Insulin enhances metabolic capacities of cancer cells by dual regulation of glycolytic enzyme pyruvate kinase M2

PK exists in four isoforms in mammals; therefore, isoform status of PK in HepG2, H1299 and PC3 cells was examined. Immunoblotting analysis in HepG2 showed that PKM2 was the predominant isoform in these cells with complete absence of PKM1 (Figure 1). Whereas, PKL isoform, which is predominant in normal liver cells, expressed at a much lower level in HepG2 cells (Figure 1). Similarly, PKM2 was the predominant isoform in H1299 and PC3 cells with negligible PKM1 expression (Additional file 1: Figure S1).

To examine the effect of insulin on PKM2 expression, serum starved HepG2 cells were treated with insulin for 0, 2, 4, 6 and 8 hours with maximum increase in PKM2 expression at 8 hours of insulin treatment (Figure 2A). Dose dependent effect of insulin on PKM2 expression revealed that 1 nM insulin (near physiological concentration) substantially induced PKM2 expression, however, maximum increase in PKM2 expression was observed with 100 nM insulin for 8 hours (Figure 2B); suggesting to choose 100 nM insulin for further experiments. A two fold increase in PKM2 mRNA and protein expression was observed upon 100 nM insulin treatment for 8 hours (Figure 2C-E), whereas, PKL expression remained unaltered under similar conditions (Figure 2D). Insulin treatment up-regulated PKM2 in H1299 (lung cancer) and PC3 (prostate cancer) cells also (Additional file 2: Figure S2), replicating the observations made in HepG2 cells.

PI3K/mTOR signalling was inhibited to find out if insulin-induced PKM2 up-regulation is PI3K dependent. PI3K is the chief mediator of insulin signalling in liver and is frequently mutated in liver and other cancers [26‐28]. As a result of PI3K/mTOR inhibition, PKM2 expression was down-regulated (Figure 2C-E). These results confirmed that insulin regulates PKM2 expression through PI3K/mTOR pathway. To elucidate how PI3K/mTOR promoted PKM2 expression, the role of HIF1α protein, a transcription factor that binds to hypoxia response element (HRE) in PKM2 gene to induce its expression, was examined [29, 30]. Insulin induced HIF1α expression in a PI3K/mTOR sensitive manner, as was evident from the decrease in HIF1α protein after treatment with PI3K/mTOR inhibitors (Figure 2D). Further, to corroborate this observation, HIF1α was silenced as well as induced, using siRNA and cobalt chloride (CoCl₂) treatment, respectively, to observe the effect on PKM2 expression [31, 32]. Expectedly, HIF1α silencing led to a decreased PKM2 expression; whereas CoCl₂-induced HIF1α accumulation increased PKM2 expression (Figure 2F), confirming that HIF1α is crucial for PKM2 up-regulation. These results demonstrated that insulin upregulates PKM2 expression through PI3K/mTOR dependent induction of HIF1α transcription factor.

PKM2 activity is crucial in determining the glycolytic flux; it was pertinent therefore to examine the potential role of insulin in altering PKM2 activity. Interestingly, ~30% drop in PKM2 activity was observed in HepG2 cells within 15 minutes after insulin treatment with no further decrease on increasing the duration of treatment (Figure 3A, Additional file 3: Figure S3). However, PI3K/mTOR inhibition did not reverse insulin-induced decrease in PKM2 activity (Figure 3A). Insulin-treated H1299 and PC3 cells also exhibited decrease in PKM2 activity (Additional file 4: Figure S4). To analyse if the insulin-induced drop in PKM2 activity is due to subunit dissociation, which governs the overall activity of PKM2 [9, 11, 15, 33, 34], glycerol density gradient centrifugation was performed using HepG2 cell lysate. Density gradient centrifugation is known to separate different oligomeric forms of an enzyme with highest quaternary form at the bottom of the gradient and the magnitude of activity representing the respective content of an oligomeric form [35]. PKM2 activity was measured in fractions collected from top to bottom of the glycerol gradient. Observation of two peaks suggested two oligomeric forms of PKM2 and the vertical shift in peaks indicated subunit dissociation (Figure 3B). As evident from relative peak areas, a significant fall in peak II with concomitant increase in peak I, compared to control peaks, was observed upon insulin treatment (Figure 3C). Dose dependent insulin treatment showed that even 1nM insulin treatment decreased PKM2 activity significantly with maximum decrease on 100 nM insulin (Figure 3D). Results, thus, suggested that decrease in PKM2 activity was due to subunit dissociation upon insulin treatment. As expected, PI3K/mTOR inhibition did not abolish the effect of insulin on PKM2 oligomeric status (Additional file 5: Figure S5). In an attempt to understand how insulin promoted subunit dissociation, we checked the tyrosine 105 phosphorylation of PKM2, which is known to diminish its activity by disrupting the formation of active tetramer [33], but, no change in phosphorylation was observed following insulin treatment (Additional file 6: Figure S6 and Additional file 7: Figure S7). These results suggested the role of some other insulin-induced factors in PKM2 activity regulation.

To find out the factor responsible for inhibition of PKM2 activity in HepG2 cells, the possible role of insulin-induced ROS was analysed. ROS is known to regulate the activity of several proteins through oxidation, we thus investigated if insulin-induced-ROS affected PKM2 activity through its possible oxidation. Insulin treatment increased the in-vitro production of ROS by ~60%, as assessed by fluorescence spectrophotometry (Figure 4A). Pretreatment with ascorbate (Asc) or N-acetyl-L-cysteine (NAC), well-known ROS scavengers [36‐41], reduced insulin-induced-ROS markedly by ~50-70%. This observation was replicated in H1299 and PC3 cells (Additional file 8: Figure S8). Interestingly, ascorbate or NAC pretreatment, followed by incubation with insulin, reversed the insulin-induced decrease in PKM2 activity in HepG2 (Figure 4B), H1299 and PC3 (Additional file 4: Figure S4) cells, suggesting the role of insulin-induced-ROS in PKM2 activity inhibition. Further, dose-dependent increase in ROS production by insulin correlated with dose-dependent decrease in PKM2 activity (compare Figures 3D and 4C). To understand how ROS contributed to reduction in PKM2 activity, the possibility of ROS-induced- oxidation of PKM2 was studied. Addition of dithiothreitol (DTT), a strong reducing agent, to activity reaction mixture abolished insulin-induced-decrease in PKM2 activity (Figure 4D). In fact, addition of DTT increased PKM2 activity to the levels similar to that of control without insulin treatment (Figure 4D and B). These data indicated the role of ROS-induced-oxidation of PKM2 [34], resulting in its activity-inhibition.

PKM2 expression has been suggested to promote aerobic glycolysis [12, 30], which is characterized by high glucose uptake and lactate production even in presence of oxygen, and is believed to be the hallmark of nearly all cancer cells [4]. Both glucose uptake and production of lactate were substantially enhanced in cells treated with insulin as compared to untreated control (Figure 5A). Further, inhibition of PI3K/mTOR pathway reduced aerobic glycolysis which is consistent with earlier reports (Figure 5A) [42, 43]. To show that PI3K/mTOR dependent PKM2 up-regulation contributed to insulin-induced augmentation in aerobic glycolysis, PKM2 was knocked down in cells with hyperactive PI3K/mTOR signalling (to mimic insulin-induced activation of PI3K/mTOR), followed by an assessment of glucose uptake and lactate production, approximately 48 hours after transfecting control shRNA or PKM2-shRNA (to knock down PKM2 expression). Interestingly, silencing of PKM2 partially inhibited glucose uptake and lactate production, signifying that PKM2 is required for aerobic glycolysis (Figure 5B-C). Moreover, PKM2 knockdown also retarded cellular proliferation (Figure 5D), consistent with our previous observation [44].

Characteristic accumulation of glycolytic intermediates, as a result of decreased PKM2 activity, is another key feature of cancer cells and has been shown to be important for anabolic synthesis required for tumor growth [8, 11]. To explore the implications of insulin-induced suppression of PKM2 activity on glycolytic pooling, intracellular levels of fructose-1,6-bisphosphate (FBP) and phosphoenolpyruvate (PEP), the glycolytic metabolites upstream of PKM2, were measured [15]. Consequently, an increased build-up of both metabolites was observed compared to control (Figure 6A). Glycolytic pooling is known to facilitate pentose phosphate pathway for macromolecular synthesis [8, 9, 13]. Therefore, intracellular levels of NADPH, which is produced as a result of PPP and provides reducing power for macromolecular synthesis [45], were determined. Accumulation of NADPH increased significantly (P < 0.05) in insulin treated cells (Figure 6B).

To demonstrate that decreased PKM2 activity is contributing to the observed accumulation of FBP, PEP and NADPH, PI3K/mTOR pathway was inhibited as it increases glucose uptake which might have contributed to observed accumulation. PI3K/mTOR inhibition did not completely block the insulin-induced pooling of FBP, PEP and NADPH. However, pretreatment with ROS scavengers NAC or ascorbate abolished the insulin-induced glycolytic pooling and accumulation of NADPH. These results indicated that insulin promoted accumulation of glycolytic intermediates and NADPH through decreased PKM2 activity.

HepG2 and PC3 cell lines were procured from the National Centre for Cell Science, Pune, India. H1299 cells were a kind gift from Dr. Uttam Pati, School of Biotechnology, Jawaharlal Nehru University. All the cell lines were maintained in DMEM (Sigma) with 10% heat-inactivated FBS (Biowest, France), 1% penicillin/streptomycin (Sigma) at 37°C and 5% CO₂ in a humified atmosphere (Heraeus, UK). Cells were grown in monolayer and passaged routinely two-three times a week. For drug treatment; LY294002 (Sigma) and rapamycin (Sigma) were dissolved in DMSO; single use aliquots were stored at −80°C For insulin treatment: cells were seeded in triplicate at a density of 0.4 million cells/well of six well plates, serum starved for 20–24 hours and then stimulated with insulin (Sigma) in presence or absence of inhibitors. DMSO treated cells were used as mock control. For PKM2 knockdown: short hairpin RNA (shRNA) constructs in lentiviral pGIPZ vector were purchased from Open Biosystems. Control and shRNA containing pGIPZ vectors were transfected using polymer based transfection reagent- Arrestin from Thermo-Scientific, as per manufacturer’s instructions. Knock down efficiency was then checked by real time PCR and Western blotting. For HIF1α silencing: pre-designed siRNA was used and transfected using Lipofectamine (Invitrogen) as per manufacturer’s protocol and efficiency was checked by Western blotting. For proliferation assay: cells were counted before and after 48 hours of transfection, using a hemocytometer.

Cellular RNA was extracted from cell lines using TRIzol (Sigma), according to manufacturer’s protocol. RNA quality was analyzed by A₂₆₀/A₂₈₀ absorbance and by electrophoresis on a 1.2% agarose formaldehyde gel. 3–4 μg of total RNA was reverse transcribed into single stranded DNA using cDNA preparation kit (Applied Biosystems, USA). Commercially available Taqman gene expression assay (Applied Biosystems, USA) was used for quantitating mRNA levels of PKM2. β-actin was used as endogenous control. Real time PCR was carried out on ABI Prism 7000 Sequence Detection System (Applied Biosystems). ΔΔC_t (Cycle threshold) method of relative quantification was used to calculate fold change in gene expression by SDS 1.1 RQ software (Applied Biosystems).

Whole cell lysate was prepared by incubating cells, on ice for 30 minutes, in buffer containing 50 mM Tris pH 7.2, 150 mM NaCl, 0.5% sodium deoxycholate, 10% glycerol, 1% Triton X-100, 0.1% SDS, 1 mM DTT, 1 mM PMSF, 5 mM NaF, 1 mM NaV, phosphatase inhibitor cocktail (Sigma), 4 μg/ml aprotinin, 4 μg/ml leupeptin and 4 μg/ml pepstatin (Sigma). The lysate was centrifuged at high speed in a cooling centrifuge (CM 12, Remi, India) for 30 minutes and supernatant was collected in pre-chilled fresh tubes. Protein concentration was estimated using BCA method as per manufacturer protocol (Thermo Scientific). Proteins were separated on 8% SDS-PAGE, transferred to nitrocellulose membrane (mdi) at 4°C (wet transfer) and probed with primary antibodies. Membrane was incubated with appropriate secondary antibody for one hour at room temperature and proteins were detected using Luminata forte (Millipore). Primary antibodies used were: anti-HIF1α (Novus Biologicals, USA), anti PKM2, anti-phospho-PKM2 (Tyr105), anti-AKT, anti-phospho-AKT, anti-phosphoS6, anti-S6 protein and anti-β-actin (Cell Signalling Technology).

For activity, cells were lysed in buffer as described previously [12]. Activity was measured using NADH/lactate dehydrogenase (LDH) coupled assay. Decrease in OD at 340 nm due to oxidation of NADH was monitored using a double beam spectrophotometer (UV-1800, Shimadzu). Reaction was started by adding 2 μg cell lysate to mixture containing 50 mM Tris pH 7.5, 100 mM KCl, 5 mM MgCl₂, 1.25 mM ADP, 0.5 mM PEP, 0.28 mM NADH and 8 units of LDH. Specific activity per mg of cell lysate was calculated as:

U/mg = specific enzyme activity per mg of protein.

OD₃₄₀ = Change in absorbance due to oxidation of NADH in one minute at 340 nm wavelength.

For glycerol gradient experiment, 500 μg of cell lysate protein was loaded on the top of 11-25% glycerol gradient and centrifuged at 45000 rpm for 16 hours at 4°C in SW55Ti rotor (Beckman Coulter) and rest of the procedure was followed as described [16].

Metabolite extract was prepared from 20 million cells in 0.5 ml of chilled 90% ethanol containing 0.2% formic acid and centrifuged at 15000 rpm in a refrigerated centrifuge. Supernatant was dried using nitrogen flow and then reconstituted in 0.2 ml of MilliQ water. PEP was assessed using NADH/LDH coupled assay as mentioned above with 30 ng of recombinant PKM2. FBP was measured as described [51]. For both FBP and PEP, concentration was determined against standard curve. NADPH was analyzed using kit from BioVision- USA, as per the manufacturer’s protocol. For glucose and lactate: media was collected from wells; spun down at high speed to remove any cell debris, deproteinized using TCA, pH was adjusted between 7.0-7.5 and then glucose uptake was analyzed using glucose assay kit (Sigma), according to manufacturer’s protocol. Lactate and was analysed using kit (BioVision, USA) as per manufacturer’s instructions. All the measurements were normalized to cell numbers.

Cells pretreated with or without ascorbate or NAC (Sigma) were incubated with DCFH-DA for 30 minutes in dark at 37°C, followed by insulin stimulation. Cells were then washed with PBS, trypsinized and resuspended in 0.5 ml PBS, kept on ice, in dark until ROS analysis by fluorescence spectrophotometry. Excitation and emission wavelengths used were 500 nm and 510 nm respectively.

Each experiment was performed in triplicate. All experiments were repeated at least 3 times. Significance was calculated using student’s t-test. P value less than 0.05 was considered statistically significant.