Long-term treatment with EGFR-TKIs induces acquired resistance with downregulation of EGFR signaling in lung cancer cells
Gefitinib-resistant NSCLC HCC827 (HCC827 GR) cells were established via long-term exposure of the cells to stepwise escalation of gefitinib. HCC827 GR cells exhibited resistance to gefitinib at a level similar to that in H1975 cells (T790M mutation). Other lung cancer cell lines, H1993 (NSCLC) and H292 (mucoepidermoid lung cancer) cells, also exhibited resistance to erlotinib after long-term treatment (H1993 ER and H292 ER, respectively) (Figure 1A). Osimertinib, a third-generation EGFR-TKI, strongly inhibited H1975 cell proliferation but marginally affected cell proliferation in the lung cancer cell lines with acquired EGFR-TKI resistance (HCC827 GR, H1993 ER, and H292 ER) (Figure1B). Because various mutations in the EGFR tyrosine kinase domain associated with gefitinib resistance have been reported (6), DNA sequencing of EGFR exons 18 to 21 was performed for HCC827, HCC827 GR, H1975, H1993, H1993 ER, H292, and H292 ER cells. An exon 19 deletion mutation and inherited gefitinib resistance, T790M, were found in HCC827 and H1975 cells, respectively (Figure 1C). However, mutations were not detected in all the lung cancer cells with acquired EGFR-TKI resistance tested (Figure 1C). Furthermore, EGF-induced phosphorylation of EGFR (Tyr1068) and its downstream kinases, phosphorylation of AKT (Ser473) or p44/p42 MAPK (Thr202/Tyr204), was not observed or was weaker in the three acquired-resistance cells compared with their parental cell lines. In addition, the EGFR tyrosine kinase-inhibiting effects of gefitinib, erlotinib, and osimertinib were weaker (Figure 1D). Conversely, EGFR phosphorylation and its downstream signals were successfully suppressed by osimertinib but not by gefitinib or erlotinib in H1975 cells (T790M)(Figure 1D). The results showed that the acquisition of EGFR-TKI resistance by lung cancer cells led to relatively minimal effects on EGFR activity, and these cells subsequently exhibited decreased sensitivity to EGFR-TKIs, even osimertinib.
Glycolysis activity is decreased in lung cancer cells with acquired EGFR-TKI resistance
EGFR signaling promotes aerobic glycolysis in triple-negative breast cancer (TNBC) (7), and the inhibition of EGFR reverses the Warburg effect and reactivates oxidative phosphorylation (OXPHOS) in NSCLC cells (8). Because EGFR signaling is suppressed in the three acquired-resistance lung cancer cell lines (HCC827 GR, H1993 ER, and H292 ER), we examined glucose metabolism in these cancer cells. The 2-DG uptake assay showed that glucose uptake in all three resistant cancer cell lines was decreased (Figure 2A), and the glucose consumption assay showed that the amount of glucose remaining in the culture medium was higher in HCC827 GR cells compared with HCC827 parental cells (Figure 2B). Using the XFp analyzer with a glycolysis stress kit, changes in the extracellular acidification rate (ECAR) representing glycolysis were estimated. After exposed to the same amount of glucose (10 mM), HCC827 GR and H1993 ER cells showed less changes compared with their parental cells (Figure 2C). Furthermore, expression levels of enzymes involved in glycolysis (HK1, HK2, and GAPDH) were lower in the two resistant cell lines (HCC827 GR and H1993 ER) (Figure 2D), whereas the dimer (inactive) form of Tyr105-phosphorylated PKM2 was slightly reduced in HCC827 GR and H1993 ER cells (Figure 2D). Notably, H292 ER cells lacked both HK1 and HK2, first-step enzymes in glycolysis, and the Tyr105-phosphorylated PKM2 level was upregulated (Figure 2D), indicating that glycolysis dysfunction in the H292 ER cells was mainly due to HK1/2 defects. The mRNA expression associated with the glycolysis gene set was further analyzed using RNA sequencing for both HCC827 and HCC827 GR cells. However, mRNA expression patterns associated with glycolysis in the two cell types did not exhibit any specific trends (Figure 2E). Although metabolic differences were not clearly identified at the mRNA or protein level in glycolytic enzymes, the metabolic data indicate that glycolysis capacity was decreased in the lung cancer cells with acquired EGFR-TKI resistance.
Reactivation of mitochondrial OXPHOS function in lung cancer cells with acquired EGFR-TKI resistance
Because glycolysis is the main energy process using glucose in rapidly growing cancers, glycolysis is considered a hallmark of cancer (Warburg effect) (9). However, the role of mitochondrial OXPHOS in cancer progression has also been studied (4,5,10,11). Because EGFR inhibition induces the re activation of mitochondrial OXPHOS in NSCLC cells (8), and the lung cancer cells with acquired EGFR-TKI resistance exhibited decreased EGFR kinase signaling activities (Figure 1), OXPHOS activities in the three lung cancer cell lines with acquired EGFR-TKI resistance were compared with those in their parental cell lines. Using the XFp analyzer with a cell mito-stress test kit, oxygen consumption rate (OCR) changes were monitored after treatment with OXPHOS modulators (1.5 mM oligomycin, 0.5 mM trifluoromethoxy carbonylcyanide phenylhydrazone (FCCP), and a mixture of 0.5 mM rotenone and 0.5 mM antimycin A). ATP-linked mitochondrial respiration increased in HCC827 GR and H1993 ER cells compared with their parental cells (Figure 3A). In H292 ER cells, maximal respiration in mitochondria increased but ATP-associated respiration did not change significantly compared with the parental cells (Figure 3A). Then, the ATP production ratio was compared between glycolysis and OXPHOS using the XFp analyzer with an ATP real-time rate assay kit. ATP production in the HCC827 cells was highly dependent on glycolysis (88.75%), but not on OXPHOS (11.25%). However, HCC827 GR cells had more ATP production that depended on mitochondrial OXPHOS (27.14%) than did HCC827 cells (Figure 3B). Mitochondrial membrane potential is induced by a proton pump in OXPHOS and regarded as an essential component during mitochondrial ATP production. Mitochondrial membrane potential in HCC827 and HCC827 GR cells was analyzed via staining with tetramethylrhodamine methyl ester (TMRM, red dots). As shown in Figure 3C, total integrated red fluorescence intensity versus cell confluence was significantly higher in HCC827 GR cells. Conversely, the basal intracellular level of lactate, the end product of glycolysis, was significantly decreased in HCC827 GR cells compared with HCC827 cells (Figure 3D). Inhibition of OXPHOS can cause a compensative increase in glycolysis resulting in the conversion of pyruvate to lactate. The lactate level increased in phenformin (OXPHOS inhibitor)-treated HCC827 GR cells; however, a difference in HCC827 cells was not detected (Figure 3D). These results indicate that mitochondrial OXPHOS activity was reactivated in the NSCLC cell lines with acquired EGFR-TKI resistance (HCC827 GR and H1993 ER). Next, we assessed morphological changes in the mitochondria as well as the protein and mRNA expression of OXPHOS subunits. Transmission electron microscopy images showed no differences in the size and number of mitochondria in both HCC827 and HCC827 GR cells (Figure 3E). The expression of OXPHOS subunit proteins (complex 2, 3, 5) was slightly increased in HCC827 GR cells compared with HCC827 cells; however, expression in H1993 ER and H292 ER cells was not significantly different from that in their parental cells (Figure 3F). The distribution of expression levels of different mRNAs associated with OXPHOS subunits were viewed as a MA plot. Significant fold-changes between HCC827 and HCC827 GR cells were not found (Figure 3G). The data indicated that increased OXPHOS capacity in EGFR-TKI-resistant NSCLC is associated with enhanced mitochondrial function but not with changes in related gene expression.
Inhibition of proliferation in lung cancer cells with acquired EGFR-TKI resistance is caused by phenformin
Biguanides, the most prescribed anti-diabetic agents, have been recognized for their anti-cancer effects, and many clinical trials are currently in progress (12–15). The repositioning of biguanides as anti-cancer agents is attracting much attention due to the cost benefit and minimal safety issues. In a recent study, the combination of osimertinib with phenformin delayed osimertinib resistance in a preclinical NSCLC model (16). We hypothesized that phenformin, an OXPHOS complex 1 inhibitor, would selectively inhibit the proliferation of lung cancer cells with acquired EGFR-TKI resistance that mainly employ mitochondrial OXPHOS. A prototype biguanide, phenformin, more effectively inhibited the proliferation of HCC827 GR and H1993 ER cells than their parental cell lines (Figure 4A). However, the growth-inhibiting effect of phenformin was not enhanced in H292 ER cells compared with H292 cells (Supplemental Figure 1A), which is consistent with the data showing no difference in ATP-associated OCR change (Figure 3A). Next, the potential mechanism by which phenformin inhibits the proliferation of EGFR-TKI-resistant NSCLC cells was assessed. Because OXPHOS is required for aspartate biosynthesis in proliferating cells (17,18), intracellular aspartate level was measured after exposing HCC827 and HCC827 GR cells to phenformin. The intracellular aspartate level decreased to a greater extent in HCC827 GR cells than in HCC827 cells after exposure to phenformin (Figure 4B). Inhibition of aspartate biosynthesis occurs due to an imbalance in the NAD+/NADH ratio. When the cellular NAD+/NADH ratio was measured, phenformin-mediated NAD+/NADH imbalance was only observed in HCC827 GR cells (Figure 4C). a-Ketobutyrate (AKB) is a representative electron acceptor that participates in regenerating NAD+ (17). The addition of both AKB and aspartate partially alleviated the growth-inhibiting effects of phenformin in HCC827 GR cells (Figure 4D). To assess if mitochondrial OXPHOS complex I is a target of phenformin, the effects of rotenone, a potent OXPHOS complex 1 inhibitor, were investigated. As expected, rotenone exerted strong growth-inhibiting effects in HCC827 GR cells, and its anti-proliferative effect was reversed with AKB treatment (Figure 4E). To evaluate the anti-cancer effects of phenformin in vivo, Balb/c nude mice were implanted with HCC827 or HCC827 GR cells. In xenografts inoculated with HCC827 GR cells, oral administration of phenformin (300 mg/kg/day) significantly reduced tumor growth derived from HCC827 GR cells. Conversely, phenformin administration did not significantly affect tumor growth in xenografts inoculated with HCC827 cells (Figure 4F and 4G).
Reversal of anti-cancer effects of phenformin is caused by glycolysis reactivation in lung cancer cells with acquired EGFR-TKI resistance
Because the reactivation of OXPHOS in most of the lung cancer cell lines with acquired EGFR-TKI resistance was observed, we hypothesized that phenformin sensitivity could be diminished by the restoration of glycolysis (Warburg effect) in the resistant cell types. Hexokinase (HK) is a first-step glycolytic enzyme that converts glucose into glucose 6-phosphate. Among the five HK isoforms, HK2 is highly expressed and functions as the predominant form in cancer cells (19). Because HK2 and HK1 expression was absent in H292 ER cells (Figure 2D), we hypothesized that HK enzyme deficiency is a key event for the metabolic shift to OXPHOS in H292 ER cells. HK2-overexpressing H292 ER cells (H292 ER-HK2) were established by transfection with a flag-tagged human HK2 overexpression vector (Figure 5A). XFp analysis with a glycolysis stress test kit showed that ECAR changes representing glycolysis and its capacity were greater in H292 ER-HK2 cells compared with mock-transfected cells (Figure 5B). The ATP production ratio from glycolysis also increased from 63.55% to 83.22% (Figure 5C). Furthermore, the inhibitory effects of phenformin on the proliferation of H292 ER cells were significantly decreased by HK2 overexpression, indicating that enhanced glycolysis reduces the anti-cancer effects of phenformin in lung cancer cells with acquired EGFR-TKI resistance (Figure 5D).