Upregulation of cytosolic phosphoenolpyruvate carboxykinase is a critical metabolic event in melanoma cells that repopulate tumors

Introduction

Stem cell-like cancer cells (SCLCC) are a self-renewing, highly tumorigenic subpopulation of cancer cells, playing crucial roles in initiation, promotion, and progression of tumorigenesis. In vivo imaging of unperturbed tumors has visualized SCLCCs in mouse and zebrafish models (1, 2). Further experimental evidences demonstrate the existence of SCLCCs in murine brain, skin, and intestinal tumors (3–5). These tumorigenic SCLCCs are of considerable clinical importance, because they are resistant to cytotoxic therapeutics and most likely responsible for treatment failure and cancer recurrence. Meanwhile, a poor prognosis in cancer patients has been reported to link to SCLCCs (6), emphasizing the importance of targeting SCLCCs in tumor treatment. To achieve the goal of better targeting SCLCCs their intrinsic, especially metabolic features need to be better elucidated and explained. This is because that reprogrammed energy metabolism is fundamental for cancer cell growth, survival, differentiation and migration (7). Currently, the identification of SCLCCs through conventional methods that depend on cell surface markers often lacks specificity and is thus unreliable (8). Recently, we developed a mechanical method to select and grow SCLCCs from the general population of tumor cells by culturing single tumor cells in 3D soft fibrin gels, and found that as few as 10 selected cells are sufficient to grow tumors in immunocompetent mice (9). We thus functionally define these soft fibrin gel-selected cells as tumor-repopulating cells (TRC) (9, 10). Using this method to generate TRCs, here we identify that cytosolic phosphoenolpyruvate carboxykinase (PEPCK-C), also known as PCK1, is a novel metabolic feature of melanoma TRCs.

Materials and Methods

Results and Discussion

Hepatocytes abundantly express phosphoenolpyruvate carboxylkinase cytosolic forms (PCK1), to produce glucose via the gluconeogenic pathway. However, malignant hepatocytes lose gluconeogenesis by striking downregulation of PCK1 in favor of tumor glycolysis (14). Stem cell-like cancer cells have been shown to rely on glycolysis for glucose metabolism (15, 16), which promoted us to speculate a further downregulation of PCK1 in TRCs. Surprisingly, hepatic TRCs were found to upregulate PCK1 expression when compared with control H22 cells cultured in rigid plate (Fig. 1A and 1B). In addition to H22 tumor cells, B16 melanoma cells and EL4 lymphoma cells were examined and both B16 and EL4 TRCs upregulated PCK1 (Fig. 1A and 1B). CD133 is commonly used to mark SCLCC including melanoma (17). We found that PCK1 expression was much higher in CD133⁺ B16 cells, compared to CD133⁻ B16 cells (Supplementary Fig. S2A). PCK1 was also found to be highly expressed in mouse embryonic stem cells (mESCs) as well as mouse mesenchymal stem cells (mMSCs) (Fig. 1A), indicating that the upregulation of PCK1 expression might be a unique feature of undifferentiated cells. In addition, using PCK1 promoter-EGFP constructs further confirmed the expression of PCK1 in cultured B16 TRC spheroids (Fig. 1C). However, after being placed back to rigid plates to induce TRC differentiation (10), these spheroid cells gradually lost PCK1 expression (Fig. 1C), further suggesting that the expression of PCK1 depends on the undifferentiated status of cells. Consistently, the PCK1 protein expression in TRCs and stem cells was higher than that in controls on rigid plastic (Fig. 1D). Moreover, we isolated single melanoma tumor cells from fresh melanoma tissues of two patients, respectively, and seeded them into soft 3D fibrin gels. In both cases, we found that the expression of PCK1 was upregulated in human primary melanoma TRCs (Supplementary Fig. S2B and 2C). Meanwhile, nine paraffin-embedded melanoma samples were analyzed by immunohistochemical staining. We found that PCK1 was highly expressed in three samples but lowly expressed in the left samples (Supplementary Fig. S2D). However, whether the expression levels of PCK1 are correlated to disease progression needs further study. Together, these data suggested that PCK1 might be an unusual metabolic marker for tumorigenic cells.

Given that PCK1 is the key enzyme in catalyzing the rate-limiting step in gluconeogenesis, a further test was conducted to see whether tumorigenic cells used PCK1 to carry out gluconeogenesis. Fructose-1,6-bisphosphatase and glucose-6-phosphatase (G6Pase), two enzymes to convert fructose 1,6-bisphosphate to fructose and 6-phosphate glucose-6-phosphate to glucose, respectively, were not expressed by those TRC cells (Supplementary Fig. S3A), suggesting that PCK1 cannot be used for gluconeogenesis in TRCs. Cell growth and proliferation are fed through glucose metabolism. Here, the rates of glucose consumption were found to be much higher in B16 TRCs, compared to the corresponding bulk tumor cells (Fig. 2A left). However, knockdown of PCK1 expression by siRNA resulted in the decrease of glucose consumption in B16 TRCs (Fig. 2A right), suggesting that PCK1 promote the use of glucose in TRCs. Carbon flow from glucose to lactate facilitates glucose consumption. Here, B16 TRCs exhibited higher lactate release rates compared with control, and the knockdown of PCK1 decreased their release of lactate (Fig. 2B). Although PCK1 could not mediate gluconeogenesis in TRC cells, whether PCK1 induced a retrograde carbon flow in TRC cells was unclear. Here, we tested the biosynthesis of serine, considering its derivation from glycerate-3-phosphate. We found that the knockdown of PCK1 resulted in slight decrease of serine levels in B16 TRCs (Fig. 2C left). However, glycine, the metabolite of serine, was significantly decreased after PCK1 knockdown (Fig. 2C right). In parallel, we found the levels of glycerol- 3-phosphate, a production of glyceroneogenesis, were increased in TRCs, compared to bulk B16 cells (Fig. 2D left). Similarly, knockdown of PCK1 resulted in the decrease of G-3-P in TRCs (Fig. 2D right). Together, these data suggest that the expression of PCK1 enhances the carbon flow from glucose to lactate as well as the flow toward glycerol and serine in TRCs.

Tumor cells consistently express Glucose transporter 1(GLUT1) that initiate glucose metabolism. Here, we additionally analyzed the expression of GLUT1, and found that GLUT1 was decently expressed in TRCs (Supplementary Fig. S3B). PCK1 knockdown did not affect its mRNA expression and cellular membrane localization (Supplementary Fig. S3C). Besides, other enzymes involving glycolysis, Krebs cycle and lipid metabolism, including hexokinase 2, phosphofructokinase, pyruvate kinase isozyme type M2, lactate dehydrogenase, pyruvate Dehydrogenase alpha, citrate synthase, succinate dehydrogenase, ATP citrate-lyase and fatty acid synthase were analyzed. However, we did not find the differential expression between TRCs and control tumor cells (Supplementary Fig. S4). In addition, knockdown of PCK1 seemed not to affect the expression of those enzymes (Supplementary Fig. S4). Thus, PCK1 enhancing glucose carbon flow in TRCs seems not to be attributable to increasing glucose taking up or regulating enzyme expression.

To further investigate the role of the above unexpected upregulation of PCK1 in TRCs, we treated the 3D soft fibrin gel-cultured TRCs with 3-MPA, an inhibitor of PCK1, to inhibit the enzymatic activity of PCK1. We found that the inhibition of PCK1 enzymatic activity resulted in decreases in the size and number of spheroids of TRCs (Supplementary Fig. S5A). Moreover, we harvested the second generation of B16 TRCs by culturing the first generation of TRCs in another 3D soft fibrin gel to generate purer TRCs, and transfected them with PCK1 siRNA. Consistently, the growth of B16 TRCs in the soft gels was significantly inhibited by PCK1 knockdown (Fig. 3A). We did not observe that PCK1 knockdown affected B16 TRC apoptosis (Supplementary Fig. S5B). Interestingly, when we added glycine and glycerol-3- phosphate to PCK1 siRNA-transfected TRC culture system, respectively, we found the impaired growth of TRCs could be partially rescued (Supplementary Fig. S5C). On the other hand, when we forcedly overexpressed PCK1 (PCK1-OE) in TRCs, the increased growth was found in soft 3D fibrin gels, compared to the control TRCs (Fig. 3B). These data together suggest that PCK1 plays an important role in promoting tumorigenic TRC growth.

To further dissect the role of PCK1 in TRCs, we cultured B16 TRCs in 2D rigid dishes to induce TRC differentiation. Consistent with the previous result of Fig. 1C, the expression of PCK1 was gradually down-regulated along time (Supplementary Fig. S6A). Given tumorigenic cells with a small number that can repopulate the whole tumor in vivo, we speculated that targeting tumorigenic cells might cause shrinkage of tumor cell population in culture. When culturing bulk tumor cells in 2D rigid dish in the presence or absence of PCK1 inhibitor 3-MPA, we found that although 3-MPA under given concentration did not affect tumor cell proliferation, the number of bulk tumor cells was indeed significantly decreased along the culture time (Supplementary Fig. S6B and 6C), implying a potential critical role of PCK1 in B16 tumorigenic cells. Moreover, when we put the same number 3-MPA-treated tumor cells back to 3D soft fibrin gels, the spheroid formation was significantly increased, relative to the control (Fig. 3C). This unexpected result might be due to the inhibition of PCK1 activity that also inhibited TRC differentiation, resulting in more TRC enrichment in the population. In addition, we constructed PCK1 siRNA-expressing vectors. When B16 cells were transfected with these plasmids, tumor cell clone formation was not observed in the presence of selection drugs. However control siRNA-transfected tumor cells easily formed clones, further supporting that PCK1 is required for maintaining tumorigenic cells. To validate the above data in vivo, PCK1 siRNA-transfected B16 TRCs (5 × 10²) were i.v. or s.c. injected to mice. Three weeks later, most control siRNA-transfected TRCs formed tumor nodules in the lungs and grew much larger tumors in the skin (Fig. 3D). By contrast, PCK1 siRNA-transfected TRCs did not form lung tumor nodule and grew much smaller tumors in skin (Fig. 3D). Together, these data suggest that PCK1 confers B16 TRCs the ability to repopulate a tumor in vivo.

Finally, we investigated how PCK1 expression was regulated in TRCs in 3D soft fibrin gels. B16 tumor cells use αvβ3 integrin to sense extracellular mechanical force in 3D fibrin gels (9). Here, we found that blockade of αVβ3 integrin with the inhibitor downregulated PCK1 expression in TRCs in a dose-dependent manner; however, blockade of β1 integrin did not affect PCK1 expression (Fig. 4A). αvβ3 integrin engagement may trigger integrin-linked kinase (ILK)/extracellular signal-regulated kinase (ERK) signaling pathways (18). We found that the inhibition of ERK1/2 pathway by U0126 did not downregulate the expression of PCK1 (Fig. 4B). However, blockade of PI3K decreased the expression of PCK1 (Fig. 4B). Intriguingly, inhibition of AKT, the classical downstream signal molecule of PI3K, did not affect the expression of PCK1 (Fig. 4B). Thus, αVβ3 integrin engagement probably activates non-classical PI3K pathway to induce PCK1 expression in 3D soft fibrin gels. DNA methylation is an important event that regulates gene expression (19). The analysis of methylation of PCK1 promoter region showed high methylated sequences in TRCs and rigid dish-cultured B16 cells and no difference was found between them (Supplementary Fig. S7). Besides DNA methylation, histone methylation i.g., histone H3 lysine9 (H3K9) methylation is also very important in regulation of gene expression. G9a and SUV39h1 are two methyltransferases that mediate the methylation of H3K9 (20). Knockdown of G9a or SUV39h1 by siRNA resulted in the upregulation of PCK1 in TRCs (Fig. 4C), suggesting that H3K9 methylation downregulates PCK1 expression. To clarify whether PI3K signaling pathway is linked to H3K9 methylation so to regulate PCK1 expression, we additionally blocked PI3K pathway and detected the expression of G9a and SUV39h1. We found that PI3K inhibitor did not affect the expression of G9a, but caused the upregulation of SUV39h1 expression (Fig. 4D), suggesting a potential link between PI3K signaling pathway and H3K9 methylation. On the basis of the above data, we proposed that in 3D soft fibrin gels, TRCs use αVβ3 integrin to transduce extracellular mechanical signaling, leading to activating PI3K and subsequently influencing histone methylation, so to regulate the expression of PCK1.

In summary, in the present study we use biomechanical method to select and amplify melanoma tumorigenic cells (TRCs) to study their metabolic features. Our data show that the key gluconeogenic enzyme cytosolic phosphoenolpyruvate carboxykinase, by virtue of its differential expression in B16 TRCs, not only promotes TRC glycolysis but also mediates the retrograde carbon flow to serine and glycerol-3-phosphate, thus conferring TRCs with the ability to repopulate a tumor. We describe this novel metabolic pathway in a schematic (Supplementary Fig. S8). All in all, these findings disclose that upregulation of PCK1 is a critical metabolic feature of tumorigenic TRCs, thus providing a potential target for melanoma treatment.

Supplementary Material