The capture of phosphoproteins by14-3-3 proteins mediates actionsof insulinShuai Chen, Silvia Synowsky, Michele Tinti and Carol MacKintosh

MRC Protein Phosphorylation Unit, College of Life Sciences, University of Dundee, Dundee DD1 5EH, UK

Review

How does signalling via PI3K–PKB (AKT)–mTORC1–

p70S6K and ERK–p90RSK mediate wide-ranging physio-logical responses to insulin? Quantitative proteomicsand biochemical experiments are revealing that thesesignalling pathways induce the phosphorylation of largeand overlapping sets of proteins, which are thencaptured by phosphoprotein-binding proteins named14-3-3 s. The 14-3-3 s are dimers that dock onto dual-phosphorylated sites in a configuration with specialsignalling and mechanical properties. They interact withthe Rab GTPase-activating proteins AS160 and TBC1D1to regulate glucose uptake into target tissues in re-sponse to insulin and energy stress. Dynamic patternsin the 14-3-3-binding phosphoproteome are providingnew insights into how insulin triggers coherent shifts inmetabolism that are integrated with other cellular re-sponse systems.

Established and emerging actions of insulinInsulin is best known for suppressing blood glucose levelsby driving glucose from the bloodstream into muscle andliver glycogen and into fat in adipose tissue, and forinhibiting glucose production from the liver. However,these actions are only part of an orchestrated shift in globalmetabolism induced by insulin action. Nutrients and ionssurge into our bloodstream after a meal, and insulin directsthe assimilation of glucose, metal ions, amino acids andlipids according to the demands of specialized tissues inways that ensure whole-body homeostasis, electrochemicalbalance of ions across membranes, and stoichiometricbalance in fluxes through metabolic pathways.

Although the main targets for insulin-regulated glucosehomeostasis are muscle, fat and liver, insulin also preparesother organs in their response to food. For example, it wasrecently discovered that insulin induces cytoskeletal remo-delling in kidney glomerular podocytes [1] and thus readiesthe kidney for increased urine filtration following a meal.Insulin influences vascular endothelium in ways that mayaffect tissue blood flow, and regulates fuel use and cellsurvival in the heart [2–5]. Insulin signalling to the hypo-thalamus, pituitary, bone, pancreas and reproductive sys-tems is also being recognised as critical to the whole-bodydistribution of resources [6–9].

Corresponding author: MacKintosh, C. (c.mackintosh@dundee.ac.uk).

1043-2760/$ – see front matter � 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.tem.2011.

The discovery that insulin has more widespread influ-ences than was previously realised implies that complica-tions of diabetes such as neuropathies, nephropathy, bonedisorders and heart disease might not be solely attribut-able to hyperglycaemia, but could be due in part to deregu-lated insulin action in the damaged tissues [10]. In anycase, an understanding of how insulin exerts its diverseeffects on different tissues is of enormous practical signifi-cance given the growing worldwide burden of insulin re-sistance and type 2 diabetes.

The immediate effects of insulin binding to its cognatetyrosine kinase receptor are relatively well understood.Phosphatidyinositol 3-kinase (PI3K)–protein kinase B(PKB, also known as AKT)– mammalian target of rapa-mycin complex 1 (mTORC1)–p70 ribosomal S6 kinase(p70S6K) signalling is activated, and in some cell typesinsulin also activates the Ras–Raf– extracellular signal-regulated kinase (ERK)–p90 ribosomal S6 kinase(p90RSK) pathway [11,12]. However, too few downstreamtargets are known to explain how these core signallingpathways stimulate diverse physiological responses toinsulin. Here, we review the roles of a family of phospho-protein-binding proteins named 14-3-3 s in mediating in-sulin responses. We outline how 14-3-3 proteins work andfocus on contributions of 14-3-3 s to key regulatory steps ininsulin-regulated glucose homeostasis. Recent develop-ments in 14-3-3 affinity capture and proteomics technolo-gies are also highlighted, which show promise as a way toidentify new intracellular targets and help to explain thewider actions of insulin.

14-3-3 dimers dock onto specific pairs ofphosphorylated sitesThe 14-3-3 s dock onto and regulate hundreds of phospho-proteins inside eukaryotic cells, including proteins thatare deregulated in diabetes, cancer and neurological dis-orders (Figure 1). Mammals have seven 14-3-3 genes(gamma, epsilon, beta, zeta, sigma, theta, tau), and theirpeculiar name refers to the discovery of the protein iso-forms as a cluster of spots in position 14-3-3 on an earlytwo-dimensional starch gel separation of bovine brainextract (Box 1) [13]. The numbers 14-3-3 have no function-al significance; perhaps a more meaningful number forthese proteins is ‘two’, for they are dimers that bind twophosphorylated residues. Often a 14-3-3 dimer docks ontoa pair of tandem phosphorylated sites on a target protein,

07.005 Trends in Endocrinology and Metabolism, November 2011, Vol. 22, No. 11 429

ITB2

Metabolic

MEFV

GP1BA IGF1R

TY3H

IRS1

Tumorigenesis Psychiatric/behaviour

TPH2 TAU

P53

CHK1

PRLR

MDM2

CTNB1

AKP13

BRAF

ABL1

GLI2

LRRK2 ACHA4

SLIK1COF1

CDN1B

Cardiovascular

TRENDS in Endocrinology & Metabolism

Figure 1. The current 14-3-3 diseasome encompasses tumourigenesis, cardiovascular, metabolic and neurological disorders. Some 480 human disease gene sets described

in the Genetic Association Database were recently mapped to indicate the degree of gene sharing among diseases [68]. We extracted from this disease map the gold

standard proteins with defined 14-3-3-binding phosphorylation sites [15,69]. The 14-3-3 binders map predominantly in four disease classes: tumourigenesis (11 proteins:

P53, BRAF, MDM2, CDN1B, AKP13, IRS1, PRLR, CTNB1, CHK1, ABL1, GLI2), cardiovascular (6 proteins: GP1BA, TY3H, ITB2, MEFV, IRS1, CDN1B), metabolic (5 proteins:

IRS1, GP1BA, TY3H, IGF1R, MEFV) and psychiatric or behaviour (11 proteins: TAU, IGF1R, GP1BA, P53, LRRK2, TY3H, ACHA4, MEFV, COF1, SLIK1, TPH2). The protein

overlap between these classes is illustrated here by Visant graphing (http://visant.bu.edu). This map is just a starting point, with indications that the concentration of 14-3-3-

binding proteins in the tumourigenesis, cardiovascular, metabolic and neurological disease classes will increase. For example, AS160 and TBC1D1, mutations of which are

found in subtypes of diabetes and obesity, respectively [56,61,62], were not in the Genetic Association Database when the disease map was prepared [68]. Moreover, further

proteins linked to these four disease classes are among those that exhibit affinity for 14-3-3 s in high-throughput proteomics studies; these await biochemical studies to

define the specificity of their interactions with 14-3-3 [23,24,26,27].

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

which is a configuration that confers interesting signallingand mechanical properties [14–16].

The act of docking onto two phosphorylated residuesturns a 14-3-3 dimer into a type of biochemical microcon-troller called a logic gate (Figure 2a). Logic gates arefundamental components of digital electronic circuits,where they perform OR and AND operations on two ormore inputs to produce a single output. The concept isuseful because an increasing number of proteins are beingidentified in which two tandem phosphorylated 14-3-3-binding sites can be phosphorylated by distinct proteinkinases [15]. One is the BCL-XL/BCL2-associated death

430

promoter (BAD) for which the 14-3-3 has been proposed asan OR gate because phosphorylation of either site is suffi-cient to recruit 14-3-3, which in turn promotes cell survivalby preventing BAD binding to the anti-apoptotic BCL-XL[17]. The 14-3-3 would be an AND gate if both sites must bephosphorylated, and by different kinases, for a 14-3-3 tobind and exert its action. Such devices would filter outsignalling noise because only the correct combination ofinputs would trigger an output. In biological systems, thedigital logic is likely to be more ‘fuzzy’ than in electroniccircuits, because the phosphorylation and binding kineticsoften differ for the two 14-3-3 interaction sites. Later in this

P

Kinase 2 Kinase 1

P

(a)

Mechanical action

(c)

(b)

‘AND’ ‘OR’

Hypotheticalclip function PP

Mask

PP

Lever

PP

Adapter

P P

Proliferativestimuli

Ca2+

Insulin & growth factors

Various stimuliin different cell types

PMA,platelet activators,

etc

Energydeprivation

PKA

Pims

PKC

AMPK

Akt/PKB

TRENDS in Endocrinology & Metabolism

Figure 2. 14-3-3 dimers are logic gates and mechanical devices that dock onto two CAMK- and AGC-phosphorylated sites. (a) Analogy between a 14-3-3 dimer and digital

logic gates. Digital AND and OR gates, which integrate two inputs into a single output, are depicted on the right. The left-hand side shows how the act of docking onto two

sites, which may be phosphorylated by different kinases, turns the 14-3-3 dimer into a digital logic gate whose output is a mechanical action. (b) The 14-3-3 s are dimeric

regulatory proteins that dock onto two phosphorylated residues, which are usually on the same target protein. Mechanically, this means that 14-3-3 can act as levers to force

a conformational change in a target, can mask a functional domain within the central groove, and can potentially clip back flanking regions to present a functional domain

for external interactions. When the two phosphorylated 14-3-3-binding sites are on different proteins, the 14-3-3 dimer would be an adapter that stabilises the interaction

between the two targets. 14-3-3 s are remarkable for having hundreds of phosphoprotein partners. (c) Phosphorylated 14-3-3-binding sites are created by CAMK and AGC

kinases. The frequency plot at the bottom was generated by aligning the sequences around more than 200 14-3-3-binding sites, centred on the phosphorylated 14-3-3-

binding residue, which can be serine or threonine [15]. Most 14-3-3-binding sites have arginine somewhere in the -3 to -5 positions relative to the phosphorylated site, and

just under half have a proline at +2. Such sites are generally phosphorylated by basophilic kinases in the AGC and CAMK sections of the human kinome, including central

mediators of insulin and nutrient signaling.

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

review we consider cases for which fuzzy 14-3-3 AND gatesmight integrate insulin signalling with other cellularsignals.

In electronic circuits, the output of a logic gate is anelectrical impulse. In intracellular circuits, a 14-3-3 dock-ing event triggers a mechanical action [18]. The two phos-phorylated 14-3-3-binding sites on target proteins aregenerally located within disordered regions and/or strad-dle a functional domain [15]. Their positions can give cluesabout how 14-3-3 dimers modulate the conformations andinteractions of their targets (Figure 2b). 14-3-3 s can act aslevers, whereby one site acts as a fulcrum for the 14-3-3dimer to exert a force on the other site, which causes a

conformational change in the target. For example, a 14-3-3dimer alters the conformations of phosphorylated tyrosinehydroxylase and the GTPase-activating protein RGS3 toactivate these enzymes [19,20]. In other cases, 14-3-3dimers mask a subcellular targeting sequence or a func-tional domain. For example 14-3-3 binds to PKB-phosphor-ylated FOXO4, which shields the DNA binding interfaceof this transcription factor and also interferes with itsnuclear localization [21]. In this way, 14-3-3 inhibits tran-scriptional activation when FOXO4 is phosphorylated inresponse to insulin and other stimuli. In principle, a 14-3-3dimer could clip back flanking regions to better present afunctional domain for external interactions, but there are

431

Box 1. 14-3-3 proteins

� 14-3-3 s are phosphoprotein-binding proteins with many regula-

tory roles in all eukaryotes.

� Dimers of curved L-shaped monomers (�30 kDa) form a central

groove that docks onto two phosphorylated sites on targets.

� AGC and CAMK kinases phosphorylate many of the known 14-3-3-

binding sites.

� The two phosphorylated docking sites must be spatially compa-

tible with fitting into either side of the central groove (�35 A

across).

� Different eukaryotic lineages have evolved sets of 14-3-3 isoforms

(e.g. gamma, epsilon, beta, zeta, sigma, theta and tau in

mammals).

� All 14-3-3 isoforms share a similar mode of action, but may differ

in their ability to form hetero- and homodimers, in affinities for

different targets, in expression patterns and in their post-

translational regulation.

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

no compelling examples of this. A 14-3-3 dimer may also actas an adapter linking two phosphorylated proteins, and aconvincing case of this mode of action is the stabilisationand activation of the hexomeric plasma membrane protonATPase in plants by phosphorylation-dependent interac-tions with three 14-3-3 dimers [22].

AGC and CAMK protein kinases phosphorylate many14-3-3-binding sites14-3-3 s interact with many phosphoproteins [23] and pro-vide a common mechanism for linking signalling pathwaysto cell metabolism, growth, proliferation and morphology.With so many targets and over 500 human protein kinases,it seems daunting to sort out which kinase phosphorylateswhich 14-3-3-binding sites and when. Fortunately, the taskis simplified because much of the kinome can be disregarded:14-3-3 s do not bind to phosphorylated tyrosines, nor gener-ally to phosphorylated serines and threonines that arefollowed by proline residues. This means that neithertyrosine kinases nor proline-directed enzymes, such asmitogen-activated protein (MAP) kinases and cyclin-depen-dent protein kinases, create docking sites for 14-3-3 s. ModeI RXX(pS/pT)XP and mode II RX(F/Y)X(pS)XP sequencemotifs showed optimal binding to 14-3-3 s in a screen of aphosphopeptide library [16]. In a recent survey of defined14-3-3-binding sites in mammalian proteins, mode IRXX(pS/pT)XP motifs dominate, although the +2 prolineresidue occurs in less than half (Figure 2c). Interestingly,these 14-3-3-binding sequences overlap with the specifici-ties of the protein kinase A/protein kinase G/protein kinaseC (AGC) group and Ca2+/calmodulin protein kinase (CAMK)group of protein kinases [15], which exhibit individual pre-ferences for phosphorylating motifs with basic residues inthe -3 to -5 positions relative to the phosphorylated residue.These basophilic kinases include proviral integration ofMMLV (PIM) kinases, protein kinase C (PKC) isoforms,the energy-sensing AMP-activated protein kinase (AMPK),and the insulin-activated PKB, SGK, p70S6K and p90RSKs(Figure 2c).

14-3-3 capture and release combined with differentialproteomicsThere are many proteins whose binding to 14-3-3 s isdynamically regulated by insulin, growth factors and

432

nutrients. Their identification is aided by recent develop-ments in high-resolution mass spectrometry and quantita-tive proteomics; thus, proteins within complex mixturescan be identified and quantified, and their post-transla-tional modification status can be analysed. For example,14-3-3 affinity capture can be used to isolate 14-3-3-bindingproteins from lysates of two or three sets of cells exposed todifferent agonists and/or inhibitors. Proteins that bindspecifically to the immobilised 14-3-3 s can be releasedby competitive elution with a synthetic 14-3-3-bindingphosphopeptide. After digestion of the proteins, the pep-tides are reacted with formaldehyde and borohydride todimethylate every primary amine, with the chemicals foreach preparation labelled with different stable isotopes togenerate a ‘light’, ‘intermediate’ or ‘heavy’ dimethyl moie-ty. Mass spectrometry is then used to identify the sequenceof the peptides, and to determine the ratio of the light,intermediate and heavy forms of each peptide. In this way,proteins whose peptides are more abundant in, for exam-ple, the 14-3-3-captured sample from insulin-stimulatedthan from unstimulated cells or inhibitor-treated cells canbe identified [24,25]. Because the dimethyl labelling step iscarried out after proteins are proteolytically digested, thismethod has potential to identify insulin-regulated proteinsisolated from their physiological target tissues. Anotherstrategy uses a similar principle, except that the differen-tial isotope label is introduced into proteins by metabolicincorporation of ‘light’, ‘intermediate’ or ‘heavy’ stableisotopically labelled amino acids that are fed to cells inculture (SILAC) [26,27]. Other quantitation methods, suchas differential isobaric tag for relative and absolute quan-titation (iTRAQ) labelling of protein and peptide samplesand even label-free methods, are also possible, and eachmethod has its advantages and disadvantages [27,28].

14-3-3 phosphoproteomics data indicate new regulatorymechanisms for insulinThe first experiments using 14-3-3-phosphoproteomicstechnology have identified many proteins that bind to14-3-3 s in response to signalling pathways that are acti-vated by insulin [24,26,27]. The insulin-regulated 14-3-3-binding sites have been verified for fewer proteins. Some ofthese proteins have poorly defined functions, includingcoiled-coil domain protein 6 (CCDC6), the E3 ubiquitinligase zinc-finger, RING-finger 2 (ZNRF2), and the sterilealpha motif (SAM) and SRC homology 3 (SH3) domainprotein SASH1 [24]. Intriguingly, it was found that asingle-nucleotide polymorphism in the SASH1 gene isassociated with diabetic nephropathy genes in AfricanAmericans in a genome-wide study, so study of this proteinmay reveal new disease mechanisms [10]. Other insulin-regulated 14-3-3-binding proteins provide more immediateinsights into the known actions of the hormone. Theseinclude proteins that regulate insulin-stimulated glucoseuptake into tissues, namely the Rab GTPase-activatingproteins AS160 (TBC1D4) and TBC1D1; the myosinMYO1C; the cardiac glycolytic activator PFKFB2; thetranscriptional regulator and E3 SUMO-protein ligasePIAS2 (Miz1); transcription factors FOXO1, FOXO3,FOXO4 and capicua; ECD3, which is involved in decappingmRNA as a step in its degradation; cell cycle and apoptotic

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

control proteins including CDKN1B (p27, Kip1), beta-cate-nin, BAD and BIM; and components of the insulin signal-ling PI3K–PKB–mTOR pathway itself, namely IRS2(insulin receptor substrate 2), PRAS40 (AKT1S1) andrictor [3,17,21,24,26,27,29–40]. Overall, a network of 14-3-3–phosphoprotein interactions is emerging that is pro-viding mechanistic insights into how insulin stimulatesglucose assimilation, activates cardiac glycolysis, and reg-ulates transcription, protein synthesis, feedback control ofinsulin signalling, cytoskeletal dynamics and other intra-cellular events.

It should be emphasised that none of the above-men-tioned proteins are regulated solely by insulin. Insulinsignalling and other pathways converge on several 14-3-3-binding sites, and some of these proteins are also modu-lated by energy and nutrient levels [41]. An intriguing ideais that 14-3-3 dimers as fuzzy AND logic gates mightintegrate insulin signals with other inputs. For example,there is an interplay between insulin and amino acids instimulating phosphorylation of the 14-3-3-binding site(s)on PRAS40 [42,43]. In addition, 14-3-3 interacts with thecardiac PFKFB2 via a high-affinity PKB-phosphorylatedsite and a second site that can be phosphorylated by bothPKB and AMPK, which is activated when the energy orATP status of cells is low, for example during ischaemia[3,44]. Insulin enhances the ability of cardiomyocytes tosurvive ischaemia, and a functional role for 14-3-3 isindicated by the finding that the PKB-mediated increasesin fructose-2-6-bisphosphate levels and glycolysis areprevented by a cell-permeable 14-3-3-binding phosphopep-tide in cells containing cardiac PFKFB2 [3].

Another key physiological process in which insulin sig-nalling is integrated with other inputs is the uptake ofglucose into muscles [45]. The goal of improving glucosecontrol in individuals means that the underlying mecha-nisms are under intense research, and the next sectionoutlines the contributions of two 14-3-3-binding proteins,the related Rab GTPase activating proteins (Rab-GAPs)AS160 (TBC1D4) and TBC1D1. Both proteins are regulat-ed by multiple phosphorylations [46,47] and details ofrelevant kinases and the roles of some sites are stillcontroversial, a situation that is being resolved with im-proved reagents for phosphosite identification and quanti-fication. Readers should therefore note that we do notattempt to be comprehensive here, but present a personal14-3-3-centred perspective with a testable hypothesis.

Differential regulation of AS160 and TBC1D1Insulin promotes glucose uptake into muscles and fatthrough GLUT4 glucose transporters [48]. In unstimulatedcells, GLUT4 is localised to internal storage vesicles thatmust fuse with the plasma membrane before glucose canflow into cells. GLUT4 trafficking to the plasma membraneis stimulated by insulin signalling mainly via the PI3K–

PKB pathway, and impairment of this process is an earlysign of insulin resistance and type 2 diabetes [46,48]. Inskeletal and cardiac muscle, GLUT4 trafficking is alsoactivated by contraction, which activates protein kinases,including AMPK, which is activated by lowering ATPlevels [49]. There is considerable interest in understandinghow exercise, as well as drugs that activate AMPK, restore

glucose homeostasis in individuals with insulin resistance[50].

GLUT4 trafficking requires active GTP-bound Rab pro-teins. AS160 and the closely related Rab-GAP, TBC1D1,promote GTP hydrolysis by Rabs. It has been proposed thatthese Rab-GAPs are somehow inactivated by phosphoryla-tion via insulin and AMPK signalling, which allows therelevant Rabs to be loaded with GTP to facilitate GLUT4trafficking [51–54]. AS160 and TBC1D1 have similar do-main architectures, and both proteins have two 14-3-3-binding sites that are located within clusters of multisitephosphorylation in a part of the protein that also has so-called PTB domains (Figure 3). Beyond this superficialsimilarity, however, some key regulatory details forAS160 and TBC1D1 are distinct.

AS160 is named for its discovery as an AKT (PKB)substrate, and the effects of overexpression of a nonpho-sphorylatable mutant [AS160(4P)] first implicated thisprotein in GLUT4 trafficking [53,55]. Furthermore, a re-cent genetic analysis identified patients with severe insu-lin resistance during puberty who carry a premature stopmutation in one allele of AS160 [56], which underscores theneed to define precisely how AS160 regulates glucosehomeostasis. Insulin stimulates phosphorylation of severalsites of AS160 and induces its binding to 14-3-3 s mainlythrough PKB-phosphorylated Thr642, with support fromphospho-Ser341, which can be phosphorylated by p90RSKand PKB [29,37,53]. A form of AS160 that binds 14-3-3constitutively, generated by genetic mutation, could over-ride the inhibitory effect of AS160(4P) on GLUT4 translo-cation in adipocytes [29,37,53]. In addition, insulin-stimulated fusion of GLUT4 vesicles with the plasmamembrane was prevented by blocking the AS160–14-3-3interaction in cell-free assays, which further indicates afunctional role for 14-3-3 in this process [57]. It was ini-tially proposed that Thr642 on AS160 was a convergencepoint for phosphorylation by both PKB and AMPK [58].However, in vitro and cell culture studies suggest thatwhereas AMPK may phosphorylate other residues, AMPKdoes not seem to be a physiological Thr642 kinase [29,59].Therefore, 14-3-3 binds to phospho-Thr642 and phospho-Ser341 in response to insulin and growth factors, and notAMPK activators (Figure 3).

In contrast to AS160, AMPK has a clear role in mediat-ing 14-3-3 binding to the related TBC1D1 [60]. TBC1D1was originally identified as a candidate gene underlyingsevere familial female obesity [61] and its natural deletionin mice resulted in leanness and protection against diet-induced obesity [62]. TBC1D1 and AS160 share a similarsubstrate preference towards Rabs, at least in vitro [63],and overexpression of a form of TBC1D1 mutated at threeof its phosphorylation sites interfered with GLUT4 trans-location in 3T3-L1 adipocytes [64].

In cultured L6 myotubes and intact skeletal muscles,whereas AS160 is phosphorylated on Ser341 and Thr642and binds to 14-3-3 s in response to insulin, TBC1D1 isphosphorylated on Ser237 and binds to 14-3-3 s in responseto acute exercise [65] and pharmacological AMPK activa-tors [60,66]. Insulin and growth factors also stimulate thephosphorylation of a second 14-3-3-binding site at Thr596of TBC1D1, which is similar to Thr642 of AS160; thus,

433

14-3-3 dimer

AS160

GLUT4 trafficking

Glucose homeostasis

InsulinAMPK activators

14-3-3 dimer

TBC1D1

PKB/p90RSK

S341S237 T596

S585?S566

S565S507S263S235

1 96 153 301 373 721 800757

994 1168

S318S570

S588S666

S751

1121 1299918876

8324393671801211

R363XR125W

55aa insert

T568

T642

PKBAMPK PKB

PTB1

PTB2 Rab GAP

CB

D

TRENDS in Endocrinology & Metabolism

PTB1

PTB2 Rab GAP

CB

D

Figure 3. Hypothesis of complementary regulation of glucose homeostasis by AS160 and TBC1D1 in skeletal muscles. Domains, motifs and disease-associated mutations

(the latter in red) are shown in bar diagrams of AS160 and TBC1D1 (see the text for sources). PTB1 and PTB2 are phosphotyrosine interaction domains (by sequence,

although their interaction partners have not been reported); CBD is a calcium-regulated region; and Rab GAP is the Rab GTPase-activating domain. In the hypothesis shown

here, AMPK controls GLUT4 trafficking through Ser237 phosphorylation and consequent docking of 14-3-3 onto the TBC1D1 protein, whereas the insulin–PKB (AKT)

pathway regulates surface translocation of GLUT4 via Thr642 phosphorylation and consequent 14-3-3 binding of the AS160 protein. This hypothesis does not take into

account that both TBC1D1 and AS160 also receive other regulatory inputs via the other phosphorylations that are clustered close to the 14-3-3-binding sites on these

proteins. Phosphorylated residues are indicated by numbers above the bar diagrams, with the 14-3-3-binding sites in yellow. R125W is an obesity-linked mutation of

TBC1D1 [61], and R363W is a diabetes-linked mutation of AS160 [56].

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

TBC1D1 seems to be a dual-input 14-3-3 logic gate(Figure 3). The functional significance of this moleculararrangement is not yet clear, however, and, at least inmuscles, phosphorylation of Thr596 alone is insufficient for14-3-3 binding [60,65].

Complementary roles of AS160 and TBC1D1 for insulinand AMPK activatorsWe propose that 14-3-3 binding to AS160 regulates glucosehomeostasis in response to insulin, whereas 14-3-3 bindingto TBC1D1 regulates this process in response to AMPK-activating stimuli (Figure 3). To test this hypothesis ofcomplementary roles for AS160–14-3-3 and TBC1D1–14-3-3 complexes in GLUT4 regulation, mice with knock-inmutations in the 14-3-3-binding phosphorylation sites inAS160 and TBC1D1 must be generated. Unlike studiesusing ectopic overexpression, introduction of a point mu-tation of a phospho-Ser/Thr residue to a nonphosphoryla-table residue using knock-in genetic technology generallyresults in expression of mutant proteins at endogenouslevels [67] (provided that mutations do not cause unexpect-ed outcomes such as destabilisation of the protein), withoutexerting dominant-negative effects or competition fromwild-type proteins.

First, a mouse model with Thr649 on AS160 (equivalentto Thr642 on human AS160) substituted by alaninewas generated to prevent insulin-stimulated binding to14-3-3 in vivo [67]. The most striking phenotype of theAS160-Thr649Ala knock-in mice is their altered glucose

434

homeostasis, including glucose intolerance and reducedinsulin sensitivity. Despite elevated total GLUT4 proteinlevels in skeletal muscles and fat tissues of the knock-ins,GLUT4 is not efficiently localised to the cell surface, whichresults in a reduced glucose uptake rate in muscle in re-sponse to insulin [67]. Together, these findings confirm thatinsulin-stimulated Thr649 phosphorylation and 14-3-3binding to AS160 in muscle has a key role in glucosehomeostasis.

Further experiments are needed to address the com-plexity of the glucose uptake process, including genetictests of the roles of the AMPK and insulin-regulated 14-3-3-binding sites on TBC1D1. For wider health implications,we need to know if the knock-in mice display alteredphysiological responses to high-fat diets and ageing. Willthe phenotype of TBC1D1 mice provide an insight into theobesity of individuals with mutations or polymorphisms inthis protein or gene? Would genetic crosses between theknock-in AS160 and TBC1D1 mice and ‘diabetes loci’reveal synergism with respect to insulin resistance?Answers to such questions would help to determine wheth-er it would be useful to devise drugs that mimic the effectsof insulin and exercise on AS160 and TBC1D1 as means toimproving blood glucose management.

The future: 14-3-3-phosphoproteome barcodesMore general lessons may also be learned from this narrowfocus on AS160 and TBC1D1. Earlier in this review, wealluded to the possibility of using 14-3-3 capture and

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

phosphoproteomics to identify the commonalities and dif-ferences in insulin targets in specialised organs of the body.Another exciting issue would be the potential of developingthese technologies to track which signalling pathways areactivated in cells and tissues under normal conditions andin response to therapeutic drugs. As exemplified by thecontrast between AS160 and TBC1D1, each protein has itsown signalling signature, meaning its own pattern ofresponsiveness to one, two or many signalling pathways.In turn, this means that insulin, growth factors and otherstimuli, which activate signalling pathways to differentextents, will each induce the phosphorylation and 14-3-3binding of distinct, but overlapping, subsets of targets. Inother words, each stimulus will have its own ‘barcode’within the 14-3-3 phosphoproteome in individual cellsand tissues. Definition of the commonalities and differ-ences in insulin responses in specialised organs and cells ofthe body, and any further overlaps between the insulin-regulated 14-3-3 phosphoproteome and emerging geneticmaps of diabetes and other diseases, will be importantpriorities for future metabolic research (Figure 3).

References1 Welsh, G.I. et al. (2010) Insulin signaling to the glomerular podocyte is

critical for normal kidney function. Cell Metab. 12, 329–3402 Mouton, V. et al. (2010) Heart 6-phosphofructo-2-kinase activation by

insulin requires PKB (protein kinase B), but not SGK3 (serum- andglucocorticoid-induced protein kinase 3). Biochem. J. 431, 267–275

3 Pozuelo Rubio, M. et al. (2003) 14-3-3 s regulate fructose-2,6-bisphosphate levels by binding to PKB-phosphorylated cardiacfructose-2,6-bisphosphate kinase/phosphatase. EMBO J. 22, 3514–3523

4 Singhal, A.K. et al. (2010) Role of endothelial cells in myocardialischemia-reperfusion injury. Vasc. Dis. Prev. 7, 1–14

5 Yu, Q. et al. (2011) Insulin says NO to cardiovascular disease.Cardiovasc. Res. 89, 516–524

6 Brothers, K.J. et al. (2010) Rescue of obesity-induced infertility infemale mice due to a pituitary-specific knockout of the insulinreceptor. Cell Metab. 12, 295–305

7 Ferron, M. et al. (2010) Insulin signaling in osteoblasts integrates boneremodeling and energy metabolism. Cell 142, 296–308

8 Fulzele, K. et al. (2010) Insulin receptor signaling in osteoblastsregulates postnatal bone acquisition and body composition. Cell 142,309–319

9 Suzuki, R. et al. (2010) Diabetes and insulin in regulation of braincholesterol metabolism. Cell Metab. 12, 567–579

10 McDonough, C.W. et al. (2011) A genome-wide association study fordiabetic nephropathy genes in African Americans. Kidney Int. 79,563–572

11 Chen, S. and Mackintosh, C. (2009) Differential regulation of NHE1phosphorylation and glucose uptake by inhibitors of the ERK pathwayand p90RSK in 3T3-L1 adipocytes. Cell Signal. 21, 1984–1993

12 Cohen, P. (2006) The twentieth century struggle to decipher insulinsignalling. Nat. Rev. Mol. Cell Biol. 7, 867–873

13 Aitken, A. (2006) 14-3-3 proteins: a historic overview. Semin. CancerBiol. 16, 162–172

14 Gardino, A.K. et al. (2006) Structural determinants of 14-3-3 bindingspecificities and regulation of subcellular localization of 14-3-3-ligandcomplexes: a comparison of the X-ray crystal structures of all human14-3-3 isoforms. Semin. Cancer Biol. 16, 173–182

15 Johnson, C. et al. (2010) Bioinformatic and experimental survey of 14-3-3-binding sites. Biochem. J. 427, 69–78

16 Yaffe, M.B. et al. (1997) The structural basis for 14-3-3:phosphopeptidebinding specificity. Cell 91, 961–971

17 She, Q.B. et al. (2005) The BAD protein integrates survival signaling byEGFR/MAPK and PI3K/Akt kinase pathways in PTEN-deficient tumorcells. Cancer Cell 8, 287–297

18 Yaffe, M.B. (2002) How do 14-3-3 proteins work? Gatekeeperphosphorylation and the molecular anvil hypothesis. FEBS Lett.513, 53–57

19 Obsilova, V. et al. (2008) The 14-3-3 protein affects the conformation ofthe regulatory domain of human tyrosine hydroxylase. Biochemistry47, 1768–1777

20 Rezabkova, L. et al. (2010) 14-3-3 protein interacts with and affects thestructure of RGS domain of regulator of G protein signaling 3 (RGS3).J. Struct. Biol. 170, 451–461

21 Silhan, J. et al. (2009) 14-3-3 protein masks the DNA binding interfaceof forkhead transcription factor FOXO4. J. Biol. Chem. 284, 19349–

1936022 Ottmann, C. et al. (2007) Structure of a 14-3-3 coordinated hexamer of

the plant plasma membrane H+-ATPase by combining X-raycrystallography and electron cryomicroscopy. Mol. Cell 25, 427–440

23 Johnson, C. et al. (2011) Visualization and biochemical analyses of theemerging mammalian 14-3-3 phosphoproteome. Mol. Cell. ProteomicsDOI: 10.1074/mcp.M110.005751

24 Dubois, F. et al. (2009) Differential 14-3-3 affinity capture reveals newdownstream targets of phosphatidylinositol 3-kinase signaling. Mol.Cell. Proteomics 8, 2487–2499

25 Hsu, J.L. et al. (2006) Dimethyl multiplexed labeling combined withmicrocolumn separation and MS analysis for time course study inproteomics. Electrophoresis 27, 3652–3660

26 Larance, M. et al. (2010) Global phosphoproteomics identifies a majorrole for AKT and 14-3-3 in regulating EDC3. Mol. Cell. Proteomics 9,682–694

27 Yip, M.F. et al. (2008) CaMKII-mediated phosphorylation of the myosinmotor Myo1c is required for insulin-stimulated GLUT4 translocationin adipocytes. Cell Metab. 8, 384–398

28 Gouw, J.W. et al. (2010) Quantitative proteomics by metabolic labelingof model organisms. Mol. Cell. Proteomics 9, 11–24

29 Geraghty, K.M. et al. (2007) Regulation of multisite phosphorylationand 14-3-3 binding of AS160 in response to IGF-1, EGF, PMA andAICAR. Biochem. J. 407, 231–241

30 Julien, L.A. et al. (2010) mTORC1-activated S6K1 phosphorylatesRictor on threonine 1135 and regulates mTORC2 signaling. Mol.Cell. Biol. 30, 908–921

31 Kovacina, K.S. et al. (2003) Identification of a proline-rich Aktsubstrate as a 14-3-3 binding partner. J. Biol. Chem. 278, 10189–10194

32 Dissanayake, K. et al. (2011) ERK/p90(RSK)/14-3-3 signalling has animpact on expression of PEA3 Ets transcription factors via thetranscriptional repressor capicua. Biochem. J. 433, 515–525

33 Easley, C.A., IV et al. (2010) mTOR-mediated activation of p70 S6Kinduces differentiation of pluripotent human embryonic stem cells. CellReprogram. 12, 263–273

34 Fang, D. et al. (2007) Phosphorylation of beta-catenin by AKT promotesbeta-catenin transcriptional activity. J. Biol. Chem. 282, 11221–

1122935 Harthill, J.E. et al. (2002) Regulation of the 14-3-3-binding protein p39

by growth factors and nutrients in rat PC12 pheochromocytoma cells.Biochem. J. 368, 565–572

36 Qi, X.J. et al. (2006) Evidence that Ser87 of BimEL is phosphorylatedby Akt and regulates BimEL apoptotic function. J. Biol. Chem. 281,813–823

37 Ramm, G. et al. (2006) A role for 14-3-3 in insulin-stimulated GLUT4translocation through its interaction with the RabGAP AS160. J. Biol.Chem. 281, 29174–29180

38 Sekimoto, T. et al. (2004) 14-3-3 suppresses the nuclear localization ofthreonine 157-phosphorylated p27Kip1. EMBO J. 23, 1934–1942

39 Treins, C. et al. (2010) Rictor is a novel target of p70 S6 kinase-1.Oncogene 29, 1003–1016

40 Wanzel, M. et al. (2005) Akt and 14-3-3eta regulate Miz1 to control cell-cycle arrest after DNA damage. Nat. Cell Biol. 7, 30–41

41 Lee, T.Y. et al. (2011) RegPhos: a system to explore the protein kinase-substrate phosphorylation network in humans. Nucleic Acids Res. 39,D777–D787

42 Nascimento, E.B. and Ouwens, D.M. (2009) PRAS40: target ormodulator of mTORC1 signalling and insulin action? Arch. Physiol.Biochem. 115, 163–175

43 Oshiro, N. et al. (2007) The proline-rich Akt substrate of 40 kDa(PRAS40) is a physiological substrate of mammalian target ofrapamycin complex 1. J. Biol. Chem. 282, 20329–20339

44 Marsin, A.S. et al. (2000) Phosphorylation and activation of heart PFK-2 by AMPK has a role in the stimulation of glycolysis during ischaemia.Curr. Biol. 10, 1247–1255

435

Review Trends in Endocrinology and Metabolism November 2011, Vol. 22, No. 11

45 Bryant, N.J. et al. (2002) Regulated transport of the glucosetransporter GLUT4. Nat. Rev. Mol. Cell Biol. 3, 267–277

46 Sakamoto, K. and Holman, G.D. (2008) Emerging role for AS160/TBC1D4 and TBC1D1 in the regulation of GLUT4 traffic. Am. J.Physiol. Endocrinol. Metab. 295, E29–E37

47 Rowland, A.F. et al. (2011) Mapping insulin/GLUT4 circuitry. Traffic12, 672–681

48 Huang, S. and Czech, M.P. (2007) The GLUT4 glucose transporter. CellMetab. 5, 237–252

49 Lund, S. et al. (1995) Contraction stimulates translocation ofglucose transporter GLUT4 in skeletal muscle through amechanism distinct from that of insulin. Proc. Natl. Acad. Sci.U.S.A. 92, 5817–5821

50 Hardie, D.G. (2007) AMP-activated protein kinase as a drug target.Annu. Rev. Pharmacol. Toxicol. 47, 185–210

51 Eguez, L. et al. (2005) Full intracellular retention of GLUT4requires AS160 Rab GTPase activating protein. Cell Metab. 2, 263–

27252 Larance, M. et al. (2005) Characterization of the role of the Rab

GTPase-activating protein AS160 in insulin-regulated GLUT4trafficking. J. Biol. Chem. 280, 37803–37813

53 Sano, H. et al. (2003) Insulin-stimulated phosphorylation of a RabGTPase-activating protein regulates GLUT4 translocation. J. Biol.Chem. 278, 14599–14602

54 Stockli, J. et al. (2008) Regulation of glucose transporter 4 translocationby the Rab guanosine triphosphatase-activating protein AS160/TBC1D4: role of phosphorylation and membrane association. Mol.Endocrinol. 22, 2703–2715

55 Kane, S. et al. (2002) A method to identify serine kinase substrates. Aktphosphorylates a novel adipocyte protein with a Rab GTPase-activating protein (GAP) domain. J. Biol. Chem. 277, 22115–22118

56 Dash, S. et al. (2009) A truncation mutation in TBC1D4 in a family withacanthosis nigricans and postprandial hyperinsulinemia. Proc. Natl.Acad. Sci. U.S.A. 106, 9350–9355

57 Koumanov, F. et al. (2011) PTB-domain constructs of AS160 inhibitinsulin-stimulated GLUT4 vesicle fusion with plasma membrane. J.Biol. Chem. 286, 16574–16582

436

58 Cartee, G.D. and Wojtaszewski, J.F. (2007) Role of Akt substrate of 160kDa in insulin-stimulated and contraction-stimulated glucosetransport. Appl. Physiol. Nutr. Metab. 32, 557–566

59 Treebak, J.T. et al. (2010) Identification of a novel phosphorylation siteon TBC1D4 regulated by AMP-activated protein kinase in skeletalmuscle. Am. J. Physiol. Cell Physiol. 298, C377–C385

60 Chen, S. et al. (2008) Complementary regulation of TBC1D1 and AS160by growth factors, insulin and AMPK activators. Biochem. J. 409, 449–

45961 Stone, S. et al. (2006) TBC1D1 is a candidate for a severe obesity gene

and evidence for a gene/gene interaction in obesity predisposition.Hum. Mol. Genet. 15, 2709–2720

62 Chadt, A. et al. (2008) Tbc1d1 mutation in lean mouse strain confersleanness and protects from diet-induced obesity. Nat. Genet. 40, 1354–

135963 Roach, W.G. et al. (2007) Substrate specificity and effect on GLUT4

translocation of the Rab GTPase-activating protein Tbc1d1. Biochem.J. 403, 353–358

64 Peck, G.R. et al. (2009) Insulin-stimulated phosphorylation of the RabGTPase-activating protein TBC1D1 regulates GLUT4 translocation. J.Biol. Chem. 284, 30016–30023

65 Frosig, C. et al. (2010) Exercise-induced TBC1D1 Ser237phosphorylation and 14-3-3 protein binding capacity in humanskeletal muscle. J. Physiol. 588, 4539–4548

66 Pehmoller, C. et al. (2009) Genetic disruption of AMPK signalingabolishes both contraction- and insulin-stimulated TBC1D1phosphorylation and 14-3-3 binding in mouse skeletal muscle. Am.J. Physiol. Endocrinol. Metab. 297, E665–E675

67 Chen, S. et al. (2011) Mice with AS160/TBC1D4-Thr649Ala knockinmutation are glucose intolerant with reduced insulin sensitivity andaltered GLUT4 trafficking. Cell Metab. 13, 68–79

68 Zhang, Y. et al. (2010) Systematic analysis, comparison, andintegration of disease based human genetic association data andmouse genetic phenotypic information. BMC Med. Genomics 3, 1

69 Dzamko, N. et al. (2010) Inhibition of LRRK2 kinase activity leads todephosphorylation of Ser910/Ser935, disruption of 14-3-3 binding andaltered cytoplasmic localization. Biochem. J. 430, 405–413