Influence of Inositol Pyrophosphates on Cellular Energy Dynamics

See allHide authors and affiliations

Science  11 Nov 2011:
Vol. 334, Issue 6057, pp. 802-805
DOI: 10.1126/science.1211908


With its high-energy phosphate bonds, adenosine triphosphate (ATP) is the main intracellular energy carrier. It also functions in most signaling pathways, as a phosphate donor or a precursor for cyclic adenosine monophosphate. We show here that inositol pyrophosphates participate in the control of intracellular ATP concentration. Yeasts devoid of inositol pyrophosphates have dysfunctional mitochondria but, paradoxically, contain four times as much ATP because of increased glycolysis. We demonstrate that inositol pyrophosphates control the activity of the major glycolytic transcription factor GCR1. Thus, inositol pyrophosphates regulate ATP concentration by altering the glycolytic/mitochondrial metabolic ratio. Metabolic reprogramming through inositol pyrophosphates is an evolutionary conserved mechanism that is also preserved in mammalian systems.

Adenosine triphosphate (ATP) plays essential roles in cellular energetic metabolism. However, other “energy-rich” molecules may also function as ATP does. Inositol pyrophosphates, such as IP7 and IP8, have one or more energy-rich pyrophosphate moieties that have a free energy of hydrolysis comparable to that of ATP (1). Many cellular roles have been attributed to inositol pyrophosphates, from controlling telomere length to modulating insulin signaling (25), but the widespread effects on cellular functions have generated doubts about their real signaling roles (4, 5). The concentrations of inositol pyrophosphates are dynamically controlled: once synthesized, they are rapidly used and recycled back to their precursor IP6. Thus, a considerable amount of ATP is invested in maintaining their steady-state concentrations (6), which raises the possibility that these molecules may link basic metabolism with signaling (3, 5). To assess this possibility, we tested how ATP metabolism is affected by inositol pyrophosphates.

We measured ATP concentration, [ATP], in mutant yeast (Saccharomyces cerevisiae) strains with altered inositol polyphosphate metabolism (Fig. 1, A and B). In IP6-kinase mutants (kcs1Δ) in various genetic backgrounds, all of which had no detectable inositol pyrophosphates, the [ATP] was high (Fig. 1, A and B, and fig. S1); it was 3.5 times that of wild-type (WT) cells when normalized by cell density (Fig. 1, A and B) or weight and 5 times WT when normalized by cell number (fig. S2). Analysis of [ATP] in mutants of various inositol phosphate kinases confirmed a correlation between the absence of inositol pyrophosphates and an increase in [ATP] (Fig. 1B). Overexpression of mammalian inositol hexakisphosphate kinase 1 (IP6K1) in WT yeast resulted in an increased concentration of IP7 and IP8 (fig. S3A) and a decreased [ATP] (Fig. 1C). Mouse embryonic fibroblasts (MEFs) derived from IP6K1 knockout mice (ip6k–/–) had a decreased concentration of IP7 (7) and an increased [ATP] compared with WT MEFs (fig. S4A). The interplay between ATP production and use determines the intracellular [ATP] and influences the adenylate pool (AP) and adenylate energy charge (AEC), which represents the relative saturation of the AP in phosphoanhydride bonds. When [ATP] rises, adenosine diphosphate concentration, [ADP], or adenosine monophosphate concentration, [AMP], (or both) drop. Both [ADP] and [AMP] were significantly lower in kcs1∆ cells than in WT cells (fig. S5, B and C), so that both ATP/ADP and ATP/AMP ratios were increased (fig. S5, D and E). Thus, kcs1∆ yeast had a larger AP and a larger AEC (Fig. 1, D and E), with the AEC being close to one, which is indicative of slow metabolism (8).

Fig. 1

Effects of inositol pyrophosphates on cellular bioenergetic status. (A) Relative [ATP] of WT and kcs1Δ yeasts in W303 genetic background. (B) Analyses of [ATP] of yeasts deficient in the various enzymes of the inositol polyphosphate biosynthetic pathway in DDY1810 background; arg82Δ, ipk1kcs1∆, and vip1kcs1∆ do not have inositol pyrophosphates (23, 24); ipk1Δ and vip1∆ with altered inositol polyphosphates profile have inositol pyrophosphates (23, 24). (C) Relative [ATP] of WT yeast transformed with pGST or pGST-IP6K1 plasmid. (D) Analyses of adenylate pool and (E) adenylate energy charge of WT and kcs1Δ yeast. Data are means ± SD of three independent experiments. Statistical analyses to compare WT with each mutant were performed using a two-tailed and homoscedastic t test (*P < 0.05; **P < 0.01; ***P < 0.001).

Most of the ATP in aerobic eukaryotic cells is generated by oxidative phosphorylation in mitochondria (9). To assess mitochondrial activity, we cultured cells on nonfermentable carbon sources: ethanol (Fig. 2A), glycerol (fig. S6A), or galactose (fig. S6B), which are metabolized only by respiring mitochondria (10, 11). To our surprise, kcs1Δ and kcs1ΔvipΔ yeast were unable to grow in these conditions, which indicated that yeasts without inositol pyrophosphates do not have functional mitochondria. Respirometry measurements on WT and kcs1Δ yeast cultured in glucose showed that, during the logarithmic phase of growth—when fermentation predominates over oxidative phosphorylation, the mitochondria of WT cells consumed oxygen at a rate four times that of kcs1Δ yeast (Fig. 2B). This difference was even greater during the stationary phase, when the supply of glucose is exhausted and WT yeast shifts from fermentation to oxidative phosphorylation; kcs1Δ yeast failed to make this diauxic shift (fig. S7).

Fig. 2

Effects of inositol pyrophosphates on mitochondrial mass and function. (A) Yeast strains of DDY1810 or W303 genetic background were plated on synthetic medium containing 1% glucose or 2% ethanol and incubated at 30°C for 48 and 96 hours, respectively. (B) Respiration of yeast in logarithmic growth at identical cell densities under basal conditions and after treatments with oligomycin or FCCP. Means ± SD; n = 4 (**P < 0.01; ***P < 0.001; paired sample t test). (C) Respiration of WT and ip6k1–/– MEFs at the same cell density under basal condition and after treatments with oligomycin or FCCP. (Right) Means ± SD ; n = 5 Statistical analyses paired sample t test (**P < 0.01). (D) Confocal images of yeast after staining with MTG. (E) Flow cytometric analyses of WT and kcs1∆ yeast and (F) mammalian WT and ip6k1–/– MEFs after incubation with MTG. Representative histograms of three independent experiments.

Mitochondrial oxygen uptake is associated with the translocation of protons out of the mitochondrial matrix, which generates a membrane potential (Δψm) that can be used to synthesize ATP. The amount of oxygen used for ATP synthesis by the mitochondria can be estimated by inhibiting the ATP synthase with oligomycin, which reduced respiration by 50% in WT yeast but had no effect on kcs1Δ cells (Fig. 2B). We measured the maximum respiratory capacity of mitochondria by uncoupling the respiratory chain with trifluorocarbonylcyanide phenylhydrazone (FCCP) (12). As shown in Fig. 2B and figs. S7 and S8A, in both logarithmic and stationary phase, the maximum respiratory capacity of WT cells was much greater than that of kcs1Δ cells, and inhibition of cytochrome c oxidase with 500 μM KCN reduced the oxygen consumption in both cell types, which indicated that the measured oxygen consumption is mitochondrial. Respirometry measurements in ip6k–/– MEF cells (7) also showed reduced mitochondrial function (Fig. 2C and fig. S8B). Under all conditions investigated, the mitochondria of ip6k–/– MEFs consumed oxygen at a rate ~55% of that of WT cells, which supports a conserved role of inositol pyrophosphates in regulating energy metabolism. We visualized yeast mitochondria by confocal microscopy after staining the cells with MitoTracker Green FM (MTG). Although mitochondria were present in both WT and kcs1Δ cells, the mitochondria in mutant cells accumulated less MTG (Fig. 2D), as confirmed by flow cytometry (Fig. 2E). Furthermore, kcs1Δ cells accumulated less tetramethylrhodamine methyl ester (TMRM), which showed that they had a lower Δψm (fig. S9A). Staining ip6k–/– and WT MEFs with MTG and TMRM also revealed that the ip6k–/– MEFs accumulated less of these dyes than WT MEFs (Fig. 2F and fig. S9B). Protein immunoblot of cytochrome c oxidase subunit IV (Cox-IV) confirmed that ip6k–/– MEFs have less of the respiratory chain protein, consistent with reduced respiration (fig. S10).

The high concentration of ATP in mutant cells with inactive mitochondria is paradoxical and suggests that the mutants have enhanced glycolysis and decreased ATP-consuming metabolic processes. Thus, the reduced proliferation rate of kcs1Δ cells may result from decreased biosynthesis of macromolecules, such as fatty acids, nucleotides, and heme, that mainly depend on enzymes in mitochondria. The intracellular redox state depends principally on the ratio of the concentrations of nicotinamide adenine dinucleotide to the same reduced, [NAD+] to [NADH], which, in addition to the ratio of [ATP] to [ADP], largely determines the metabolic state of a cell. Both kcs1Δ yeast and ip6k–/– MEFs showed an increase in [NAD+]/[NADH] compared with that of WT cells (Fig. 3A and fig. S4B), and overexpression of IP6K1 in yeast reduced [NAD+]/[NADH] (Fig. 3B). These findings support a conserved role of inositol pyrophosphates in influencing energy metabolism. The oxidation of NADH to NAD+ is essential for glyceraldehyde 3-phosphate dehydrogenase (GAPDH), a redox-sensitive enzyme, to drive glycolysis (fig. S11). The mitochondrial electron transport chain regenerates NAD+ during oxidative phosphorylation, but with repressed mitochondrial function, NAD+ is produced by the conversion of acetaldehyde to ethanol during fermentation (fig. S11). Thus, in kcs1∆ cells, the high concentration of NAD+ might result from increased glycolytic flux and fermentation.

Fig. 3

Enhanced glycolytic flux and fermentation in kcs1Δ yeast. (A) Increase in kcs1∆ redox homeostasis expressed by the [NAD+]/[NADH]. (B) WT yeast transformed with pGST-IP6K1 has decreased [NAD+]/[NADH]. Measurements of glucose consumption (C) and ethanol concentration (D) in the medium in WT and kcs1∆ yeast cultures. Enzymatic activities in cell lysates from WT and kcs1∆ yeast of (E) GAPDH, (F) PGK1, and (G) ADH2. Means ± SD of three independent experiments. Statistical analyses were performed using a two-tailed and homoscedastic t test (*P < 0.05; **P < 0.01; ***P < 0.001).

The glucose uptake rate could reflect the rate of glycolysis and was greater in kcs1∆ cells than in WT cells (Fig. 3C). Conversely, the concentration of ethanol in the medium of kcs1∆ cells was higher than that in WT cultures (Fig. 3D). The enzymatic activities of the key glycolytic enzymes GAPDH and phosphoglycerate kinase (PGK1) were high in kcs1∆ cells (Fig. 3, E and F), whereas the activity of alcohol dehydrogenase II (ADH2), which catalyzes the oxidation of ethanol to acetaldehyde, was low (Fig. 3G and fig. S11).

Yeasts display tight transcriptional regulation of several glycolytic genes (13). Quantitative polymerase chain reaction (QPCR) analyses revealed increased expression of GAPDH, PGK1, and alcohol dehydrogenase I (ADH1) genes in kcs1Δ cells (Fig. 4A and fig. S12), with no change in ADH2 gene expression. Overexpression of IP6K1 decreased PGK1 gene expression (fig. S3B). The transcriptional machinery controlling glycolytic genes consists of RAP1, GCR1, and GCR2 transcription factors (14, 15). The promoters of all glycolytic genes contain a CT-box (a Gcr1p-binding site) and an RPG-box (a Rap1p-binding site). The binding of GCR1 and RAP1 to these sites is facilitated by GCR2 through a conformational change in GCR1, the stimulation of GCR1 phosphorylation, or both (14, 15). Mass spectrometric analyses of GCR1, performed by the PhosphoPep project (16), showed that the protein has four adjacent serines (S515 to S518) with an ambiguous double-phosphorylation pattern, compatible with either two phosphoserines or with one serine-pyrophosphate (fig. S13A) (16). This GCR1 region also contains acidic amino acids, resembling an IP7 pyrophosphorylation consensus site (17, 18). Incubation of radiolabeled IP7 with either glutathione S-transferase (GST)–GCR1 or with GST-GCR1 (S515-8A), in which the four serines were mutated to alanines, revealed pyrophosphorylation of the WT but not of the mutant protein (fig. S13, B and C). We expressed GST-GCR1 or GST-GCR1 (S515-8A) in kcs1∆ and WT yeast (Fig. 4B left) and found that the expression of GCR1 inhibited the growth of only the mutants (Fig. 4B right). The kcs1Δ cells expressing GST-GCR1(S515-8A) grew at a rate similar to that of control cells (Fig. 4B right). Overexpression of GST-GCR1(S515-8A), which cannot be phosphorylated, reduced GCR1 activity and partially reversed the kcs1Δ phenotype, as confirmed by a 20% reduction of ATP concentration in these cells compared with that of control transformed cells. Because protein pyrophosphorylation can destabilize protein-protein interactions (19), we tested GST-GCR2 association with endogenous GCR1 (model in fig. S14). We observed a stronger GCR1-GCR2 interaction in kcs1Δ compared with WT cells (Fig. 4C). The indirect interaction between endogenous RAP1 and GCR2 was also increased in the mutant strain (Fig. 4D). In chromatin immunoprecipitation (ChIP) analysis, we observed an increase in association of GST-GCR2 with the PGK1 promoter in kcs1Δ compared with WT cells (Fig. 4E). These data suggest that inositol pyrophosphates influence the transcription of glycolytic genes.

Fig. 4

Effect of inositol pyrophosphates on RAP1–GCR1-GCR2 transcriptional complex. (A) QPCR analyses of GAPDH, PGK1, ADH1, and ADH2 in kcs1Δ versus WT yeast. (B) (Left) Immunoblots confirming the expression of WT and mutated GCR1; (right) growth assay, spotting serial dilution of yeasts expressing WT and mutated GCR1. (C) Analyses of the direct interaction between GCR1 and GCR2. After precipitating (PD) of GST-GCR2, GCR1 tandem affinity purification tagged (TAP-GCR1) was detected with antibody to CBPs (calmodulin-binding domains). Inputs of TAP-GCR1 and heat shock protein HSP60 controls are also shown. (D) Study of the indirect interaction between GCR2 and RAP1. RAP1 protein was detected after the precipitation of GST-GCR2. (E) ChIP assay of the promoter region of PGK1 gene in WT and kcs1∆ yeasts transformed with pGST and pGST-GCR2. The relative fold difference in the fractional enrichment of PGK1 promoter was determined as described in the Materials and Methods. Means ± SD of three independent experiments. Statistical analyses were performed using a two-tailed and homoscedastic t test (**P < 0.01; ***P < 0.001).

Inositol pyrophosphates have been implicated in some human diseases, including cancer, diabetes, and obesity (2022). For example, ip6k1–/– mice are resistant to diet induced obesity, perhaps because functional mitochondria are required for the synthesis of fatty acids (20). Therefore, drugs that regulate inositol pyrophosphates, and thereby energy metabolism, might have therapeutic potential.

Supporting Online Material

Materials and Methods

Figs. S1 to S14

References (2537)

References and Notes

  1. Acknowledgments: We thank R. Bhandari, A. Resnick, A. Riccio, and M. Raff for reading the manuscript and M. Bennett for his contribution in the initial phase of the project. This work was supported by the Medical Research Council (MRC) funding to the Cell Biology Unit and by a Human Frontier Science Program Grant (RGP0048/2009-C). A.G. was supported by the British Heart Foundation grant PG/10/72/28449 and MRC grant G1001704.

Stay Connected to Science

Navigate This Article