Long noncoding RNA OVAAL enhances nucleotide synthesis through pyruvate carboxylase to promote 5‐fluorouracil resistance in gastric cancer

Abstract 5‐Fluorouracil (5‐FU) is widely used in gastric cancer treatment, yet 5‐FU resistance remains an important clinical challenge. We established a model based on five long noncoding RNAs (lncRNA) to effectively assess the prognosis of gastric cancer patients; among them, lncRNA OVAAL was markedly upregulated in gastric cancer and associated with poor prognosis and 5‐FU resistance. In vitro and in vivo assays confirmed that OVAAL promoted proliferation and 5‐FU resistance of gastric cancer cells. Mechanistically, OVAAL bound with pyruvate carboxylase (PC) and stabilized PC from HSC70/CHIP‐mediated ubiquitination and degradation. OVAAL knockdown reduced intracellular levels of oxaloacetate and aspartate, and the subsequent pyrimidine synthesis, which could be rescued by PC overexpression. Moreover, OVAAL knockdown increased sensitivity to 5‐FU treatment, which could be reversed by PC overexpression or repletion of oxaloacetate, aspartate, or uridine. OVAAL overexpression enhanced pyrimidine synthesis to promote proliferation and 5‐FU resistance of gastric cancer cells, which could be abolished by PC knockdown. Thus, OVAAL promoted gastric cancer cell proliferation and induced 5‐FU resistance by enhancing pyrimidine biosynthesis to antagonize 5‐FU induced thymidylate synthase dysfunction. Targeting OVAAL‐mediated nucleotide metabolic reprograming would be a promising strategy to overcome chemoresistance in gastric cancer.


| INTRODUC TI ON
Gastric cancer is the third most common cause of cancer-related death worldwide. 1,2 Chemotherapy of GC is based on combination with fluorouracil. Despite advances in surgical treatment, the survival rates of GC have not been effectively improved, which is mainly related to intrinsic or acquired chemoresistance. 3 The underlying molecular mechanisms of GC progression and chemoresistance are largely unclear. Thus, there is an urgent need to gain an in-depth understanding of the molecular mechanisms of GC tumorigenesis and chemoresistance.
Long noncoding RNAs are a class of noncoding transcripts longer than 200 nucleotides in length and have been shown to regulate several crucial biological functions. [4][5][6] Mechanistically, lncRNA regulates these cellular processes by interacting with and regulating macromolecules, including DNA, RNA, and proteins. 7 Increasing evidence has shown that lncRNAs influence tumorigenesis and cancer progression and drug resistance. For example, the lncRNAs HOTAIR, MALAT1, and GAS5 are considered to play important roles in tumor initiation and development. [8][9][10] Long noncoding RNA CRAL acts as competing endogenous RNA to decrease cisplatin resistance of GC through the microRNA-505/CYLD/AKT axis. 11 Our group revealed that lncRNA BDNF-AS promoted endocrine resistance of breast cancers by activating mTOR signaling. 12 However, lncRNAs that regulate chemoresistance in GC are still largely unknown.
Metabolic reprograming is a hallmark of cancer. 13 Long noncoding RNAs reprogram cancer metabolism by modulating the key metabolic regulators. Our previous study discovered that lncRNA HIFAL and HISLA drove hypoxia-inducible factor-1α mediated transactivation and glycolysis. 14,15 Metabolic reprograming not only drives cancer development but also chemoresistance of cancers. Our group revealed that activation of nonoxidative pentose phosphate pathway in breast cancer induced chemoresistance. 16 JAK/STAT3 transcriptionally upregulates CPT1B to enhance the fatty acid β-oxidation, which promotes chemoresistance by regulating breast CSCs. 17 DECR1 regulated redox homeostasis by balancing intracellular levels of saturated and unsaturated phospholipids to participate in the resistance of androgen receptor antagonist in prostate cancer. 18 In the present study, we undertook a bioinformatics analysis of the TCGA GC dataset and identified the lncRNA OVAAL, whose overexpression was associated with poor outcome of GC patients and resistance to 5-FU treatment. We further revealed that OVAAL prevented proteasomal degradation of PC, thus accelerated the production of oxaloacetate from pyruvate and the following accumulation of malate and aspartate, which enhanced the pyrimidine production for cell proliferation. We found that OVAAL overexpressed GC cells were more resistance to 5-FU-induced pyrimidine metabolism abnormalities and thus become resistance to 5-FU treatment.

| Patients and clinical samples
Thirty-four primary GC samples and paired nontumor samples (cohort 1) were collected from patients at Sun Yat-Sen Memorial Hospital from January 2015 to January 2016. One hundred and twenty primary GC samples with clinicopathologic data and prognosis data (cohort 2) were collected from GC patients who underwent surgery at Sun Yat-Sen Memorial Hospital from January 2010 to January 2015. The nontumor samples were collected at least 5 cm from the tumor.

| Bioinformatics analysis
RNA sequencing and corresponding clinical data of GC patients were obtained from the public TCGA database. In total, 1978 lncR-NAs were identified by differential expression analysis between the tumor and normal tissues (log2|fold change| > 1 and p-value <0.05).
We further undertook univariable Cox regression analysis to identify prognosis-related lncRNAs. We ranked these candidate lncRNAs by p-value of univariable Cox regression for subsequent analysis. We then constructed an lncRNA-based risk score model by considering each lncRNA's HR and AIC. Finally, five lncRNAs were used for establishing a risk score model and calculated their risk score using the following formula: in which Coef i and i represent the coefficient of the multivariate regression Cox analysis and expression level of each lncRNA, respectively. We used the median score as the cut-off value to divide patients into high-and low-risk groups; survival analysis and ROC curves were applied to evaluate this risk model. All the analyses were carried out using R 3.5.2.

| Animal experiment
To evaluate the effects of OVAAL knockdown and PC overexpression under OVAAL knockdown on tumorigenicity and 5-FU therapy in vivo, 5 × 10 6 BGC823 cells expressing shCtrl or shO-VAAL1/2 and 5 × 10 6 shOVAAL BGC823 cells transfected with vector or pcDNA3.1-PC were subcutaneously injected into the right dorsal flank of 4-week-old male BALB/C nude mice (Vital River).
Intraperitoneal injection of either 5-FU (15 mg/kg) or PBS was carried out every 3 days.  (Table S1). The IgG (Abcam) controls were assayed simultaneously to confirm that the RNA specifically interacted with PC.

| Statistics
Statistical analysis was undertaken using SPSS version 22.0 (SPSS).  Figure S1A,B), we screened out 151 prognosis-associated lncR-NAs by Kaplan-Meier analysis and univariate Cox regression analysis ( Figure 1A). To further narrow down to the most relevant lncRNAs in GC, we constructed a prognosis risk model using five lncRNAs ( Figure 1B), considering higher HRs and smallest AIC.
Survival analysis showed that the high risk score was significantly associated with the poor OS ( Figure 1C) and higher mortality rate of GC patients ( Figures 1D and S1C). The univariate and multivariate regression analyses also revealed that this risk model had the highest HR among the clinicopathologic characteristics and was an independent prognostic factor in the TCGA cohort (Table   S2). The ROC curve showed that the 5-year value of area under the ROC was 0.739, indicating that our model was effective in GC progression prediction ( Figure 1E Figure 1H), which was consistent with the TCGA data ( Figure 1I). Moreover, high level of OVAAL, detected by qRT-PCR in 120 GC tissues (cohort 2), was significantly associated with low OS and CSS ( Figure 1J,K), larger tumor size, and lymph node metastasis (Table S3). The univariate and multivariate regression model also discovered that high OVAAL expression independently predicted poor OS and CSS in GC patients, with even higher HR than the local invasion (T stage) or distance metastasis (M stage) (Tables S4-S7) 5-Fluorouracil is widely applied in GC treatment as a basic chemotherapy drug. The sensitivity to 5-FU treatment markedly impacts the survival of GC patients. Interestingly, we discovered that the level of OVAAL was associated with the efficacy of 5-FU. Seventyeight patients were treated with 5-FU-based adjuvant chemotherapy after surgery in cohort 2. High expression of OVAAL correlated with poor CSS in these patients ( Figure 1L). These results suggested that a high OVAAL level was associated with tumor progression and efficacy of 5-FU treatment in GC.

| OVAAL promotes proliferation and increases resistance to 5-FU in GC cells
To functionally validate the above findings, we first tested the ex- Graphs represent mean ± SD of experimental triplicates. *p < 0.05; **p < 0.01; ***p < 0.001. OD, optical density 3.3 | OVAAL enhances pyrimidine nucleotide synthesis to promote proliferation and increase 5-FU resistance 5-Fluorouracil induces cancer cell death by impairing pyrimidine metabolism and DNA synthesis. 19 Growing evidence has suggested that lncRNAs promoted metabolism reprogramming to accelerate cancer progression. [20][21][22] Therefore, we tested whether lncRNA OVAAL induced 5-FU resistance of GC by regulating cancer cell metabolism, especially pyrimidine metabolism. Mass spectrometry analysis of the metabolite profiling showed that knockdown of OVAAL caused dramatic decrease of oxaloacetate in BGC823 cells ( Figure 3A).
Oxaloacetate is known to transform to citrate by condensing with acetyl CoA derived from a second pyruvate. Otherwise, oxaloacetate can undergo a transamination reaction to form aspartate. Consistent with the above knowledge, we observed the decrease of citrate ( Figure 3A), as well as a significant decrease of aspartate among all amino acids ( Figure 3B). However, glycolytic metabolites were not affected significantly following OVAAL knockdown ( Figure S3A).
In cancer cells, de novo synthesis of nucleotides is the major source of nucleotide supplies. [23][24][25] Aspartate is essentially required for de novo synthesis of nucleotides ( Figure S3B). 26 In purine biosynthesis, aspartate acts as nitrogen donor for the purine ring of inosine monophosphate, and the generated hypoxanthine is further converted into AMP or GMP. 27 More importantly, the whole structure of aspartate is used for pyrimidine ring synthesis, which is conjugated with carbamoyl phosphate to transform into carbamoyl aspartate and then converted into pyrimidine ring by dehydration and dehydrogenation. Thus, we hypothesized that the nucleotide synthesis, especially for pyrimidine nucleotide synthesis, would decrease following OVAAL knockdown because oxaloacetate and aspartate decreased following OVAAL knockdown. As expected, silencing OVAAL resulted in the reduction of the nucleosides pool in BGC823 cells, especially pyrimidine nucleosides UMP/UTP and CMP/CTP ( Figure 3C). Moreover, silencing OVAAL reduced intracellular levels of dTMP ( Figure 3D), which was in line with the report that dUMP could be transformed to dTMP by thymidylate synthase, therefore adding 5-FU would interfere with the dUMP to dTMP transformation, disturbing pyrimidine metabolism and inducing cancer cell death. 28 These results supported that OVAAL promoted NTPs, especially the pyrimidine nucleotide synthesis, to increase intracellular levels of dTMP, thus protecting GC from 5-FU-induced cell death.
To further confirm that OVAAL induced 5-FU resistance by enhancing pyrimidine metabolism, we directly replenished oxaloacetate, aspartate, or nucleotides to rescue the OVAAL-silencing phenotype. We found that repletion of oxaloacetate, aspartate, and uridine rescued the OVAAL silencing-induced decrease of 5-FU resistance ( Figures 3E-J and S4A-F), whereas repletion of adenosine, guanosine, and cytidine could not ( Figure S5A-C). These data supported that OVAAL could reprogram pyrimidine metabolism, therefore endowing GC cells with 5-FU resistance.

| OVAAL enhances pyrimidine metabolism and induces 5-FU resistance by interacting with pyruvate carboxylase
Next, we interrogated the mechanism of how OVAAL decreased the oxaloacetate levels and reprogrammed pyrimidine metabolism in GC.
Like many other lncRNAs, we hypothesized that OVAAL might exert its function by interacting with proteins in the related pathway. 29 We undertook RNA pulldown assays in BGC-823 cells to identify the protein interacting with OVAAL. The proteins pulled down by the sense probe but not antisense probe were identified by mass spectrometry ( Figure 4A). Pyruvate carboxylase was the top scored protein identified by mass spectrometry ( Figure 4B). The binding of OVAAL with PC was verified by western blot analysis following the RNA pull-down assay ( Figure 4C). In line with these results, the RIP assays showed the interaction between OVAAL and PC in BGC823 and AGS cells ( Figure 4D).

| OVAAL promotes pyruvate carboxylase activity by increasing its stability
To identify which region of OVAAL interacted with PC, we con- and F6, 751-1489 nt). The RNA pull-down assay using these fragments as probes showed that F3 and F6 could physically bind to PC ( Figure 5A), which suggested that 751-1000 nt was the region in OVAAL that interacted with PC.
We further constructed the specific fragment (Fs, 751-1000 nt) of OVAAL and the mutant lacking the Fs fragment (ΔFs). The functional assays and western blot analysis showed that overexpression of Fs, but not ΔFs, was sufficient to rescue the IC 50 of 5-FU ( Figure 5B,C) and the protein level of PC after OVAAL knockdown ( Figure 5D). These data together suggested fragment 751-1000 nt was crucial for the function of OVAAL.
Next, we explored how OVAAL regulated PC function. We  Figure 5I). These results supported that OVAAL increased the PC level by regulating its proteasome-dependent degradation, but not synthesis. We further confirmed these results by the ubiquitination assay, which showed that knocking down OVAAL increased the level of PC ubiquitination in BGC823 cells ( Figure 5J).
Consistently, overexpressed OVAAL decreased the level of PC ubiquitination ( Figure 5K). These results illustrated that OVAAL interacted with PC and prevented the ubiquitin-mediated degradation of PC protein.
In order to explore the underlying mechanism of how OVAAL protects PC from protein degradation, we used UbiBrowser 2.0 (http:// ubibr owser.ncpsb.org.cn) to predict the potential E3 ligase of PC. 31 UbiBrowser 2.0 predicted that HSC70 was the potential protein that interacted with PC to induce ubiquitin degradation. To validate the prediction, we used the Co-IP assay followed by mass spectrometry to compare PC binding proteins in control and OVAAL knockdown cells ( Figure 6A). In PC-immunoprecipitated complexes, HSC70 was detected in OVAAL knockdown BGC-823 cells by mass spectrum ( Figure 6B), but not in control cells, indicating that OVAAL weakened the interaction between PC and HSC70. The Co-IP experiment also confirmed the interaction between PC and HSC70 in BGC823 cells ( Figure 6C). In addition, knockdown of OVAAL strengthened the interaction between PC and HSC70 ( Figure 6D), without influencing the protein level of HSC70 ( Figure S8A). Moreover, knockdown of HSC70 significantly decreased PC ubiquitination ( Figure 6E) and increased PC protein levels ( Figure 6F,G). HSC70 may function as molecular chaperon and CHIP is an important E3 ligase, which forms the HSC70/CHIP complex to modify substrate targeting and facilitate protein degradation. 32,33 As expected, the Co-IP assay validated the interaction between HSC70 and CHIP in BGC823 cells ( Figure 6H).
Knockdown of OVAAL did not influence the protein level of CHIP ( Figure S8B). However, knockdown of CHIP could significantly decrease the ubiquitination of PC ( Figure 6I) and upregulate the protein level of PC ( Figure 6J). These data indicated that OVAAL prevented the interaction between PC and HSC70, thereby further blocking the ubiquitination and degradation of PC by CHIP. Together, these results suggested that knockdown of OVAAL might be a promising strategy to overcome resistance of 5-FU as well as to prevent GC progression ( Figure 8).  In summary, our data revealed that lncRNA OVAAL stabilized pyruvate carboxylase and accelerated oxaloacetate-aspartate production and thus enhanced pyrimidine biosynthesis, which promoted GC cell proliferation and the resistance to 5-FU. Our finding also indicated that the level of OVAAL could serve as potential biomarker

D I SCLOS U R E
The authors have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.