-Jones. 1994. Effects

of hypoxia on membrane potential and intracellularcalcium in rat neonatal carotid body type I cells.J. Physiol. 476: 423–428.

Ion Channel Regulation by AMPK

Documents