visfatin is similar in visceral and subcuta-neous adipose tissue. Similar results have also been obtained in two different strains of rat8. Further studies are required before the matter is settled conclusively, but it does seem possible that visfatin may yet receive a third name, one more indicative of its true physiological actions. Thus far, the insulin-

mimetic role of visfatin has further com-plicated an already complex story. It is our hope that this year will see the completion of the crucial experiments required to define the physiological role or roles of this inter-esting molecule.

1. Bosello, O. & Zamboni, M. Obes. Rev. 1, 47–56 (2000).

2. Fukuhara, A. et al. Science 307, 426–430 (2005).3. Samal, B. et al. Mol. Cell Biol. 14, 1431–1437

(1994).4. Chen, M.P. et al. J. Clin. Endocrinol. Metab (doi:

10.1210/jc.2005-1475).5. Hug, C. & Lodish, H. F. Science 307, 366–367 (2005).6. Rongvaux, A. et al. Eur. J. Immunol. 32, 3225–3234

(2002).7. Berndt, J. et al. Diabetes 54, 2911–2916 (2005).8. Kloting, N. & Kloting, I. Biochem. Biophys. Res.

Commun. 332, 1070–1072 (2005).

The tangled path to glucose productionMichihiro Matsumoto & Domenico Accili

Liver glucose production is crucial to survival during fast and is abnormally elevated in diabetes. Studies of the transcriptional coactivator Torc2 redefine the mechanism by which cAMP signaling affects fasting-induced glucogenesis.

During fast, the liver is the main provider of glucose to the brain and contributes to main-taining constant levels of plasma glucose. Hepatic production of glucose arises from two sources: breakdown of glycogen (glycogenoly-sis) and gluconeogenesis from lactate, pyruvate, glycerol and amino acids. Fasting promotes glucose production through cAMP-dependent mechanisms. Feeding inhibits glucose produc-tion through insulin. The cAMP response is dependent on phosphorylation of the cAMP response element binding (CREB) protein and recruitment of the coactivators Cbp/p300 to promoters of gluconeogenic genes. In Nature, Koo and colleagues report that the mechanism of cAMP-induced glucose production is in fact much more complex, and identify the CREB coactivator Torc2 as the main mediator of the fasting response1.

Hepatic glucose production is the result of a concerted process that integrates metabo-lite fluxes to and from the liver, hormonal cues and neurotransmitter release. The clas-sic work of Exton and colleagues in perfused livers showed that glucagon mediates glucose production through cAMP2. This effect does not require new protein synthesis and occurs within seconds. Insulin’s inhibitory effect is equally rapid and results in the suppression of phosphoenolpyruvate carboxykinase (encoded by Pck1)3 and glucose-6-phosphatase (encoded by G6pc) expression.

The identification of hormone-regulated transcription factors that mediate gluconeo-genesis has proven exceedingly difficult. The

cAMP response has classically been viewed as the result of CREB phosphorylation, leading to its recruitment, along with the coactivators p300/Cbp, to the Pck1 and G6pc promoters (Fig. 1)4. The opening salvo of the Koo et al. report is in this respect rather startling. They show that phosphorylation of CREB is induced equally by cAMP and insulin, indicating that phosphorylation is permissive, but not suffi-cient, to explain the induction of gluconeo-genic genes by fasting. However, a different CREB coactivator—Torc2—is specifically dephosphorylated in response to cAMP and translocates to the nucleus to activate CREB-dependent transcription of Pck1 and G6pc. Moreover, the authors show that Torc2 acts upstream of Pgc1α, a known coactivator of fasting-induced glucose production (Fig. 1)5.

After activation of gene expression, the plot thickens, as Torc2 is rapidly inactivated through phosphorylation by the salt-inducible kinase Sik1. This is consistent with the finding that the glucogenetic response is self-limiting and declines upon prolonged exposure to cAMP6. Sik1, in turn, is activated by AMP-dependent protein kinase (AMPK), but not by insulin, ruling out its participation in insulin inhibi-tion of gluconeogenesis (Fig. 1). AMPK is also activated in response to feeding and promotes utilization of nutrients. Thus, the Sik1-Torc2 pathway may explain the suppression of gluco-neogenesis by AMPK agonists, such as the drug metformin. It remains to be seen whether fast-ing-induced changes in AMPK activity are of sufficient magnitude to account for dephos-phorylation of Torc2, and to identify the relevant phosphatases.

Several transcription factors and coactivators have a role in glucose production. In addition to the PKA-CREB cascade, another important

pathway depends on the forkhead protein Foxo1, acting in concert with the coactivator Pgc1α5. Foxo1 is required for induction of G6pc in response to cAMP and for the insulin-mediated inhibition of this process (Fig. 1)7. It is unclear how Foxo1 and CREB signaling interact; one possibility is that they compete for a limited pool of Pgc1α8. Unlike other actors on this crowded stage, Foxo1 is exclusively reg-ulated by insulin through phosphorylation and does not require synthesis of new protein, pro-viding a rapid-response mechanism to mediate hormone-dependent changes.

The authors are in the Department of Medicine,

Columbia University Medical Center, New York,

New York 10032, USA.

E-mail: da230@columbia.edu

Figure 1 Pathways of gluconeogenesis. The involvement of several transcription factors and coactivators has been suggested in the nutritional and hormonal control of gluconeogenesis and are listed above the DNA double strand. Green symbols represent mediators of the cAMP (fasting) response, red symbols represent mediators of the insulin response and brown symbols represent mediators of the AMPK (feeding) response. Akt inhibits glucose production by phosphorylation of Foxo1 and Cbp. Sik inhibits glucose production through phosphorylation of Torc2. Foxa2 is also involved in glucose production during fasting, but its post-translational regulation is disputed. C/ebpα is the main promoter of gluconeogenesis in the early postnatal phase, but does not appear to be hormonally regulated.

AMPK Sik1

PKA

Akt

CREBCbpTorc2

Foxo1Pgc1α

C/ebpαFoxa2

Sim

on F

enw

ick

NATURE MEDICINE VOLUME 12 | NUMBER 1 | JANUARY 2006 33

N E W S A N D V I E W S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e

Author’s CommentsI suppose the biggest surprise for us during this study came at the mechanistic level. We have worked on cAMP-dependent transcription for 20 years, all the time working under the notion that Ser133 phosphorylation of CREB and the subsequent recruitment of the coactivator CBP were the principal events by which cAMP stimulated transcription. So it was a surprise that TORC, instead of CBP, was more critical in turning on gluconeogenesis. The second surprise was the apparent disconnect between fasting and energy depletion pathways. During fasting, glucagon triggers gluconeogenesis through induction of the cAMP pathway, leading to activation of CREB and TORC; but superimposed on this level of regulation, cellular depletion of ATP stores blocks gluconeogenesis through activation of AMPK. AMPK phosphorylates and inhibits TORC at the same site that fasting signals (glucagon) dephosphorylate and activate it. On the basis of these observations, we are currently searching for hormonal signals that utilize this pathway to modulate gluconeogenesis and activation of the CREB-TORC pathway. This work may also address the extent to which hepatic gluconeogenesis is tied to adiposity. Insulin exerts dominant effects in blocking gluconeogenesis. The surprise for us was that despite the prominent role of TORC in the activation of this pathway, insulin does not influence activation of TORC. So insulin probably inhibits regulators downstream of TORC (such as PGC-1α and FOXO). We were also surprised at how effective knockdowns of crucial activators such as TORC are at lowering blood glucose production during fasting. In contrast, it appears far more difficult to induce hyperglycemia by overexpression of the same components in liver, probably because many protective signals (such as IRS2 and Sik1) are in place to protect against fasting hyperglycemia. I suppose the biggest question is this: given all these protective mechanisms that seem difficult to surmount even with the crude tools we use (such as adenoviruses), why do people get type 2 diabetes? There is still much to investigate in the future.

Marc Montminy, The Salk Institute

Sirt1: a metabolic master switch that modulates lifespanIngo B Leibiger & Per-Olof Berggren

Sirt1, an enzyme that removes acetyl groups from specific nuclear proteins, has been linked to the regulation of aging. It is now clear that Sirt1 also controls hepatic glucose metabolism by serving as a sensor of the metabolic status in hepatocytes.

A growing body of evidence suggests that restricted caloric intake extends the lifespan of a number of organisms including yeast, worms, flies and even mammals, and reduces

the incidence of age-related diseases such as cancer, cardiovascular disease and diabetes in animal models1. So, a key question to answer is how the molecular mechanisms that regu-late lifespan relate to metabolism. In a recent publication, Rodgers et al.2 report that Sirt1, a nuclear deacetylase that is involved in aging, also controls hepatic glucose metabolism through the regulation of PPARγ coactivator 1

The authors are at the Rolf Luft Research Center for

Diabetes and Endocrinology, Karolinska Institutet,

SE-171 76 Stockholm, Sweden.

E-mail: per-olof.berggren@ki.se

(PGC-1α), a transcriptional coactivator that controls metabolism at the level of gene tran-scription3.

Sirt1 is one of seven mammalian orthologs of the yeast protein silent information regulator 2 (Sir2), a NAD+-dependent histone deacetylase that was discovered for its role in chromatin remodeling associated with gene silencing and the prolongation of lifespan in yeast. Whereas

The forkhead protein Foxa2, a distant relative of Foxo1, also regulates glucose production9, but its ability to respond to hormonal signals is disputed10. C/ebpα is a member of the bZIP family of tran-scription factors and is thought to be involved in several metabolic processes. C/ebpα knockout mice show complete sup-pression of glucose production with ensuing hypoglycemia, but this phenotype appears to reflect developmental, rather than hormonal regulation of gluconeogenesis, as C/ebpα is the main glucogenetic transcription factor activated in the crucial postnatal period in rodents11. So far, it remains unclear how the nutritional and hormonal status affects C/ebpα activity on Pck1 and G6pc. Similarly, the transcription factor Hnf4α, mutations of which can cause a rare form of diabe-

tes, activates Pck1, but does not respond to hormonal cues12. There is also evidence for Akt-dependent inactivation of Cbp, which may participate in turning off the cAMP response (Fig. 1)13.

As indicated in Figure 1, all of these proteins interact with CREB, or bind to closely spaced cis-acting elements in regulatory regions of glucogenic genes. This eclectic cast of seem-ingly competing and somewhat redundant mechanisms should not befuddle the reader. Hepatic glucose production is a fundamen-tal physiological defense mechanism, and a network of checks and balances has been developed throughout evolution to ensure its smooth operation. It is also an attractive target for diabetes treatment, as fasting hyperglycemia in diabetes tightly correlates with hepatic glu-cose production. Unfortunately, tuning it down

without turning it off and causing hypoglyce-mia has proven exceedingly difficult. Perhaps the Torc2 pathway will provide a new therapeu-tic lead to combat the diabetes epidemic.

1. Koo, S.H. et al. Nature 437, 1109–1111 (2005).2. Exton, J.H. & Park, C.R. J. Biol. Chem. 243, 4189–

4196 (1968).3. Granner, D., Andreone, T., Sasaki, K. & Beale, E. Nature

305, 549–551 (1983).4. Chrivia, J.C. et al. Nature 365, 855–859 (1993).5. Puigserver, P. et al. Nature 423, 550–555 (2003).6. Sasaki, K. et al. J. Biol. Chem. 259, 15242–15251

(1984).7. Nakae, J., Kitamura, T., Silver, D.L. & Accili, D. J. Clin.

Invest. 108, 1359–1367 (2001).8. Herzig, S. et al. Nature 426, 190–193 (2003).9. Wolfrum, C., Asilmaz, E., Luca, E., Friedman, J.M. &

Stoffel, M. Nature 432, 1027–1032 (2004).10. Zhang, L., Rubins, N.E., Ahima, R.S., Greenbaum, L.E.

& Kaestner, K.H. Cell Metab. 2, 141–148 (2005).11. Wang, N.D. et al. Science 269, 1108–1112 (1995).12. Hall, R.K., Sladek, F.M. & Granner, D.K. Proc. Natl.

Acad. Sci. USA 92, 412–416 (1995).13. Zhou, X.Y. et al. Nat. Med. 10, 633–637 (2004).

34 VOLUME 12 | NUMBER 1 | JANUARY 2006 NATURE MEDICINE

N E W S A N D V I E W S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e