A glucose-responsive transcription factor that

regulates carbohydrate metabolism in the liver

Hiromi Yamashita*†, Makoto Takenoshita*, Masaharu Sakurai*, Richard K. Bruick*, William J. Henzel‡,
Wendy Shillinglaw‡, David Arnot‡, and Kosaku Uyeda*§
*Dallas Veterans Affairs Medical Center and Department of Biochemistry, University of Texas Southwestern Medical Center, 4500 South Lancaster Road,
Dallas, TX 75216; and ‡Genentech Inc., South San Francisco, CA 94080

Communicated by Steven L. McKnight, University of Texas Southwestern Medical Center, Dallas, TX, June 6, 2001 (received for review April 6, 2001)

Carbohydrates mediate their conversion to triglycerides in the liver tribution, specificity, and carbohydrate responsiveness of ChRE-
by promoting both rapid posttranslational activation of rate- BP’s DNA-binding activity are consistent with those of the
limiting glycolytic and lipogenic enzymes and transcriptional in- putative glucose-responsive transcription factor. Furthermore,
duction of the genes encoding many of these same enzymes. The ChREBP contains several consensus phosphorylation sites that
mechanism by which elevated carbohydrate levels affect transcrip- may regulate the activity of this protein. These data suggest that
tion of these genes remains unknown. Here we report the purifi- ChREBP is likely responsible for mediating carbohydrate in-
cation and identification of a transcription factor that recognizes duction of transcription of the LPK gene, and possibly other
the carbohydrate response element (ChRE) within the promoter of genes, in the liver and may influence long-term fat accumulation
the L-type pyruvate kinase (LPK) gene. The DNA-binding activity of resulting from a high-carbohydrate diet.
this ChRE-binding protein (ChREBP) in rat livers is specifically
induced by a high carbohydrate diet. ChREBP’s DNA-binding spec- Materials and Methods
ificity in vitro precisely correlates with promoter activity in vivo. Materials. Okadaic acid was purchased from Boehringer Mann-
Furthermore, forced ChREBP overexpression in primary hepato- heim. The cAMP-dependent protein kinase (PKA) catalytic
cytes activates transcription from the L-type Pyruvate kinase pro- subunit and PKA inhibitor were purchased from New England
moter in response to high glucose levels. The DNA-binding activity Biolabs. The protein phosphatase 2A catalytic subunit was
of ChREBP can be modulated in vitro by means of changes in its purchased from Bio-Rad. All other chemicals were reagent
phosphorylation state, suggesting a possible mode of glucose- grade and obtained from commercial sources.
responsive regulation. ChREBP is likely critical for the optimal
long-term storage of excess carbohydrates as fats, and may con- Preparation of Nuclear Extracts. Rats were starved for 48 h and
tribute to the imbalance between nutrient utilization and storage then fed 24 h with either the National Institutes of Health
characteristic of obesity. high-sucrose diet lab chow (by weight: 20% casein, 60.2%
sucrose, 15% cellulose, 2.25% minerals, and 2.25% vitamins)
(16), or a high-fat diet without starch (31% casein, 30.5%
U ntil recently, evolutionary pressures have favored the ability
to efficiently store nutrients as fat during periods of abun-
dant food supply as a safeguard against periodic famine (1).
cellulose, 7% minerals, 1.5% vitamins, 1.5% corn oil, 1.5%
peanut oil, and 27% lard) (17). Liver nuclear extracts were
Coupled with dramatic changes in modern lifestyle and food prepared according to Hattori et al. (18) with minor modifica-
consumption, these ‘‘thrifty genes’’ may now contribute to health tions (13). Fresh rat livers (⬇10 g) were homogenized with a
defects suffered by as many as half of the American population Potter–Elvehjem homogenizer in extraction buffer (10 mM
presently overweight or obese (2). The liver is the principal organ Hepes, pH 7.9兾10 mM KCl兾0.1 mM EDTA兾0.74 mM spermi-
responsible for the conversion of excess dietary carbohydrates to dine兾1 mM DTT兾0.5 mM PMSF) containing 0.3 M sucrose. The
triglycerides. Within minutes, elevated glucose levels in the liver homogenate was mixed with 2 vol of cushion buffer (extraction
lead to posttranslational activation of several key enzymes of buffer containing 2.2 M sucrose), layered on top of cushion
glycolysis and lipogenesis, including fructose-6-phosphate 2- buffer, and centrifuged at 75,000 ⫻ g for 30 min at 4°C. Pelleted
kinase兾fructose-2,6-bisphosphatase, fatty acid synthase, acetyl- nuclei were resuspended in 5 vol of nuclear lysis buffer [10%
CoA carboxylase, and L-type pyruvate kinase (LPK). A high- (vol兾vol) glycerol兾10 mM Hepes, pH 7.9兾420 mM NaCl兾0.1 mM
carbohydrate diet also induces transcription of many of the genes EDTA兾3 mM MgCl2兾5 mM DTT兾0.5 mM PMSF兾0.5 ␮g/ml
encoding these enzymes, thereby promoting long-term storage pepstatin A兾1 mM benzamidine兾0.2 ␮g/ml leupeptin) and ho-
of sugars as triglycerides (3, 4). While the expression of some mogenized with 10 strokes of a hand-held Dounce homogenizer.
lipogenic enzymes is enhanced by insulin, which is secreted in The nuclear extracts were centrifuged at 27,000 ⫻ g for 25 min
response to high blood sugar levels, optimal transcription of most at 4°C, and polyethylene glycol (PEG; Mr ⫽ 8,000) was added to
lipogenic genes requires elevated carbohydrate levels (reviewed the supernatant solution to 25% saturation. After centrifugation
in refs. 5 and 6). LPK gene transcription is stimulated by glucose, at 27,000 ⫻ g for 10 min, the pellet was dissolved in a buffer
independent of insulin, in cultured hepatocytes expressing active containing 20% glycerol, 20 mM Hepes (pH 7.9), 100 mM KCl,
glucokinase (7). The mechanism by which carbohydrates regu- 0.2 mM EDTA, 5 mM DTT, and 0.5 mM PMSF and cleared by
late transcription of these genes in the liver remains unknown. centrifugation at 27,000 ⫻ g for 5 min at 4°C. Protein concen-
Previous studies have shown that many lipogenic genes contain tration was determined by using the Bradford method (19).
carbohydrate response elements (ChREs) within their promot-
ers that mediate glucose responsiveness (4, 8–11). These se-
Abbreviations: ChRE, carbohydrate response element; ChREBP, ChRE-binding protein; LPK,
quences have been used to characterize several ChRE-binding L-type pyruvate kinase; PKA, cAMP-dependent protein kinase A; WT, wild type; USF,
proteins (12–15). However, the identity of the carbohydrate- upstream stimulatory factor; GRBP, glucose response element binding protein.
responsive transcription factor has remained elusive, as the †Present address: Okayama Prefectural University, Okayama 719-11, Japan.
expression兾activity of these candidate proteins does not corre- §To whom reprint requests should be addressed. E-mail: KUyeda6400@aol.com.
late with glucose-induced transcription regulated by the ChRE. The publication costs of this article were defrayed in part by page charge payment. This
In this study, we report the purification and identification of article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
a ChRE-binding protein, designated ChREBP. The tissue dis- §1734 solely to indicate this fact.

9116 –9121 兩 PNAS 兩 July 31, 2001 兩 vol. 98 兩 no. 16 www.pnas.org兾cgi兾doi兾10.1073兾pnas.161284298

Table 1. The sequences of the wild-type (WT) and mutated FBS (Life Technologies). After attachment, synthetic liposomes
LPK ChREs were used to transfect hepatocytes with 0.5 ␮g of the luciferase
Oligonucleotide Sequence reporter plasmid and 0.1 ␮g of the internal control plasmid
pRL-TK. After 14 h the medium was replaced by DMEM
WT 5⬘-GGGCGCACGGGGCACTCCCGTGGTTCC-3⬘ supplemented with 100 nM dexamethasone, 10 nM insulin, 10%
3⬘-CCCGCGTGCCCCGTGAGGGCACCAAGG-5⬘ FBS, and 10 mM lactate, 5.5 mM glucose, or 27.5 mM glucose.
The cells were incubated for an additional 12 h before deter-
M1 5⬘-GGGCGaACGGGGCACTCCCGTtGTTCC-3⬘ mination of luciferase activity by using the Dual-Luciferase
3⬘-CCCGCtTGCCCCGTGAGGGCAaCAAGG-5⬘ Reporter Assay System (Promega).

M2 5⬘-GGGCGCcCGGGGCACTCCCGgGGTTCC-3⬘ Purification of ChREBP. ChREBP DNA-binding activity was mon-

3⬘-CCCGCGgGCCCCGTGAGGGCcCCAAGG-5⬘ itored by gel-shift analysis with 32P-labeled WT ChRE oligonu-
cleotide (Table 1). Crude nuclear extract was prepared from the
M3 5⬘-GGGCGCAaGGGGCACTCCCtTGGTTCC-3⬘ livers of 800 male Sprague–Dawley rats re-fed a high-
3⬘-CCCGCGTtCCCCGTGAGGGaACCAAGG-5⬘ carbohydrate diet for 24 h after starvation for 48 h. Livers were
homogenized in 2 vol of buffer (300 mM sucrose兾10 mM Hepes,
M4 5⬘-GGGCGCACtGGGCACTCCaGTGGTTCC-3⬘ pH 7.9兾10 mM KCl兾0.1 mM EDTA兾0.74 mM spermidine兾1 mM
3⬘-CCCGCGTGaCCCGTGAGGtCACCAAGG-5⬘ DTT兾0.5 mM PMSF). The homogenate was mixed with 2 vol of
buffer containing 2.2 M sucrose, layered on top of buffer
M5 5⬘-GGGCGCACGtGGCACTCaCGTGGTTCC-3⬘ containing 2.2 M sucrose, and centrifuged at 75,000 ⫻ g for 45
3⬘-CCCGCGTGCaCCGTGAGtGCACCAAGG-5⬘ min at 4°C. Pelleted nuclei were homogenized in 2.5 vol of lysis
buffer (10% glycerol兾10 mM Hepes, pH 7.9兾10 mM KCl兾0.1
M6 5⬘-GGGCGCACGGtGCACTaCCGTGGTTCC-3⬘ mM EDTA兾3 mM MgCl2兾5 mM DTT兾0.5 mM PMSF兾0.5 ␮g/ml
3⬘-CCCGCGTGCCaCGTGAtGGCACCAAGG-5⬘ pepstatin A兾1 mM benzamidine兾0.2 ␮g/ml leupeptin). After
addition of 0.1 vol of 4.2 M NaCl, the homogenate was gently
M3兾5 5⬘-GGGCGCAtGcGGCACTCgCaTGGTTCC-3⬘ stirred for 1 h at 4°C and then centrifuged at 100,000 ⫻ g for 30
3⬘-CCCGCGTaCgCCGTGAGcGtACCAAGG-5⬘ min at 4°C. The supernatant was diluted to 100 mM NaCl and
cleared by centrifugation at 100,000 ⫻ g for 30 min. The nuclear
S⫺1 5⬘-GGGCGCACGGGGCCTCCCGTGGTTCCT-3⬘ extract was applied to a DE-52 DEAE-cellulose column equil-
3⬘-CCCGCGTGCCCCGGAGGGCACCAAGGA-5⬘ ibrated with buffer A (20 mM Hepes, pH 7.9兾1 mM EDTA兾10%
glycerol兾1 mM DTT) containing 0.1 M KCl. Proteins were
S⫹1 5⬘-GGGCGCACGGGGCACTTCCCGTGGTTCC-3⬘ eluted from the column with buffer A containing 0.2 M KCl.
3⬘-CCCGCGTGCCCCGTGAAGGGCACCAAGG-5⬘ Active fractions were pooled and subsequently applied to a
DNA-cellulose (native DNA) column equilibrated with buffer A
The E-boxes are in boldface type. Mutations within the E-box are indicated
by lowercase letters. S⫺1 and S⫹1 indicate the deletion or addition, respec-
containing 0.2 M KCl. The column was washed with buffer A
tively, of one base between the two E-boxes. containing 0.5 M KCl and eluted with buffer A containing 0.7 M
KCl. The active fractions were pooled, and (NH4)2SO4 was
added to a final concentration of 0.436 g兾ml and stirred for 30
Gel-Shift Assays. Gel mobility-shift assays were performed as min at 4°C. After centrifugation at 100,000 ⫻ g for 20 min at 4°C,
the precipitate was resuspended in buffer A containing 50 mM

CELL BIOLOGY
described by Liu et al. (8). Double-stranded oligonucleotides
(Table 1) were prepared by mixing equal amounts of the KCl and competitor DNAs [poly(dI-dC) and CATCAGGC-
complementary single-stranded DNAs in 50 mM NaCl, heating CATCTGGCCCCTTGTTATTAA at 50 ␮g兾ml and 10 ␮g兾ml,
to 70°C for 15 min, and cooling to room temperature. The respectively]. The sample was applied to a DNA-affinity column
annealed oligonucleotides were labeled with 32P in the presence displaying double-stranded CATGGGCGCACGGGGCACTC-
of [␥-32P]ATP (Amersham Pharmacia) and T4 polynucleotide CCGTGGTTCC DNA. The column was washed with buffer A
kinase (New England Biolabs). Binding reactions were carried containing 0.4 M KCl and nonspecific competitor DNAs and
out in reaction mixture containing 20 mM Hepes (pH 7.9), 50 eluted with buffer A containing 0.8 M KCl. The DNA-affinity
mM KCl, 5 mM DTT, 0.2 mM EDTA, 0.5 mM PMSF, 10% column step was repeated. The active fractions were pooled,
glycerol, 1 ␮g兾␮l poly(dI-dC) (Amersham Pharmacia), and 1 resolved by SDS兾PAGE, and visualized by silver staining. The
␮g兾␮l nuclear extract. The reaction mixture was incubated at 100-kDa band was excised from poly(vinylidene difluoride)
(PVDF) membrane and digested, and individual peptides were
room temperature for 30 min, and DNA–protein complexes were
sequenced as described in ref. 21. Four peptides were found to
separated by electrophoresis in a nondenaturing 4.5% polyacryl-
match human AF156603㛭1.
amide gel.
Results
Primary Hepatocytes Culture and Transfection. The construction of
A High-Carbohydrate Diet Induces ChREBP DNA-Binding Activity in Rat
luciferase reporter plasmids was described in ref. 13. The mouse Livers. It has been shown that glucose-induced transcription of
ChREBP coding region (GenBank accession no. AF156604) was the LPK gene is mediated by a ChRE located within its promoter
amplified by PCR and ligated into the pcDNA3.1兾V5HisB (4, 8–11). The ChRE identified within the promoter of the LPK
expression vector (Invitrogen). gene contains two imperfect E-box motifs in a head-to-tail
Primary hepatocytes were prepared from male Sprague– orientation separated by 5 bp (WT; Table 1). This sequence was
Dawley rats by using the collagenase (Life Technologies, Grand used to examine DNA-binding activity in liver nuclear extracts
Island, NY) perfusion method (20) and plated in collagen- prepared from rats fed either a high-carbohydrate or high-fat
coated six-well tissue culture plates (CMS兾Fisher, Houston) at diet for 24 h after starvation. One of the DNA-binding com-
a density of 1 ⫻ 106 cells per well in glucose-free Dulbecco’s plexes observed in nuclear extracts, designated ChREBP, was
modified Eagle’s medium (DMEM; Life Technologies) supple- enriched 2–3-fold only in rats fed the high-carbohydrate diet
mented with 100 nM dexamethasone, 10 nM insulin, 100 (Fig. 1A, lanes 2 and 3). Conversely, formation of the previously
units兾ml penicillin, 100 ␮g兾ml streptomycin, and 10% dialyzed described DNA-binding complex containing the upstream stim-

Yamashita et al. PNAS 兩 July 31, 2001 兩 vol. 98 兩 no. 16 兩 9117

 ulatory factor (USF) (12) did not vary as a function of diet
(Fig. 1 A).

ChREBP’s DNA-Binding Specificity in Vitro Correlates with Promoter

Activity in Vivo. The relative affinities of the observed DNA-
binding proteins for the native LPK ChRE sequence and mu-
tated variants (Table 1) are shown in Fig. 1B. Mutation of the
modified E-box motifs to perfect palindromes (sequence M5)
enhanced formation of the ChREBP DNA-binding complex,
whereas further deviation from the canonical E-box motif
significantly impeded complex formation (sequences M1–4,
M6). Unlike formation of USF and GRBP complexes (13),
formation of the ChREBP complex was particularly sensitive to
changes in the spacing between the two E-box motifs (sequences
S⫺1 and S⫹1). Importantly, only complex formation between
the variant sequences and ChREBP correlated precisely with the
ability of the ChRE variants to mediate glucose-induced tran-
scription of a luciferase reporter in primary hepatocytes (Fig.
1C). The agreement between ChREBP’s DNA-binding speci-
ficity and the relative ability of the various DNA sequences to
confer carbohydrate-responsive transcriptional activation sug-
gests that this complex might contain the long-sought glucose-
regulated transcription factor.

ChREBP Is a Member of the Basic Helix–Loop–Helix Leucine Zipper

(bHLH-ZIP) Family of Transcription Factors Observed Only in the Liver.
ChREBP was purified to homogeneity from the livers of rats fed
a high-carbohydrate diet and shown to consist of a single
polypeptide with an apparent molecular mass of 100 kDa. The
amino acid sequences of four peptides generated by tryptic
digestion were determined. Each peptide was found to match
the Williams–Beuren syndrome critical region 14 protein
(WBSCR14), encoded by one of the 17 genes heterozygously
deleted from patients suffering from the neurodevelopmental
disorder Williams–Beuren syndrome (22). The contributions of
most of the deleted genes, including WBSCR14, to the WBS
phenotypes are not known.
ChREBP is a member of the basic helix–loop–helix leucine
zipper (bHLH-ZIP) family of transcription factors known to
recognize E-box motifs within their target promoters. Northern
analysis identified two ChREBP isoforms (6 kb and 4 kb) present
in the liver, kidney, and small intestine (Fig. 2A). These results
are consistent with the ChREBP expression pattern observed in
the mouse (22). However, gel-shift analysis of nuclear extracts
prepared from various tissue sources demonstrated that
ChREBP DNA-binding activity is observable only in liver nu-
clear extracts (Fig. 2B). ChREBP’s limited expression profile is
consistent with previous studies demonstrating that LPK is
expressed exclusively in the liver (23) and ␤ cells of the pancreas
(24). The failure to detect the ChREBP mRNA in the pancreas
is likely due to the fact that ␤ cells represent only 1% of the cells
in this organ. Conversely, the USF-containing complex was
observed in many tissues that do not express LPK, as expected
from USF’s ubiquitous expression pattern (25). Although several
other unidentified complexes are observed, only the ChREBP-
Fig. 1. (A) Gel-shift analysis of hepatic nuclear factors that recognize the WT containing complex is induced by a high-carbohydrate diet and
LPK ChRE. Rats were starved for 48 h (lane 1) and subsequently re-fed a
is specific to the site of LPK expression.
high-fat diet for 24 h (lane 2) or a high-carbohydrate diet for 24 h (lane 3).
Nuclear extracts were prepared from the rat livers and incubated with a
radiolabeled oligonucleotide corresponding to the WT LPK ChRE. Lanes 2 and
3 contain equal protein loads. Arrows indicate the positions of the DNA- with the WT and the mutated oligonucleotides and examined by gel-shift
binding complexes containing USF or the glucose response element binding analysis. (C) Mutations to the LPK ChRE affect glucose-responsive transcription
protein (GRBP) as determined by supershift of the DNA-binding complex with of a downstream luciferase reporter gene. Construction of luciferase reporters
antibodies specific for each protein. (B) Mutations to the LPK ChRE differen- driven by the WT and variant LPK promoter sequences (Table 1) was described
tially affect the formation of the various DNA-binding complexes. The se- in ref. 13. Primary hepatocytes were transfected with the reporter constructs
uences of the mutated oligonucleotides are shown in Table 1 (Md and Mi and the fold activation of luciferase expression in response to high glucose
correspond to S⫺1 and S⫹1, respectively). Liver nuclear extracts from rats r-fed (27.5 mM) was determined. Values represent the mean ⫾ standard error of
a high-carbohydrate diet for 24 h after 48-h starvation were incubated four experiments.

9118 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.161284298 Yamashita et al.

Fig. 2. (A) Tissue distribution of ChREBP as determined by Northern blot analysis. Message levels were visualized from total RNA (25 ␮g) with a 32P-radiolabeled
probe derived from the ChREBP cDNA. (B) ChREBP activity is detected only in liver nuclear extracts. Nuclear extracts were prepared from rats re-fed a high
carbohydrate diet for 24 h after 48-h starvation and incubated with the 32P-radiolabeled WT oligonucleotide (Table 1) followed by gel-shift analysis. Arrows
indicate the positions of the DNA-binding complexes containing USF or GRBP as determined by supershift of the DNA-binding complex with antibodies specific
for each protein.

ChREBP Activates Transcription from the LPK Promoter in Response to activates PKA and may result in phosphorylation of ChREBP.
Glucose in Hepatocytes. To determine whether ChREBP can Addition of a negatively charged phosphoryl group to the basic
mediate glucose-activated transcription in cells, primary hepa- region would be expected to disrupt DNA-binding activity,
tocytes were cotransfected with the luciferase reporter gene thereby inactivating the transcription factor. Conversely, excess
driven by a 200-bp fragment of the LPK promoter containing the carbohydrates would lower cAMP levels and could also lead to
ChRE (13) and an expression vector featuring ChREBP under the activation of a protein phosphatase that hydrolyzes the
the control of the constitutively active cytomegalovirus (CMV)
promoter. When transfected cells were cultured in the absence
of glucose (10 mM lactate) or in the presence of a low concen-

CELL BIOLOGY
tration of glucose (5.5 mM), expression of the ChRE-driven
reporter remained low and was unresponsive to forced overex-
pression of ChREBP (Fig. 3, dark bars). Supplementation of the
culture medium with 27.5 mM glucose did lead to a 2-fold
induction of reporter expression in cells transfected with empty
vector (Fig. 3, open bars). This reporter induction reflects the
presence of the endogenous glucose-responsive transcription
factor that recognizes the LPK ChRE (Fig. 1C). Forced expres-
sion of ChREBP was able to further induce expression of the
reporter an additional 2-fold above the endogenous levels. The
modest level of reporter induction, above endogenous ChREBP
activity, in response to glucose is typical of the glucose-
responsive transcriptional induction of many glycolytic enzymes
in vivo. These results indicate that ChREBP may function as a
glucose-responsive transcription factor in vivo.

ChREBP’s DNA-Binding Activity Is Inhibited by Phosphorylation in

Vitro. The mechanism by which excess glucose regulates the
DNA-binding activity of ChREBP remains unclear. However,
cAMP is known to inhibit glucose-responsive signaling pathways
(6, 26, 27). Interestingly, the repressive effects of cAMP on LPK
gene transcription have been reported to be localized to the Fig. 3. Forced expression of ChREBP activates expression from the LPK
ChRE despite the complete lack of similarity to known cAMP- promoter in response to glucose in vivo. Primary hepatocytes were cotrans-
fected with 0.5 ␮g of the luciferase reporter construct driven by the LPK
regulated promoter elements (6, 9). ChREBP contains three
promoter and 1.5 ␮g of the ChREBP cDNA under the control of the constitu-
consensus PKA phosphorylation sites, including one located tively active cytomegalovirus (CMV) promoter (filled bars) or with the empty
within the basic region of the putative DNA-binding domain expression vector (open bars). After transfection, the cells were incubated for
(Fig. 4A). When circulating carbohydrate levels are low, elevated 12 h in medium containing 10 mM lactate, 5.5 mM glucose, or 27.5 mM
glucagon levels promote increased cAMP production, which glucose. Values represent the mean ⫾ standard error of five experiments.

Yamashita et al. PNAS 兩 July 31, 2001 兩 vol. 98 兩 no. 16 兩 9119

Fig. 4. (A) ChREBP contains several recognizable protein motifs, including a bipartite nuclear localization signal (NLS), a basic helix–loop– helix leucine zipper
(bHLH-ZIP) motif, and three consensus PKA phosphorylation sites (P) including one (sequence RRIT) within the bHLH region. (B) (Left) DNA-binding activity of
ChREBP can be regulated by phosphorylation in vitro. Incubation of active ChREBP (lane 1) for 20 min with both ATP (1 mM) and PKA (0.1 unit兾␮l) abolished
DNA-binding activity (lane 2) as measured by gel-shift analysis with the 32P-radiolabeled WT oligonucleotide. Omission of ATP from the reaction mixture prevents
ChREBP inactivation (lane 3). PKA-dependent ChREBP inactivation is blocked by 0.2 unit兾␮l protein kinase inhibitor (PKI) (lane 4). After PKI addition, DNA-binding
activity of phosphorylated ChREBP can be restored by addition of 0.025 unit兾␮l protein phosphatase 2A (PP2A) (lane 5). Addition of the PP2A inhibitor okadaic
acid (OKA) (10 nM) prevents reactivation of the ChREBP DNA-binding activity. (Right) Incubation of the protein with [␥-32P]ATP in the presence of PKA leads to
radiolabeling of ChREBP observed after SDS兾PAGE.

phosphoryl group from inactive ChREBP. Dephosphorylation expression, regulation, and selectivity of ChREBP are consistent
of inactivated ChREBP would restore DNA-binding activity and with its assignment as a carbohydrate-responsive transcription
induce transcription of genes required for carbohydrate metab- factor. ChREBP activity is induced by consumption of a high-
olism. To test this model, purified ChREBP was incubated with carbohydrate diet, but not a high-fat diet, after starvation (Fig.
the catalytic subunit of PKA in the presence of ATP, resulting 1 A). ChREBP’s DNA-binding specificity precisely correlates
in phosphorylation of ChREBP and a 90% reduction in its with promoter activity in vivo (Fig. 1 B and C). Gel-shift analysis
DNA-binding activity (Fig. 4B). ChREBP inactivation required using oligonucleotides corresponding to the related ChREs from
ATP and was inhibited by addition of a protein kinase inhibitor, other carbohydrate-responsive genes such as S14 (28) and fatty
PKI (Fig. 4B). DNA-binding activity of inactivated ChREBP was acid synthase (29) also resulted in the formation of a complex
restored by the addition of the catalytic subunit of protein with mobility properties identical to the ChREBP–LPK ChRE
phosphatase 2A (PP2A). Reactivation of phosphorylated complex (data not shown). ChREBP’s DNA-binding activity was
ChREBP by PP2A is blocked by the phosphatase inhibitor observable only in nuclear extracts prepared in the liver (Fig. 2).
okadaic acid (Fig. 4B). These data are consistent with the model This expression pattern is consistent with the expression pattern
that ChREBP activity is regulated by changes in its phosphor- of the LPK target gene. Significantly, forced ChREBP expres-
ylation state. sion in liver hepatocytes activates transcription from the LPK
promoter specifically in response to high glucose levels but not
Discussion low glucose levels or lactate (Fig. 3).
The identity of the glucose-responsive ChRE-binding activity Analysis of the primary sequence of the ChREBP protein
has until now remained elusive. Although the USF1兾USF2 reveals the presence of several putative domains typical of
heterodimer has been shown to bind the ChRE in vitro, the transcription factors (Fig. 4A). The ChREBP carboxyl terminus
expression, regulation, and selectivity of USF do not correlate contains a basic helix–loop–helix motif that likely mediates
with the glucose response (12). The same is true for GRBP (13), binding of the factor to the E-box motifs within the ChRE. In
which shares mobility properties with the ChRE-binding protein addition, there are regions of homology to proline-rich domains,
purified by Yamada et al. (14) and the ChRE-binding protein leucine zipper domains, and leucine zipper-like domains (22)
reported by Koo and Towle (15), which does not vary with diet that could mediate additional protein–protein interactions, in-
and was not regulated by phosphorylation. Alternatively, the cluding dimerization. The amino terminus features a bipartite

9120 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.161284298 Yamashita et al.

nuclear localization signal that likely mediates nuclear import. phosphatase could activate ChREBP, providing an additional
The role of these domains in mediating the function of this factor point of regulation that could be mediated by a glucose metab-
requires further investigation. olite. If the model proposed in this study is correct, identification
Elucidation of the mechanism by which elevated carbohydrate of the particular protein phosphatase(s) responsible for
levels in the liver induce transcription by means of ChREBP ChREBP activation in vivo should provide additional insight into
continues to be the subject of much study. Glucose must first be the nature of a long-sought glucose metabolite responsible for
metabolized in the liver to promote carbohydrate-responsive the glucose signaling and the molecular mechanism by which it
transcription of the LPK gene (reviewed in ref. 6). However, the induces ChREBP activity. Putative phosphorylation sites also
identity of important metabolic intermediates and the pathways reside near the nuclear localization signal (NLS) and may hint at
by which they regulate transcription remain unknown. Identifi- other layers of regulation. We close by hypothesizing that
cation of ChREBP as a critical downstream component of this ChREBP-dependent induction of genes involved in carbohy-
pathway should provide new insights into these regulatory drate metabolism and lipogenesis is likely required for efficient,
long-term storage of excess dietary carbohydrates. Inhibition of
mechanisms. For example, the presence of several potential
ChREBP activation would be expected to attenuate excess fat
phosphorylation sites suggests that the activity of this transcrip-
accumulation resulting from a high-carbohydrate diet and pro-
tion factor could be regulated by changes in its phosphorylation vide novel opportunities to address the health consequences
state, particularly within the DNA-binding domain. As we show stemming from obesity and diabetes.
here, phosphorylation of ChREBP by PKA inactivates DNA-
binding activity in vitro, consistent with a model in which cAMP We thank Dr. Steven L. McKnight and Dr. Bonnie C. Miller for helpful
serves an intracellular mediator to inactivate this transcription advice. R.K.B. is supported by a National Research Service Award from
factor in vivo. In addition, this model predicts that a protein the National Institutes of Health.

1. Neel, J. N. (1999) Nutr. Rev. 57, S2–S9. 16. Casazza, J. P. & Veech, R. L. (1986) Biochem. J. 236, 635–641.
2. Troiano, R. P. & Flegal, K. M. (1999) Int. J. Obes. Relat. Metab. Disord. 23, 17. Francone, O. L., Griffaton, G. & Kalopissis, A.-D. (1992) Am. J. Physiol. 263,
S22–S27. E615–E623.
3. Goodridge, A. G. (1987) Annu. Rev. Nutr. 7, 157–185. 18. Hattori, M., Tugores, A., Veloz, L., Karin, M. & Brenner, D. A. (1990) DNA
4. Granner, D. & Pilkis, S. (1990) J. Biol. Chem. 265, 10173–10176. Cell Biol. 9, 777–781.
5. Girard, J., Ferré, P. & Foufelle, F. (1997) Annu. Rev. Nutr. 17, 325–352. 19. Bradford, M. M. (1976) Anal. Biochem. 72, 248–254.
6. Towle, H. C., Kaytor, E. N. & Shih, H.-M. (1997) Annu. Rev. Nutr. 17, 405–433. 20. Berry, M. N. & Friend, D. S. (1969) J. Cell Biol. 43, 506–520.
7. Doiron, B., Cuif, M.-H., Kahn, A. & Diaz-Guerra, M.-J. (1994) J. Biol. Chem. 21. Wang, K. H., Brose, K., Arnot, D., Kidd, T., Goodman, C. S., Henzel, W. &
269, 10213–10216. Tessier-Lavigne, M. (1999) Cell 96, 771–784.
8. Liu, Z., Thompson, K. S. & Towle, H. C. (1993) J. Biol. Chem. 268, 12787– 22. de Luis, O., Carmen Valero, M. & Pérez Jurado, L. A. (2000) Eur. J. Hum.
12795. Genet. 8, 215–222.
9. Bergot, M. O., Diaz-Guerra, M. J., Puzenat, N., Raymondjean, M. & Kahn, A. 23. Imamura, K. & Tanaka, T. (1972) J. Biochem. 71, 1043–1051.
(1992) Nucleic Acids Res. 20, 1871–1877. 24. Noguchi, T., Yamada, K., Yamagata, K., Takenaka, M., Nakajima, H., Imai, E.,
10. Cuif, M. H., Porteu, A., Kahn, A. & Vaulont, S. (1993) J. Biol. Chem. 268, Wang, Z. & Tanaka, T. (1991) Biochem. Biophys. Res. Commun. 181, 259–264.
13769–13772. 25. Sirito, M., Lin, Q., Maity, T. & Sawadogo, M. (1994) Nucleic Acids Res. 22,
11. Thompson, K. S. & Towle, H. C. (1991) J. Biol. Chem. 266, 8679–8682. 427–433.
12. Kaytor, E. N., Shih, H.-M. & Towle, H. C. (1997) J. Biol. Chem. 272, 7525–7531. 26. Viollet, B., Kahn, A. & Raymondjean, M. (1997) Mol. Cell. Biol. 17, 4208–4219.
13. Hasegawa, J., Osatomi, K., Wu, R. F. & Uyeda, K. (1999) J. Biol. Chem. 274, 27. Gourdon, L., Lou, D. Q., Raymondjean, M., Vasser-Cognet, M. & Kahn, A.
1100–1107. (1999) FEBS Lett. 459, 9–14.
14. Yamada, K., Tanaka, T. & Noguchi, T. (1999) Biochem. Biophys. Res. Commun. 28. Shih, H. M., Liu, A. & Towle, H. C. (1995) J. Biol. Chem. 270, 21991–21997.

CELL BIOLOGY
257, 44–49. 29. Foufelle, F., Lepetit, N., Bosc, D., Delzenne, N., Morin, J., Raymondjean, M.
15. Koo, S.-H. & Towle, H. C. (2000) J. Biol. Chem. 275, 5200–5207. & Ferre, P. (1995) Biochem. J. 308, 521–527.

Yamashita et al. PNAS 兩 July 31, 2001 兩 vol. 98 兩 no. 16 兩 9121