|Home | About | Journals | Submit | Contact Us | Français|
Histone acetyltransferases (HATs) and histone deacetylases (HDACs) conduct many critical functions through nonhistone substrates in metazoans, but only chromatin-associated nonhistone substrates are known in Saccharomyces cerevisiae. Using yeast proteome microarrays, we identified and validated many nonchromatin substrates of the essential nucleosome acetyltransferase of H4 (NuA4) complex. Among these, acetylation sites (Lys 19 and 514) of phosphoenolpyruvate carboxykinase (Pck1p) were determined by tandem mass spectrometry. Acetylation at Lys 514 was crucial for enzymatic activity and the ability of yeast cells to grow on non-fermentable carbon sources. Loss of Pck1p activity blocked the extension of yeast chronological life span caused by water starvation. In human hepatocellular carcinoma (HepG2) cells, human Pck1 acetylation and glucose production was dependent on TIP60, the human homolog of ESA1. Our results demonstrate a novel regulatory function for the NuA4 complex in glucose metabolism and life span by acetylating a critical metabolic enzyme.
Acetylation and deacetylation of histone lysine residues are among the best characterized post-translational modifications, and control many chromatin-related processes (Kouzarides, 2007; Millar and Grunstein, 2006). There are over 20 histone acetyltransferases (HATs) and histone deacetylases (HDACs) in S. cerevisiae (Lee and Workman, 2007; Shahbazian and Grunstein, 2007); among which only Esa1p, the catalytic enzyme of NuA4 is essential for cell viability (Allard et al., 1999; Clarke et al., 1999; Smith et al., 1998). Esa1, with Epl1p and Yng2p forms the “piccolo NuA4” core that acetylates the N-terminal tails of histones H2A, H4 and Htz1p in vivo (Babiarz et al., 2006; Boudreault et al., 2003; Keogh et al., 2006; Millar et al., 2006). NuA4 is critical for regulating gene transcription, limiting the spread of silent heterochromatin, maintaining cell cycle progression and repairing DNA double-strand breaks. The NuA4 complex is highly conserved from yeast to humans (Auger et al., 2008); the TIP60 complex, the human homolog of NuA4, is important for efficient DNA damage repair and maintenance of apoptotic competence (Ikura et al., 2000). Several recent large-scale yeast genetic interaction screens have implied functional relationships between HATs/HDACs and many extranuclear proteins (Lin et al., 2008; Mitchell et al., 2008). Previous gene expression studies showed that although global downregulation of acetylation profoundly affected gene expression (Takahashi et al., 2006), dysfunction of the individual major HATs and HDACs resulted in only minor transcriptional alterations of distinct gene sets (Huisinga and Pugh, 2004; Robyr et al., 2002; Zhang et al., 2004), suggesting that these enzymes regulate specific extranuclear functions through mechanisms other than transcriptional control, such as (de)acetylation of nonhistone substrates. Most recently, several chromatin-associated proteins were found to be acetylated in yeast, including a cohesin subunit (Smc3p) acetylated by Eco1p (Ben-Shahar et al., 2008; Unal et al., 2008; Zhang et al., 2008), an RSC chromatin remodeling complex subunit (Rsc4p) acetylated by Gcn5p (Choi et al., 2008; VanDemark et al., 2007), and a subunit of NuA4 (Yng2p) acetylated by Esa1p (Lin et al., 2008). However, nonchromatin substrates of HATs/HDACs have not been reported in yeast despite the identification of many such acetylation substrates in higher organisms (Yang and Seto, 2008a; Yang and Seto, 2008b). Full understanding of the molecular functions of these enzymes requires the identification of nonchromatin substrates.
The enzymatic activities of HATs and HDACs are closely related to the metabolic condition of the cells, whereas acetylation and deacetylation of metabolic enzymes also plays a pivotal role in adaptation of the cells to changing energy status (Schwer and Verdin, 2008). In yeast, histone acetylation and global gene transcription is directly regulated by abundance of a key metabolic intermediate, acetyl-CoA, which in turn is determined by nucleocytoplasmic acetyl-CoA synthetase activity (Takahashi et al., 2006). In mammals, the cytoplasmic acetyl-CoA synthetase 1 (AceCS1) is deacetylated and activated by SIRT1, an NAD+ dependent HDAC (Hallows et al., 2006). A large number of cytoplasmic and mitochondrial metabolic enzymes were recently reported to be subject to (de)acetylation in mammals (Kim et al., 2006), but the responsible (de)acetylases remain largely unknown.
The intersection of metabolism and protein (de)acetylation is further reflected by the links between sirtuins and life span regulation in yeast, worms, fruit flies, and possibly mammals (Guarente and Picard, 2005; Schwer and Verdin, 2008). In yeast, limiting the concentration of either glucose or amino acids in the medium can lengthen both replicative life span (RLS, defined as the number of mitotic divisions occurring in a mother cell before senescence) and chronological life span (CLS, defined as the period of time for a cell to survive in stationary phase) (Fabrizio et al., 2005; Lin et al., 2000; Powers et al., 2006). SIR2 plays an antagonistic role in the extension of CLS under water starvation (Fabrizio et al., 2005). Although the detailed mechanism is largely unknown, the effect of SIR2 on CLS is partly mediated by SIR2-dependent inactivation of alcohol dehydrogenase 2 (ADH2) that depletes extracellular ethanol, a proposed central mediator of chronological longevity in yeast (Fabrizio et al., 2005). Although ethanol was shown to be critical in this longevity paradigm, and the requirement for deacetylase activity in blocking extreme life span extension was demonstrated, no relevant Sir2 substrate(s) were identified.
Here, we report 13 novel substrates of the NuA4 complex identified by a protein acetylation microarray approach. In depth analysis of one substrate, Pck1p, provides the first evidence of a metabolic enzyme (de)acetylation state controlling the gluconeogenic flux in the yeast and possibly human cells. Remarkably, Pck1p acetylation is necessary and sufficient for extension of yeast chronological longevity under water starvation.
To fully understand the molecular functions of the NuA4 complex, we sought its substrates using yeast proteome microarrays (Lu et al., 2008; Zhu et al., 2001). First, we developed an in vitro acetylation reaction on a pilot microarray containing histones H3, H4, and bovine serum albumin (BSA). To optimize reaction conditions and surface chemistry (Tao et al., 2007), we performed the assay using both the NuA4 and Spt-Ada-Gcn5-acetyltransferase (SAGA) complexes and [14C]-acetyl-CoA as a labeling reagent on three surfaces at various reaction temperatures and time. NuA4 and SAGA complexes showed preference for H4 and H3, respectively, as expected (Sterner and Berger, 2000) on the FAST (nitrocellulose) slide (Figure S1). No detectable signals were observed for BSA. We then carried out acetylation reactions in duplicate with NuA4 complex on proteome microarrays containing 5,800 yeast proteins (Zhu et al., 2001). We found that 91 proteins were readily acetylated by NuA4 (Figure 1A, and Table S1 for a full list of candidate substrates). Gene Ontology (GO) analysis indicated that these candidate substrates were enriched for nucleosome assembly and histone binding activities, and depleted for mitochondrial and membrane-associated proteins, as expected. However, significant proportions (P < 10−4) of the candidate substrates are cytoplasmic proteins such as Pck1p. Therefore, we chose 20 candidates involved in regulating metabolism (Glk1p, Gph1p, Gsy2p, Pck1p, Sip2p, Sip5p, and Tap42p), transcription (Nnt1p and Ume1p), cell cycle progression (Cdc34p, Ctf4p, Rpt5p and Skp1p), RNA processing (Brx1p and Prp19p), and stress response (Atg3p, Atg11p, Hsp104p, Imp2′p and Plp2p) for further validation.
We tested the ability of the NuA4 complex to acetylate these candidate substrates using standard in vitro activity assays (Smith et al., 1998) and confirmed 13 substrates (Atg3p, Atg11p, Brx1p, Cdc34p, Gph1p, Hsp104p, Nnt1p, Pck1p, Prp19p, Rpt5p, Sip2p, Sip5p and Tap42p), as detected by immunoprecipitation (IP) with a monoclonal anti-pan-acetyl-lysine antibody (Li et al., 2007) (Figure 1B and Figure S2A), followed by immunoblotting with an anti-TAG antibody. Because in vitro enzymatic assays could suffer from promiscuity caused by non-physiologic conditions created in the experiments, we further validated the substrates with in vivo experiments performed in an ESA1 Ts (esa1-531) mutant (Lin et al., 2008). We found that all the in vitro substrates were also acetylated in the wild type (WT) cells; their acetylated fractions dramatically diminished in esa1-531 at nonpermissive temperature (Figure 1C and Figure S2B). Furthermore, the in vivo acetylation signals in the IP assay could be efficiently competed with lysine-acetylated BSA (Ac-K-BSA), indicating the specificity of the detected acetylation signals (Figure S3). Together, these data suggested that Gph1p, Pck1p, Sip2p, Sip5p, Tap42p, Nnt1p, Cdc34p, Rpt5p, Brx1p, Prp19p, Atg3p, Atg11p, and Hsp104p are in vivo substrates of Esa1p.
More than half the confirmed in vivo substrates of NuA4 are enzymes (Atg3p, Gph1p and Pck1p) or effectors (Atg11p, Sip2p, Sip5p and Tap42p) involved in signaling pathways responsive to nutrient availability and energy status. To explore the possible link between NuA4 acetylation and metabolism, we focused on Pck1p for several reasons. First, it is a cytosolic non-chromatin substrate. Second, it is a well-studied enzyme (Baly et al., 1985; Hanson and Reshef, 1997). Finally, it seemed to provide a potential insight into a new and poorly understood connection between acetylation and metabolism. Moreover, the important role of the human enzyme in diabetes (Burgess et al., 2007; Permutt and Hattersley, 2000) and prolonged life span in mice overexpressing PCK1 (Hakimi et al., 2007) made it interesting. PCK1 encodes the intricately regulated phosphoenolpyruvate carboxykinase (PEPCK), catalyzing the rate-limiting step in gluconeogenesis by converting oxaloacetate (OAA) and ATP to phosphoenolpyruvate (PEP), ADP, and CO2. We applied tandem mass spectrometric (MS/MS) analysis and identified lysines 19 (K19) and 514 (K514) as potential acetylation sites for the NuA4 complex (Figure S4). A K514 arginine mutant (K514R) was generated to test association of acetylation of Pck1p with NuA4 activity. In vitro and in vivo assays showed that NuA4-dependent acetylation of overproduced Pck1p (Figure 2A–2C) or endogenously expressed (Figure 2D) was almost completely dependent on K514, suggesting that NuA4 acetylates Pck1p at K514. Conversely, acetylation at K19 is far less abundant than K514 in vivo (Figure 2E), and a K19R mutation did not affect efficient acetylation by NuA4 in vitro (Figure 2F).
Because acetylation is reversible, we applied a similar in vivo approach to search for Pck1p deacetylase. A substantial increase of acetylated Pck1p was observed in the sir2Δ mutant, but not in hda1Δ or rpd3Δ mutants (Figure 3A). SIR2 encodes an NAD+-dependent HDAC involved in chromatin silencing (Rusche et al., 2003), while the other two encode major HDACs governing global histone deacetylation and transcriptional control (Vogelauer et al., 2000). Moreover, affinity-purified Sir2p, but not Hda1p (Figure S5), effectively deacetylated purified Pck1p in vitro in an NAD+-dependent reaction (Figure 3B). Acetylation of endogenous Pck1p increased significantly in the sir2Δ mutant (Figure 2D); in summary, Esa1p and Sir2p specifically catalyze reversible (de)acetylation at K514 of Pck1 in vivo.
PCK1 mRNA and protein are highly induced during a diauxic shift, i.e., the glycolysis to gluconeogenesis shift as glucose is depleted (Schuller, 2003; Zaman et al., 2008). To determine whether (de)acetylation of Pck1p controls its induction, we compared endogenous protein levels of chromosomally tagged Pck1p in esa1-531, sir2Δ, gcn5Δ (the catalytic subunit of the SAGA complex), and hda1Δ mutants with the WT strain. Except for gcn5Δ, none of these mutations significantly affected Pck1p abundance (Figure 3C). Consistent with this finding, neither K514R nor K514Q mutations significantly affected Pck1p abundance (Figure 3C). We checked PCK1 mRNA levels in HAT and HDAC deletion mutants during a diauxic shift, and observed that PCK1 mRNA increased ~8-fold in esa1-531 grown at nonpermissive temperature, and decreased slightly in sir2Δ (Figure 3D). Interestingly, PCK1 mRNA increased 3-fold in hda1Δ, and decreased significantly in gcn5 (Figure 3D), suggesting that Hda1p and Gcn5p repress and activate PCK1 transcription via chromatin modification, respectively, on non-fermentable carbon sources (NFCSs). These results confirm previous functional genomics studies of yeast HDACs, in which Hda1p deacetylates (and probably suppresses) Hda1-affected subtelomeric domains in subtelomeric chromatin containing gluconeogenic genes (Robyr et al., 2002), and clarifies the previously unexplained finding that yeast SAGA is essential for growth on ethanol (van Voorst et al., 2006).
In addition, we observed that C-terminally GFP-tagged Pck1p localized to both nucleus and cytoplasm, independent of (de)acetylation status (Figure S6). Because Pck1p functions as a homotetramer in vivo (Ravanal et al., 2003), we also examined Pck1p oligomerization status by size exclusion chromatography; its oligomerization was normal in esa1-531 and sir2Δ (Figure S7).
To determine whether (de)acetylation of Pck1p was physiologically relevant, we monitored viability of various mutants on NFCSs. Gluconeogenesis is essential for yeast growth on NFCSs, such as ethanol and glycerol, allowing replenishment of sugar phosphate stores important for biomass generation (Schuller, 2003). Deletion of PCK1 is lethal when ethanol is the only available carbon source (Figure 4A). Adding glycerol and ethanol (GE) bypasses the need of Pck1p in the early stages of the pathway and rescues pck1Δ lethality (Figure 4A). Similar phenotypes were observed for esa1-531 grown at a semi-permissive temperature but not at permissive temperature (Figure S8), and for the pck1-K514R mutant grown on either ethanol or GE, suggesting acetylation at K514 of Pck1p is crucial for function (Figure 4A). There was no additive effect on the growth defect in the esa1-531/pck1-514R double mutant on NFCSs (Figure S9). Consistent with this result, substitution of K514 with glutamine (K514Q) almost completely rescued the esa1-531 growth defect on ethanol or GE (Figure 4A). Moreover, deletion of SIR2 did not rescue pck1-K514R growth defects on ethanol or GE, suggesting that K514 is the only Pck1p lysine residue targeted by Sir2p (Figure S10). Overexpression of SIR2 by introducing an extra copy of SIR2 on a centromeric plasmid (2X SIR2 in Figure S11A) does not significantly affect growth on glucose, ethanol or GE (Figure S11B). Although both GCN5 and HDA1 regulate PCK1 at the mRNA level, only deletion of GCN5 showed similar growth defects, consistent with its effect on Pck1p abundance (Figure 3C, 3E and and4B).4B). Interestingly, both pck1-K514Q and sir2Δ enhanced growth on high concentrations of ethanol (Figure 4C), indicating the contribution of gluconeogenesis to ethanol tolerance in yeast.
These results prompted us to investigate whether acetylation directly affects Pck1p enzymatic activity. An enzyme assay (Figure 4D) was used to assess the ability of Pck1p to convert PEP and ADP to OAA and ATP, coupled with the conversion of OAA to malate by malate dehydrogenase; in this standard assay for Pck1p activity the latter activity is detected spectrophotometrically via NADH consumption (Baly et al., 1985). We observed that Pck1p purified from esa1-531 was defective in the reaction under a wide range of ADP concentrations (Figure 4E and 4F). While pck1-K514R diminished Pck1p enzymatic activity (Figure 4E; the curves of pck1-K514R and esa1-531 overlapped), pck1-K514Q did not cause significant changes in activity (Figure S12), but partially restored the impaired enzymatic activity of Pck1p purified from esa1-531 grown at the non-permissive temperature (Figure 4E), indicating that acetylation at K514 of Pck1p is important in maintaining maximal activity. Consistent with the insignificant contribution of K19 to the overall Pck1p acetylation status, pck1-K19R minimally affected activity (Figure S13). Finally, the Km and Vmax of Pck1p purified from esa1-531 differ substantially from those from WT cells (Figure 4F).
The enzyme-substrate relationship between Sir2p and Pck1p and Pck1p’s ability to convert ethanol to glucose in postdiauxic cells suggest a possible connection between gluconeogenesis and longevity. To test this hypothesis, we investigated whether Pck1p is involved in CLS extension caused by water starvation. The CLS of various mutants in the BY4742 background was assessed by holding saturated cultures in SC medium containing 2% glucose, a standard culture condition to propagate yeast cells in the laboratory, and in water, a severe form of starvation, causing entry into a low metabolic state and a dramatic extension of CLS (Fabrizio et al., 2005; Fabrizio and Longo, 2003). In agreement with a previous study (Fabrizio et al., 2005), the chronological life span of a sir2Δ mutant was extended significantly compared to WT in water but not in SC medium containing glucose (Figure 5A). By contrast, knockout of PCK1 dramatically reduced CLS in both water and glucose containing medium (Figure 5A). K514 acetylation and hence enzymatic activity of Pck1p was involved since the pck1-K514R mutation, like pck1Δ, shortened CLS both in water and glucose-containing medium; whereas the pck1-K514Q mutation, just like sir2Δ, extends CLS in water (Figure 5A). Moreover, deletion of SIR2 did not alter the life span of pck1Δ, pck1-K514R and pck1-K514Q mutants, indicating PCK1 was epistatic to SIR2 in this pathway and implicating it as the key Sir2p substrate controlling this form of longevity (Figure 5B). The esa1-531 mutant has an even shorter CLS than pck1Δ in both 2% glucose and water at 30° C, a semi-permissive temperature; at the permissive temperature there is little difference (25° C, Figure S14). This finding suggests that additional NuA4 substrates besides Pck1p control yeast CLS. In contrast, Pck1p acetylation status had minimal effect on replicative life span of yeast cells grown in YPD (Figure S15), thus Pck1p is not involved in the replicative pathway.
On the basis of previous findings that uptake of extracellular ethanol mediates extension of CLS, and that extracellular ethanol is depleted faster in long-lived mutants such as sir2Δ (Fabrizio et al., 2005), we measured the ethanol concentration in the CLS conditioned medium. Interestingly, extracellular ethanol concentration was inversely proportional to CLS. Ethanol concentration remained high up to day 7 in short-lived mutants including esa1-531 at semi-permissive temperature (30° C), pck1Δ and pck1-K514R; whereas in long-lived mutants including sir2Δ and pck1-K514Q extracellular ethanol was depleted as early as day 3 (Figure 5C). Similar to the effect on CLS, PCK1 was epistatic to SIR2 in the mechanism controlling consumption of extracellular ethanol, since deletion of SIR2 resulted in only minimal changes of ethanol concentration in pck1Δ, pck1-K514R and pck1-K514Q mutants (Figure 5C).
We further examined the resistance of these mutants to oxidative stress resultingfrom exposure to high level of H2O2, generating reactive oxygen species (Lin et al., 2002; Scherz-Shouval and Elazar, 2007). Consistent with a previous study (Fabrizio et al., 2005), a sir2Δ mutant in stationary phase (day 3 cells) relative to the exponential growth phase (day 1 cells) was more resistant to H2O2 stress (Figure S16). Although pck1Δ was less resistant overall, the sensitivity of pck1-K514R and pck1-K514Q to oxidative stress was similar to the WT (Figure S16) regardless of the growth phase, suggesting that (de)acetylation status and enzymatic activity of Pck1p play minimal role in this function.
Two isozymes of PEPCK exist in mammalian cells: cytosolic (PCK1) and mitochondrial (PCK2) (Hanson and Reshef, 1997). The mammalian Pck1 (PEPCK-C) catalyzes the first committed step of gluconeogenesis in liver and kidney (Rognstad, 1979), and induces glyceroneogenesis in liver and adipose tissue (Reshef et al., 2003); the physiologic role of Pck2 (PEPCK-M) has not been well established (Hanson and Reshef, 1997). Both mammalian isozymes show low sequence identity with yeast Pck1p (Figure S17), and utilize GTP instead of ATP (Utter and Kolenbrander, 1972). However, both ATP- and GTP-dependent enzymes share similar active site sequence consensus motifs, suggesting conserved catalytic mechanism (Matte et al., 1997).
To examine this hypothesis, plasmid-borne full-length coding regions of both human PEPCK genes were introduced into the yeast pck1Δ mutant, and growth of transformants on NFCSs was examined. As illustrated in Figure 6A, only human PCK1 but not PCK2 could partially complement the function of yeast PCK1, suggesting that human Pck1 shares gluconeogenic mechanism with yeast Pck1p. To further investigate the role of acetylation in controlling human Pck1 enzymatic activity, we used short hairpin RNAs (shRNAs) to efficiently knockdown TIP60 and SIRT1, the human homologs of yeast ESA1 and SIR2, respectively, in human hepatocellular carcinoma (HepG2) cells (Figure S18A–B). Transcription of human PCK1 depends on SIRT1 in primary hepatocytes (Rodgers et al., 2005). Consistent with this, knockdown of SIRT1 reduced the endogenous human Pck1 level (~50%) in HepG2 cells (Figure 6B). By IP, we found that human Pck1 was acetylated in HepG2 cells (Figure 6C), but not in human embryonic kidney 293T cells (data not shown). Whereas knockdown of TIP60 did not affect the endogenous human Pck1 level (Figure 6B), it dramatically diminished the acetylation fraction of human Pck1 compared with GFP knockdown control (Figure 6C). However, knockdown of SIRT1 showed no detectable effect on human Pck1 acetylation (Figure 6C) suggesting a distinct deacetylation mechanism in mammalian cells. Treatment with HDAC inhibitors such as trichostatin A (TSA, a class I/II HDAC inhibitor) or nicotinamide (NAM, a class III HDAC inhibitor) (Neugebauer et al., 2008) increased the acetylation fraction of human Pck1 in HepG2 cells (Figure 6D). Consistent with the hypothesis that acetylation controls enzymatic activity of human Pck1 in vivo, the amount of glucose secreted into glucose-free medium significantly declined (~40%) in TIP60 knockdown HepG2 cells, but knockdown of SIRT1 had no significant effect (Figure 6E).
We report a proteome-wide search for nonhistone substrates of the essential yeast NuA4 complex. We have validated 13 in vivo substrates among 20 candidates, 7 of which are cytoplasmic proteins involved in metabolic pathways responding to extracellular nutrient sources and intracellular energy status. It is noteworthy that several functions (e.g., gluconeogenesis and autophagy) of NuA4 metabolic protein substrates are induced by limiting extracellular nutrient sources (Schuller, 2003; Takeshige et al., 1992; Zaman et al., 2008). Considering that acetylation of Pck1p by NuA4 is important for its gluconeogenic activity, it is reasonable to speculate that NuA4 might act as a central hub to help yeast cells meet the challenge of nutrient deprivation by acetylating these nonhistone substrates, and activating a metabolic network.
Several functions are important in the extension of yeast life span under calorie restriction or starvation, including stress response (Fabrizio et al., 2005), autophagy (Powers et al., 2006), decreased ribosome biogenesis and protein translation (Kaeberlein et al., 2005), respiration (Lin et al., 2002), and changes in carbon metabolism (Ashrafi et al., 2000). The target of rapamycin (TOR) signaling pathway is regarded as a central negative regulator of these functions (Kaeberlein et al., 2005; Powers et al., 2006) to block extension of life span. Other genes such as SIR2 might also be important because of additive effects on life span extension when both the TOR pathway and SIR2 function are disrupted compared to either single mutation (Fabrizio et al., 2005). Our finding that acetylation control of Pck1p gluconeogenic activity to convert ethanol to glucose is crucial for longevity extension provides evidence that NuA4 might be a central positive regulator that opposes TOR in CLS extension under water starvation. NuA4 might further modulate TOR function by acetylating Tap42p, a downstream effector in TOR pathway (Duvel et al., 2003) uncovered as a substrate in this study. It also provides a new insight into how SIR2 plays an antagonistic role in extending CLS. Considering that transcription of PCK1 is repressed by even trace amounts of glucose (as low as 0.001% in the medium) (Yin et al., 2000), the lack of effect on RLS by deleting PCK1 or altering the acetylation status of its encoding protein is unsurprising (Figure S15). However, NuA4 might regulate RLS through acetylating Nnt1p, the yeast putative nicotinamide N-methyltransferase with a role in rDNA silencing and life span determination (Anderson et al., 2003). Interestingly, mice that overexpress PCK1 in skeletal muscle live longer and are more energetic at an older age (Hakimi et al., 2007), suggesting a potential role of PCK1 in regulating longevity in mammals.
Gluconeogenesis is extensively and elaborately regulated to maintain blood glucose level within a narrow range during fasting to ensure a steady and continuous glucose supply in mammals (Burgess et al., 2007). As a crucial enzyme in gluconeogenesis, mammalian PCK1 transcription is tightly controlled by various transcription factors in response to varied physiological stimuli, including dietary carbohydrate content, nutritional status and various hormones (Beale et al., 2004; Hanson and Reshef, 1997). Fasting also induces PCK1 expression in liver and adipose tissue (Hagopian et al., 2003; Hagopian et al., 2008; Raab et al., 2005). Failure of this exquisite transcriptional control mechanism causes dysregulated overexpression of PCK1 in liver and inappropriately high levels of hepatic gluconeogenesis in all forms of diabetes (Burgess et al., 2007; Permutt and Hattersley, 2000). Previous studies have focused mainly on transcriptional regulation of PCK1. However, our result suggests that HATs and HDACs not only regulate PCK1 transcription (by Gcn5p and Hda1p) in yeast, but also regulate Pck1p activity post-translationally both in yeast and humans (Figure 7). Whereas acetyl-CoA synthetase activates acetate produced by non-fermentable sugar metabolism to acetyl-CoA during diauxic shift in yeast (Haurie et al., 2001; van den Berg et al., 1996), fasting activates expression of SIRT1 in mammals, which in turn promotes the production of acetyl-CoA by deacetylating and activating AceCS1 (Schwer and Verdin, 2008). The elevated intracellular acetyl-CoA concentration favors the transcription of PCK1 and the stimulatory acetylation of Pck1p. Our results suggest that acetylation of Pck1p increases its enzymatic activity through mechanisms other than regulating macromolecular abundance, subcellular localization or quaternary structure of the protein. Based on the crystal structure of the Escherichia coli ortholog (Delbaere et al., 2004), which shares high level of amino acid sequence homology with yeast Pck1p (Figure S19), acetylation at K514 might cause a conformational change of the C-terminal mononucleotide-binding domain in favor of substrate binding or catalysis of the phosphoryl-transfer reaction. Despite sharing minimal primary structural identity with yeast Pck1p (Ravanal et al., 2003) (Figure S17), we found that human Pck1 is also activated by TIP60-dependent acetylation. Unlike yeast Pck1p, which is deacetylated exclusively by Sir2p, human Pck1p might be deacetylated by multiple deacetylases. Upregulation of Pck1p transcription and activity, and allosteric modulation of the activity of pyruvate carboxylase (another key enzyme of the gluconeogenesis pathway) by acetyl-CoA (Pronk et al., 1996) might together provide a “triple-switch” to ensure a prompt adaptation of the gluconeogenic flux to energy status in yeast and possibly human cells (Figure 7).
In summary, we have found novel extracellular functions of yeast NuA4 complex in regulating gluconeogenesis and chronological life span. PEPCK gluconeogenic activity, etiologically important in human diabetes, is acetylated in both yeast and humans. Characterization of the structural basis of this acetylation control mechanism, as well as identification of the HDACs responsible for reversing the acetylation of human Pck1 in human liver will be crucial for further clarifying how blood glucose level is maintained within a remarkably narrow range. Besides Pck1p, understanding how other metabolic protein substrates are regulated by NuA4 in response to starvation and their possible roles in extension of chronological life span will also be an important next step. Lastly a similar approach using a human proteome microarray will help identify the HATs/HDACs responsible for the increasing number of acetylated human proteins.
A 200 μL reaction mixture containing 50 mM sodium butyrate, 5 mM DTT and 5 mM PMSF, 0.025 μCi [14C]-Acetyl-CoA (GE Healthcare, 0.05 mCi/mL, 8 mCi/mmol), with or without 4 μg of NuA4, was prepared in reaction buffer (250 mM Tris-HCl, pH7.5, 25% glycerol, 0.5 mM EDTA, and 250 mM KCl). A yeast proteome microarray (Zhu et al., 2001) was incubated with the reaction mixture at 30°C for 1 h, and then subjected to three 15 min washes with 50 mM NaHCO3-Na2CO3 (3:1) (pH9.3). The acetylation signals were then visualized by autoradiography after exposure to X-ray film for 20 d, scanned and saved as TIFF files. The TIFF images were processed by assigning an artificial green color to each brightened spot and analyzed using GenePix software to determine relative acetylation levels. After subtraction of backgrounds, protein spots with median signal intensity greater than an arbitrary cutoff value of 3,000 (determined based on visual inspection) were chosen as candidate substrates of NuA4 for further validation studies.
GST-tagged Pck1p driven by the GAL1 promoter was purified from WT, esa1-531, pck1-K514R and the esa1-531pck1-K514R double mutant after protein induction at 37°C (non-permissive temperature) for 2 h, or from WT, hda1Δ, rpd3Δ, and sir2Δ strains after protein induction at 30°C for 4 h, immunoprecipitated by incubating with 30 μL protein A-sepharose beads (GE Healthcare) conjugated with 6 μg of mouse monoclonal anti-acetylated-lysine (Cell Signaling, 9681S) in the presence of a cocktail of HDAC inhibitors (100 μM Trichostatin A, 50 mM nicotinamide, and 50 mM sodium butyrate) at 4°C for 3 h. To detect acetylation of endogenous C-terminally Myc-tagged Pck1p from WT, esa1-531, pck1-K514R and sir2Δ mutants, whole cell extracts were precleared by incubating with 30 μL protein A-sepharose beads at 4°C for 90 min, followed by IP as described above. After extensive washes with 500 μL of washing buffer (50 mM HEPES, pH 7.0, and 100 mM NaCl) at 4°C for four times (10 min each time), the immunoprecipitate was denatured in 20 μL of 2xSDS sample buffer, resolved by SDS-PAGE, transferred to nitrocellulose membranes and probed with rabbit polyclonal anti-GST antibody (Chemicon, AB3282) or mouse monoclonal anti-Myc antibody (Santa Cruz, SC-40) for purified or endogenous Pck1p, respectively. 150 μg of acetylated BSA (Ambion, AM26141) was used for competition of binding, and an equal amount of unacetylated BSA (New England BioLabs, B9001S) was used in parallel samples as control.
The activity of purified Pck1p was assayed as described previously (Baly et al., 1985). In brief, enzymatic activity of Pck1p was assayed using a coupled system in which the oxaloacetate produced in the reaction was reduced to malate by NADH in the presence of malate dehydrogenase. The NADH consumption rate was measured spectrophotometrically at 340 nm. Equal amount (50 μg) of purified GST-tagged enzyme was added to a 1 mL reaction mixture (10 mM potassium phosphoenolpyruvate, 10 mM NaHCO3, 2 mM of MnCl2, 1 mM ADP, 12 mM sodium aspartate, 80 mM Tris-HCl, pH 8.8, 14 units malate dehydrogenase, and 0.15 mM NADH) for each reaction. Samples without NADH served as a blank control, and samples without Pck1p as a negative control.
Chronological life span of yeast cells grown in either synthetic complete medium (with a 4-fold excess of the supplements histidine, leucine, tryptophan and uracil) containing 2% glucose, 2% glycerol and ethanol, or water was measured as described (Fabrizio and Longo, 2007). In either culture condition, colony-forming units (CFUs) were measured every 48 h. The number of CFUs on day 3 was considered the initial survival (100%). The ethanol concentration in medium collected from day 1, 3, 5 and 7 yeast cultures was analyzed with a colorimetric ethanol assay kit (BioAssay Systems, DIET-500).
HepG2 cells were cultured in minimal essential medium (MEM, GIBCO) containing 10% fetal bovine serum (GIBCO), 100 units/mL penicillin and 100 μg/mL streptomycin at 37°C in a humidified atmosphere containing 5% CO2. The shRNA (obtained from The RNAi Consortium human lentiviral shRNA library) targeting GFP, TIP60 (TRCN0000020314, TRCN0000020315 and TRCN0000020316) or SIRT1 (TRCN0000018979, TRCN0000018980 and TRCN0000018981) were used for transfection of the packaging HEK 293T cells with helper vectors using Lipofectamine™ 2000 transfection reagent (Invitrogen) according to the manufacturer’s instructions. Medium containing lentiviral particles and 8 μg/mL polybrene (Sigma-Aldrich) was used to infect HepG2 cells. Infected HepG2 cells were then selected in MEM in the presence of 1.25 μg/mL puromycin (Sigma-Aldrich). Knockdown efficiency of TIP60 and SIRT1 was assayed by immunoblotting and RT-PCR. When treated with histone deacetylase inhibitors (HDACi), TSA (100 μM; Sigma-Aldrich) or NAM (5 mM; Sigma-Aldrich) was added in the culture medium 6 hours prior to harvest cells for preparation of whole-cell extracts.
The amount of glucose secreted by HepG2 cells in media was assayed as described previously (Yoon et al., 2001). HepG2 cells expressing shRNAs of GFP control, TIP60 or SIRT1 were cultured in six-well plates (1.4 × 106 cells per well) in MEM with 10% FBS. The medium was replaced with 2 mL of glucose production buffer consisting of glucose-free MEM (pH 7.4), without phenol red, supplemented with 2 mM sodium pyruvate and 20 mM sodium lactate. After a 3 h incubation, half of the glucose production buffer was collected and the glucose concentration was measured with a colorimetric glucose assay kit (Sigma-Aldrich, GAGO20). The readings were normalized to the total protein content determined from the whole cell extracts.
We thank E. M. Cooper, L. Dai, A. Norris and K. A. O’Donnell for gifts of reagents; J. Liu, J. T. Stivers, H. Takahashi, S. Taverna, J. Zhang and members of the Zhu and Boeke labs for valuable discussions throughout the course of the work. This work was supported in part by NIH Roadmap grant “Technology Center for Networks and Pathways” (U54 RR 020839) to J.D.B., S.L.B. and H.Z., the Keck Foundation to H.Z., and NIH grants (R01CA126832) to Y.Z. and (R01EY017589) to J.Q.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.