Paclitaxel Resistance in Ovarian Cancers Relies on a PGAM1 Mediated Glycolytic Metabolism


 Background: Paclitaxel, as an anti-microtubule drug, has been recommended to be the first-line chemotherapy agent for ovarian cancer for about two decades. However, most patients with advanced stage relapse within two years and ultimately die of relapsed cancer displaying increased chemoresistance. The mechanisms underlying paclitaxel resistance remain ncompletely understood.Methods: The paclitaxel-resistant ovarian cancer cell line SKOV3-TR30 and its parental cell line SKOV3 were analyzed by proteinomics. The glycolytic enzyme PGAM1-modulated glycolytic flux including pyruvic acid production and lactic acid production, and mitochondrial function was investigated using the assays kit. The correlations between PGAM1 expression, overall survival and progression free survival were evaluated for ovarian cancer patients. To elucidate the underlying mechanisms involved in paclitaxel resistance, PGAM1 gene knockout, gene overexpression and/or manipulation of glycolytic products were employed, followed by Western blot, immunostaining, and cell viability assay.Results: PGAM1 was highly expressed in the paclitaxel resistant ovarian cancer cell line SKOV3-TR30, as compared to its parental cell line SKOV3. A high expression of PGAM1 was ignificantly correlated with a reduced overall survival and progression free survival in ovarian cancer patients. Under paclitaxel exposure, SKOV3 cells showed a stronger mitochondrial function than SKOV3-TR30 cells, suggesting that under the stress, SKOV3-TR30 cells had responded with enhanced glycolysis rather than undergoing oxidative phosphorylation. Down-regulation of PGAM1 in SKOV3 TR30 cells resulted in decreased paclitaxel resistance. Up-regulation of PGAM1 in SKOV3 cells led to enhanced paclitaxel resistance. Analysis of the lycolytic flux revealed that PGAM1-mediated pyruvic acid or lactic acid production could control the capabilities of ovarian cancer cell resistance to paclitaxel.Conclusions: Our study demonstrated that PGAM1 could act as a modulator linked to paclitaxel sensitivity in ovarian cancer cells. Blocking PGAM1 may provide a new approach to reverse paclitaxel resistance in ovarian cancer patients.


Background
Ovarian cancer is one of the most common malignancies and has the highest mortality of any female reproductive system disease. According to American cancer statistics, the overall 5-year survival rate of ovarian cancer patients is maintained between 44% and 46%, but becomes only 27-29% for advanced stage patients (1)(2)(3)(4)(5)(6). A short-term complete response to first-line chemotherapy is achieved in nearly three quanters of patients with advanced disease. However, most patients relapse within two years and ultimately die of relapsed cancer displaying increased chemoresistance. Although abundant research has recently focused on new agents such as bevacizumab and Poly (adenosine diphosphate-ribose) polymerase (PARP) inhibitors (7)(8)(9), primary surgery followed by combined chemotherapy of paclitaxel and carboplatin still remains the standard treatment strategy (10,11). Paclitaxel, as an anti-microtubule drug, has been recommended to be the first-line chemotherapy agent for ovarian cancer for about two decades. In this way, overcoming paclitaxel resistance becomes a key goal to improve the prognosis of ovarian cancer patients.
To gain insight into the mechanisms by which paclitaxel resistance occurs in ovarian cancer we have made use of proteomic analysis between the paclitaxel-resistant ovarian cancer cell line, SKOV3-TR30, and its parental cell line, SKOV3 (12). In screening for differentially expressed proteins, Phosphoglycerate mutase 1 (PGAM1), a non rate-limiting enzyme in glycolysis (13), was identified to be highly expressed in the SKOV3-TR30 cell line. PGAM1 reversibly catalyzes a unique step during glycolysis, controlling the metabolite levels of its substrate, 3-phosphoglycerate (3-PG), and product, 2phosphoglycerate (2-PG), in the later stages of the glycolytic pathway (14)(15)(16). PGAM1 has been reported to be highly expressed in lung cancer (17), polymorphic gliomas (18), and it is positively correlated with poorer prognosis of patients with gliomas (19). 4 Higher PGAM1 activity has also been observed in the hepatocellular carcinoma cell line, HepG2, as compared to the normal immortalized human hepatocyte cell line L02 (20). It seems that PGAM1-mediated glycolysis plays a potent role in modulating cell growth and/or apoptosis.
Enhanced glycolysis occurs in most human cancer cells and is considered to be related to chemoresistance. In normal cells, glucose is catabolized into pyruvic acid, which then enters the tricarboxylic acid cycle for either oxidative phosphorylation or, less usually, is anaerobically converted into lactic acid. In cancer cells, aerobic glycolysis is preferred regardless of oxygen concentration (21,22). With enhanced glycolysis in cancer cells, more ATPs are generated. This is believed to be one of the mechanisms by which rapid growing cells withstand external stresses such as hypoxia and drug exposure. In this way, cancer cells exposed to drug treatments can escape extermination through a series of adaptive alterations, including those of the glycometabolism (23). In leukemia, for example, cancer cells residing at higher oxygen tensions still remain highly glycolytic comparing to most of the surrounding normal tissue cells (24,25). Zhang et al also recently reported that glycometabolic adaptation could mediate drug resistance in leukemia cells (26). The involvement of glycolysis and biosynthesis in the regulation of paclitaxel resistance in ovary cancer cells remain elusive.
In the present study, we show that paclitaxel resistance in ovarian cancers rely on a PGAM1 mediated glycolytic metabolism. Overexpression of PGAM1 in SKOV3 ovarian cancer cells leads to increased glycolytic flux and paclitaxel sensitivity, and PGAM1 knock down in SKOV3-TR30 cells results in decreased glycolytic flux and paclitaxel sensitivity. Importantly, we found that a high-expression of PGAM1 was correlated with poorer prognosis including PFS and OS, suggesting that PGAM1 acts as an oncoprotein linked to sensitivity in ovarian cancer patients. Blocking PGAM1 may provide a new approach to reverse paclitaxel resistance in ovarian cancer patients.

Cell culture
The human ovarian adenocarcinoma cell line SKOV3 was purchased from the American Type Culture Collection (ATCC, HTB-77) and maintained in McCoy's 5A medium (Corning, 10-050-CV) supplemented with 10% fetal bovine serum (FBS) at 37 ℃ and 5% CO2. The paclitaxel-resistant cell sub-line SKOV3-TR30 acquired paclitaxel resistance by being exposed to increased concentrations of paclitaxel (27).
SKOV3-TR30 was maintained in complete medium with 10nM paclitaxel daily and 30nM paclitaxel for 3 days at a 21-day-intervals to keep the paclitaxel resistance at 27fold over the parental cell line SKOV3. Paclitaxel resistance was detected every 3 months.

Reagents and antibodies
Paclitaxel was purchased from the Bristol-Myers Squibb Company. Pyruvic acid and lactic acid were purchased from Merck. The primary antibodies used for Western blotting were anti-PGAM1 (ab96622) and anti-beta-actin (Proteintech, 60008-1-Ig).
The antibody anti-PGAM1 (ab2220) was used for immunohistochemistry. HRP-labeled secondary anti-mouse (Proteintech, SA00001-1) and anti-rabbit (Proteintech, SA00001-2) were purchased from Proteintech Group Inc. University. Anti-PGAM1 antibody (ab2220, 1:150) was used for the IHC following the protocol provided by Abcam. The scoring method and analysis of IHC were performed as previously described.

Plasmids and siRNA transfection
The pCMV5(+)-PGAM1 plasmid was constructed as follows. A full length of PGAM1 mRNA (NM_002620.1) was synthesized and cloned into a pCMV5 vector by Genscript.
A pCMV5(+) vector was designated as negative control. Cells were grown to 70-80% confluence before plasmid transfection. The ratio of X-tremeGENE HP DNA

Cell viability detection
For cell viability detection for paclitaxel-sensitivity, cells were seeded at 4000 cells per well in 96-well plates after transfection with plasmid or siRNA. SKOV3 and SKOV3-TR30 cells were then exposed to paclitaxel at various final concentrations (0, 5, 10, 15,   20, 25, 30 nM for SKOV3 and 0, 50, 100, 150, 200, 250, 300 nM for SKOV3-TR30) for 48 and 72 hours. Each concentration was conducted in triplicate. Cell viability was then detected using a Cell Counting Kit-8 (Dojindo laboratories, CK04) using a Varioskan Flash microplate reader (Thermo Scientific) for spectral analysis at a wavelength of 450nm.

Pyruvic acid and lactic acid production detection
Cells were seeded at 4000 cells per well in 96-well plates and cultured for the appropriate time. Culture medium was then piped out to detect pyruvic acid (using pyruvic acid assay kit, sigma-aldrich, MAK071) or lactic acid (using lactic acid assay kit, sigma-aldrich, MAK064) concentrations according to the manufacturer's instructions. Cell viability was detected using a Cell Counting Kit-8 (CCK-8) as previously mentioned. The increased pyruvic acid concentration was divided by the absorbance of CCK-8. Separate experiments were done for three replicates and the P value was then calculated.

Immunofluorescence staining of functional mitochondria
SKOV3 and SKOV3-TR30 cells were seeded at 5×104 cells/ml into each well of μslide 8 well (ibiTreat, ibidi, 80826). The cells were exposed to 10nM, 20nM and 30nM paclitaxel for 24 hours after cell adherence. TBST was used for washing after 2% paraformaldehyde in PBS for fixation and permeabilization for 15 minutes. Functional mitochondria were stained by Mitobright for 10 minutes (Dojindo Molecular Technologies catalog number MT06). TBST was then used for washing again. Nuclei were stained using fluoroshield with DAPI (Abcam, ab104139). Representative images were photographed using a confocal fluorescence microscope.

Western blot
Proteins were extracted from whole-cell lysate. Appropriate amounts of proteins were 8 loaded on 12% SDS-PAGE and transferred to 0.22um PVDF membranes. The membranes were blocked with 5% Bovine Serum Albumin in TBST and incubated in primary antibodies at 4℃ overnight and then in secondary antibodies for 1 hour at room temperature. The bands were detected with an EZ-ECL kit (BI biological industries, 20-500-120) and scanned using Imagequant LAS400 mini (GE Healthcare).

Statistical analysis
All statistics were conducted using the SPSS version 20.0 statistical software package (IBM Corp, USA). The differences in proteomic analysis between SKOV3 and SKOV3-TR30 were estimated using a Student's t-test. The correlations between the IHC of PGAM1 protein and clinicopathologic parameters were evaluated using chisquare tests. PFS and OS curves were determined using the Kaplan-Meier method and differences in survival were compared using a log-rank test. The influence of PGAM1 expression on survival was assessed using the Cox regression analysis. Cell viability was evaluated using a Student's t-test. A P-value of less than 0.05 (two-tailed was considered as statistically significant.

Identification of PGAM1 differentially expressed between paclitaxel-sensitive and paclitaxel-resistant ovarian cancer cells
We previously obtained a SKOV3-TR30 cell line that had acquired paclitaxel resistance with a 27-fold increase over its parental SKOV3. We then carried out a MALDI-TOF-TOF following 2-D DIGE (two-dimensional fluorescence difference in gel electrophoresis) quantitative proteomic analysis between the two cell lines. In this mining we observed 57 differentially expressed protein gel spots between SKOV3/SKOV3-TR30 with ratios of more than 1.5-fold. 49 of these were then successfully authenticated. Figure 1a shows a partial image of one of the three replicate gels. DeCyder software analysis showed that the SKOV3/SKOV3-TR30 ratios ranged from -7.93-fold to 9.42-fold. Gel pots were then picked and digested by tryptan. Protein  .008), compared to the patients with PGAM1 low-expression (Fig. 2b) Cox regression analysis showed that the expression of PGAM1, together with FIGO stage, ascitic fluid volume, primary surgery, and chemosensitivity were significantly correlated with PFS, though not with OS (Table S1). These data suggest that the increased expression of PGAM1 protein might be used as a predictor of poorer prognosis in ovarian cancer patients.

SKOV3-TR30 cells manifest a larger glycolytic flux than SKOV3 cells
PGAM1 actively coordinates biosynthesis and glycolysis, in which many glycolytic intermediates including pyruvic acid and lactic acid are produced and are used as precursors for the anabolic pathway [19,20]. We therefore examined the glycolytic flux

PGAM1 modulates paclitaxel sensitivity in ovarian cancer cells through the alteration of pyruvic acid production
Given the higher expression of PGAM1 and elevated glycolytic flux in the SKOV3-TR30 cells than that in the parental SKOV3 cells, we considered whether paclitaxel 12 sensitivity is regulated by PGAM1 mediated glycolysis. Using Western blot, we firstly examined PGAM1 expression patterns in the parental SKOV3 cells exposed to different concentrations of paclitaxel. PGAM1 expression was up-regulated subsequent to an increased dosage of paclitaxel, with a peak at 20nM in SKOV3 cells (Figure 5a).
Beyond a paclitaxel concentration of 20nM, the cells quickly died. Secondly, we analyzed if the alteration of the glycolysis flux influences PGAM1 expression in the parental SKOV3 cells. Different amounts of pyruvic acid were added into the culture medium and PGAM1 expression was again measured using Western blot. Figure 5b showed that PGAM1 was up-regulated with additional pyruvic acid and achieved a peak at a concentration of 0.1mmol/L in SKOV3 cells. We then up-regulated PGAM1 by expressing pCMV5(+)-PGAM1 plasmid in the SKOV3 cells (Figure 5c), where an increased pyruvic acid production ( Figure 5d) and a decreased paclitaxel sensitivity was detected (Figure 5e and Figure 5f).
Finally, we attenuated PGAM1 in SKOV3-TR30 cells by shRNA and subsequently evaluated the effects on glycolytic flux and paclitaxel sensitivity. As Figure 6a shows, down-regulation of PGAM1 in SKOV3-TR30 cells resulted in significantly declined pyruvic acid production ( Figure 6b) and paclitaxel resistance at 48 or 72 hours (Figure 6c and Figure 6d).

Discussion
Paclitaxel is a preferred chemotherapeutic agent for ovarian cancer, however, paclitaxel resistance has represented a long term challenge for treatment strategy. Here we demonstrate that PGAM1 modulates paclitaxel resistance in ovarian cancer cells via pyruvic acid production (Figure 6e). We also found that a high expression of PGAM1 in cancer cells of ovary cancer patients is significantly correlated with poorer prognosis.
Aerobic oxidation of glucose catabolism is a main pathway for normal cells to obtain energy under conditions of sufficient oxygen supply. During this process, glucose is eventually decomposed into carbon dioxide and water, generating abundant ATP. Conversely, cancer cells prefer anaerobic oxidation, even under the conditions of sufficient oxygen supply (28,29). In this case, glucose is catabolized into pyruvic acid, 13 and primarily decomposed into lactic acid, leading to lower levers but faster ATP generation to meet the biosynthetic demands for rapid proliferation (30). The increased expressions of many enzymes, including PGAM1, have been demonstrated to be involved in this process.
Upregulated PGAM1 in human cancers contributes to biosynthesis regulation.
Inhibition of PGAM1 by shRNA or the small molecule inhibitor PGMI-004A results in that there are four key steps controlling glycolytic flux, namely glucose import, hexokinase, phosphofructokinase, and lactate export (33). It is interesting that in our study that among the 125 (35%) out of 356 successfully mined proteins that participated in metabolic processes, PGAM1 was the only remarkable enzyme involved in the glycolic pathways by which paclitaxel resistance could be manipulated in ovarian cancer cells.
In the following study, we collected 180 primary epithelial ovarian cancer tissues and detected PGAM1 protein by immunohistochemical staining. The results demonstrated that PGAM1 high-expression was significantly related to paclitaxel resistance in ovarian cancer patients and thus revealed poorer prognosis including PFS and OS. This suggested that PGAM1 may act as an oncoprotein that links with paclitaxel resistance in ovarian cancer patients.
As tumors progress, they first become hypoxic and then acidic， and then malignant progression is accelerated. Tumors become resistant to therapeutic strategies, including both radiotherapy and chemotherapy (30,34 In our study, we measured PGAM1 expression and found an increased PGAM1 expression following ovarian cancer cells being exposed to paclitaxel. We enforcedly

Conclusions
In paclitaxel-induced paclitaxel-resistant SKOV3 ovarian cancer cell SKOV3-TR30, PGAM1 is high-expressed, which modulates paclitaxel resistance in ovarian cancer cells via pyruvic acid and/or lactate production. High expression of PGAM1 in ovary cancer cells could be a potential chemotherapeutic target. Our findings may provide a new approach to overcome paclitaxel resistance for ovarian cancer patients.    Comparison of pyruvic acid (a) and lactic acid (b) production between SKOV3 and SKOV3-TR30 cells when cultured for 24h, 48h and 72h. SKOV3 and SKOV3-TR30 cells were cultured without any paclitaxel in the culture medium (c), or exposed to 10nM (d), 20nM (e) and 30nM (f) paclitaxel for 24 hours. Representative images were photographed using a confocal fluorescence microscope after immunofluorescence staining. g The fluorescence density was measured using the software Image J and was significantly increased under the exposure of 10nM paclitaxel (P=0.012), 20nM paclitaxel (P=0.002) and 30nM paclitaxel (P=0.038). DAPI: staining for the nuclei.