Forward chemical genetic approach identifies new rolefor GAPDH in insulin signalingJaeki Min1,4, Yun Kyung Kim1,4, Patricia G Cipriani2, Mira Kang1, Sonya M Khersonsky1, Daniel P Walsh1,Ji-Young Lee1, Sherry Niessen3, John R Yates III3, Kristin Gunsalus2, Fabio Piano2 & Young-Tae Chang1

Insulin and insulin-like growth factor have an essential rolein growth, development and the maintenance of metabolichomeostasis, including glucose uptake from the bloodstream.Researchers have identified mutations in insulin receptorsthat cause severe insulin resistance1, and a temperature-sensitive daf-2 (a gene encoding an insulin receptor–likeprotein) mutant in Caenorhabditis elegans has served as aninsulin resistance model2. Here we report a forward chemicalgenetic approach with a tagged library that we used toidentify a small molecule, GAPDH segregator (GAPDS), thatsuppresses the dauer formation induced by the daf-2 mutant.Like insulin, GAPDS increased both glucose uptake and theconcentration of phosphatidylinositol-3,4,5-trisphosphate(PtdIns(3,4,5)P3) in mammalian preadipocytes. Usingaffinity matrices and RNA interference, we identifiedglyceraldehyde-3-phosphate dehydrogenase (GAPDH) asa GAPDS target. We discovered that GAPDH stimulatesphosphatase activity against not only PtdIns(3,4,5)P3 butalso PtdIns(4,5)P2. These results suggest that GAPDH isboth an active regulator in the phosphoinositide-mediatedsignaling pathway and a potential new target for insulinresistance treatment.

The forward chemical genetics uses chemical library screening toidentify phenotypic changes by targeting specific proteins. Thoughpromising, the current chemical genetic approach contains intrinsi-cally difficult steps in target identification. One of the difficulties isconjugating the active lead compound to an affinity matrix withoutreducing the lead’s activity, which requires laborious structure-activityrelationships (SAR) studies. To avoid this well-known problem, wehave developed a tagged library carrying a built-in linker from thebeginning of the synthesis and demonstrated its usefulness in severalexamples3–5. In contrast to our previous tagged library, which con-tained a free N-terminal linker (TG), we developed a new library inwhich we introduced benzoyl capping (TG-Bz) with the aim ofincreasing the bioavailability of the compounds through higherhydrophobicity and reducing general toxicity induced by the freeamine group (Fig. 1a). We used the new library (TG-Bz library,1,120 members, Supplementary Fig. 1 online) to identify compounds

that can regulate the insulin pathways in C. elegans, whose DAF-2pathway has been used as an insulin resistance model system6–9.

In C. elegans, the DAF-2 signaling pathway controls food uptake,metabolism, growth and life span through the regulation of age-1, anortholog of mammalian phosphatidylinositol-3-OH kinase (PI(3)K)(Supplementary Fig. 2 online). daf-2 (strain e1370) mutants carryinga point mutation in the kinase domain of the DAF-2 protein causeconstitutive dauer formation (Fig. 1b) at a restrictive temperature(25 1C, normal phenotype at the permissive temperature of 16 1C)8,9.Mutagenic and RNA interference (RNAi)-based screens suppressingdaf-2 (e1370) have identified regulators of the insulin pathway10. Weused a similar approach to identify small molecules that can overcomethe dauer-forming phenotype of daf-2 (e1370). To identify smallchemical rescuers of daf-2 (e1370), we soaked early L2 larval stageworms (grown at 16 1C) in a 100 mM TG-Bz library compoundsolution and incubated them at 25 1C for 4 d. One compound,GAPDS (1; Fig. 1a), was identified as the strongest chemical rescuer(18 ± 4.4% (s.d.) non-dauer from two independent experiments, n ¼80; Fig. 1b), whereas another compound, E3 (2; Fig. 1a), did not elicitany rescue and was chosen as a negative control in subsequentanalyses. This result suggests that GAPDS is able to either restore orbypass the lack of activity of DAF-2.

To validate the relevance of this activity in C. elegans to amammalian system, we tested whether GAPDS mimics insulin activityby inducing 3T3-L1 preadipocytes to uptake glucose11. We monitoredglucose uptake using a fluorescent derivative of glucose, 2-NBD-glucose (3)12, and observed a dose-dependent response with GAPDS(15 min) and an apparently low response with the control compoundE3 (Fig. 1c). Cotreatment of GAPDS and insulin (100 nM) resulted inan additive effect, but at the highest concentration of GAPDS(10 mM), the glucose uptake effect reached plateau. To furtherinvestigate the mechanism of GAPDS function, we examinedPtdIns(3,4,5)P3 concentrations by immunostaining upon chemicaltreatment. GAPDS (10 mM, 1 min) created a 2.2-fold increase in thePtdIns(3,4,5)P3 concentration (Fig. 1d,e). Together these results sup-port the idea that GAPDS stimulates insulin signaling throughelevation of PtdIns(3,4,5)P3 concentrations.

To identify targets of GAPDS, we immobilized the free-amino version of GAPDS (4) on activated agarose Affi-Gel 10 beads

Received 29 March; accepted 11 September; published online 19 November 2006; doi:10.1038/nchembio833

1Department of Chemistry and 2Department of Biology, New York University, New York, New York 10003, USA. 3Department of Cell Biology, The Scripps ResearchInstitute, La Jolla, California 92037, USA. 4These authors contributed equally to this work. Correspondence should be addressed to Y.-T.C. (yt.chang@nyu.edu).

NATURE CHEMICAL BIOLOGY VOLUME 3 NUMBER 1 JANUARY 2007 5 5

L E T T ERS©

200

6 N

atur

e P

ublis

hing

Gro

up h

ttp

://w

ww

.nat

ure.

com

/nat

urec

hem

ical

bio

log

y

(Bio-Rad) (Supplementary Methods online). It is noteworthy that aSAR analysis was not necessary for making the affinity matrix becausewe simply replaced the preexisting linker tag (the benzoyl group) withthe agarose beads. We used agarose beads treated with only ethano-lamine as a negative control matrix. After incubation of the resins withC. elegans extracts and washing with buffer, the total proteins boundto the resin were directly digested by trypsin, and the resultingpeptides were analyzed by LC-MS/MS13.

Cross-comparison of peptides eluted from the GAPDS resin withthose eluted from the control resins led to identification of 11 putativebinding partners for GAPDS (Supplementary Table 1 online). Toidentify the functionally relevant target protein, we performed anRNAi knockdown experiment against 14 genes that correspond to the11 GAPDS-specific binding proteins (14 gene candidates: GAPDH hasfour genes, gpd-1, gpd-2, gpd-3 and gpd-4). These tests identified twoof the four genes that encode GAPDH in C. elegans (K10B3.8, gpd-2:3.6%, n ¼ 250 and K10B3.7, gpd-3: 1.7%, n ¼ 238), whereas all otherRNAi knockdowns scored zero dauer escape (n 4 200).

GAPDH is one of the key enzymes in the glycolytic pathway.Although new functions of this enzyme have been discovered overthe last few decades, a direct regulating function of GAPDH on theinsulin signaling pathway has not been reported previously14. Whentested against purified rabbit muscle GAPDH in vitro, GAPDSinhibited the glycolytic enzymatic function of GAPDH with a half-maximal inhibitory concentration (IC50) of 15 mM, whereas E3 didnot show significant inhibition at concentrations higher than 50 mM.In comparison, a known GAPDH inhibitor, 6-hydroxydopamine(6-OHDA, 5)15, showed an IC50 of 8 mM in our enzymatic assay.Notably, 6-OHDA did not show an increase in glucose uptake inpreadipocytes (data not shown), which implies that the glycolyticactivity of GAPDH is not directly related to insulin signaling.

Next, we tested the effect of GAPDS on GAPDH complex status, asGAPDH forms a homotetramer for full enzymatic activity in thecell14. GAPDH (100 mg ml–1) was incubated with GAPDS and then

chemically cross-linked before SDS gel analysis16. GAPDH was almostdisassembled into monomers at low micromolar concentrations ofGAPDS, whereas neither E3 nor 6-OHDA and another GAPDHbinder, (–)-deprenyl (6)17, caused a significant change even at thehighest concentration (Fig. 2 and Supplementary Fig. 3 online). Wealso confirmed that GAPDS segregates tetrameric GAPDH to mono-mers in 3T3-L1 cell lysate by immunoprecipitation with a monoclonalantibody (clone 6C5) that specifically binds to GAPDH monomer18

(Supplementary Fig. 4 online). This segregation effect was suppressedby excess NAD+ (2 mM) in accordance with the previous report19.Therefore, we speculate that GAPDS mimics insulin signaling throughthe dissociation of the GAPDH tetramer into monomers, rather thanby inhibition of its glycolytic enzyme activity.

We next wanted to explore the ways in which GAPDH may belinked to insulin signaling. As described above, PtdIns(3,4,5)P3

a

N

N

N

HN

R1 R2

OO

HN

Bz

TG-Bz

N

N

N

HN

NH

NH

OO

HN

OH

OH

Bz

GAPDS (1)

N

N

N

HN

NH

NH

OO

HN

Bz

N

E3 (2)

b0.1 mm

Dauer control GAPDS

c 0 µM

0.4 µM

2 µM

10 µM

E30

10

Glu

cose

upt

ake

(RF

U)

20

30

GAPDS GAPDS withinsulin (100 nM)

d

0

1

2

3 * *

PIP

3 sig

nal i

nten

sity

Contro

l

Insu

lin

GAPDS E3

e

Control Insulin GAPDS E3

20 µm

Figure 1 Insulin-mimicking effects of GAPDS. (a) The chemical

structures of a representative TG-Bz library compound, GAPDS

and E3. (b) Inhibition of dauer formation by GAPDS. (c) Glucose

uptake by E3, GAPDS and insulin treatment. The error bars ares.d. and represent the range of four experiments. RFU, relative

fluorescence units. (d) Quantification of increased cellular

PtdIns(3,4,5)P3 signal upon treatment of insulin, GAPDS and E3,

measured as the intensity of immunofluorescence signal relative

to the intensity of anti-PtdIns(3,4,5)P3 immunofluorescence and

normalized to the DMSO control staining. Error bars show

s.d. from 12 images from three independent experiments

(*P o 0.0005). (e) Representative images of d.

Tetramer

Monomer

0.4 2 10

GAPDS

0.4 2 10

E3

0.4 2 10

6-OHDA

(µM)

Figure 2 GAPDH segregation by GAPDS. Purified GAPDH from rabbit

muscle was incubated with each compound for 30 min at 4 1C and then

cross-linked. Cross-linked samples were loaded on an SDS-PAGE and

visualized by silver staining.

L E T TERS

56 VOLUME 3 NUMBER 1 JANUARY 2007 NATURE CHEMICAL BIOLOGY

© 2

006

Nat

ure

Pub

lishi

ng G

roup

htt

p:/

/ww

w.n

atur

e.co

m/n

atur

eche

mic

alb

iolo

gy

concentration was elevated by GAPDS. The elevated concentrations ofPtdIns(3,4,5)P3 may be due to enhanced conversion of phosphatidyl-inositol bisphosphate (PIP2) to PtdIns(3,4,5)P3 or suppression ofPtdIns(3,4,5)P3 conversion to PIP2. We first checked the effect ofGAPDS on PI(3)K, one of the best known insulin signaling enzymes,which converts PtdIns(4,5)P2 into PtdIns(3,4,5)P3. However, thetyrosine phosphorylation level of the p85 subunit of PI(3)K, anindication of PI(3)K activation, did not show any change withGAPDS treatment (data not shown).

The other possibility is that suppression of lipid phosphataseactivity results in an increase in the half-life of PtdIns(3,4,5)P3. Todetermine whether phosphate hydrolysis activity was present, we usedfluorescent analogs of PtdIns(3,4,5)P3 (7), PtdIns(4,5)P2 (8) andPtdIns(3,4)P2 (9) as substrates and monitored the reactions withGAPDH by TLC20,21. In this test, purified GAPDH alone did not showany hydrolysis activity, which is evidence against the possibility thatGAPDH itself has phosphatase activity on phosphoinositides. Inpreadipocyte cell lysates, PtdIns(3,4,5)P3 and PtdIns(4,5)P2 weredephosphorylated to generate PIP2 and PIP as products, butPtdIns(3,4)P2 dephosphorylation was negligible (Fig. 3a). The con-version of PtdIns(3,4,5)P3 was diminished by the addition of GAPDSbut not E3 (Fig. 3b). To further test for a direct role of GAPDH, celllysates in which GAPDH was immunodepleted were compared withuntreated cell lysates. We observed a decreased PtdIns(3,4,5)P3 con-version to PIP2 in GAPDH-immunodepleted cell lysates; conversionwas restored upon addition of purified GAPDH. When GAPDH andGAPDS were added together, restoration of the phosphatase activitydid not occur, which indicates that GAPDH is an active regulator ofPtdIns(3,4,5)P3 concentrations. Notably, PtdIns(4,5)P2 conversion toPIP showed a pattern similar to that of PtdIns(3,4,5)P3, but the effectwas stronger (Fig. 3c–e). Finally, we immunostained PtdIns(4,5)P2 inwhole cells to determine whether GAPDS results in a change inPtdIns(4,5)P2 concentrations in vivo. Congruent with the TLC data,GAPDS showed a substantial increase in PtdIns(4,5)P2 concentrationsrelative to the control and E3 treatment (Fig. 3f,g and Supplementary

Fig. 5 online). It is noteworthy that insulin treatment decreased thePtdIns(4,5)P2 concentrations, as reported previously22.

PtdIns(3,4,5)P3 is a well established key regulator of the insulinsignaling pathway. PtdIns(4,5)P2 is the main precursor ofPtdIns(3,4,5)P3 and is itself an important regulator of insulin signalingin that it is a stimulator of the actin polymerization that mediatesoptimal movement and fusion of glucose transporter 4 (GLUT4)-containing vesicle membranes to the cell surface, which eventuallyfacilitate glucose uptake and reverse insulin resistance22,23. Therefore,GAPDS activity has a dual role in activating the insulin signalingpathway by elevating concentrations of both PtdIns(3,4,5)P3 andPtdIns(4,5)P2.

In summary, the present work describes a forward chemical geneticapproach using the daf-2 mutant in C. elegans to discover a small-molecule activator of the insulin signaling pathway and identify theprotein target (Supplementary Fig. 2). The discovered compound,GAPDS, rescues the dauer phenotype of daf-2 mutants of C. elegansand increases glucose uptake in mammalian preadipocytes. Wepropose that GAPDS stimulates insulin signaling by segregating thetetrameric GAPDH into monomers, and thereby suppresses theintrinsic GAPDH activation of phosphoinositide’s phosphatase acti-vity. Though GAPDH has long been recognized as having a keyrole in glycolysis, recent studies have shown additional functions ofthe protein, including uracil DNA glycosylase function, trans-cription activation, nuclear RNA export, DNA repair and kinase/phosphatase activity14; GAPDH also catalyzes membrane fusion andaugments calcium signaling24,25. This study proposes that GAPDH alsofunctions as an active regulator of phosphoinositide’s phosphatasesand suggests GAPDH as a new target for diabetes, obesity andaging research.

METHODSC. elegans primary screening. daf-2 mutants (strain e1370) were grown on

nematode growth media (NGM) agar plates with Escherichia coli (strain OP50)

at 16 1C8. Twenty synchronized early L2 animals were transferred into 96-well

PIP

PIP2

PIP3

Ptdlns

(3,4

,5)P 3

Ptdlns

(4,5

)P 2

Ptdlns

(3,4

)P 2 1 2 3 4 5 6 7

– –– + + +– – + +– – – +– – – –

–– – –– + –– – +

1 2 3 4 5 6 7

– –– + + +– – + +– – – +– – – –

–– – –– + –– – +

GAPDH depletionGAPDH add-backGAPDSE3

Origin

Solventfront

Ptdlns(3,4,5)P3 Ptdlns(4,5)P2 2

1.5

1

0.5

0Hyd

roly

zed

PIP

2H

ydro

lyze

d P

IP

1.5

1

0.5

0

0

1 2 3 4 5 6 7

1

1

2

2

3 4 5 6 7

Ant

i-Ptd

lns(

4,5)

P2

sign

al in

tens

ity

*

**

*

Contro

l

Insu

lin

GAPDS E3E3GAPDSInsulinControl

20 µm

a b c d

e

f gFigure 3 GAPDH and GAPDS effects

on phosphoinositide concentrations.

(a) Dephosphorylation of phosphoinositides by

preadipocyte cell lysates after 1 h of incubation.

(b,c) Analysis of chemical and GAPDH effects in

phosphatase activity against PtdIns(3,4,5)P3 (b)

and PtdIns(4,5)P2 (c) after 2 h of incubation.

(d,e) Quantitative analysis of hydrolyzed PIP2 (d)

and PIP (e) for b and c, calculated by comparing the

relative intensity of fluorescence signal from hydrolyzed spot over total spot. Error bars show s.d. from two independent experiments. (f) Immunostaining of

PtdIns(4,5)P2 upon treatment of insulin, GAPDS and E3. (g) Immunofluorescence intensity normalized to the DMSO control staining for f. Error bars shows.d. from 15 images from three independent experiments. *P o 0.01, **P o 0.0005 versus the control.

L E T T ERS

NATURE CHEMICAL BIOLOGY VOLUME 3 NUMBER 1 JANUARY 2007 5 7

© 2

006

Nat

ure

Pub

lishi

ng G

roup

htt

p:/

/ww

w.n

atur

e.co

m/n

atur

eche

mic

alb

iolo

gy

plates containing 100 mM of compound, 2% OP50 culture and M9 buffer

(200 ml final), and then incubated at 25 1C for 4 d. The number of young

adult–size animals was counted in solution, and pictures of the phenotypic

change were taken after transferring the animals to NGM agar plates.

Glucose uptake measurements. 3T3-L1 preadipocytes were grown in DMEM

(500 ml) containing 1% antibiotics and 10% heat-inactivated newborn calf

serum overnight in 48-well plates. Cells were washed with phosphate-buffered

saline (PBS) and further incubated in PBS for 5 min before compound

treatment. The PBS was replaced by a solution (200 ml) of test compound

along with 2-NBD-glucose (100 mM) (Molecular Probes), and the cells were

incubated for 15 min at 37 1C. After washing with PBS, cells were lysed with

potassium phosphate buffer (100 ml, 0.1 M, pH 10) containing 1% Triton X-

100, and then DMSO (50 ml) was added. A portion of the lysed solution

(120 ml) was transferred to a 96-well microtiter plate, and fluorescence

was measured in a fluorescence microplate reader (lex ¼ 466 nm and

lem ¼ 540 nm).

Immunostaining. Cells grown in the 96-well microtiter plate were incubated

with GAPDS (10 mM), insulin (400 nM) or E3 (10 mM) in PBS (100 ml) for

1 min, and DMSO was used as a control. Cells were fixed with 4%

formaldehyde and stained with fluorescein isothiocyanate (FITC)-labeled anti-

body to PtdIns(3,4,5)P3 (Echelon Biosciences Inc) or antibody to PtdIns(4,5)P2

(Assay Design Inc) as described26. Fluorescence was visualized by a Leica 2000

fluorescence microscope (lex ¼ 488 nm and lem ¼ 510 nm).

Affinity matrix with C. elegans extract. Worms (strain e1370) grown at 16 1C

were harvested with extract buffer (HEPES (50 mM, pH 7.4), NaCl (150 mM),

EDTA (2 mM)) and washed twice with extract buffer. The washed worm pellet

was resuspended in extract buffer including protease inhibitor cocktail (Sigma,

P8340) and then mechanically homogenized. After addition of Triton X-100

(1% final concentration), the lysates were centrifuged at 4 1C (14,000 r.p.m.,

30 min). Protein concentration was measured using a Bradford assay and

adjusted to 1 mg ml–1. Affinity matrix resins (50 ml) were washed three times

with ‘‘bead’’ buffer (HEPES (50 mM, pH 7.4), NaCl (250 mM), EDTA (5 mM),

protease inhibitor cocktail). The resins were resuspended in bead buffer

(150 ml), and then an equal volume of the protein extract (1 mg ml–1) was

added to the resin suspension; the final mixture was shaken at 4 1C for 1 h.

After the supernatant was removed, the beads were washed five times with bead

buffer. Protein identification by mass spectrometry was performed

as described13.

RNAi treatment. Bacteria containing clones for RNAi were spotted in each

NGM agar plate containing IPTG (6 mM) and ampicillin (50 mg ml–1). To

synchronize, 20 gravid adult worms were placed in each duplicated plate and

allowed to lay eggs for 2 h at 16 1C, and then removed. After 30 h of incubation

at 16 1C, the plates were transferred to 25 1C for dauer formation. Phenotypic

changes were monitored after 4 d.

Fluorescent phosphoinositide phosphatase assays. Fluorescent phosphoinosi-

tide TLC assays were performed as recently described27. All fluorescent

phosphoinositide substrates used in this study were obtained from Echelon

Research Laboratories. Phosphatase assays using cell lysates of 3T3-L1 pre-

adipocytes were conducted in assay buffer (23 ml, 50 mM ammonium carbo-

nate, pH 8.0 and 2 mM DTT) containing cell lysates (10 ml, 1 mg ml–1 protein

concentration) at 25 1C for 1 h. Reactions were initiated by the addition of a

C6-NBD6-phosphoinositide substrate (1 mg). The compound effects for the

phosphatase activity of cell lysates were examined with 100 mM of GAPDS or

E3 in 23 ml of assay buffer containing cell lysates (5 ml) at 25 1C for

2 h. Reactions were initiated by the addition of C6-NBD6-PtdIns(3,4,5)P3 or

C6-NBD6-PtdIns(4,5)P2. Phosphatase assays using GAPDH-immunodepleted

cell lysates were monitored for 2 h at 25 1C in 25 ml of assay buffer containing

substrates with or without exogenous GAPDH (1 mM). The compound effects

were investigated with GAPDS (100 mM) in the presence of GAPDH (1 mM)

under the same conditions. The reactions were monitored every 20 min using

TLC. Silica gel TLC plates (Sorbent Technologies) were soaked in a 1.2%

solution of potassium oxalate, air-dried in a fume hood for 10 min and placed

in a baking oven for 1 h at 180 1F. The mixtures of fluorescent phosphoinositide

assays were spotted onto the TLC plate, dried for 10 min in a fume hood and

developed in CHCl3/MeOH/acetone/glacial acetic acid/water (70:50:20:20:20).

Fluorescent lipids were visualized and scanned using an Alpha Array 7000

(Alpha Innotech).

Note: Supplementary information and chemical compound information is available onthe Nature Chemical Biology website.

ACKNOWLEDGMENTSWe gratefully acknowledge the support of the US National Institutes of Health(NIH) (R01-CA096912), and the US National Science Foundation equipmentgrants for the NMR (MRI-0116222) and the capillary LC ion trap massspectrometer (CHE-0234863). Components of this work were conductedin a Shared Instrumentation Facility constructed with support from ResearchFacilities Improvement grant C06 RR-16572 from the US National Centerfor Research Resources and the US NIH. J.R.Y. is supported by US NIHgrant P41RR11823-10.

COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.

Published online at http://www.nature.com/naturechemicalbiology

Reprints and permissions information is available online at http://npg.nature.com/

reprintsandpermissions/

1. Taylor, S.I. et al. Mutations in the insulin receptor gene. Endocr. Rev. 13, 566–595(1992).

2. Hanover, J.A. et al. A Caenorhabditis elegans model of insulin resistance: alteredmacronutrient storage and dauer formation in an OGT-1 knockout. Proc. Natl. Acad.Sci. USA 102, 11266–11271 (2005).

3. Khersonsky, S.M. et al. Facilitated forward chemical genetics using a tagged triazinelibrary and zebrafish embryo screening. J. Am. Chem. Soc. 125, 11804–11805(2003).

4. Williams, D. et al. Identification of compounds that bind mitochondrial F1F0 ATPaseby screening a triazine library for correction of albinism. Chem. Biol. 11, 1251–1259(2004).

5. Uttamchandani, M. et al. Microarrays of tagged combinatorial triazine libraries in thediscovery of small-molecule ligands of human IgG. J. Comb. Chem. 6, 862–868(2004).

6. Carroll, P.M., Dougherty, B., Ross-Macdonald, P., Browman, K. & FitzGerald, K. Modelsystems in drug discovery: chemical genetics meets genomics. Pharmacol. Ther. 99,183–220 (2003).

7. Choy, R.K. & Thomas, J.H. Fluoxetine-resistant mutants in C. elegans define a novelfamily of transmembrane proteins. Mol. Cell 4, 143–152 (1999).

8. Gems, D. et al. Two pleiotropic classes of daf-2 mutation affect larval arrest, adultbehavior, reproduction and longevity in Caenorhabditis elegans. Genetics 150,129–155 (1998).

9. Kimura, K.D., Tissenbaum, H.A., Liu, Y.X. & Ruvkun, G. daf-2, an insulin receptor-likegene that regulates longevity and diapause in Caenorhabditis elegans. Science 277,942–946 (1997).

10. Ashrafi, K. et al. Genome-wide RNAi analysis of Caenorhabditis elegans fat regulatorygenes. Nature 421, 268–272 (2003).

11. Nagashima, H. & Matsumura, F. 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD)-induceddown-regulation of glucose transporting activities in mouse 3T3–L1 preadipocyte.J. Environ. Sci. Health B 37, 1–14 (2002).

12. Itoh, Y., Abe, T., Takaoka, R. & Tanahashi, N. Fluorometric determination of glucoseutilization in neurons in vitro and in vivo. J. Cereb. Blood Flow Metab. 24, 993–1003(2004).

13. Cheeseman, I.M. et al. Implication of a novel multiprotein Dam1p complex in outerkinetochore function. J. Cell Biol. 155, 1137–1145 (2001).

14. Sirover, M.A. New insights into an old protein: the functional diversity of mammalianglyceraldehyde-3-phosphate dehydrogenase. Biochim. Biophys. Acta 1432, 159–184(1999).

15. Hayes, J.P. & Tipton, K.F. Interactions of the neurotoxin 6-hydroxydopaminewith glyceraldehyde-3-phosphate dehydrogenase. Toxicol. Lett. 128, 197–206(2002).

16. Bitan, G., Lomakin, A. & Teplow, D.B. Amyloid beta-protein oligomerization: prenu-cleation interactions revealed by photo-induced cross-linking of unmodified proteins.J. Biol. Chem. 276, 35176–35184 (2001).

17. Carlile, G.W. et al. Reduced apoptosis after nerve growth factor and serum withdrawal:conversion of tetrameric glyceraldehyde-3-phosphate dehydrogenase to a dimer.Mol. Pharmacol. 57, 2–12 (2000).

18. Grigorieva, J.A., Dainiak, M.B., Katrukha, A.G. & Muronetz, V.I. Antibodies tothe nonnative forms of D-glyceraldehyde-3-phosphate dehydrogenase: identification,purification, and influence on the renaturation of the enzyme. Arch. Biochem.Biophys. 369, 252–260 (1999).

19. Constantinides, S.M. & Deal, W.C., Jr. Reversible dissociation of tetrameric rabbitmuscle glyceraldehyde 3-phosphate dehydrogenase into dimers or monomers byadenosine triphosphate. J. Biol. Chem. 244, 5695–5702 (1969).

L E T TERS

58 VOLUME 3 NUMBER 1 JANUARY 2007 NATURE CHEMICAL BIOLOGY

© 2

006

Nat

ure

Pub

lishi

ng G

roup

htt

p:/

/ww

w.n

atur

e.co

m/n

atur

eche

mic

alb

iolo

gy

20. Pagliarini, D.J., Worby, C.A. & Dixon, J.E. A PTEN-like phosphatase with a novelsubstrate specificity. J. Biol. Chem. 279, 38590–38596 (2004).

21. Taylor, G.S. & Dixon, J.E. An assay for phosphoinositide phosphatases utilizingfluorescent substrates. Anal. Biochem. 295, 122–126 (2001).

22. Chen, G. et al. Protective effect of phosphatidylinositol 4,5-bisphosphateagainst cortical filamentous actin loss and insulin resistance induced by sustainedexposure of 3T3–L1 adipocytes to insulin. J. Biol. Chem. 279, 39705–39709(2004).

23. Strawbridge, A.B. & Elmendorf, J.S. Phosphatidylinositol 4,5-bisphosphate reversesendothelin-1-induced insulin resistance via an actin-dependent mechanism. Diabetes54, 1698–1705 (2005).

24. Glaser, P.E., Han, X. & Gross, R.W. Tubulin is the endogenous inhibitor of theglyceraldehyde 3-phosphate dehydrogenase isoform that catalyzes membrane fusion:implications for the coordinated regulation of glycolysis and membrane fusion. Proc.Natl. Acad. Sci. USA 99, 14104–14109 (2002).

25. Patterson, R.L., van Rossum, D.B., Kaplin, A.I., Barrow, R.K. & Snyder, S.H. Inositol1,4,5-trisphosphate receptor/GAPDH complex augments Ca2+ release via locallyderived NADH. Proc. Natl. Acad. Sci. USA 102, 1357–1359 (2005).

26. Niswender, K.D. et al. Immunocytochemical detection of phosphatidylinositol 3-kinaseactivation by insulin and leptin. J. Histochem. Cytochem. 51, 275–283 (2003).

27. Taylor, G.S. & Dixon, J.E. PTEN and myotubularins: families of phosphoinositidephosphatases. Methods Enzymol. 366, 43–56 (2003).

L E T T ERS

NATURE CHEMICAL BIOLOGY VOLUME 3 NUMBER 1 JANUARY 2007 5 9

© 2

006

Nat

ure

Pub

lishi

ng G

roup

htt

p:/

/ww

w.n

atur

e.co

m/n

atur

eche

mic

alb

iolo

gy