Simultaneous inhibition of PFKFB3 and GLS1 selectively kills KRAS- transformed pancreatic cells

Abstract

Activating mutations of the oncogenic KRAS in pancreatic ductal adenocarcinoma (PDAC) are associated with an aberrant metabolic phenotype that may be therapeutically exploited. Increased glutamine utilization via glutaminase-1 (GLS1) is one such feature of the activated KRAS signaling that is essential to cell survival and proliferation; however, metabolic plasticity of PDAC cells allow them to adapt to GLS1 inhibition via various mechanisms including activation of glycolysis, suggesting a requirement for combinatorial anti-metabolic approaches to combat PDAC. We investigated whether targeting the glycolytic regulator 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 (PFKFB3) in combination with GLS1 can selectively prevent the growth of KRAS-transformed cells. We show that KRAS- transformation of pancreatic duct cells robustly sensitizes them to the dual targeting of GLS1 and PFKFB3. We also report that this sensitivity is preserved in the PDAC cell line PANC-1 which harbors an activating KRAS mutation. We then demonstrate that GLS1 inhibition reduced fructose-2,6-bisphosphate levels, the product of PFKFB3, whereas PFKFB3 inhibition increased glutamine consumption, and these effects were augmented by the co-inhibition of GLS1 and PFKFB3, suggesting a reciprocal regulation between PFKFB3 and GLS1. In conclusion, this study identifies a novel mutant KRAS-induced metabolic vulnerability that may be targeted via combinatorial inhibition of GLS1 and PFKFB3 to suppress PDAC cell growth.

1. Introduction

Activating mutations of KRAS underlie majority of pancreatic ductal adenocarcinoma (PDAC) cases [1] and are associated with profound metabolic changes in PDAC cells, including dependence on enhanced glutamine metabolism [2] and glycolytic activity [3] for proliferation and survival. Understanding the precise metabolic hubs of KRAS signaling may reveal targetable vulnerabilities for the treatment of PDAC tumors. However, additional alterations incurred during PDAC development may endow PDAC cells with unique capabilities to resist single metabolic perturbances [4]. In support of this notion, although PDAC cells harboring a KRAS mutation exhibit increased glutamine metabolism, they are relatively insensitive to the inhibition of glutaminase-1 (GLS1) [5]dthe primary glutaminase isozyme in tumor cells. Therefore, combinatorial metabolic targeting may be required to combat KRAS-driven PDAC tumors, a malignancy with the five-year survival of a paltry 10 % [6]. The bifunctional 6-phosphofructo-2-kinase/fructose 2,6- bisphosphatase (PFKFB) enzymes play an essential role in regulating glycolytic flux in cells, as their product fructose 2,6- bisphosphate (F2,6BP) is a key allosteric activator of phosphofructokinase 1 (PFK1), a major control point of glycolysis [7]. Activity of the third isozyme of the family, PFKFB3, has been causatively linked with malignant properties, including proliferation and survival [8]. Although PFKFB3 has been known to be essential for the oncogenic transformation of mouse lung fibro- blasts by the HRAS isoform [9], the association of PFKFB3 with RAS signaling in a more relevant cell model of human tumors has only been recently made. Wang et al. [10] showed that the oncogenic transformation of immortalized pancreatic duct cells with the most common mutant KRAS variant observed in PDAC, i.e. KRAS with aspartate substitution at codon 12, KRASG12D, increases PFKFB3 protein stability.

However, it is unknown whether targeting PFKFB3 activity can selectively attenuate KRAS-associated metabolic and proliferative phenotype. Importantly, despite the relevance of PFKFB3 and GLS1 in regulation of glycolysis and glutamine meta- bolism, respectively, which are frequently co-activated in tumor cells, the potential effect of PFKFB3 and GLS1 co-inhibition on cell growth remains unknown. Using human pancreatic duct cells sequentially immortalized and transformed with a mutant KRAS [11], we show that PFKFB3 is required for glycolytic activity and proliferation induced by KRAS, and the co-inhibition of GLS1 and PFKFB3 selectively suppress the growth of KRAS-transformed pancreatic duct cells and the PDAC model PANC-1 cells harboring a KRAS mutation.

2. Materials and methods

2.1. Cell culture and reagents

Immortalized human nestin-positive pancreatic duct cells (HPNE) and KRASG12D-transformed counterpart (HPNE/KRAS) has been previously described [11]. These cells were cultured in DMEM supplemented with 5 % fetal bovine serum, 2 mM glutamine, 10 ng/ ml epidermal growth factor, non-essential amino acids, and 750 ng/ ml puromycin. PANC-1 cells (ATCC cat#CRL-1469) were cultured in DMEM supplemented with 10 % fetal bovine serum, 2 mM gluta- mine, and 1 % penicillin/streptomycin. BxPC-3 cells (ATCC cat#CRL- 1687) were cultured in RPMI-1640 media supplemented with 10 % FBS and 1 % penicillin/streptomycin. KRAS inhibitor (SAH-SOS1A) was purchased from EMD Millipore (cat#5.38092.0001). AZ PFKFB3 26 was purchased from MedChemExpress (cat#HY-101971). CB- 839 was purchased from EMD Millipore (cat#5.33717.0001).

2.2. siRNA transfection

A mammalian non-targeting control (ThermoFisher Sci. cat#4390846) and validated PFKFB3-specific (ThermoFisher Sci. cat#s10358 and s10359) and GLS1-specific siRNA molecules (ThermoFisher Sci. cat#s5838) were used in transfections. Trans- fections were carried out using Lipofectamine RNAiMAX (Ther- moFisher Sci. cat#13778150).

2.3. RNA isolation and real-time quantitative PCR

RNA was extracted using GeneJET RNA purification kit (Ther- moFisher) and reverse-transcribed using an mRNA to cDNA syn- thesis kit (ThermoFisher Sci. cat#4787406). Real-time quantitative PCR was performed using Gene Expression Master Mix (Thermo- Fischer Sci.cat#4369016) and TaqMan® probes (ThermoFischer Sci. cat#s: PFKFB1, Hs00997227_m1; PFKFB2, Hs01015408_m1; PFKFB3, Hs00998698_m1; PFKFB4, Hs00190096_m1; GLS1, Hs010104028_m1 and b-actin, Hs01060665_g1) in StepOnePlus (ThermoFischer Sci.). For quantitation, the comparative CT method was employed using formula 2—DDCT [12].

2.4. SDS-PAGE and western blot

SDS-PAGE and Western blotting were performed following standard protocols [13]. The following primary antibodies were used: Anti-PFKFB2 (Cell Signaling cat#13045), anti-PFKFB3 (Ther- moFisher sci. cat#PA5-21931), anti-GLS1 (MyBioSource cat#MBS2026708), anti-b-actin (Cell Signaling cat#3700), anti-ERK (Cell Signaling cat#4695), anti-pERK (Cell Signaling cat#4370), and anti-GAPDH (Cell Signaling cat#5174). Immunoreactive bands were developed with Luminata Forte (EMD Millipore) and visualized in ChemiDoc MP imaging system (Bio-Rad).

2.5. Fructose 2,6-bisphosphate analysis

F2,6BP levels in cell lysates were measured spectrophotomet- rically as described elsewhere [14], except that potato tuber pyrophosphate-dependent phosphofructokinase (PPi-PFK) was replaced by the hetero-hexameric PPi-PFK from Citrus sinensis obtained recombinantly to high purity as recently described [15].

2.6. Metabolic analyses

Glucose uptake and glycolytic activity were determined as described elsewhere [16]. For glucose uptake, cells were incubated for 30 min in glucose-free media and for additional 15 min after adding 2-[1e14C]deoxyglucose (0.1 mCi/ml; PerkinElmer). Counts were measured on Tri-Carb 2910 liquid scintillation analyzer (Per- kinElmer). For glycolytic activity, cells were incubated in medium supplemented with [5-3H]glucose (1 mCi; PerkinElmer) for 60 min. Separation of 3H2O was achieved by an evaporation technique in a sealed system. 3H2O formed was then measured and compared to [5-3H]glucose and 3H2O standards. Glutamine in media was determined by LC-MS/MS (TandemGold, Zivak Technologies) using an amino acid assay kit (#ZV-3002-0200-10) following manufac- turer’s instructions. Glutamine consumption was calculated by subtracting glutamine remaining in media from glutamine con- centration in fresh media (2 mmol/ml).

2.7. Cell proliferation and viability assays

Trypan blue-excluding cells were counted under an inverted microscope (Accu-Scope, China) using a standard hemocytometer to determine the total number of viable cells. Viability assays after small molecule inhibitor treatments were performed using crystal violet [17] and sulforhodamine B [18] assays.

2.8. Colony formation assay

One thousand cells were seeded in 12-well plates, and inhibitors were added the following day, as indicated in figure legends. Cells were allowed to grow for 10 days. Cells were fixated with methanol in —20 ◦C for 10 min and stained with 0.2 % crystal violet.

2.9. Statistical analyses

All experiments were performed three times. Data are expressed as the mean ± s.d. of triplicate (duplicate for colony formation assays and F2,6BP) measurements of a single experiment. Statistical significance was assessed by the two-tailed t-tests and p < 0.05 was considered significant.

3. Results and discussion

3.1. KRAS transformation increases PFKFB3 expression and F2,6BP levels

We utilized the previously described immortalized human pancreatic nestin expressing cells without/with a mutant activating KRASG12D (HPNE vs. HPNE/KRAS) [11], in order to determine the specific cellular consequence of KRASG12D transformation on PFKFB3 expression and activity. Wang et al. [10] recently showed that KRASG12D-driven transformation of HPNE increased PFKFB3 protein stability by inducing p38g-mediated phosphorylation. However, it remains to be determined whether KRASG12D-driven transformation affects transcription of PFKFB3 and other PFKFB isoforms and, importantly, whether these changes affect steady- state F2,6BP levels. We believe that determining F2,6BP levels is important, as it can validate the use of inhibitors of the kinase function of PFKFBs. To determine the effect of KRASG12D trans- duction on mRNA expression levels of PFKFB isoforms expressed in tumor cells, real-time qPCR analysis was conducted in exponen- tially growing HPNE and HPNE/KRAS cells and found that PFKFB3 mRNA levels were higher and PFKFB2 mRNA levels were lower in HPNE/KRAS cells, as compared with HPNE cells.

In line with the qPCR data, PFKFB3 protein expression was elevated and PFKFB2 protein expression was diminished in HPNE/KRAS cells relative to HPNE cells, as assessed by Western blot (Fig. 1b). To determine whether high PFKFB3 expression in HPNE/KRAS cells remains dependent on KRAS signaling, the cells were treated with an established inhibitor of KRAS, SAH-SOS1A [19], and 24 h later, PFKFB3 protein expression was analyzed by Western blot. As illustrated in Fig. 1c, the inhibition of KRAS in HPNE/KRAS cells reduced the PFKFB3 protein to a level comparable with HPNE cells, suggesting that the elevated PFKFB3 expression in HPNE/KRAS cells is linked to an active KRAS signaling. We confirmed the inhibition of KRAS signaling by analyzing a phosphorylated form of ERK (phospho (T202/Y204)-ERK1/2) (Fig. 1c), a well-known effector kinase of RAS signaling. We next analyzed steady-state F2,6BP levels in these cells and found that KRASG12D-driven transformation led to an approximately 70 % increase in F2,6BP levels (Fig. 1d). Among the PFKFB isozymes, PFKFB3 displays the highest relative kinase activity (ki- nase/bisphosphatase activity ratioz760) [20]. Further increase in F2,6BP levels by the profound increase in PFKFB3 protein expres- sion may have been offset by a decrease in the PFKFB2 protein in HPNE/KRAS cells (Fig. 1b). Regardless, data obtained support the conclusion that KRASG12D transduction induces PFKFB3 expression to increase F2,6BP production in HPNE cells.

3.2. PFKFB3 silencing reduces F2,6BP levels, glucose uptake, glycolytic activity and proliferation induced by mutant KRAS

Stimulation of the glycolytic activity and proliferative capacity of HPNE cells by mutant KRAS has been reported [21]. However, it remains unknown whether PFKFB3 has a role in KRASG12D -medi- ated induction of the glycolytic activity and growth of HPNE cells. We first confirmed that HPNE/KRAS cells displayed higher glucose uptake, glycolytic activity and proliferative phenotype relative to HPNE cells. Next, to determine the requirement of PFKFB3 for KRASG12D -induced metabolic and proliferative phenotype, we transfected HPNE and HPNE/KRAS cells with PFKFB3- specific siRNA molecules, and 48 h later, measured F2,6BP levels, glucose uptake, glycolytic activity, and proliferation. Real-time qPCR analysis validated the efficacy of used siRNAs in depleting PFKFB3 mRNA levels, and consistent with the higher basal PFKFB3 mRNA levels in HPNE/KRAS cells, PFKFB3 expression was reduced to a greater extent in HPNE/KRAS cells relative to HPNE cells with siRNA silencing (Fig. 2d). We also confirmed the specificity of siR- NAs used for the PFKFB3 isozyme relative to other isozymes by real- time qPCR. Both PFKFB3 siRNAs led to a greater decrease in F2,6BP levels in HPNE/KRAS cells than HPNE cells, and reduced glucose uptake, glycolytic activity, and proliferation only in HPNE/KRAS cells (Fig. 2feh), suggesting KRAS renders HPNE cells dependent on PFKFB3 to increase and maintain glycolytic activity and proliferation. These findings not only confirm PFKFB3's stimulatory role in glycolysis, as well as in proliferation [22], but also ties PFKFB3 to a transformative phenotype induced by the most frequently mutated RAS isoform in human cancers.

3.3. Combinatorial targeting of PFKFB3 and GLS1 activites selectively kills KRASG12D-transformed HPNE cells and PDAC cells

GLS1 is a validated therapeutic target [5] and a clinical grade inhibitor of GLS1, CB-839, is currently being evaluated in clinical trials in many types of cancers, including mutant KRAS-harboring non-small cell lung cancer and colorectal cancer (clinical- trial.gov# NCT03965845). However, CB-839 is not currently being pursued as a potential anti-cancer agent in PDAC patients, pre- sumably because of the relative insensitivity and adaptability of PDAC cells to GLS1 inhibition and glutamine withdrawal for sur- vival and proliferation [5]. Given the association of KRAS signaling with enhanced glycolysis and the glycolytic regulator PFKFB3, we investigated whether abolishing PFKFB3 activity would augment the anti-proliferative effect of GLS1 inhibition by CB-839 in KRAS- transformed pancreatic cells. To curtail PFKFB3 kinase activity, we used a specific inhibitor recently developed by AstraZeneca called AZ PFKFB3 26 (hereafter denoted as AZP3), which displays a favorable pharmacokinetic and pharmacodynamic profile in pre- clinical mouse models [23].

HPNE and HPNE/KRAS cells were treated with vehicle (DMSO), AZP3 (1, 2, and 3 mM), CB-839 (5 mM), or, with combinations of both, and 72 h later, the effect on cell growth was assessed by sulforhodamine B assay. While AZP3 and CB-839 individually caused less than 50 % decrease in cell growths in both cells, the combination of CB-839 and AZP3 led to a potent, dose-dependent decrease in HPNE/KRAS growth, reaching a near complete cytotoxicity with the 3 mM AZP3 combination (Fig. 3a). This combination also suppressed HPNE growth, however the effect was significantly less pronounced (Fig. 3a). Similar results were obtained with crystal violet assay (Supplemental Fig. 2). We then conducted siRNA-mediated silencing experiments to rule out the possibility that the phenotype obtained with the inhibitors could be due to potential off-target effects. Consistent with the results obtained with pharmacological inhibitors, siRNA-mediated co- silencing of PFKFB3 and GLS1 mRNAs led to a greater anti- proliferative effect in HPNE/KRAS cells than in HPNE cells (Fig. 3b), although the growth suppressive effect of the siRNA approach was less drastic than the pharmacological inhibitors, which may be due to insufficient silencing of existing mRNA iso- forms. Taken together, these findings suggest that mutant KRAS- mediated metabolic rewiring increases the dependency of HPNE cells on GLS1 and PFKFB3 for survival and proliferation.

Although mutant KRAS is a key driver of metabolic changes in pancreatic tumor cells, additional genetic/non-genetic alterations incurred during the progression of pancreatic tumorigenesis may impart insensitivity to GLS1/PFKFB3 blockade in PDAC cells. To investigate this, we utilized the PDAC cell line model PANC-1 harboring the KRASG12D allele and found that the co-treatment of PANC-1 cells with AZP3 (3, 10 and 20 mM) and CB-839 (10 mM) exhibited a much stronger anti-proliferative effect than either treatment alone (Fig. 3c). We then performed colony formation assays with PANC-1 cells, with half the maximal AZP3 and CB-839 concentrations used in viability assays (10 mM AZP3 and 5 mM CB- 839) and found that the combination completely suppressed PANC-1 growth as single colonies (Fig. 3d). Importantly, the CB- 839/AZP3 combination did not exhibit a greater anti-proliferative effect than either inhibitor alone in the PDAC cell line BxPC3, which does not have a KRAS activating mutation. We also examined whether AZP3 displays a similar cu- mulative effect as combination with gemcitabine and paclitaxel, which are commonly used chemotherapeutics in PDAC management, and found no superior efficacy of combinations over indi- vidual treatments in PANC-1 cells (Supplemental Fig. 4). Taken together, these data suggest that the KRASG12D-induced metabolic vulnerability is preserved in PANC-1 cells and a GLS1/PFKFB3 blockade may be an effective strategy to prevent the growth of PDAC cells with KRAS mutation. Although this treatment strategy should be tested in more cell line models with and without KRAS mutation before drawing any far-reaching conclusions with respect to predicting tumor response to CB-839/AZP3 combination, our findings reveal a novel metabolic vulnerability that may be utilized in the management of PDAC and potentially other cancers with KRAS mutation and warrant the test of this approach in various in vitro and in vivo settings.

3.4. GLS1 inhibition reduces F2,6BP levels whereas PFKFB3 inhibition increases glutamine consumption in KRAS-transformed cells

To gain a mechanistic insight into the cooperation of PFKFB3 and GLS1 in the growth of KRAS-transformed pancreatic cells, we analyzed PFKFB3, GLS1 protein levels, and measured glutamine uptake and intracellular F2,6BP levels in HPNE, HPNE/KRAS and PANC-1 cells treated with DMSO, AZP3 (3 mM for HPNE, 20 mM for PANC-1), CB-839 (5 mM for HPNE, 10 mM for PANC-1), or with the combination of both. We found that, intriguingly, while both AZP3 and CB-839 strikingly reduced PFKFB3 protein levels in HPNE/KRAS cells as compared with HPNE cells and increased PFKFB3 protein levels in PANC-1 cells, AZP3 led to a near complete loss of the GLS1 protein in both HPNE and HPNE/KRAS cells, and CB-839 markedly diminished GLS1 protein levels only in HPNE/KRAS cells (Fig. 3e). Notably, in PANC-1 cells, while the level of GLS1 protein expression was unaffected by either AZP3 or CB-839 treatments, it was significantly reduced by the CB-839/AZP3 co-treatment (Fig. 3e). These data indicate that not only do AZP3 and CB-839 affect levels of their target proteins but also, importantly, they reciprocally regulate each other's target levels depending on the context, although further studies are needed to mechanistically account for the observed effects.

While AZP3 treatment resulted in a differential decrease in F2,6BP levels in HPNE/KRAS cells compared with HPNE cells, which is consistent with the induction of PFKFB3 expression upon mutant KRAS transduction (Fig. 1b), CB-839 also led to a comparable level of decrease in F2,6BP levels both in HPNE and HPNE/KRAS cells (Fig. 3f), which is in line with the effect of CB-839 on PFKFB3 protein levels (Fig. 3e). Notably, CB-839/AZP3 co-treatment further reduced F2,6BP levels only in HPNE/KRAS cells, which is consistent with the selective cytotoxicity of CB-839/AZP3 combinatorial treatment toward HPNE/KRAS cells relative to HPNE cells. Similar reduction in F2,6BP levels with the combination treatment as compared with either agent alone was obtained in PANC-1 cells (Fig. 3f). We speculate that further reduction in F2,6BP levels upon combinatorial treatment may account for the observed cytotoxicity in HPNE/KRAS and PANC-1 cells, supporting the notion that KRAS- associated oncogenicity may introduce a metabolic vulnerability that can be effectively targeted by CB-839 and AZP3. Interestingly, while GLS1 inhibition by CB-839 had no effect on glutamine con- sumption by HPNE/KRAS cells and had a moderate effect in PANC-1 cells, PFKFB3 inhibition by AZP3 strongly induced glutamine consumption in HPNE/KRAS and PANC-1 cells (Fig. 3g). Importantly, the CB-839/AZP3 combination led to a further increase in gluta- mine consumption. Taken together, these data suggest that PFKFB3 and GLS1 reciprocally regulate each other’s level/activity and that coordinated activities of glycolysis and glutamine metabolism via PFKFB3 and GLS1 may be selectively essential to the survival of KRAS-transformed pancreatic cells. It remains to be determined, however, if the increase in glutamine consumption by PFKFB3 inhibition is a compensatory response to a block in glutamine utilization by GLS1.

Although the association of glutamine availability and PFKFB3 activity has been previously reported [24], the current study, to our knowledge, is first to demonstrate that PFKFB3 inhibition increases glutamine consumption whereas GLS1 inhibition reduces F2,6BP production, and that the simultaneous inhibition of GLS1 and PFKFB3 activities results in a robust cytotoxicity in a transformed cell model. Rationale for exploiting PFKFB3 as a metabolic target in the context of HRas activation has been previously provided [9]. Not only do our findings further emphasize PFKFB3 as a viable target in the context of a highly relevant human KRAS cell model but also suggest that targeting PFKFB3 may be an effective approach to overcome the adaptation to GLS1 inhibition in PDAC cells with mutant KRAS [5].