Phosphofructokinase 1 Glycosylation Regulates Cell Growth and Metabolism

See allHide authors and affiliations

Science  24 Aug 2012:
Vol. 337, Issue 6097, pp. 975-980
DOI: 10.1126/science.1222278


Cancer cells must satisfy the metabolic demands of rapid cell growth within a continually changing microenvironment. We demonstrated that the dynamic posttranslational modification of proteins by O-linked β-N-acetylglucosamine (O-GlcNAcylation) is a key metabolic regulator of glucose metabolism. O-GlcNAcylation was induced at serine 529 of phosphofructokinase 1 (PFK1) in response to hypoxia. Glycosylation inhibited PFK1 activity and redirected glucose flux through the pentose phosphate pathway, thereby conferring a selective growth advantage on cancer cells. Blocking glycosylation of PFK1 at serine 529 reduced cancer cell proliferation in vitro and impaired tumor formation in vivo. These studies reveal a previously uncharacterized mechanism for the regulation of metabolic pathways in cancer and a possible target for therapeutic intervention.

Rapid physiological response is critical for the growth of tumors, which must satisfy the metabolic demands of increased cell proliferation in the face of a dynamically changing microenvironment. Cancer cells reprogram their cellular metabolism to generate molecules such as adenosine triphosphate (ATP), nucleotides, lipids, and reduced nicotinamide adenine dinucleotide phosphate (NADPH) to facilitate macromolecular synthesis and maintain redox homeostasis (1, 2). Although mutations in tumor suppressors and oncogenic pathways contribute to these altered metabolic phenotypes (13), such changes are slower and relatively static. In contrast, posttranslational modifications such as protein phosphorylation allow cells to respond rapidly and reversibly to a wide range of signals.

The dynamic posttranslational glycosylation of proteins with O-linked β-N-acetylglucosamine (O-GlcNAcylation) serves as a nutrient sensor to couple metabolic status to the regulation of signaling pathways (47). O-GlcNAc transferase (OGT) catalyzes the transfer of N-acetylglucosamine from uridine diphospho-N-acetylglucosamine (UDP-GlcNAc) to serine or threonine residues of many intracellular proteins, including signaling proteins important for insulin resistance (5), oncogenes and tumor suppressors (6), and transcriptional coactivators that control gluconeogenesis (7). O-GlcNAc levels are increased in multiple tumor types, and OGT silencing inhibits breast tumor growth and prostate cancer metastasis (6, 8, 9).

O-GlcNAcylation can be rapidly induced (10) and is dynamically sensitive to changes in cellular UDP-GlcNAc concentrations (4). UDP-GlcNAc is biosynthesized from several key metabolites, including glucose, glutamine, acetyl–coenzyme A, uridine, and ATP (11, 12). Consequently, UDP-GlcNAc may serve as a functional reporter of the status of various metabolic pathways. Indeed, UDP-GlcNAc and the hexosamine biosynthesis pathway couple growth factor–induced glutamine uptake to glucose metabolism through N-glycosylation of the interleukin-3 (IL-3) receptor (11). However, the fates of UDP-GlcNAc, particularly the functional consequences of dynamic O-GlcNAcylation in the regulation of cell metabolism, remain unknown.

To test whether O-GlcNAcylation directly couples nutrient sensing to cellular metabolism, we modulated O-GlcNAc concentrations and measured the effects on aerobic glycolysis. Global abundance of O-GlcNAc was increased by two- to fourfold in human lung cancer H1299 cells by overexpressing OGT or after pharmacological inhibition of β-N-acetylglucosaminidase (O-GlcNAcase or OGA), the glycosidase that removes O-GlcNAc, with O-(2-acetamido-2-deoxy-d-glucopyranosylidene)amino-N-phenylcarbamate (PUGNAc; Fig. 1A). Increasing the abundance of O-GlcNAc resulted in decreased rates of glucose metabolism relative to those of untreated cells under both normoxic and hypoxic conditions, as measured by the conversion of 5-3H-glucose to 3H2O, which is catalyzed by enolase in the penultimate step of glycolysis (Fig. 1B and fig. S1). Enhancing O-GlcNAcylation also led to reduced lactate production and lowered cellular concentrations of ATP (Fig. 1B). Similar effects were observed in other cells, including invasive human lung cancer A549 cells and human embryonic kidney 293T cells (fig. S2). Because glycolytic flux is 5 to 15 times higher in cancer cells than flux through other central pathways in cancer cells (13), small alterations in glycolysis can result in substantial changes in the relative flow of branching pathways (14). To assess whether OGT-dependent glycosylation of protein substrates contributes to these effects, we stably depleted H1299 and 293T cells of OGT through the expression of short hairpin RNA (shRNA) (fig. S3). The inhibition of OGA in these OGT-deficient cells had no significant effect on glucose metabolism, lactate production, or ATP production (fig. S3).

Fig. 1

Effects of O-GlcNAcylation on cellular metabolism and glycosylation of PFK1. (A) O-GlcNAcylation in H1299 cell lysates, as determined by immunoblotting for O-GlcNAc after treatment of cells with the OGA inhibitor PUGNAc or OGT overexpression. WB, Western blot. (B) Glycolytic rate, lactate production, and relative ATP levels in untreated (Cont), PUGNAc-treated, and OGT-overexpressing H1299 cells (n = 5 experiments). (C) PFK1 activity in untreated (Cont), PUGNAc-treated, and OGT-overexpressing 293T cells (n = 5 assays). (D) Detection of PFK1 glycosylation by protein immunoblotting after chemoenzymatic labeling of O-GlcNAc residues with UDP-GalNAz and the enzyme GalT, followed by reaction with an alkyne-biotin derivative, streptavidin precipitation, and elution of the biotinylated proteins. GalT or UDP-GalNAz was removed to confirm selective labeling of O-GlcNAc. (E) Detection of glycosylated PFK1 in 293T cells stably expressing Flag-tagged PFK1 by chemoenzymatic labeling with a 5-kD mass tag (18) and immunoblotting with a Flag antibody. Cells were untreated (Cont), PUGNAc-treated, or transfected to overexpress OGT. (F) Induction of PFK1 glycosylation under hypoxic conditions. H1299 cells stably expressing Flag-tagged PFK1 were cultured under 0.5% O2 for the indicated times and rapidly lysed. Glycosylated PFK1 was detected after labeling with a 5-kD mass tag as above. (G) PFK1 glycosylation in human lung tumor (T) tissues compared to matched normal (N) tissues. O-GlcNAc–modified proteins from tissue lysates were biotinylated and detected as above. Relative PFK1 glycosylation was normalized using normal tissue for each patient. Error bars denote the standard error of the mean (SEM). Statistical analysis was performed by one-way analysis of variance (ANOVA) and Bonferroni comparison post-test in (B) and (C) and by Student’s t test in (G) (*P < 0.05, **P < 0.01, ***P < 0.001).

Nearly all of the enzymes in the glycolytic pathway are putative substrates for OGT (15). Thus, we modulated amounts of O-GlcNAc in 293T cells and assayed the activity of each enzyme in the pathway. Increased amounts of O-GlcNAc led to decreased activity of phosphofructokinase 1 (PFK1), a major regulatory enzyme that controls flux through glycolysis (16) (Fig. 1C). No change in the expression of PFK1 protein was observed (fig. S3). Enhancing the abundance of O-GlcNAc had little effect on other key regulatory points in the pathway, including hexokinase, phosphoglycerate kinase, and pyruvate kinase (fig. S4), nor did it affect other glycolytic enzymes.

To assess whether PFK1 is directly O-GlcNAcylated, we selectively labeled O-GlcNAc–modified proteins from 293T cell lysates with a non-natural azido sugar through exposure to an exogenous galactosyltransferase enzyme that specifically glycosylates terminal GlcNAc sugars (15, 17) (fig. S5). Labeled proteins were then biotinylated through [3+2] azide-alkyne cycloaddition chemistry and isolated with streptavidin-agarose beads. Immunoblotting of the purified proteins with an antibody to PFK1 showed strong O-GlcNAcylation of PFK1 (Fig. 1D), which was further enhanced by overexpression of OGT (fig. S5). We also generated a stable cell line expressing Flag-tagged PFK1 and selectively labeled O-GlcNAc–modified proteins in the lysate with a 5-kD polyethylene glycol (PEG) mass tag to shift their molecular mass (18). Immunoblotting with an antibody to Flag enabled direct visualization of both the nonglycosylated and glycosylated species of Flag-PFK1 (Fig. 1E). The population of glycosylated PFK1 significantly increased upon OGT overexpression or OGA inhibition. Moreover, PFK1 glycosylation was induced under hypoxic conditions within minutes and accumulated in a time-dependent manner on 32.3 ± 3.8% of PFK1 (Fig. 1F and fig. S6). Glycosylation was also stimulated when cells were deprived of glucose (13.3 ± 2.2%; fig. S6), consistent with previous reports that O-GlcNAc levels and OGT expression are increased by nutrient deprivation and other forms of cell stress (19, 20).

PFK1 was glycosylated in multiple cell lines from human solid tumors, including breast, prostate, liver, colon, and cervical cells, and glycosylation was greater in malignant than in nontumorigenic breast and prostate cell lines (fig. S7). Glycosylation of PFK1 also occurred in human breast and lung tumor tissues and was significantly elevated by two- to fourfold in the majority of tumors relative to tumor-adjacent normal tissues from the same patient (Fig. 1G and fig. S8). Low-stage (stages I and II) lung adenocarcinoma tumors exhibited on average a 1.8-fold increase in PFK1 glycosylation as compared to that of the matched normal tissue, whereas high-stage (stages III and IV) lung adenocarcinomas showed an average 3.2-fold increase in glycosylation.PFK1 glycosylation was not induced in rapidly proliferating normal mouse T lymphocytes and human dermal fibroblast cells, as compared to their quiescent counterparts (fig. S9). Thus, PFK1 is modified with O-GlcNAc in cancer cells both in vitro and in vivo, and glycosylation is increased specifically under conditions associated with tumorigenesis and tumor growth.

To identify the glycosylation site(s) on PFK1, we transiently overexpressed Flag-tagged PFK1 and OGT in 293T cells. After immunoprecipitation and proteolytic digestion of PFK1, O-GlcNAcylated peptides were enriched by wheat germ agglutinin lectin affinity chromatography and subjected to electron transfer dissociation mass spectrometry analysis. We identified a single site of glycosylation at Ser529, a highly conserved residue important for allosteric regulation of PFK1 by fructose-2,6-bisphosphate (F-2,6-BP) (21) (fig. S10). F-2,6-BP is the dominant activator of PFK1 at the high ATP concentrations (2 to 5 mM) found in cancer cells (22). The mutation of Ser529 to alanine (S529A) abolished the glycosylation of PFK1 in 293T cells, whereas alanine mutation of Thr527 had no effect (fig. S10).

Although no structure of human PFK1 is available, we used the Saccharomyces cerevisiae structure, which shares 82% sequence identity within the F-2,6-BP binding site, and the rabbit structure, which shares 97% sequence identity, to generate structural models of rabbit PFK1 complexed to F-2,6-BP and O-GlcNAcylated rabbit PFK1 (fig. S11; the root mean square deviation between the rabbit and yeast structures was only 1.70 Å). Ser529 formed a hydrogen bond with the 2-phosphate group of F-2,6-BP, and the O-GlcNAc moiety occupied the F-2,6-BP–binding pocket, indicating that O-GlcNAcylation might inhibit PFK1 activity by blocking binding of F-2,6-BP and disrupting PFK1 oligomerization.

To further examine the effects of O-GlcNAcylation on PFK1 activity, we expressed human Flag-tagged PFK1 (L, M, and P isoforms) in 293T cells in the presence or absence of hypoxic conditions, which enhance PFK1 glycosylation. Increasing O-GlcNAcylation of PFK1 by 25 to 33% because of hypoxia decreased the activity of all three isoforms by 21 to 36%, with the L and P isoforms being most sensitive to glycosylation (Fig. 2A and fig. S12).We then focused on the L isoform of PFK1, whose expression is enhanced in multiple cancer cells (23). Hypoxia produced no significant change in PFK1 activity when Ser529 was mutated to alanine (Fig. 2A). Similar effects on PFK1 activity were observed when O-GlcNAcylation was increased by means of other cellular treatments (fig. S13). Furthermore, glycosylation inhibited PFK1 activity across a wide ATP concentration range in the presence and absence of F-2,6-BP (fig. S14). Consistent with the importance of Ser529 in recognition of the allosteric activator, the activity of S529A PFK1 was impaired at lower F-2,6-BP concentrations (fig. S15). Therefore, we examined the effects of glycosylation in the presence of endogenous F-2,6-BP concentrations in 293T cells or 8.5 μM F-2,6-BP, which is within the physiological range for cancer cells (24). In both cases, the activity of wild-type (WT) PFK1 was significantly lower than that of S529A PFK1 when glycosylation was induced by hypoxia (fig. S16). Glycosylation thus exerts a strong inhibitory effect on PFK1 activity, and the mutation of Ser529 to Ala rescues the inhibitory effect. These results indicate that O-GlcNAcylation of PFK1 at Ser529 provides a mechanism to overcome the allosteric regulation of PFK1 by ATP and F-2,6-BP.

Fig. 2

Inhibition of PFK1 activity and oligomerization by glycosylation. (A) Relative activities of WT and S529A PFK1 (L isoform) purified from transfected 293T cells under normoxic or hypoxic conditions. Activities were measured in the presence of 100 μM F-2,6-BP and 3 mM ATP and were normalized with respect to the activity of WT PFK1 under normoxic conditions (n = 3 assays). (B) Oligomerization state of Flag-tagged PFK1 (L isoform) purified from 293T cells under normoxic or hypoxic conditions and from 293T cells overexpressing OGT or treated with PUGNAc under normoxic conditions. Complexes were resolved by nonreducing SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by Coomassie blue staining. (C) Coimmunoprecipitation (IP) of endogenous PFK1 with Flag-tagged WT or S529A PFK1 (L isoform) after OGT overexpression. Complexes were immunoprecipitated with a Flag antibody conjugated to agarose beads and analyzed by reducing SDS-PAGE, followed by immunoblotting with a PFK1 antibody. Error bars denote SEM. Statistical analysis was performed by Student’s t test (*P < 0.05).

F-2,6-BP slows the dissociation of complexes of PFK1 and promotes the association of PFK1 into tetramers and higher oligomers with enhanced catalytic activity (16). We expressed Flag-tagged PFK1 in 293T cells under normoxic or hypoxic conditions. After hypoxia treatment, a fraction of PFK1 exhibited faster mobility during native gel electrophoresis (Fig. 2B). A similar shift in mobility was observed when PFK1 glycosylation levels were increased by OGT overexpression or OGA inhibition and when PFK1 was heat-denatured (fig. S17), suggesting that this complex represents a lower oligomeric state of PFK1. We also examined the association of Flag-tagged PFK1 with endogenous PFK1 by coimmunoprecipitation. Overexpression of OGT impaired the coimmunoprecipitation of PFK1 subunits, and this effect was blocked by alanine mutation of Ser529 (Fig. 2C). Thus, O-GlcNAcylation not only inhibits the activity of PFK1 but also appears to perturb the equilibrium between different oligomeric forms.

To test the effects of PFK1 glycosylation on cellular metabolism, we depleted endogenous PFK1 and stably expressed Flag-tagged WT or S529A PFK1 in H1299 cells (henceforth referred to as WT PFK1 or S529A PFK1 rescue cells; Fig. 3A). Upon OGT overexpression, cells expressing WT PFK1 exhibited reduced glycolysis and lactate production relative to control cells (Fig. 3B). No change in glycolytic rate or lactate production was observed in cells expressing S529A PFK1 upon OGT overexpression.

Fig. 3

PFK1 glycosylation at Ser529 regulates glycolysis, increases PPP flux, and protects cells from ROS-mediated cell death. (A) Immunoblotting of H1299 cells stably expressing shRNA and rescue constructs. Flag-PFK1 levels were comparable to endogenous PFK1 levels. (B) Glycolytic rate and lactate production of H1299 cells rescued with WT or S529A PFK1 in the absence (Cont) or presence of OGT overexpression (n = 4 experiments). (C) PPP activity in WT or S529A PFK1 rescue cells in the absence (Cont) or presence of OGT overexpression, as measured by the rate of 14CO2 production from glucose via the PPP (n = 3 assays). (D) Percentage of central carbon flux from glucose to lactate flowing through the PPP in WT or S529A PFK1 rescue cells in the absence (Cont) or presence of OGT overexpression, as measured by reverse-phase triple-quadrupole liquid chromatography mass spectrometry. (E) NADPH and GSH levels in WT or S529A PFK1 rescue cells in the absence (Cont) or presence of OGT overexpression. NADPH and GSH concentrations are shown relative to WT PFK1 Cont and were assessed using a colorimetric assay and the thiol probe monochlorobimane, respectively (n = 4 experiments). (F) NADPH and GSH levels in WT or S529A PFK1 rescue cells under hypoxic conditions (n = 3 experiments). (G) Cellular ROS levels induced by varying concentrations of diamide in untreated H1299 cells (Cont) and H1299 cells overexpressing OGT, as measured by a fluorimetric assay. (H) Percentage of cell death induced by varying concentrations of hydrogen peroxide in untreated (Cont) and OGT-overexpressing H1299 cells, as measured by lactate dehydrogenase levels (n = 4 assays). Error bars denote SEM. Statistical analysis was performed by Student’s t test (*P < 0.05) for all figures.

Suppressing glycolysis can redirect metabolic flux down the oxidative pentose phosphate pathway (PPP) (25, 26), providing cells with pentose sugars for nucleotide and nucleic acid biosynthesis, as well as NADPH to combat oxidative stress (2527). We observed both increased total and proportional flux through the oxidative PPP pathway, as measured by the amount of released 14CO2 from [1-14C]-glucose, and by relative accumulation of singly versus doubly [13C]-labeled lactate from a [1,2-13C]-glucose feed, when O-GlcNAcylation was enhanced in WT PFK1 rescue cells (Fig. 3, C and D). In contrast, PPP flux remained unaffected in S529A PFK1 rescue cells; however, it was increased as compared to that of untreated WT PFK1 rescue cells, possibly because of inhibitory effects of the S529A mutation on PFK1 activity (fig. S15).

PPP flux generates NADPH, which maintains a pool of reduced glutathione (GSH) and combats reactive oxygen species (ROS)–mediated cell death (24, 28). Consistent with increased PPP flux, enhancing abundance of O-GlcNAc by OGT overexpression in WT PFK1 rescue cells led to 1.6-fold and 4-fold increases in amounts of NADPH and GSH, respectively (Fig. 3E). Blocking glycosylation of PFK1 at Ser529 prevented the increase in NADPH and partially prevented the increase in GSH. Amounts of NADPH and GSH were also increased under hypoxic conditions in WT PFK1 rescue cells as compared to those in S529A PFK1 rescue cells (Fig. 3F). Furthermore, untargeted metabolite profiling by high-resolution flow-injection mass spectrometry (29) revealed enhanced steady-state concentrations of GSH, amino acids, and nucleotide precursors in WT PFK1 rescue cells relative to those in S529A PFK1 rescue cells (table S1). We measured the sensitivity of H1299 cells to ROS-mediated cell death upon overexpression of OGT. Enhancing O-GlcNAcylation prevented the increase in ROS levels induced by diamide (Fig. 3G) and protected the cells from hydrogen peroxide–mediated cell death (Fig. 3H). Thus, increases in PPP flux induced by PFK1 glycosylation might help promote cancer cell survival.

Cells expressing the S529A mutant proliferated more slowly than cells expressing WT PFK1 under hypoxic conditions, which is consistent with reduced flux through the PPP (Fig. 4A). The proliferation rate of WT PFK1-expressing cells was enhanced further upon OGT overexpression, whereas that of S529A PFK1-expressing cells was unchanged. OGT overexpression increased, whereas OGT depletion decreased, cell proliferation under hypoxic conditions (Fig. 4B). Depletion of PFK1 abolished these effects, indicating that O-GlcNAcylation stimulates cell proliferation through a PFK1-dependent mechanism (although decreasing PFK1 expression may blunt the differences in cell growth).

Fig. 4

PFK1 glycosylation contributes to cell proliferation and tumor growth. (A) Cell proliferation rates under hypoxic conditions of WT and S529A PFK1 H1299 rescue cells with and without OGT overexpression, as measured by the amount of cellular ATP (n = 3 experiments). (B) Cell proliferation rates under hypoxic conditions of H1299 cells infected with lentiviruses containing scrambled or PFK1 shRNA constructs after no treatment (Cont), OGT knockdown, or OGT overexpression (n = 3 experiments). (C) Tumor formation in nude mice injected with WT or S529A PFK1 H1299 rescue cells with and without OGT overexpression. (Left, top) Dissected tumors after 7 weeks of growth in mice injected with WT cells on the left flank and S529A cells on the right flank. (Left, bottom) PFK1 glycosylation in tumor lysates originating from WT or S529A H1299 rescue cells after labeling with a 5-kD mass tag and immunoblotting with Flag antibody. (Right) Masses of the dissected tumors. Each x represents the mass from one mouse; the horizontal red line indicates the mean tumor mass. Error bars denote SEM. Statistical analysis was performed by one-way ANOVA and Bonferroni comparison post-test in (B) and (C) and by Student’s t test in (A) (*P < 0.05, **P < 0.01, ***P < 0.001; N.S., not significant).

We injected immunocompromised mice (nu/nu) with WT PFK1 or S529A PFK1 rescue cells in the presence or absence of OGT overexpression (fig. S18) and assayed for tumor formation. Mice injected with S529A PFK1 rescue cells showed decreased tumor mass as compared to mice injected with WT PFK1 rescue cells (Fig. 4C). Moreover, overexpression of OGT in WT PFK1 rescue cells enhanced tumor growth but had no significant effect on S529A PFK1 rescue cells. Protein immunoblot analysis confirmed that the Flag-tagged WT or S529A PFK1 proteins were retained in the tumors and that WT PFK1 was O-GlcNAcylated (Fig. 4C). Under these conditions, glycosylation of PFK1 at Ser529 provides a critical growth advantage to tumor cells in vivo.

We demonstrated that O-GlcNAc glycosylation directly regulates glycolysis and reroutes metabolic flux through pathways critical for cancer cell proliferation and survival. O-GlcNAc can simultaneously sense and redirect flow through essential metabolic pathways, specifically through the modulation of PFK1 activity, adding a previously unrecognized mode of regulation to this glycolytic enzyme. Furthermore, because UDP-GlcNAc represents a key point of pathway integration and enables the cell to monitor the balance between glucose and glutamine uptake (11, 12), O-GlcNAcylation of PFK1 may provide an important mechanism to link the availability of both carbon and nitrogen sources for the cell to the production of metabolites necessary for sustaining rapid tumor growth.

Hypoxia also activates PFK1 glycosylation, thereby redirecting a larger fraction of the glucose flux through the PPP and increasing biosynthetic precursors, as well as reducing power from increased NADPH and GSH, to impart a growth advantage to cancer cells. Our results suggest that dynamic physiological inhibition of PFK1 may be a major regulatory point for central carbon flow. Blocking PFK1 glycosylation to enhance its activity and reset cellular metabolism toward normal cell growth could therefore provide a new strategy to combat cancer.

Supplementary Materials

Materials and Methods

Figs. S1 to S18

Table S1

References (3044)

PDB Coordinates of Lowest-Energy Glycosylated PFK1 Structure

References and Notes

  1. Acknowledgments: We thank P. Qasba for the Y289L GalT construct, R. Abrol for computational modeling advice, R. Diamond for cell cycle analysis, and L. Cantley and M. Vander Heiden for useful discussions. This work was supported by the National Institutes of Health (grant R01 GM084724 to L.C.H-W), the Department of Defense Breast Cancer Research Program (grant W81XWH-10-1-0988 to W.Y), and a Tobacco-Related Disease Research Program Postdoctoral Fellowship (19FT-0078 to W.Y). We also thank Agios Pharmaceuticals for financial support in providing patient tumor and matched tissue samples.
View Abstract

Navigate This Article