Unlike the growth factor dependence of normal cells, cancer cells can maintain growth factor–independent glycolysis and survival through expression of oncogenic kinases, such as BCR-Abl. Although targeted kinase inhibition can promote cancer cell death, therapeutic resistance develops frequently, and further mechanistic understanding is needed. Cell metabolism may be central to this cell death pathway, as we have shown that growth factor deprivation leads to decreased glycolysis that promotes apoptosis via p53 activation and induction of the proapoptotic protein Puma. Here, we extend these findings to show that elevated glucose metabolism, characteristic of cancer cells, can suppress protein kinase Cδ (PKCδ)–dependent p53 activation to maintain cell survival after growth factor withdrawal. In contrast, DNA damage–induced p53 activation was PKCδ independent and was not metabolically sensitive. Both stresses required p53 Ser18 phosphorylation for maximal activity but led to unique patterns of p53 target gene expression, showing distinct activation and response pathways for p53 that were differentially regulated by metabolism. Consistent with oncogenic kinases acting to replace growth factors, treatment of BCR-Abl–expressing cells with the kinase inhibitor imatinib led to reduced metabolism and p53- and Puma-dependent cell death. Accordingly, maintenance of glucose uptake inhibited p53 activation and promoted imatinib resistance. Furthermore, inhibition of glycolysis enhanced imatinib sensitivity in BCR-Abl–expressing cells with wild-type p53 but had little effect on p53-null cells. These data show that distinct pathways regulate p53 after DNA damage and metabolic stress and that inhibiting glucose metabolism may enhance the efficacy of and overcome resistance to targeted molecular cancer therapies. Cancer Res; 70(20); 8066–76. ©2010 AACR.
Developing hematopoietic cells normally require input from growth factor signaling pathways to support basal glucose metabolism for cell survival and proliferation (1, 2). In contrast, cancer cells often become independent of cell extrinsic growth factors and gain autonomous control over metabolism and survival (3). In particular, cancer cells adopt the metabolic program of aerobic glycolysis (4) that is characterized by increased glucose uptake, glycolytic flux, and lactate production and is reminiscent of growth factor–stimulated cells (5). It is now clear that aerobic glycolysis can be directly initiated by growth factor signals and oncogenes known to cause hematologic malignancies, including Notch (6), Akt (7, 8), and BCR-Abl (9, 10). The extent to which this metabolic phenotype affects cell survival or oncogenesis, however, remains unclear.
Cancer cells can become growth factor independent through the activation of pathways that mimic growth factor signaling, and inhibition of these pathways has proven an effective way to eliminate cancer cells. The BCR-Abl fusion protein, for example, can maintain glucose uptake (9) and cell survival (11). The tyrosine kinase inhibitor imatinib, which is widely used to treat BCR-Abl–positive leukemias, blocks the survival signal in these cells, causing decreased glycolysis and cell death (10). Tyrosine kinase inhibitors are also used to treat several solid cancers, including breast, colorectal, and lung cancer, but development of resistance to these small-molecule inhibitors represents an obstacle to long-term remission (12–14). Further insight into how loss of growth signals leads to cell death may provide direction to improve these important clinical tools.
It is now clear that decreased metabolism may initiate cell death upon inhibition of growth signals. The AMP-activated protein kinase (AMPK) and the lipid-sensitive protein kinase C (PKC) family of proteins are each sensitive to metabolic cues and may affect apoptosis (15, 16). Additionally, cellular metabolism can directly regulate the Bcl-2 family proteins, as loss of glucose uptake on growth factor withdrawal (17) leads to degradation of the prosurvival Bcl-2 protein Mcl-1 (16) and induction of the proapoptotic BH3-only protein Puma (17). Inhibition of glucose metabolism can lead to apoptosis, however, only when proapoptotic Bcl-2 family proteins Bax (5, 18) and Bim or Puma are present (17), indicating that metabolic pathways that influence cell death must converge on Bcl-2 family proteins.
We recently showed that aerobic glycolysis can prevent p53 activation and Puma induction in growth factor withdrawal (17). We sought here to determine how p53 is metabolically regulated and how this pathway may contribute to imatinib-induced death of BCR-Abl–expressing cells. Although p53 was required for Puma induction and cell death in response to growth factor withdrawal and DNA damage, elevated glucose metabolism attenuated p53 activation and Puma induction only after cytokine withdrawal. Importantly, imatinib decreased glucose metabolism, but maintenance of aerobic glycolysis attenuated p53 activation and cell death, whereas inhibition of glycolysis enhanced imatinib sensitivity via p53. Thus, glucose metabolism can itself suppress a specific pathway of p53 activation and may contribute to oncogenesis and sensitivity to targeted therapies.
Materials and Methods
Control, Glut1/HK1, and Bcl-xL–expressing FL5.12 cells were cultured as described in RPMI 1640 (Mediatech) with 10% FetalClone III serum (Thermo Scientific) and 0.5 ng/mL recombinant murine interleukin (IL)-3 (eBioscience; ref. 17). K562 cells were cultured in RPMI 1640 with 10% fetal bovine serum (FBS; Gemini Bio-Products), 2 mmol/L glutamine, 10 mmol/L HEPES, and 55 nmol/L 2-mercaptoethanol. Nalm-1 cells (American Type Culture Collection) were cultured in RPMI 1640 with 15% FBS. Murine T cells were isolated via negative selection (StemSep) and cultured in RPMI 1640 supplemented with 10% FBS with 10 ng/mL recombinant murine IL-2 (eBioscience). T cells were activated on 5 μg/mL anti-CD3ε and anti-CD28 antibody-coated plates (BD Pharmingen) for 48 hours, followed by 24 hours in 10 ng/mL IL-2. For growth factor withdrawal, cells were washed three times with PBS and cultured in the presence or absence of IL-3 or IL-2. Some cells were resuspended in medium containing 0.4 μmol/L etoposide (Sigma), 10 μmol/L compound C (EMD Biosciences), 10 μmol/L rottlerin (Calbiochem), 1 μmol/L imatinib (LC Laboratories), 5 mmol/L 2-deoxyglucose (2-DOG; Sigma), or an equal volume of vehicle control.
Samples were prepared as previously described (17). Primary antibodies were Puma (Cell Signaling), p53 (1C12; Cell Signaling), actin (Sigma), Bim (BD Pharmingen), phospho-p53 Ser15 (Cell Signaling), PKCδ (BD Pharmingen), hemagglutinin (HA; Roche), lamin B2 (Invitrogen), glyceraldehyde-3-phosphate dehydrogenase (6C5; Santa Cruz Biotechnology), and COOH-terminal acetylated p53 (generously provided by Dr. Tso-Pang Yao, Duke University, Durham, NC). Secondary antibodies were horseradish peroxidase–labeled anti-rabbit (Cell Signaling), horseradish peroxidase–labeled anti-rat (BD Pharmingen), and fluorescent-labeled anti-mouse (LI-COR) antibodies. Blots were imaged using SuperSignal West Pico or Femto Chemiluminescent Substrate (Thermo Scientific) or the Odyssey IR imaging system (LI-COR). Images were uniformly contrasted, and some were separated digitally for ease of viewing (indicated by white spaces).
Transfection and plasmids
Transient transfections were performed via nucleofection (Amaxa Kit V, Lonza), and short hairpin RNA (shRNA) plasmids were constructed as previously described (17). p53, Puma, and Bim were knocked down using previously reported shRNA sequences (17). The shRNA sequence for PKCδ was GAAGATGAAGGAGGCACTCAGCCTCGAGCCTGAGTGCCTCCTTCATCTTC. HA-tagged, wild-type human pcDNA3.1-p53 was expressed for acetylation studies.
Cell viability analysis
Cell viability was assessed via propidium iodide exclusion on a FACScan flow cytometer (BD Biosciences) using FlowJo software (TreeStar), as described (17). Means ± SDs are shown. Data are representative of three or more independent experiments.
Rates of glycolysis were measured as previously described (5), and means plus SDs are shown.
Cells were cultured in the presence or absence of IL-3 or etoposide for 8 hours or imatinib for 10 hours. Luciferase activity was measured using a Dual Luciferase Assay kit (Promega), following the manufacturer's instructions as previously described (17). Values are means plus SDs of triplicate samples. Data are representative of three independent experiments.
Quantitative real-time PCR analysis
A p53 signaling real-time PCR (RT-PCR) SuperArray (SA Biosciences) was performed according to the manufacturer's instructions. Quantitative RT-PCR (qRT-PCR) was performed as previously described for Puma and β2-microglobulin (17). Additional primer sequences were as follows: Gadd45, ATTTCACCCTCATCCGTGCGTTCT and ATGAATGTGGGTTCGTCACCAGCA; Mdm2, GTTGTTAAAGTCCGTTGGAGCGCA and CAATGTGCTGCTGCTTCTCGTCAT; Bax, CCGGCGAATTGGAGATGAACT and CCAGCCCATGATGGTTCTGAT; p21, AATCCTGGTGATGTCCGACCTGTT and GTGACGAAGTCAAAGTTCCACCGT; Zmat3, ATGACTACTGTAAGCTGTGCGA and GGCCACCATCTTAAATTCACTCC. Values are means plus SDs of three independent experiments.
Cell fractionations were performed as previously described (19).
p53 is required for efficient cell death after both cytokine withdrawal and DNA damage in activated T lymphocytes
Inhibition of growth signals leads to decreased glucose metabolism and apoptosis via p53 activation and induction of Puma and Bim (2, 17, 20). It remained unclear how decreased glycolysis stimulated p53 and if this regulatory pathway was common to other p53-responsive stresses. We initially examined activated primary T lymphocytes, which undergo a dramatic increase in glycolytic metabolism on activation that is characteristic of aerobic glycolysis and requires IL-2 (Fig. 1A). Purified p53+/+ or p53−/− T cells were activated and cultured in IL-2, and cell death was observed after IL-2 withdrawal or treatment with the DNA-damaging chemotherapeutic agent etoposide (Fig. 1B). In both cases, p53−/− T cells exhibited a persistent survival advantage over control cells, showing a contribution by p53 to these cell death pathways.
p53 likely regulated cell death by control of proapoptotic proteins, and activated wild-type T cells upregulated Puma expression on both IL-2 withdrawal and etoposide treatment (Fig. 1C). In contrast, p53-deficient T cells failed to upregulate Puma in response to either IL-2 withdrawal or etoposide treatment. Bim has also been implicated in T-cell apoptosis (20), and p53 may have regulated cell death through modification of Bim. To determine if p53 might also regulate cell death in the absence of Bim, activated Bim−/− and Bim−/−p53−/− T cells were analyzed (Fig. 1D). Importantly, p53 deficiency provided protection from growth factor withdrawal and DNA damage, regardless of Bim expression, and Bim−/−p53−/− T cells were highly resistant to apoptosis on IL-2 withdrawal. Thus, p53 is critical to the induction of apoptosis following growth factor deprivation or DNA damage, independent of Bim.
Increased glucose metabolism attenuates p53-dependent Puma induction and cell death after growth factor withdrawal but not after DNA damage
The IL-3–dependent and nontransformed hematopoietic precursor cell line FL5.12 also undergoes apoptosis following growth factor deprivation (17) or DNA damage (21). Importantly, these cells retain wild-type p53 (17), yet can be readily transformed to cause leukemias in vivo (22), and thus offer a model to examine p53 regulation. To address the effect of glucose metabolism on p53 directly, we established cells that stably coexpress the glucose transporter Glut1 and hexokinase 1 (HK1; ref. 18). Elevated HK1 activity and exogenous Glut1, which localized largely to the cell surface even after IL-3 withdrawal (18), drove glucose uptake and metabolism even in the absence of growth factor (Fig. 2A). DNA damage, however, did not reduce glucose metabolism, and glycolytic rates were modestly elevated in control and Glut1/HK1-expressing cells. Importantly, maintenance of glucose metabolism delayed cell death on growth factor withdrawal (Fig. 2B; ref. 16) but did not provide protection against DNA damage by etoposide (Fig. 2B).
Consistent with selective protection from apoptosis, Puma induction in Glut1/HK1 cells was largely suppressed on growth factor withdrawal, but not following DNA damage (Fig. 2C). This was mediated by p53, as shRNA knockdown of p53 suppressed Puma induction (Fig. 2D) and attenuated cell death after both IL-3 deprivation and etoposide treatment (Supplementary Fig. S1). These data support a role for p53 to promote apoptosis after multiple cell stresses and show selective metabolic regulation of p53 and Puma induction after cytokine withdrawal.
Glucose metabolism selectively regulates p53 transcriptional activity after cytokine withdrawal but not after DNA damage
The differential patterns of Puma induction shown above suggested distinct modes of p53 transcriptional regulation. Cells were therefore transfected with a p53 luciferase reporter construct to assess p53-dependent transcription. After cytokine withdrawal, p53 activity increased in control cells, but this was prevented by maintenance of glycolysis in Glut1/HK1-expressing cells (Fig. 3A). In contrast, p53 transcriptional activity increased equivalently in both control and Glut1/HK1 cells after DNA damage (Fig. 3B). p53 can upregulate different target genes in response to various stresses (23), and differential regulation might activate p53 to induce distinct sets of target genes. To examine this possibility, control cells were cultured in the presence or absence of IL-3 or the presence of etoposide, and mRNA levels for a panel of p53 target genes were analyzed using a qRT-PCR array focused on p53 signaling pathways. Candidate genes were subsequently validated using qRT-PCR (Fig. 3C). Treatment with etoposide led to increased levels of the p53 targets Puma, Mdm2, Bax, p21, and Zmat3, whereas IL-3 deprivation led to upregulation of Puma and Gadd45. Similarly, activated T cells deprived of IL-2 also had increased Puma and Gadd45 mRNA levels, but unchanged or decreased levels of Mdm2 mRNA (Fig. 3D). Thus, p53 activation pathways are differentially sensitive to metabolic status and can induce distinct sets of target genes.
Glucose metabolism regulates posttranslational modification of p53 after cytokine withdrawal but not after DNA damage
To allow differential responses to growth factor withdrawal or DNA damage, p53 may be subjected to distinct patterns of modification. In the DNA damage response pathway, phosphorylation dissociates p53 from the ubiquitin ligase Mdm2 to allow p53 stabilization. Despite increased p53 transcriptional activity, however, no increases in total p53 protein were observed in either FL5.12 cells or primary T cells after cytokine withdrawal (Supplementary Fig. S2). As expected, p53 was stabilized in both cell types on etoposide treatment (Supplementary Fig. S2). We next examined whether cytokine withdrawal also induced p53 phosphorylation. Indeed, growth factor withdrawal led to phosphorylation of p53 on Ser18 (mSer18, corresponding to human Ser15), albeit to a lower level than that induced by DNA damage (Fig. 4A). Similarly, p53 was phosphorylated on mSer18 after both cytokine withdrawal and DNA damage in activated primary T cells (Fig. 4A). Phosphorylation was seen specifically at this site, but not at mSer23 (hSer20) or mSer389 (hSer392), after cytokine withdrawal (Supplementary Fig. S3). This was distinct from the pattern of p53 modification induced by DNA damage, where phosphorylation at these additional sites was readily detected.
To directly address the role of p53 S18 phosphorylation, T lymphocytes were examined from p53 S18A knock-in mice, in which S18 is mutated to alanine (24). Activated p53 S18A T cells were partially resistant to cell death after cytokine withdrawal or DNA damage induced by etoposide or γ-irradiation (Fig. 4B). Importantly, Puma induction was also attenuated in activated S18A T cells after both cytokine withdrawal and DNA damage (Fig. 4B). These data show that phosphorylation of p53 at Ser18 is required for maximal activation of p53 and Puma induction.
Cells that maintained glucose uptake and metabolism may have suppressed p53 transcriptional activity through inhibition of p53 modification. To examine this possibility, control and Glut1/HK1 cells were cultured in the presence or absence of IL-3 or the presence of etoposide, and phosphorylation at mSer18 of p53 was examined. Whereas control cells showed p53 phosphorylation after IL-3 withdrawal, Glut1/HK1 expression strongly suppressed this phosphorylation (Fig. 4C). In contrast, both control and Glut1/HK1 cells showed strong phosphorylation on mSer18 after etoposide treatment.
The intermediate phenotype of p53 S18A T cells suggested that additional metabolically regulated mechanisms also contribute to p53 activation. COOH-terminal acetylation can also enhance p53 transcriptional activity (25). To examine p53 acetylation, cells were transfected with HA-tagged human p53 and cultured in the presence or absence of IL-3. HA-p53 was then immunoprecipitated and probed with antisera recognizing pan–COOH-terminal acetylated human p53 (26). Control cells showed an increase in p53 COOH-terminal acetylation after IL-3 deprivation, consistent with p53 activation. However, Glut1/HK1-expressing cells showed no increase in p53 acetylation (Fig. 4D). Although the hierarchy of these modifications in regulating p53 remains unclear, these results show that p53 is phosphorylated and acetylated after cytokine withdrawal, and these modifications are prevented by aerobic glycolysis.
PKCδ is required for p53 phosphorylation after cytokine withdrawal
Selective regulation of p53 modification suggested that nutrient-sensitive kinases mediate p53 activation when cells are deprived of growth signals through a pathway distinct from the DNA damage response of ATM and ATR. Of likely candidates, AMPK has been characterized as a glucose-sensitive kinase and can phosphorylate p53 at mSer18 in the context of glucose limitation (27). However, we failed to detect an increase in activating phosphorylation of AMPK after cytokine withdrawal (Fig. 5A). Furthermore, pharmacologic inhibition of AMPK with compound C had no effect on p53 phosphorylation or Puma induction upon growth factor withdrawal (Fig. 5A). These findings suggested that an AMPK-independent pathway may sense altered metabolism in growth factor deprivation.
Members of the PKC family can also respond to changes in glucose metabolism (15, 16), and recent work has implicated PKCδ in cytokine withdrawal–induced cell death (28). Indeed, cells transiently transfected with PKCδ shRNA showed markedly reduced p53 phosphorylation and Puma induction (Fig. 5B) and failed to increase p53 transcriptional activity (Fig. 5C) on growth factor withdrawal. In contrast, PKCδ was dispensable for p53 phosphorylation and Puma induction after DNA damage (Fig. 5B). PKCδ may have promoted cell death indirectly by suppressing glucose metabolism on cytokine withdrawal, which Glut1/HK1 cells could overcome. However, knockdown of PKCδ did not significantly affect glycolytic rate (Supplementary Fig. S4). Importantly, PKCδ contributed to cell death, as PKCδ shRNA protected FL5.12 cells against IL-3 withdrawal, and the PKCδ inhibitor rottlerin attenuated the death of IL-2–dependent CTLL-2 T cells after cytokine deprivation (Fig. 5D).
PKCδ may have phosphorylated p53 directly (29, 30) or through indirect regulation of metabolic stress pathways. However, we were unable to detect direct phosphorylation in vitro, and fractionation experiments showed that PKCδ and p53 resided primarily in different cellular compartments (Supplementary Fig. S5). Furthermore, PKCδ deficiency provided further protection beyond that observed with knockdown of p53 alone or in combination with Bim knockdown (Fig. 5E; Supplementary Fig. S6). This suggested that PKCδ contributed to cell death after cytokine withdrawal through additional p53-independent mechanisms. Thus, PKCδ is essential, possibly as a nutrient sensor and signal initiator, to promote cell death after growth factor withdrawal through both p53-dependent and p53-independent pathways.
Increased glucose metabolism confers resistance to imatinib
Oncogenic kinases can render cells growth factor independent (11), and expression of the p190 isoform of BCR-Abl in FL5.12 cells allowed cells to survive and proliferate in the absence of IL-3 (Supplementary Fig. S7). Like growth factor withdrawal, the BCR-Abl kinase inhibitor imatinib decreased glycolysis (Fig. 6A) and induced apoptosis in control cells expressing p190 (Fig. 6B). Similar to growth factor withdrawal, shRNA-mediated knockdown of p53 or Puma (Supplementary Fig. S8) dramatically reduced cell death after imatinib treatment (Fig. 6B). In addition, whereas control p190 cells induced Puma in response to imatinib, Glut1/HK1 p190 cells suppressed Puma upregulation and showed no increase in p53 transcriptional activity after imatinib treatment (Fig. 6C).
These data suggested that imatinib used a metabolically sensitive apoptotic mechanism similar to withdrawal from extrinsic growth factors. Consistent with this notion, Glut1/HK1 p190 cells were protected from imatinib-induced cell death (Fig. 6D). To examine whether this resistance was dependent on increased glycolysis, control and Glut1/HK1 cells were treated with imatinib together with the glycolysis inhibitor 2-DOG. Whereas 2-DOG treatment at this concentration alone had little effect on cell viability in either cell type, combination treatment with imatinib and 2-DOG increased cell death in both cell types and eliminated the resistance to imatinib seen in Glut1/HK1 p190 cells (Fig. 6D). Importantly, inhibition of glycolysis enhanced sensitivity to kinase inhibition even in control p190 cells. Thus, imatinib leads to p53-dependent apoptosis that can be attenuated by glucose metabolism.
Loss of p53 is clinically associated with imatinib resistance (31, 32), and our data showed a key role for metabolism in both p53 regulation and imatinib sensitivity. Therefore, we examined the response to 2-DOG and imatinb in p53 wild-type (Nalm-1; ref. 33) and p53-null (K562; ref. 34) human chronic myelogenous leukemia (CML) cell lines (Fig. 6E; Supplementary Fig. S9). Consistent with findings from BCR-Abl–expressing murine cells, Nalm-1 cells underwent apoptosis in response to imatinib that was enhanced by addition of a sublethal dose of 2-DOG. In contrast, K562 cells were more resistant to imatinib, and addition of 2-DOG did not augment cell death. These findings show that manipulation of metabolism can enhance the response to targeted therapy in human cancer cells, and this seems to depend on p53.
Oncogenic kinases can provide signals to allow cancer cells to persist independent of growth factors. Recently developed targeted therapies can block these signaling pathways to suppress glucose uptake (9, 10) and elicit apoptosis (9, 35). We previously showed that activation of p53 and induction of Puma to promote apoptosis after growth factor withdrawal could be prevented by maintenance of aerobic glycolysis. However, it remained unclear how p53 was activated and whether p53 inhibition by glucose metabolism was specific to cytokine deprivation or was applicable to other apoptotic stresses. Here, we show that growth factor deprivation of normal cells or treatment of BCR-Abl–expressing cells with imatinib led to a metabolic stress pathway that activates p53 and elicits Puma-dependent apoptosis. This pathway was distinct from the DNA damage pathway regarding the mechanism of p53 activation and target gene regulation. Although multiple proapoptotic proteins, such as the BH3-only proteins Bim and Bad, can contribute to cell death after imatinib treatment (36), our data indicate a clear role for Puma in cell death that is uniquely sensitive to cell metabolism. Together, these data show that metabolic control of p53-mediated Puma induction is a critical determinant of imatinib sensitivity.
These distinct types of cell stress seem to use specific mechanisms of p53 activation, with differential stabilization of p53 and reliance on unique upstream kinases. Unlike the DNA damage response pathway, in which p53 stabilization was apparent and considered important for p53 activity (37), the phosphorylation and activation of p53 in the context of cytokine deprivation occurred in the absence of any observable changes in p53 protein levels. In addition, distinct from the well-described ATM/ATR-mediated DNA damage pathway or AMPK-dependent glucose deprivation pathway, the loss of metabolism caused by suppression of growth signals required PKCδ to elicit cell death through both p53-dependent and p53-independent mechanisms.
Although PKCδ-dependent phosphorylation of p53 did not seem to be direct, PKCδ may serve as a metabolic sensor to initiate a pathway that culminates in p53 activation. Indeed, glucose metabolism can regulate several PKC isoforms in response to altered lipid metabolism caused by hyperglycemia (38, 39). Cells with intrinsically high rates of glucose uptake, such as growth factor–stimulated or cancerous cells, may use similar mechanisms to regulate PKCs. Consistent with this notion, we have found that PKCμ and PKCε localization are altered in Glut1/HK1-expressing cells (data not shown; ref. 16). It remains unclear, however, how PKCδ is regulated by glucose metabolism following cytokine withdrawal and why this is essential for p53 activation.
Transcriptionally, p53 can regulate the expression of numerous target genes. However, the mechanisms governing which p53 target genes are induced in response to particular stresses remain unclear. Here, we show that despite shared phosphorylation on mSer18, distinct pathways of p53 activation induced unique sets of target genes. This disparity could be explained by differences in the degree of phosphorylation or by the modification of other residues of p53 by phosphorylation or acetylation. These variations in modification likely affected recruitment of coactivators and/or corepressors of p53 activity, which could promote differential target gene selection.
Specific regulation of transcriptional cofactors may also contribute to the increase in Puma levels after cytokine withdrawal. Indeed, the multistress transcription factor CHOP can contribute to Puma upregulation following cytokine withdrawal (40) and could cooperate with p53 to induce Puma in this context. FoxO3a has also been implicated in Puma induction following cytokine withdrawal (41). Whereas FoxO3a may regulate Puma in certain contexts, the lack of Puma induction in p53−/− T lymphocytes and the reduced Puma induction in p53 S18A T lymphocytes suggest that p53 is essential for the majority of Puma induction in this cell type.
A key finding was that glycolytic metabolism provided protection against cell death after inhibition of growth signals but not DNA damage. These specific modes of p53 and apoptotic regulation by glucose metabolism may allow cells to adjust their response to stress based on nutrient availability. The high-glucose uptake of activated lymphocytes or BCR-Abl–transformed cells may suppress stress pathways during transient reductions in signaling. In contrast, initiating cell cycle arrest or apoptosis in response to damaged DNA is likely critical in nutrient-replete conditions, which may otherwise favor cell proliferation. These findings imply that inhibition of glucose metabolism may be central to the proapoptotic effect of kinase inhibitors and that further inhibition of glycolysis may augment these therapies to increase effectiveness and reduce resistance. Additionally, glycolysis inhibition failed to enhance imatinib sensitivity in p53-null K562 cells, suggesting that the outcome of metabolic manipulation may depend on the p53 status of a given tumor.
Imatinib is highly effective in inducing durable responses in patients with newly diagnosed CML in the chronic phase (42, 43). Resistance to imatinib can arise from BCR-Abl kinase mutations (44). However, a notable percentage of CML patients with no kinase domain mutations fail to respond to imatinib (13, 45, 46), and recent work showed elevated glucose metabolism in imatinib-resistant cells (47), supporting a connection between metabolism and drug resistance. Together, our findings suggest that targeted therapies may be particularly sensitive to cell metabolism, and metabolic manipulation may provide a novel means to enhance the efficacy of molecular targeted cancer treatments and overcome therapeutic resistance.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank the Rathmell lab, Dr. Tso-Pang Yao (Duke University, Durham, NC), and Dr. Kimryn Rathmell (University of North Carolina, Chapel Hill, NC) for constructive comments and suggestions. p53−/− mice were generously provided by Dr. David Kirsch (Duke University), and reagents were provided by Dr. Tso-Pang Yao.
Grant Support: NIH grants R01 CA123350 (J.C. Rathmell), F30 HL094044 (E.F. Mason), and RO1CA077735 (S.N. Jones); Leukemia and Lymphoma Society Scholar Award (J.C. Rathmell); and Alex's Lemonade Stand (J.C. Rathmell). J.C. Rathmell is the Bernard Osher Fellow of the American Asthma Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received February 18, 2010.
- Revision received July 7, 2010.
- Accepted July 23, 2010.
- ©2010 American Association for Cancer Research.