Targeting 6-Phosphofructo-2-Kinase (PFKFB3) as a Therapeutic Strategy against Cancer

Abstract

In human cancers, loss of PTEN, stabilization of HIF-1α and activation of Ras and AKT converge to increase the activity of a key regulator of glycolysis, 6-phosphofructo-2-kinase(PFKFB3). This enzyme synthesizes fructose 2,6-bisphosphate (F26BP) which is an activator of 6-phosphofructo-1-kinase, a key step of glycolysis. Previously, a weak competitive inhibitor of PFKFB3, 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO), was found to reduce the glucose metabolism and proliferation of cancer cells. We have synthesized 73 derivatives of 3PO and screened each compound for activity against recombinant PFKFB3. One small molecule, 1-(4-pyridinyl)-3-(2-quinolinyl)-2-propen-1-one (PFK15), was selected for further pre-clinical evaluation of its pharmacokinetic, anti-metabolic and anti-neoplastic properties in vitro and in vivo. We found that PFK15 causes a rapid induction of apoptosis in transformed cells, has adequate pharmacokinetic properties, suppresses the glucose uptake and growth of Lewis lung carcinomas in syngeneic mice and yields anti-tumor effects in three human xenograft models of cancer in athymic mice that are comparable to FDA-approved chemotherapeutic agents. As a result of this study, asynthetic derivative and formulation of PFK15 has undergone IND-enabling toxicology and safety studies. A phase I clinical trial of its efficacy in advanced cancer patients will initiate in 2013 and we anticipate that this new class of anti-metabolic agents will yield acceptable therapeutic indices and prove to be synergistic with agents that disrupt neoplastic signaling.

Introduction

Increased glucose uptake and metabolism are required for neoplastic cells to survive and proliferate within the micro-environment of a tumor. This nearly ubiquitous characteristic of human cancers forms the basis of a clinically useful diagnostic imaging modality that depends on the detection of pairs of gamma rays emitted indirectly by positron-emitting2-deoxy-2-[¹⁸F]fluoro-D-glucose consumed by tumor cells (i.e. ¹⁸FDG-PET imaging) (1, 2). The mechanisms that drive increased glucose uptake and metabolism in tumors include a combination of the survival response to hypoxia via the transcription factor, hypoxia inducible factor-1α (HIF-1α) (3), as well as genetic alterations such as amplification of c-MYC that can independently increase the expression and activities of multiple glucose transporters and glycolytic enzymes, including the Glut1 glucose transporter, hexokinase 2, 6-phosphofructo-1-kinase (PFK-1), pyruvate kinase M2 and lactate dehydrogenase A(4–6). Of these enzymes, PFK-1 is of particular interest since this irreversible and committed step of glycolysis serves as a metabolic sensor for the entire pathway via its allosteric inhibition by ATP, citrate and H+ ions(7, 8).

In the late 1970s, a novel allosteric activator of PFK-1, fructose 2,6-bisphosphate (F26BP), was discovered that could override the inhibition by ATP and increase glucose uptake and flux through the entire glycolytic pathway(7, 9).A family of five enzymes (PFKFB1–4, TIGAR) regulate the intracellular concentration of F26BP via a combination of fructose 6-phosphate kinase and F26BP phosphatase activities(8). Although the expression of most of these regulatory enzymes is increased by hypoxia(10), the PFKFB3 family member has been found to be a direct transcriptional target of HIF-α (11), to be stabilized by the loss of the tumor suppressor PTEN via suppressive effects on APC/Cdh1-mediated ubiquitination (12)and to be activated by oncogenic Ras and AKT (13, 14). Importantly, PFKFB3 protein expression is elevated in most tumor types(15) and heterozygous genomic deletion of the Pfkfb3 gene markedly reduces the concentration of F26BP, glucose uptake, glycolytic flux to lactate and anchorage-independent growth of LT/H-Ras^V12-transformed fibroblasts as soft agar colonies and as xenograft tumors in syngeneic mice (14, 16). Coupled to recent data that indicate that PFKFB3 is not expressed by primary neurons which, like neoplastic cells, have a very high rate of glucose uptake and metabolism (17), these studies indicate that small molecule inhibitors of PFKFB3 may have utility as anti-cancer agents.

A small molecule antagonist of PFKFB3, 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO), was recently identified using a combination of computational modeling and receptor-based in silico screening (18). 3PO causes a rapid reduction in F26BP, glucose uptake and lactate secretion followed by a reduction in the steady-state concentration of ATP and NADH, and an arrest in cell cycle progression in Jurkat T cell leukemia cells (18). Although this compound exhibits anti-tumor activity in mice (18), its pharmacokinetic properties and potency against the enzymatic activity of PFKFB3 are significantly below that required to justify testing in human subjects. In the current study, we synthesized multiple derivatives of 3PO in order to improve the pharmacokinetic properties and activity of 3PO and now report the identification of a novel PFKFB3 inhibitor termed PFK15 that has potent anti-tumor activity and that markedly reduces ¹⁸FDG uptake and the F26BP content of xenografted tumors. We also demonstrate for the first time that this novel class of anti-neoplastic agents has potent and rapid pro-apoptotic effects on transformed cells in vitro and in tumors in vivo.

Materials and Methods

Recombinant PFKFB3 Assay

Kinase reactions were performed by incubating 13ng of recombinant human PFKFB3 protein in a reaction mix containing 10μM ATP, 10μM F6P, and either DMSO vehicle control, 3PO, or PFK15 for one hour at room temperature. Kinase activity was measured with the Adapta Universal Kinase Assay (Invitrogen) per manufacturer’s instructions and fluorescent and FRET signals were measured on a Tecan Safire² plate reader. Graphs were generated using SigmaPlot software and IC₅₀ values were calculated using a three parameter Hill Equation. The KINOMEscanEDGE (San Diego, CA) list of 96 kinases can be accessed at http://www.kinomescan.com.

Cell lines and culture

Lewis Lung Carcinoma, Jurkat T cell leukemia cells, H522 human lung adenocarcinoma cells, CT26 colon adenocarcinoma cells, U-87 glioblastoma cells and BxPC-3 pancreatic cancer cells were maintained in RPMI or DMEM supplemented with 10% FCS (HyClone, Logan, UT), glucose (20mM) and glutamine (1 mM). The cell lines were obtained from the American Type Culture Collection (ATCC) and passaged for fewer than 6 months after receipt or resuscitation. ATCC conducted authentication of cell lines using a combination of sequencing, short tandem repeat profiling, and cytogenetic analysis. The morphology and behavior of each line were consistent with ATCC descriptions. Cells were plated at a density of 1 × 10⁵ cells/well in 6 well tissue culture plates and incubated in RPMI or DMEM for 24 hours (5% CO2 and 37°C). Cells then were exposed to 3PO (Chembridge) or PFK15 (Advanced Cancer Therapeutics, Louisville, KY) at the indicated concentrations and assayed for viability (48 hours), F26BP (3 hours), glucose uptake (3 hours) and intracellular ATP (3 hours). Experiments were conducted in triplicate or quadruplicate and statistical significance was assessed by the unpaired two-tail T-test.

Cell viability

Viability was determined using trypan blue exclusion. Cells were incubated in 20% trypan blue (Sigma) for 5 minutes. Cells excluding trypan blue were counted using a standard hemocytometer (Hausser Scientific) to determine the total number of viable cells. Experiments were conducted in triplicate.

F26BP measurements

Cells were triturated, washed twice with PBS, dissolved in 0.1 mM NaOH and F26BP content measured using a coupled enzyme reaction following the method of VanSchaftingen et al. (7). The F26BP concentration was normalized to total cellular protein measured by the bicinchoninic acid (BCA) assay (Pierce Biotechnology).

2-[1-¹⁴C]-deoxy-D-glucose uptake

Jurkat or H522 cells were plated in RPMI 1640 media supplemented with 10% FCS and 50μg/ml gentamycin at 100,000/ml and immediately treated with DMSO vehicle control, 3PO, or PFK15 for three hours. Cells were washed twice with pre-warmed glucose free RPMI media and incubated for 30 minutes in the presence of the PFKFB3 inhibitors. Cells were treated with 25μl of ¹⁴C-deoxy-glucose (0.1μCi/μl) for one hour, washed once with ice-cold glucose free RPMI and twice with ice cold PBS. Cells were lysed in 500μl 0.5% SDS. 400μl of lysate was added to 5ml Microscint 40 scintillation fluid (Perkin Elmer) and counts were measured on a Tri-Carb 2910 liquid scintillation analyzer (Perkin Elmer). Remaining lysate was quantitated with the BCA Assay (Pierce) per manufacturer’s instructions and protein levels measured on a Powerwave XS plate reader (Biotek). Counts were normalized to protein concentration.

ATP assay

Cells were washed (while still adherent) with cold PBS ×1, lysed with passive lysis buffer (1X, Molecular Probes, Invitrogen) added directly to the plates and immediately harvested by scraping. The lysates were flash frozen (to −80°C) and thawed (to 37°C) once to accomplish complete lysis and then centrifuged (at 4°C) for 30 seconds to clear the lysates. Intracellular ATP levels were determined using a bioluminescence assay (Molecular Probes) utilizing recombinant firefly luciferase and its substrate, D-luciferin. The luminescence was read in a TD-20/20 luminometer (Turner Designs, Sunnyvale, CA) at 560 nm. The ATP values were calculated using an ATP standard curve. The protein concentrations of the lysates were estimated using the BCA assay (Pierce Biotechnology, Rockford, IL) and ATP was expressed as nmol per mg protein.

Flow Cytometry

Jurkat cells were plated in RPMI 1640 media supplemented with 10% FCS and 50μg/ml gentamycin at 100,000 cells/ml and immediately incubated with DMSO vehicle control, 3PO, PFK15, or Etoposide (Sigma) for five hours. Cells were washed with PBS and stained with Annexin V and Propidium Iodide (BD Pharmingen). Fluorescence was measured using a FACSCalibur (BD Biosciences) and analyzed using FloJo (Tree Star).Annexin V+/PI+ (late apoptotic) and Annexin V+/PI− (early apoptotic) cells were quantified by the frequency of fluorescently labeled cells and statistical significance was assessed by the two-sample T-test (independent variable).

Pharmacokinetic Studies

The PK profile was determined in female C57Bl/6 mice following IV administration of the PFKFB3 inhibitors. Using only female mice lowered the number of animals required for meaningful results without the issues of potential gender differences in exposure. Eight time points and three animals per time point were used to determine the PK parameters calculated using WinNonLin v5.0 (Tmax, Cmax, volume of distribution, half-life, clearance, AUCs and MRT). Plasma samples were extracted using acetonitrile and analyzed by LC/MS-MS using a PhenomexSynergi Polar-RP 4micron 50×2.0 mm column eluted with a biphasic mobile phase (0.5% formic acid in acetonitrile and water).

FDG-PET Imaging

Tumor-bearing mice were anesthetized with isoflurane and 200 μCi of FDG was administered intravenously via the tail veins. The baseline micro-PET imaging of FDG in the mice was acquired by the ordered subsets expectation maximization (OSEM), using 6 subsets/4 iterations for 15-minute static scans 45 minutes after IV injection of the tracer using a Concorde Microsystems 4-ring micro-PET system. Regions of interest in the tumor and cerebellum were quantified in quadruplicate and, 24 hours later (¹⁸F half-life, 109.8 min), PFK15 (25 mg/kg) was administered i.p. and the micro-PET scan was repeated 45 minutes later.

Xenograft Studies

LLC, CT26, U-87 or BxPC-3 cells were collected from exponential growth phase culture in DMEM supplemented with 10% FCS. Cells were washed twice and re-suspended in PBS (1×10⁷ cells/ml). Groups of C57Bl6 (for LLC cells) or Balb/C athymic mice(for CT26, U-87 MG or BxPC-3 cells) (20 gm) were injected s.c. with 0.1 ml of the cell suspension (1 ×10⁶ cells). Tumor masses were determined in a blinded fashion with Vernier calipers according to the following formula: mass(mg)=(width, mm)²×(length, mm)/2 (19)and then monitored daily with microcalipers. Mice bearing xenografts (150–200 mg) were then randomized to DMSO, PFK15 (25 mg/kg i.p. every 3 days × 4), irinotecan (70 mg/kg i.v. every 3 days × 4), temozolomide (30 mg/kg p.o. 3 days on 3 days off), or gemcitabine (100 mg/kg every 3 days × 4) and microcaliper measurements were conducted daily. All data are expressed as the mean±SD of two experiments (n = 8 per group). Statistical significance was assessed by the unpaired two tail T-test. In a separate series of experiments, LLC-tumor bearing mice were euthanized after only four days of treatment and the tumors were excised, fixed in formalin, embedded in paraffin and stained with Hematoxylin and Eosin or with anti-cleaved caspase 3 (Abcam, Cambridge, MA) using standard immunohistochemical methodologies.

Results

Identification of a More Potent Small Molecule Antagonist of PFKFB3

Initially, we synthesized 73 synthetic derivatives of 3PO and then examined their relative potency against the activity of recombinant human PFKFB3 protein. We identified a 3PO derivative in which a quinolinyl ring was substituted for a pyridinyl ring, 1-(4-pyridinyl)-3-(2-quinolinyl)-2-propen-1-one (PFK15; Fig. 1A), that displayed approximately 100-fold more activity against PFKFB3 than 3PO (IC₅₀: 3PO, 22.9 μM; PFK15, 207 nM; Fig. 1B). Subsequent computational modeling suggested that the increased activity of PFK15 may be due to the interaction of the quinolinyl ring with residues that form the ADP/ATP binding site of PFKFB3 (highlighted in green; compare Fig. 1C and D). We next examined PFK15 for activity against 96 kinases using a commercially available active-site dependent competition binding assay core service (KINOMEscanEDGE). This assay quantifies the capacity of test agents to compete with an immobilized, active-site directed ligand using a DNA-tagged kinase and immobilized ligand and compound. PFK15 had no significant inhibitory effect on any of the tested kinases (range = 93–100% of control; 10 μM PFK15; see Methods to access names of 96 kinases). Additionally, up to 100 μM PFK15 was found to not inhibit the activities of purified PFK-1, hexokinase, phosphoglucoseisomerase or PFKFB4 (data not shown). These observations supported the specificity of PFK15 for PFKFB3 as well as the continued analysis of PFK15 for activity against the metabolism and viability of transformed cells.

Figure 1. Identification of a novel small molecule antagonist of PFKFB3.

A, structures of 3PO and the 3PO analogue, 1-(4-pyridinyl)-3-(2-quinolinyl)-2-propen-1-one (PFK15). B, physical screening of 3PO and PFK15 for inhibitory effects on recombinant PFKFB3 activity. C, computational modeling of 3PO docked in the substrate-binding domain of PFKFB3 (green indicates ATP/ADP-interacting residues). D, computational modeling of PFK15 docked in the substrate-binding domain of PFKFB3.

Inhibition of Cancer Cell Viability, F26BP, Glucose Uptake and ATP by 3PO and PFK15

The parent compound 3PO suppresses the metabolism and growth of Jurkat T cell leukemia cells at a relatively low concentration(18). In order to identify a lung adenocarcinoma cell line that is similarly sensitive, we employed the National Cancer Institute 60 cell line screening service which assesses the cytostatic/toxic effects of test agents on 59 cancer cell lines in vitro by measuring cellular protein content over 48 hours. H522 cells were found to be the most sensitive lung adenocarcinoma cell line (Table 1) and we thus elected to directly compare the metabolic effects of 3PO relative to PFK15 in these adenocarcinoma cells and Jurkat leukemia cells. Initially, we assessed the cytotoxic effects of 3PO relative to PFK15 on Jurkat and H522 cells after 48 hours and observed an increase in the potency of PFK15 (IC₅₀: Jurkat, 7.2 μM ([3PO], 2.42 μM [PFK15]; H522, 5.57 μM [3PO], 0.72 μM [PFK15]; Fig. 2A and B). Suppression of PFK-1 activity is expected to increase the concentration of its substrate, F6P, which is in equilibrium with glucose 6-phosphate, a potent inhibitor of hexokinase and thus glucose uptake (which is dependent on phosphorylation by hexokinase). Not surprisingly, 3PO and heterozygous genomic deletion of the Pfkfb3 gene reduce both the intracellular concentration of F26BP and glucose uptake of transformed cells in vitro and in vivo (14) (18). We examined the steady-state concentration of the substrate of PFKFB3, F26BP, as well as 2-[1-¹⁴ C]-deoxy-D-glucose (¹⁴ C-2DG) uptake and intracellular ATP, after only three hours of PFKFB3 inhibitor exposure (prior to the detection of cytotoxicity). We found that both 3PO and PFK15 reduced F26BP, glucose uptake and intracellular ATP but that PFK15 was more potent than 3PO (Fig. 2C–H). Interestingly, we found that the marked reduction of ATP caused by PFK15 could be partially rescued by pyruvate (PFK15-mediated ATP depletion [% control]: 5 mM glucose alone, 5.29 ± 0.85%; 5 mM glucose + 1 mM pyruvate, 45.62 ± 1.72%; p value = 0.0017). In contrast, the ATP depletion caused by PFK15 could not be significantly reversed by the addition of either 10 mM lactate or 4 mM glutamine (data not shown). These observations provide indirect support for the hypothesis that PFK15 predominantly targets the glycolytic pathway and that the downstream add-back of pyruvate may provide a source of carbons necessary for the TCA cycle as well as a mechanism to provide NAD+ via LDH-A.

Figure 2. Effects of 3PO and PFK15 on the viability, F26BP, glucose uptake and intracellular ATP of Jurkat T cell leukemia cells and H522 lung adenocarcinoma cells.

Jurkat or H522 cells grown in exponential phase were exposed to the indicated concentrations of 3PO or PFK15 and, after 48 hours, the effects on viability were determined (A, B). In order to identify acute effects on known metabolic effects of PFKFB3 inhibition, the Jurkat and H522 cells were exposed to 3PO or PFK15 for three hours and the F26BP (C, D), 2-[1-^14dC]-deoxy-D-glucose uptake (E, F) and intracellular ATP (G, H) were measured.

3PO and PFK15 Cause Apoptosis in Jurkat T Cell Leukemia Cells

In previous studies, we found that 3PO suppressed cell proliferation and, at >10 μM concentrations, caused cell death via undefined mechanisms(18). Since glucose deprivation can induce apoptosis (20), we postulated that the reduction in glucose uptake caused by 3PO and the more potent PFK15 might induce apoptotic death in these cells. Using 25μM etoposide as a positive control, we directly compared the effects of two concentrations of 3PO and PFK15 (3 and 20 μM to encompass both limited and extensive cell death)on early (Annexin V+/PI−) and late (Annexin V+/PI+) apoptosis using flow cytometry. After five hours of exposure, we found that whereas 20 μM 3PO and 3 μM PFK15 increased the number of cells undergoing early apoptosis (Fig. 3A and B), both 3 and 20 μM PFK15 caused a dose-dependent increase in the number of cells already undergoing late apoptosis (Fig. 3A and C). These observations provide the first demonstration that this class ofPFKFB3 inhibitors can induce apoptosis in transformed cells and suggest that PFK15 causes a more rapid progression to late apoptosis than 3PO.

Figure 3. PFKFB3 inhibitors induce apoptosis in Jurkat T cell leukemia cells.

A, Jurkat cells were exposed to the indicated concentrations of etoposide, 3PO or PFK15 for five hours, stained for Annexin V and propidium iodide and analyzed by flow cytometry. The percentages of Annexin V+/PI− (early apoptotic; B) and Annexin V+/PI+ (late apoptotic; C) cells relative to the total population were quantified. ^* p value < 0.01

PFK15 Displays Improved Pharmacokinetic Properties Relative to 3PO

A key objective of our pursuit to identify a clinical candidate that inhibits PFKFB3 has been to improve the pharmacokinetic properties of the parent compound 3PO. We intravenously administered 3PO or PFK15 to C57Bl/6 mice and then, using LC-MS/MS, assessed the plasma concentration of each of the compounds over time. PFK15 displayed a marked reduction in clearance (3PO = 2312 mL/min/kg; PFK15 = 46.2 mL/min/kg) simultaneously with increased T_1/2 (3PO = 0.3 hr; PFK15 = 5.1 hr),C_max (3PO = 113 ng/ml; PFK15 = 3053 ng/ml) and AUC_0-inf (3PO = 36 ng/hr/ml; PFK15 1804 ng/hr/ml). Microsomes are membrane vesicles isolated from hepatocyte smooth endoplasmic reticulum that contain multiple enzymes required to metabolize drugs including cytochrome P450s, flavin-containing monooxygenases and UDP glucuronyl transferases. Human microsome stability of PFK15 was prolonged relative to 3PO providing a potential mechanism for the observed improvements in pharmacokinetic properties (3PO = 1.9 min [100%]; PFK15 = 2.9 min [100%]). These data indicate that the substitution of the pyridinyl ring in 3PO with a quinolinyl ring provides a useful platform for the development of PFKFB3 inhibitors with improved pharmacokinetic properties.

PFK15 Suppresses the Growth of Lewis Lung Carcinoma (LLC) Xenografts

Intraperitoneal administration of 25 mg/kg PFK15 every three days caused <10% body mass loss and no end organ toxicity based on blood, urine and histological analyses. Since 25 mg/kg of PFK15 was easily administered in a small volume despite the limited solubility of PFK15, this non-toxic dose and schedule was selected for the efficacy testing in xenograft and syngeneic models of tumorigenesis. We initially examined the relative effects of daily intraperitoneal administration of PFK15 on the growth of established LLC tumors in vivo. Although we did not observe tumor regression as a result of PFK15 administration, we did observe a statistically significant reduction in the growth of the LLC tumors (Fig. 4A) without a major effect on body mass (Fig. 4B). Importantly, whereas LLC cells are well established to metastasize from the subcutaneous tumors to the lungs, no lung metastases were identified in the LLC-bearing mice that had been treated with PFK15 (Fig. 4C).We then examined a subset of tumors for F26BP and apoptotic cells four days after initiation of PFKFB3 inhibitor administration (n=4). We found that PFK15 reduced tumor-associated F26BP (Fig. 4D) and caused a marked increase in the number of cells that were positive for cleaved caspase 3 by immunohistochemistry (Fig. 4E and F). Taken together, these results suggest that PFK15 can reduce tumor growth and induce apoptosis in vivo, is well tolerated in terms of body mass and has the unanticipated ability to prevent either the initiation or growth of metastases.

Figure 4. PFK15 suppresses the growth, metastatic spread and glucose metabolism of LLC tumors in syngeneic mice.

Groups of eight C57Bl/6 mice (20 gm) were injected s.c. with 0.1 ml of a LLC cell suspension (1×10⁶ cells) and tumor masses were determined in a blinded fashion with Vernier calipers. Mice bearing xenografts (150–200 mg) then were randomized to DMSO or PFK15 and microcaliper measurements (A) and body mass (B) were collected daily. After 14 days, mice were euthanized and the lungs were examined for metastases (C). Tumors were extracted in a separate sub-group of mice after four days and analyzed for F26BP (D) and cleaved caspase 3 expression as a measure of apoptosis (E, F). A separate cohort of tumor-bearing mice underwent baseline micro-PET imaging of FDG uptake and, 24 hours later, were administered PFK15 (25/kg i.p. once) and the micro-PET scan was repeated in 45 minutes. Regions of interest in the tumor and cerebellum were quantified in quadruplicate (G) and a representative cut is provided in the transverse, coronal and sagittal views (H). ^*p value < 0.01

PFK15 Suppresses ¹⁸F-FDG Uptake by Lewis Lung Carcinoma (LLC) Xenografts

We postulated that PFK15 would acutely attenuate the glucose uptake of xenografted tumors in vivo. We conducted baseline ¹⁸F-FDG-PET imaging of LLC-bearing C57Bl/6 mice and, 24 hours later, injected PFK15 and then reimaged the LLC-bearing mice after 45 minutes. We observed a ~50% reduction in ¹⁸F-FDG uptake after PFK15 administration normalized to the FDG uptake in the cerebellum (neurons do not express PFKFB3 protein; Fig. 4G and H). These data demonstrate that PFK15 selectively suppresses glucose uptake by tumors in vivo and suggest that FDG uptake scans may be a useful pharmacodynamic endpoint of PFKFB3 inhibitors in clinical trials.

PFK15 Displays Anti-Tumor Activity That Is Similar to Commonly Used Chemotherapeutic Agents

Last, we directly compared the anti-tumor effects of PFK15 to three chemotherapeutic agents that are FDA-approved for the treatment of human cancers. We found that PFK15 suppressed the growth of colon and pancreatic adenocarcinomas similarly to irinotecan (Fig. 5A) and gemcitabine (Fig. 5C), respectively. However, the activity of PFK15 against glioblastoma growth was lower than that observed by temozolomide (Fig. 5B). These pre-clinical studies raise the prospect that PFK15 and close synthetic derivatives may demonstrate cytotoxic activity against human cancers in clinical trials.

Figure 5. PFK15 suppresses the growth several human xenograft tumors in athymic mice.

Groups of eight athymic mice (20 gm) were injected s.c. with CT26 (A), U-87 MG (B) and BxPC-3 (C) cells and tumor masses were determined in a blinded fashion with Vernier calipers. Mice bearing xenografts (150–200 mg) were then randomized to DMSO, the indicated standard chemotherapy or PFK15 and the tumor masses were quantified with microcalipers. ^*p value < 0.01

DISCUSSION

In this study, we describe the development of anovel small molecule antagonist of PFKFB3 that suppresses the glucose uptake and viability of human cancer cell lines in vitro and that has improved pharmacokinetic properties relative to the parent compound 3PO mice. Importantly, PFK15 rapidly reduces F26BP and glucose uptake and induces apoptosis in cancer cells in vitro and in vivo indicating that these metabolic measures can be used aspharmacodynamic endpoints in clinical trials. This study provides indirect support for a catastrophic cascade in which suppression of PFKFB3 activity reduces the steady-state concentration of F26BP which, in turn, causes suppression of PFK-1 activity and glucose uptake. Although the precise mechanism for the effect of PFK15 on glucose uptake has not been determined, we postulate that certain glycolytic intermediates upstream of PFK-1 may allosterically inhibit hexokinase or affect the glucose transporters. Importantly, the effect of PFKFB3 inhibition on glucose uptake has been well established by the observation that heterozygous genomic deletion of the Pfkfb3 gene reduces the intracellular F26BP and glucose uptake of transformed cells in vitro and in vivo (14). Given that withdrawal of glucose from transformed cells and resultant ATP depletion induces apoptosis(20), we postulate that the observed reduction of glucose uptake and intracellular ATP caused by PFK15 may cause apoptosis via similar yet undefined mechanisms. Importantly, we also believe that the functional glucose starvation that PFK15 causes via PFKFB3 inhibition also results in carbon depletion, which in turn leads to disruptions in multiple anabolic pathways that may not be readily circumvented by discrete resistance-causing mutations. Last, given the potent anti-glycolytic effects of PFK15, we also hypothesize that this class of compounds will prove useful for the targeting of resistant hypoxic cancer cells as well as the development of combinatorial strategies to suppress growth using PFK15 with ionizing radiation.

Although we found that PFK15 was more potent than 3PO at inhibiting recombinant PFKFB3 activity, glucose uptake, intracellular F26BP and ATP, the increased potency of PFK15 related to apoptosis was dramatically greater than that observed with the metabolic changes. We suspect that relatively small increments in the suppressive activity of PFK15 on glucose uptake and intracellular ATP may reach a threshold that results in a much larger effect on apoptosis depending on the cell type, availability of alternative substrates, such as glutamine and pyruvate, and on oxygen availability in vivo. An alternative and certainly plausible explanation relates to potential off-target effects of PFK15 on mitochondrial metabolism and integrity. However, ectopic expression of PFKFB3 has previously been shown to increase intracellular F26BP and protect Jurkat T cell leukemia cells from the cytotoxic effects of the parent compound, 3PO, indicating that PFKFB3 is, at a minimum, an essential enzyme target of this class of small molecules (18).

These pre-clinical studies have facilitated the design of the first-in-human phase I clinical trial of a PFK15 derivative similarly modified with a quinolinyl ring in advanced solid tumor patients. The original parent compound 3PO was recently reported to have potent activity against a highly relevant mouse model of leukemia (21) and we thus anticipate that once the maximum tolerated dose is identified, phase II trials of the optimized PFKFB3 inhibitor also will be proposed in patients suffering from leukemias. Perhaps more importantly, we expect the development of multiple phase I/II trials in which PFKFB3 inhibitors are combined with FDA-approved agents that are the standard of care for distinct cancer types as well as the rational combination of PFKFB3 inhibitors with agents that suppress angiogenesis, electron transport chain activity, glutamine metabolism and autophagy.

Recently, we demonstrated that PFKFB3 expression and F26BP synthesis are increased by stimulation of human CD3⁺ T cells and that 3PO suppresses the activation of T cells in vitro and in vivo(22). Additionally, PFKFB3 was initially identified in the mid-1990s as an inducible PFK-2 (iPFK-2) isozyme that is rapidly induced by endotoxin in human macrophages(23, 24). Given our growing understanding of the immunosuppressive effects of regulatory T cells(25) and myeloid-derived suppressor cells (26)and the differential metabolic requirements of various T cell subsets(27), we speculate that PFKFB3 inhibitors may have unidentified pleiotropic effects on the immune system which may contribute to their anti-tumor properties. Conditional Pfkfb3 knock-out mice in which the expression of Cre is controlled by transcription factors that are immune lineage-specific should enable a more complete characterization of the roles of PFKFB3 in the activation of distinct suppressor lineages of cells involved in tumor immunity.

In conclusion, we describe the identification of a novel PFKFB3 inhibitor that has acceptable pharmacokinetic properties, on-target metabolic effects and anti-tumor activity in four mouse models of cancer. Coupled with the identification of potential biomarkers for activity, we believe that the results of this study provide essential rationale for the continued testing of this novel class of targeted therapeutic agents in clinical trials.

Table I.

Effect of PFK15 on the Growth of Cancer Cell Lines Using the Sulforhodamine B Assay (GI₅₀; μM; National Cancer Institute Developmental Therapeutics Program)

Leukemia
CCRF-CEM	3.80
HL-60(TB)	2.04
MOLT-4	2.24
RPMI-8226	2.19
SR	0.23
Non-Small Cell Lung Cancer
A549/ATCC	5.37
EKVX	12.88
HOP-62	2.34
HOP-92	3.02
NCI-H23	3.16
NCI-H460	3.47
NCI-H522	1.74
Colon Cancer
COLO 205	2.45
HCC-2998	1.91
HCT-116	1.32
HCT-15	1.1
HT29	2.51
KM12	1.95
SW-620	1.7
CNS Cancer
SF-268	5.25
SF-295	13.49
SF-539	2.51
SNB-19	4.57
SNB-15	11.22
U251	2.04
Melanoma
LOX IMVI	1.66
MALME-3M	3.31
M14	3.72
MDA-MB-435	3.24
SK-MEL-2	6.46
SK-MEL-28	2.09
SK-MEL-5	3.47
UACC-257	3.98
UACC-62	2.51
Ovarian Cancer
IGROV1	4.37
OVCAR-3	3.16
OVCAR-4	2.34
OVCAR-5	5.37
OVCAR-8	1.82
NCI/ADR-RES	3.89
SK-OV-3	15.14
Renal Cancer
786–0	2.00
A498	11.75
ACHN	2.04
CAKI-1	11.48
RXF 393	1.70
SN12C	1.86
TK-10	4.07
UO-31	1.62
Prostate Cancer
PC-3	3.16
DU-145	2.24
Breast Cancer
MCF7	1.86
MDA-MB-231/ATCC	2.04
HS 578T	20.89
BT-549	2.75
T-47D	3.63
MDA-MB-468	1.66