Nutrients, glucose and SIRT1

[Guest guest]

Hi All,

This looks good.

Nature 434, 113 - 118 (03 March 2005); doi:10.1038/nature03354

Nutrient control of glucose homeostasis through a complex of PGC-1

and SIRT1

JOSEPH T. RODGERS1, CARLOS LERIN1, WILHELM HAAS3, STEVEN P. GYGI3,

BRUCE M. SPIEGELMAN2,3

& PERE PUIGSERVER1

Homeostatic mechanisms in mammals respond to hormones and

nutrients to maintain blood

glucose levels within a narrow range. Caloric restriction causes many

changes in glucose

metabolism and extends lifespan; however, how this metabolism is

connected to the ageing

process is largely unknown. We show here that the Sir2 homologue,

SIRT1—which modulates

ageing in several species1-3 —controls the gluconeogenic/glycolytic

pathways in liver in

response to fasting signals through the transcriptional coactivator

PGC-1. A nutrient

signalling response that is mediated by pyruvate induces SIRT1

protein in liver during

fasting. We find that once SIRT1 is induced, it interacts with and

deacetylates PGC-1 at

specific lysine residues in an NAD+-dependent manner. SIRT1 induces

gluconeogenic genes and

hepatic glucose output through PGC-1, but does not regulate the

effects of PGC-1 on

mitochondrial genes. In addition, SIRT1 modulates the effects of PGC-

1 repression of glycolytic

genes in res ponse to

fasting and pyruvate. Thus, we have identified a molecular mechanism

whereby SIRT1

functions in glucose homeostasis as a modulator of PGC-1. These

findings have strong

implications for the basic pathways of energy homeostasis, diabetes

and lifespan.

A key regulator of lifespan is the NAD+-dependent histone

deacetylase Sir2 (silencing

information regulator 2), which induces longevity in Saccharomyces

cerevisiae and

Caenorhabditis elegans in response to caloric restriction signals1-3.

The mammalian homologue,

SIRT1—a member of the Sir2 family called sirtuins—is known to target

MyoD, p53 and

forkhead transcription factors for deacetylation4-8. However, whether

and how SIRT1 is involved

in pathways affected directly by caloric restriction in mammals is

still unclear.

It is well established that reduced intake of dietary energy

results in metabolic

changes similar to fasting9, 10. Key among these changes are

increased fatty-acid oxidation

and hepatic gluconeogenesis. PGC-1 is a key regulator of glucose

production in the liver

of fasted and diabetic mice through activation of the entire

gluconeogenic pathway11-15.

In addition to the key hormones insulin, glucagon and

glucocorticoids, the rate of hepatic

gluconeogenesis is also controlled by nutrients, but how the nutrient

response is

controlled and whether PGC-1 is also involved remains unknown.

The fact that Sir2 controls lifespan in S. cerevisiae and C.

elegans in response to

caloric restriction1, 2 prompted us first to investigate whether

SIRT1 could also be

regulated by nutritional status in mice. As shown in Fig. 1a, SIRT1

expression in liver was not

regulated at the messenger RNA level in fasted mice; however, SIRT1

protein levels were

induced after fasting and returned to nearly control levels upon

refeeding (Fig. 1b). As

previously described11, expression of PGC-1 was upregulated at the

level of mRNA and

protein by fasting (Fig. 1a, . PGC-1 and SIRT1 increased in

correlation with induction of

the gluconeogenic gene PEPCK (phosphoenol-pyruvate kinase) (Fig. 1a).

We next investigated

whether lactate and pyruvate—which can change the ratio of NAD+/NADH

and regulate SIRT1

activity—were fluctuating in the liver. As shown in Fig. 1c, pyruvate

levels were

increased 1.7-fold in the liver of fasted mice and conversely lactate

levels were decreased

twofold. We also measured NAD+, a potent activator and substrate of

SIRT1. As shown in Fig. 1d, liver

NAD+ levels were increased by 33% by fasting and returned to control

levels after refeeding.

Interestingly, we could not detect significant changes of NADH levels

(data not shown). These data suggest that increases of SIRT1 protein

levels and NAD+ concentration might

contribute to the SIRT1 deacetylase activity in the fasted state.

To investigate further the regulation of SIRT1, we examined the

effects of cyclic AMP

or insulin, key signals of the fasted response in mammals. These two

pathways did not

affect SIRT1 protein levels in cultured hepatocytes (Supplementary

Fig. S1). In contrast,

glucose and pyruvate, known to fluctuate in fasting16, regulated

SIRT1 protein levels. As

shown in Fig. 1e, increasing concentrations of glucose decreased the

amount of SIRT1,

whereas increasing concentrations of pyruvate markedly increased it.

These changes in SIRT1

protein levels are regulated at the post-transcriptional level as

SIRT1 mRNA remained

constant (Fig. 1e). We performed pulse-chase experiments to analyse

further whether these

changes were at the level of protein synthesis or degradation. As

shown in Supplementary

Fig. S2, pulsed radiolabelled SIRT1 protein levels did not change

after chase for 6 and 12 h

under pyruvate treatment. However, an increase in SIRT1 protein

synthesis was notably

elevated by pyruvate. These results suggest that the increase of

SIRT1 protein observed in the liver of fasted mice could be mediated

by changes in glucose and/or pyruvate.

The fact that the induction pattern of SIRT1 protein correlated

with expression of

PGC-1 in fasting (Fig. 1a, suggests that these two proteins could

interact. To study

this, we performed immunoprecipitation of endogenous SIRT1 protein

from liver extracts and

precipitated PGC-1 (Fig. 2a). Moreover, an interaction between both

endogenous and/or

overexpressed proteins was also observed in cultured hepatocytes

(Supplementary Fig. S5b) and

293 cells (Supplementary Fig. S3a). To confirm that this reflected a

direct physical

interaction, we used an in vitro interaction assay with recombinant

glutathione S-transferase

(GST)–PGC-1 (Supplementary Fig. S3b). The fact that PGC-1 and SIRT1

directly interacted

suggests that PGC-1 might be a substrate for SIRT1 deacetylation. As

shown in Fig. 2b,

nicotinamide (SIRT1 inhibitor) treatment strongly induced PGC-1

lysine acetylation.

Importantly, expression of SIRT1 could overcome nicotinamide-induced

PGC-1 acetylation, but a

SIRT1 mutant

(SIRT1H355A) that lacks enzymatic activity had no effect (Fig. 2b).

PGC-1 acetylation

decreased in the fasted liver, suggesting that SIRT1 activity is

increased in this

situation (Fig. 2c). To demonstrate that SIRT1 deacetylated PGC-1

directly, we used an in vitro

deacetylase assay. As shown in Fig. 2d, PGC-1 acetylation is

decreased (60%) only upon

addition of both SIRT1 and NAD+. To map the PGC-1 lysine-acetylation

sites induced by

nicotinamide, tandem mass spectrometry analysis was performed. We

found that PGC-1 was

acetylated at 13 lysine sites (Fig. 2e and Supplementary Fig. S4).

Mutation of these 13 lysines

to arginine abolished the acetylation of PGC-1 (Fig. 2e). To test the

effect of PGC-1

acetylation on its transcriptional activity, luciferase reporter

assays using the

transcription factor HNF4 were performed. As shown in Fig. 2f,

nicotinamide treatment decreased the

transcriptional activity of wild-type PGC-1/HNF4 by 24-fold. We

tested several

combinations of PGC-1

mutants

and found that a mutant of PGC-1 (R5), in which five lysines were

substituted to

arginine, decreased the fold repression by nicotinamide to tenfold.

In addition, two different

mutants R10 and R13, containing the R5 mutations, largely decreased

this repression to

almost control levels (Fig. 2f). These results suggest that

acetylation of PGC-1 decreases

its ability to efficiently coactivate HNF4. The molecular mechanisms

underlying this

transcriptional repression by acetylation are under current

investigation.

Previous studies have focused on the hormonal regulation of PGC-1

expression11-13.

However, the above data prompted us to investigate whether the

gluconeogenic function of

PGC-1 might also be controlled through a nutrient-sensing pathway via

pyruvate and SIRT1. As

shown in Fig. 3a, the gluconeogenic genes PEPCK and G6Pase (glucose-6-

phosphatase) were

induced by pyruvate treatment. The effect of pyruvate on the

induction of G6Pase and PEPCK

was decreased by a specific short interfering RNA for PGC-1 (G6Pase

decreased by 30% and

PEPCK by 70%; Fig. 3a) but not by a scrambled siRNA (control); this

indicates that the

pyruvate regulation of these genes required PGC-1. We next used siRNA

for SIRT1

(Supplementary Fig. S6) to determine whether the effect of pyruvate

on PGC-1 was dependent on SIRT1.

Indeed, this knockdown of SIRT1 protein largely blocked the pyruvate-

induced PGC-1

increase on PEPCK and G6Pase (Fig. 3b). The pyruvate and SIRT1

effects were specific to

gluconeogenic genes

because PGC-1-targeted mitochondrial genes (cytochrome-c and -ATP

synthase) were not

significantly changed. These data indicate a requirement of SIRT1 in

the regulation of

gluconeogenic gene expression by PGC-1. Consistent with these

effects, endogenous SIRT1 and

PGC-1 were recruited to chromatin promoter fragments of the PGC-1

targets PEPCK and G6Pase.

Notably, the recruitment of both proteins was increased by pyruvate

treatment

(Supplementary Fig. S7). PGC-1 requires interaction with HNF4 to

induce gluconeogenic gene

expression14. We therefore examined whether SIRT1 was in a complex

with HNF4 and PGC-1. As shown

in Fig. 2a and Supplementary Fig. S5, endogenous or overexpressed

HNF4 coprecipitated with

SIRT1 and PGC-1. Taken together, these results might suggest that PGC-

1 and SIRT1 are in

a protein complex that includes HNF4.

The metabolic flux of glucose in hepatocytes is also affected by

the rate of

glycolysis, a pathway that is decreased during fasting. To test

whether PGC-1 and SIRT1 might also

control glycolysis, we analysed the expression of glucokinase and

liver pyruvate kinase

(LPK). As shown in Fig. 3a, b, PGC-1 and pyruvate decreased the

levels of mRNAs for

glucokinase and LPK. SIRT1 siRNA increased the expression of these

genes under pyruvate

treatment. These results suggest that PGC-1 and SIRT1 co-regulate, in

opposite directions, the

patterns of gluconeogenic and glycolytic gene expression in response

to pyruvate. We next

investigated whether this induction of gluconeogenic and glycolytic

genes by SIRT1 and

PGC-1 was reflected in glucose production per se. As shown in Fig.

4a, PGC-1 increased

glucose production threefold compared with control cells. SIRT1 siRNA

repressed the glucose

production induced by PGC-1 to control levels. This indicates that

SIRT1 is required for

PGC-1 induction of glucose production.

The data we present here show for the first time that PGC-1 and

SIRT1, which have been

studied in very different contexts, can function together to promote

adaptation to

nutrient deprivation by regulating the genetic programmes of

gluconeogenesis and glycolysis.

This provides one of the most striking examples of a biological

programme regulated through

transcriptional coactivation (Fig. 4b).

SIRT1 is in a protein complex with PGC-1 and HNF4, an essential

transcription factor

in PGC-1's gluconeogenic function14. In this protein complex SIRT1 is

likely to be the

sensor for nutrient fluctuations via NAD+ and regulates PGC-1-

dependent gene expression.

Whether other PGC-1 gluconeogenic interacting transcription factors

that are hormonally

regulated, such as FOXO1 and glucocorticoid receptor, are also

involved in SIRT1-mediated

effects is unknown. SIRT1 has mainly been linked to negative

regulation of gene expression

through protein deacetylation5-7. However, we show here that SIRT1

can act both positively

and negatively to control gene expression as a cofactor for PGC-1.

Although the mechanism

of this transcriptional regulation is unknown, it is possible that

the recruitment of a

different set of coactivators and corepressors through PGC-1/SIRT1

could explain these

opposite effects. Interestingly, it has recently been reported that

SIRT1 interacts with the transcriptional corepressor NCoR, negatively

regulating PPAR in white fat where PGC-1 is very low17.

SIRT1 has been implicated in other specific biological responses,

such as the reaction

to oxidative stress through deacetylation of p53 and FOXO1

transcription factors5-8.

Whether PGC-1 is also involved in this oxidative stress response is

an interesting question

because PGC-1 interacts with FOXO1 in hepatocytes13. It is therefore

possible that

multiple effects of SIRT1 in energy metabolism or even lifespan might

occur by means of docking

on this coactivator. Ultimately, a better understanding of these

nutritional signalling

systems may lead to new therapeutic approaches to defects in glucose

metabolism in

diabetes and ageing.

Methods

...

Cell culture and treatments Fao rat hepatocytes were cultured in

RPMI with 10% fetal

bovine serum (FBS). Treatments with nicotinamide (5 mM), insulin (10

nM) and forskolin (1

µM) (Sigma) were performed in 0.5% bovine serum albumin (BSA) for 6

h. Glucose and

pyruvate treatments were performed for 6 h in distilled PBS (Gibco)

with 0.5% BSA.

Animal experiments Four-week-old C57B1/6 mice were fed ad libitum,

fasted for 24 h and

refed for 24 h. Animals were killed and the livers were removed and

snap-frozen. Liver

whole-cell homogenates were made with RIPA buffer and used for

western blot analysis and

metabolite measurements. Liver mRNA was isolated using Trizol reagent

(Invitrogen) and used

for northern blot analysis and real-time polymerase chain reaction

(PCR).

...

Al Pater