PolG Inhibits Glycolysis by Suppressing PKM2 Phosphorylation Resulting Reduced Gastric Cancer Proliferation

Background: Gastric cancer (GC) is the fth most frequently diagnosed cancer and the third leading cause of cancer death. There is a critical need for the development of novel therapies in GC. DNA polymerase gamma (PolG) has been implicated in mitochondrial homeostasis and affects the development of a numerous of cancer types. However, its effects in GC and molecular mechanisms remain to be fully determined. Methods: GSE62254 dataset was used to predict the impact of PolG on prognostic valule in GC patients. Lentivirus-mediated transduction was used to silence PolG expression. Western blot analysis the knockdown effect. Co-immunoprecipitation analysis was performed to explore the potential molecular mechanism. Analysis of the glycolysis process in GC cells was also performed. Cell proliferation was determined using a CCK-8 (Cell Counting Kit-8) proliferation assay. Cell migration was detected using transwell method. Animal experiment was used to measure the In vivo xenograft tumor growth. Results: GC patients with low PolG expression have worse OS and pFS . PolG binds to PKM2 and affects the activation of the Tyr105 site phosphorylation, then interfering with the glycolysis of Gastric cancer cells. In vitro tumor formation experiments in mice also conrmed that PolG knockdown GC has stronger proliferation ability. PolG can suppress GC cells growth in both vivo and vitro. Conclusion: Our study reveals a potential molecular mechanism between PolG and the energy metabolic process of GC tumor cells, suggesting PolG as an independent novel potential therapeutic target for tumor therapy.


Background
One million new cases of gastric cancer (GC) were diagnosed globally in 2018 with an estimated 783,000 deaths (amounting to 1 in every 12 deaths), making it the fth most frequently diagnosed cancer and the third leading cause of cancer death [1]. Major efforts have been made in GC treatment including targeted therapies such as HER2-targeted trastuzumab, VEGFR2-targeted ramucirumab, and immune checkpoint inhibitors (ICIs) [2]. In addition to these improvements, the complex biology of GC often results in treatment failure and therapeutic resistance [3]. Therefore, it is necessary to further explore novel treatment targets, and adopt comprehensive treatment approaches to deliver better patient outcomes in GC.
Under well-oxygenated conditions, oxidative phosphorylation is the primary way of nutrient catabolism and energy production in most differentiated cells [4]. When the mitochondrial function of tumor cells is impaired, the energy supply capacity of oxidative phosphorylation is reduced which may impact the survival of tumor cells. However, due to changes in metabolic homeostasis, tumor cells have a preference to obtain energy through glycolysis under the same conditions, which is known as the Warburg effect [5,6]. DNA polymerase gamma (PolG) is the main polymerase of mitochondrial DNA. PolG can affect the stability of mitochondrial DNA and interfere with the expression of proteins synthesized by mitochondrial DNA transcription, affecting mitochondrial homeostasis [7][8][9]. Studies have reported that knockdown of PolG in bowel cancer can reduce mtDNA content, and increase glucose uptake and lactate secretion, making tumors more resistant to oxidative stress [10]. By systematically analyzing the DDR gene data of hereditary breast cancer patients, missense mutations in PolG were signi cantly related to the risk of breast cancer [11]. Curcumin can interfere with mitochondrial function by reducing the expression of PolG, thereby inhibiting the development of GC [12,13]. These ndings suggest that PolG may be closely related to the occurrence, development and prognosis of tumors, yet the speci c molecular mechanisms of PolG remain to be determined.
In this study, we explored the potential mechanism of PolG in the growth of GC tumors. Our results suggest that PolG is a tumor suppressor gene that impacts GC cell viability. PolG can also competitively bind to the phosphorylation site of PKM2, to reduce phosphorylation at the PKM2-Tyr105 site, and independently suppress the glycolysis of GC tumors to inhibit tumor growth.

Data collection and screening
Microarray data of GSE62254 were obtained from the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo/). GSE62254 data were based on the GPL570 platforms (Affymetrix Human Genome U133 Plus 2.0 Array, 300 GC patients), and 295 samples with both clinical parameters and gene expression data of GC were included in this study. Kaplan Meier-plotter (KMplotter) (http://www.kmplot.com/) was used for external validation.

Survival analysis
Survival analysis was performed by Kaplan-Meier (KM) method and log-rank test. The expression levels of hub-genes were separated according to high and low expression based on the median value. The Cox proportional-hazards regression models was applied to perform uni-and multivariate analyses were performed by. A nomogram was set up by the rms package in R, according to the nal multivariate COX regression model. The predictive accuracy was assessed by C-Index. The internal validation of nomogram was measured by a calibration curve.

Plasmid construction
A lentiviral PolG shRNA was purchased from Genechem (Shanghai, China). The shRNA sequence targeting the human PolG complementary DNA was 5′-TGTCCAGGGAGAGTTTATA-3′. A scrambled shRNA was included as a negative control (NC). The target sequence was inserted into the GV248 lentiviral vector (Genechem). The expression of the Myc-PolG plasmid was constructed by PCR and subcloned into the Myc-pCMV vector(Clontech). The Flag-PKM2 plasmid was purchased from OriGene.
Cell Culture and Transfection SGC7901, MGC803 and HEK293T cells were cultured in high-glucose DMEM with 10% FBS and 100 units/ml of penicillin/streptomycin, at 37 oC in a 5% CO2 incubator. Cells were transfection by

Western Blot Analysis
Cells were lysed with an IP lysis buffer supplemented with protease and phosphatase inhibitor cocktails for 30 min on ice. Total protein was harvested by centrifugation at 15,000 rpm for 20 min at 4 °C.
Samples were loaded to 10% polyacrylamide gels, separated by SDS-PAGE and transferred to PVDF membranes for 2.5 h at 80-120V. Membranes were incubated through prescribed antibodies at 4˚C overnight after blocked by 5% BSA in TBST for 2 h at room temperature. After 3 times wash with TBST, membranes were incubated with secondary antibody for 2 h at room temperature. At last, bands were analyzed by chemiluminescence detection (Tanon Science & Technology Co., Ltd., Shanghai, China).
Co-immunoprecipitation analysis SGC7901 cells were dissolved with IP lysis and incubated with antibody for 2 h. Then protein A/G-Sepharose was added and incubate on a mixer overnight at 4 °C. Then use centrifugation for 5 min at 700 g at 4 °C to connect Beads the next day, and cleaned in IP lysis for 3 times (each time for 10 min). The beads were then resuspended with the loading for WB.
Glucose consumption and lactate production analysis Cells were seeded in six-well culture plates. The medium was changed into phenol red-free DMEM media after 6 h. Then harvest the media after 48 h. Glucose consumption was measured between the media before and after the 48 h incubation period by an assay kit (Sigma, GAHK20). Extracellular lactate levels were normalized to the protein concentration of the samples by a lactate assay kits (Sigma, MAK065).

Cell proliferation assay
Cell counting Kit-8 (CCK8) (Abbkine, KTA1020) was used to evaluate the cell proliferation ability. Cells were seeded into 96-well plates at a density of 5 × 103 cells/well for 6 h, 24 h, 48 h, 72 h. Then the culture medium was replaced with 90 µl of basal DMEM and 10 µl CCK8. After incubation at 37 °C for 3 h, the absorbance was measured using an absorbance reader (TECAN, Switzerland) at 450 nm. Transwell migration assay SGC7901 or MGC803 cells (3 × 104) were resuspended with Serum-free DMEM media and seeded into the upper chamber while 10%FBS DMEM with added to the lower chambers (Corning, 3422). After 24 h incubation, invasive cells on the underside were xed in methanol for 10 min and stained with hematoxylin for 30 min at room temperature. Image was taken by an inverted microscope (Nikon Corp., Tokyo, Japan). Three independent experiments were performed, and ve individual elds were counted each for statistical analysis.

In vivo xenograft tumor growth
For the xenograft tumor growth assay, SGC7901 stable cell lines with PolG knockdown were injected subcutaneously into the right ank of 6-week-old male BALB/C nude mice (N = 10). NaB was intraperitoneally injected into mice in the PolG-knockdown group (200 mg/kg) (N = 5). NC group were used for comparative(N = 5). Tumors were cultivated for 14 days. All animal experiments were approved by the Committee of China Medical University.

Statistical analysis
Three independent experiments' values are expressed as the mean ± standard. Statistical signi cance was analyzed using a t-test or a one-way analysis of variance. All statistical analysis was performed using SPSS 17.0 and Prism 5.0 software. Values of P < 0.05 were de ned as statistically signi cant.

Results
Prognostic value of PolG in GC GC samples with survival data and gene expression pro les were obtained from the GSE62254 dataset as shown in Table 1. The sample characteristics of GSE62254 were consistent with randomized clinical studies of GC [3,14,15]. The prognostic value of PolG was evaluated by Kaplan-Meier analysis which showed that low expression of PolG was associated with worse overall survival (OS, log-rank P < 0.001) as well as progression-free survival (PFS, log-rank P < 0.001) (Fig. 1A). Furthermore, COX regression analysis indicated that PolG was an independent prognostic factor both in univariate (HR 0.611, 95% CI; 0.432-0.866; P = 0.006) and multivariate analysis (HR 0.679, 95% CI; 0.479-0.963; P = 0.03) ( Table 2).
External validation analysis based on KMplotter showed a favorable prognosis of PolG in GC (log-rank P = 0.004) (Fig. 1B). To further investigate the predictive value of PolG in a prognostic model, a nomogram of OS which combined the signi cant prognostic factors identi ed from multivariate analysis was adopted (Fig. 1C). The centrality-index (C-Index) of OS prediction was 0.75 (95% CI, 0.71 to 0.79). The calibration curve was applied to re ect the probability of survival at 1 and 3 years, which indicated the consistency between the nomogram prediction and the objective observations (Fig. 1D).

Gene-set enrichment analysis of PolG
The potential molecular mechanisms of high PolG expression in GC subtypes were investigated by geneset enrichment analysis. The results indicated that DNA replication and cell cycle were the most signi cantly enriched biological processes associated with PolG expression (Fig. 1E). In contrast, cytochrome P450, chemical carcinogenesis and retinol metabolism were highly enriched in samples with low expression of PolG.

Knocking down of PolG inhibits GC cells proliferation and migration
To elaborate the association of PolG with poor survival in GC patients with low expression of PolG, SGC7901 and MGC803 cells were used to explore the potential roles of PolG in GC cells. A PolG-shRNA lentiviral vector was constructed and used to establish stably transfected PolG-shRNA SGC7901and MGC803 cell lines. As shown in Fig. 2A, PolG knockdown on SGC7901and MGC803 cell proliferation was detected through cell counting Kit-8 (CCK-8). The decreased expression of PolG markedly accelerated the growth of SGC7901 and MGC803 cells. Cell proliferation was increased by 55% and 25% in SGC7901 and MGC803 cells of the shPolG group, respectively, compared to the control group (P < 0.001). The migration abilities of GC cells were signi cantly increased in both PolG-knockdown cell lines through transwell analysis(P < 0.0001, Fig. 2B); To further investigate the potential causes of these phenomena, we examined the rates of lactate production and glucose consumption of each cell line. The results showed that the decrease of PolG expression in GC cells increased both lactate production and glucose consumption (P < 0.001, Fig. 2C). These data suggest that knockdown of PolG increased the basal glycolytic rate and Warburg effect of tumor cells. Also, KEGG enrichment analysis by GSEA indicated that metabolism-related pathways were highly enriched in the group with low PolG expression (Fig. 2D) which may be due to the impact of PolG on the activity of key kinases involved in tumor metabolism.

PolG inhibits tumor glycolysis by interactions with PKM2
We detected the molecular mechanism of PolG in GC cell glycolysis. Western blot(WB) detection indicated that PKM2-Tyr105 increased signi cantly after PolG knockdown (P < 0.01, Fig. 3A). PKM2 is a main regulator for glycolysis, and PKM2-Tyr105 phosphorylation leads to less kinase activity which can promote the Warburg effect [16]. We then analyzed the potential interactions of PolG with PKM2. Endogenous and exogenous co-immunoprecipitation analysis showed that PolG interacted with PKM2 in SGC7901 cells (Figs. 3B, C). Based on these data, we hypothesized that the observed changes in cellular metabolism caused by PolG knockdown were due to interactions with PKM2, which interfered with the phosphorylation of PKM2-Tyr105.
Replenishing PolG can mimic PKM2 inhibition in reducing the proliferation of GC cells We con rmed that PolG knockdown promoted GC cells proliferation, and PolG could interact with PKM2. However, PolG knockdown may affect mitochondrial function, leading to a compensatory increase in cellular glycolysis. To further con rm that the changes in GC cell metabolism and cell behavior were due to the interactions between PolG and PKM2, we performed PolG replenishment on knocked-out cell lines.
Also, we used NaB to target PKM2 for tumor suppression [17] and to inhibit PKM2 phosphorylation in a controlled experiment.
In the preliminary experiment, we treated the SGC7901 cells with different concentrations of NaB (0-10 mM) for 24 h, and found that after the concentration was increased to 5 mM, the suppression of tumor cells entered the plateau phase, so we nally chose 5 mM as the nal stimulation concentration (Fig. 4A). Then We selectively performed PolG supplementation experiments, and added NaB (5 mM) to the medium of the PolG-knockdown cell lines. The Cell proliferation (CCK8, Fig. 4B) and migration (transwell, Fig. 4C) were shown to be signi cantly reduced. In addition, WB for PKM2-Tyr105 exhibited signi cantly reduced levels in the PolG supplementation and NaB groups (Fig. 4D).

PolG knockdown promotes GC cell growth in vivo
To explore the impact of PolG knockdown in promoting cells growth in vivo, we injected the stably PolG knockdown SGC7901 cell lines and NC into nude mice. NaB was intraperitoneally injected into mice in the PolG-knockdown group (200 mg/kg) for comparative observation [18]. PolG knockdown in SGC7901 cells was shown to be larger than the NC group, and an intraperitoneal injection of NaB was found to suppress tumor growth (Fig. 5A). Furthermore, compared with the NC and NaB groups, the PolG-knockdown group had larger tumor weights (P < 0.001, Figs. 5B and C). These data indicated that PolG knockdown promoted GC cells growth in vivo.

Discussion
Through online GC data analysis, we found that GC patients with low PolG expression have worse OS and pFS, but the molecular mechanism remains unclear. PolG knockdown in GC cells in vitro, showed increased proliferation and migration capabilities compared to control cells. KEGG enrichment analysis of GSEA suggested that these phenomena are closely related to the effect of PolG on the reprogramming of tumor cell metabolism.
In previous studies, it was shown that PolG is a mitochondrial DNA polymerase, which plays an important role in maintaining the stability of mitochondrial DNA. The abnormality of PolG could directly affect the function of the respiratory chain which is composed of proteins synthesized by transcription of mitochondrial DNA and downstream signal modulation of cellular metabolism [19][20][21]. It has also been reported that PolG promotes metabolic reprogramming in various types of tumor cells by affecting mitochondrial function in tumor cells [11][12][13]. Although more than 90% of the cellular ATP is produced by mitochondria in normal differentiated cells, tumor cells undergo the Warburg effect by relying on aerobic glycolysis as their primary energy source. Metabolic reprogramming is an important feature during tumorigenesis and development. Glycolysis is an insu cient way of ATP production that allows tumor cells to uptake more nutrients, and synthesizes organic molecules to support their proliferation and invasion [22,23].
In this study, for the rst time we con rmed the molecular interaction between PolG and the glycolysis of GC cell. This mechanism is closely related to the inhibition of phosphorylation at PKM2-Tyr105. Recently, researches shown that PKM2 is upregulated in tumor cells, and that PKM2-Tyr105 site phosphorylation can prevent its tetramer formation which directly impacts the kinase activity of PKM2 and enhances the Warburg effect [16,[24][25][26]. In the current study, both glucose consumption and lactate production were increased in PolG-knockdown GC cells. We investigated the molecular mechanism through which PolG can interact with PKM2 and affect the phosphorylation of PKM2-Tyr105. Moreover, we performed PolG replenishment for PolG-knockdown GC cells and experiments using PKM2 inhibitors (NaB). These experiments showed reduced proliferation and migration abilities. These data con rmed that PolG can suppress the energy metabolism of tumor cells by inhibiting PKM2 phosphorylation, further validating PolG as a potential therapeutic target.
However, our study did not include a comprehensive investigation on the speci c mechanism through which PolG affects phosphorylation of Tyr105 after binding to PKM2. We also did not determine the balance between the effect of PolG on PKM2 and the speci c effects on mitochondrial function. In future studies, we will continue to elucidate the potential role of PolG in GC.

Conclusions
Our present study revealed that PolG is an independent factor in the treatment of GC. PolG can interact with PKM2 and affects the activation of the Tyr105 site phosphorylation, then suppress the energy metabolism of GC cells, at last interfering the GC cells growth in both vivo and vitro. PolG may be a potential therapeutic target for GC treatment.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.