LncRNA XIST/miR-137 axis strengthens chemo-resistance and glycolysis of colorectal cancer cells by hindering transformation from PKM2 to PKM1

Abstract

BACKGROUND:

Glycolysis was an essential driver of chemo-resistance in colorectal cancer (CRC), albeit with limited molecular explanations.

OBJECTIVE:

We strived to elucidate the involvement of lncRNA XIST/miR-137/PKM axis in chemo-tolerance and glycolysis of CRC.

METHODS:

Altogether 212 pairs of tumor tissues and adjacent normal tissues were collected from CRC patients. Moreover, human CRC epithelial cell lines, including HT29, SW480, SW620 and LoVo, were purchased in advance, and their activity was estimated after transfection of si-XIST or miR-137 mimic. Furthermore, 5-FU/cisplatin-resistance of CRC cells was determined through MTT assay, and glycolytic potential of CRC cells was appraised based on oxygen consumption rate (OCR) and extracellular acidification rate (ECAR).

RESULTS:

Highly-expressed XIST were predictive of severe symptoms and unfavorable 3-year survival of CRC patients ( $P<$ 0.05). Besides, silencing of XIST not only diminished proliferative, migratory and invasive power of CRC cells ( $P<$ 0.05), but also enhanced sensitivity of CRC cells responding to 5-FU/cisplatin ( $P<$ 0.05). Glycolytic potency of CRC cells was also undermined by si-XIST, with decreased maximal respiration and maximal glycolytic capacity in the si-XIST group as relative to NC group ( $P<$ 0.05). Nevertheless, miR-137 mimic attenuated the facilitating effect of pcDNA3.1-XIST on proliferation, migration, invasion, 5-FU/cisplatin-resistance and glycolysis of CRC cells ( $P<$ 0.05). Ultimately, ratio of PKM2 mRNA and PKM1 mRNA, despite being up-regulated by pcDNA3.1-XIST, was markedly lowered when miR-137 mimic was co-transfected ( $P<$ 0.05).

CONCLUSIONS:

LncRNA XIST/miR-137 axis reinforced glycolysis and chemo-tolerance of CRC by elevating PKM2/PKM1 ratio, providing an alternative to boost chemo-therapeutic efficacy of CRC patients.



1. Introduction

Colorectal cancer (CRC), a digestive tract malignancy, accounted for over 0.9 million deaths per year around the globe, and its annual prevalence was expected to arrive at 2.5 million by 2035 [1]. It was fortunate that 90.1% of early-stage patients were able to survive CRC for $⩾$ 5 years, however, merely one case out of ten patients with metastatic CRC lived longer than 5 years [2]. Despite chemotherapy-based strategies for dealing with metastatic CRC, emergence of drug resistance marginally restricted their effectiveness in hindering tumor metastasis, which ultimately led to treatment failure [3]. As proposed by the Warburg effect theory, tumor cells primarily relied on glycolysis, rather than oxidative phosphorylation, to perform metabolic activities, whether oxygen was present or not [4]. Intriguingly, this metabolic pattern was responsible for driving invasion and chemoresistance of CRC cells [5, 6], which implied that signaling pathways that disturbed glycolysis of CRC might also matter in decelerating tumor metastasis and alleviating chemoresistance in CRC.

Non-coding genes, which covered 70%–90% of the whole genome, have sparked huge interests of researchers from home and abroad, and considerable attentions were paid to their linkage with CRC etiology [7]. Taking long non-coding RNAs (lncRNAs) for instance, expressions of lncRNA MALAT1, lncRNA HOTAIR, lncRNA CCAT1 and lncRNA CRNDE were raised markedly with exacerbation of CRC, while high expressions of lncRNA MEG3 and lncRNA RP11-462C24.1 reduced the likelihood of CRC development [8, 9, 10, 11, 12, 13]. Besides that, lncRNA XIST was associated with unfavorable 6-year overall survival (OS) and progression-free survival (PFS) of CRC patients [14]. Experiments in vitro also disclosed that CRC cells were empowered to proliferate, migrate and invade under the force of XIST, and 5-FU/doxorubicin-induced resistance in CRC cells was relievable after silencing of XIST [14, 15, 16]. Nonetheless, little convincing evidence was available to explain the role of XIST in CRC glycolysis, an energy metabolism which was pivotal in chemoresistance and metastasis of tumors [5].

Based on ceRNA hypothesis, miRNAs potentially sponged by XIST were predictable through manipulation of ENCORI software [17]. It was noteworthy that a portion of the miRNAs, including miR-497, miR-195 and miR-137, were productive in hindering deterioration of CRC [18, 19]. For example, miR-137 was reported to hamper liver metastasis of CRC by prohibiting the function of IGF-1R signaling [20]. The miR-137 also showed huge potential to sensitize tumor cells (e.g. breast cancer and lung adenocarcinoma) against chemo-drugs, yet few studies were accomplished in the context of CRC [21, 22]. Furthermore, miR-137 was found to restrain glycolysis of CRC by blocking transformation of pyruvate kinase isozyme (PKM) from PKM1 to PKM2 [23, 24], a common phenomenon during tumorigenesis [25]. It might be speculated that XIST was implicated in glycolysis of CRC by acting upon miR-137 and PKM1/PKM2 transformation, yet direct proofs were in shortage.

Hence, this investigation was carried out to verify if XIST/miR-137 axis could reinforce drug resistance of CRC by hindering transformation from PKM2 to PKM1 in glycolysis, which might benefit CRC patients who received chemotherapies.

2. Materials and methods

2.1 Cell culture

Human CRC epithelial cell lines, including HT29, SW480, SW620 and LoVo, were supplied by cell bank of Chinese Academy of Sciences (Shanghai, China), and normal epithelial cell line (i.e. NCM460) was purchased from American Type Culture Collection (ATCC). The cell lines were cultivated in 10% FBS-containing RPMI-1640 medium (Gibco, USA) under circumstances of 5% CO ${}_2$ , saturated humidity and constant temperature of 37 ${}^{\circ }$ C.

2.2 Cell transfection

After digestion by pancreatin (Beyotime Biotechnology, China), SW480 and LoVo cell lines were inoculated into 6-well plates at the concentration of 2 $\times$ 10 ${}^5$ per well. The cell lines were, respectively, transfected by si-XIST (5’-GCUGACUACCUGAGAUUUATT-3’, Genepharma, China), si-NC (5’-UUCUCCGAACGUG UCACGUTT-3’, Genepharma, China), pcDNA-XIST (Genepharma, China), miR-137 mimic (sense: 5’-UUAUUGCUUAAGAAUACGCGUAG-3’, anti-sense: 5’-ACGCGUAAUUCUUAAGCAAUAAUU-3’, Sangon, China) and miR-NC (sense: 5’-UUCUCCGAAC GUGUCACGUTT-3’, anti-sense: 5’-ACGUAGCACG UUCGGAGAATT-3’, Sangon, China) for 48 h, according to the requirement of Lipofectamine RNAiMAX transfection kit (Invitrogen, USA).

2.3 Real-time polymerase chain reaction (PCR)

Total RNAs were harvested from CRC tissues and cell lines with the assistance of Trizol kit (Takara, Japan), and the RNAs were reversely transcribed into cDNAs utilizing reverse transcription kit (catalog number: RR047A, Takara, Japan). With primers designed by Primer 5.0 software(Supplementary Table 1), cDNAs were amplified on the real-time PCR instrument (model: CFX96, Bio-Rad, USA) in line with instructions of PCR kit (catalog number: RR820A, Takara, Japan), and relative expression of genes was calculated as per 2 ${}^{-\mathrm{△}\mathrm{△}{C}^t}$ method [26]. GAPDH was designated as the internal reference for XIST, PKM1 and PKM2, and U6 was set as the internal reference for miR-137.

2.4 MTT assay to assess 5-FU/cisplatin-resistance of CRC cells

SW480 and LoVo cell lines, incubated into 96-well plates at the density of 5 $\times$ 10 ${}^3$ /well, were firstly cultivated at 37 ${}^{\circ }$ C for 24 h, before treatment by 5-FU (Tianjin Kingyork group, China) and cisplatin (Shandong Qilu pharma, China) for 48 h. Subsequently, CRC cells of each well was incubated by 20 $\mu$ l MTT solution (5 mg/ml) (Sigma, USA) for 4 h, and were then mixed by 150 $\mu$ l dimethylsulfoxide (DMSO, Sigma, USA) in the darkness for 10 min. Optical density (OD) of each sample was measured with microplate reader (model: 680, Bio-Rad, USA) at the wavelength of 490 nm, and half maximal inhibitory concentration (IC50) of CRC cells in response to 5-FU and cisplatin were calculated employing online software of Quest Graph ${}^{\text{TM}}$ IC50 Calculator (https://www.aatbio.com/tools/ic50-calculator) (AAT Bioquest Inc., USA).

2.5 CCK8 assay to evaluate proliferation of CRC cells

SW480 and LoVo cells growing at the logarithmic phase were digested and were then seeded into 96-well plates at the concentration of 5 $\times$ 10 ${}^3$ per well. After cultivation for 24 h, 48 h and 72 h, cell samples of each time point were treated by 10 $\mu$ l CCK8 reagent (DOJINDO, Japan) for 2 h. Eventually, OD values of each sample were monitored on the microplate reader (model: 680, Bio-Rad, USA) at the wavelength of 450 nm.

2.6 Transwell assay

2.6.1 Cell migration

In the first place, SW480 and LoVo cells were diluted by serum-free RPMI 1640 medium to a concentration of 1 $\times$ 10 ${}^7$ /ml. Exactly 200 $\mu$ l cell suspension was inoculated into the upper transwell chamber (Costar, USA), and culture medium that included FBS (10%, v/v) was supplemented into the lower transwell chamber. Twenty-four or thirty-six hours later, SW480 and LoVo cells in the upper transwell chamber were abandoned, while cells in the lower chamber were fixated by alcohol and dyed by hematoxylin. Ultimately, five views were randomly selected, and number of SW480 and LoVo cells was counted manually under the inverted microscope ( $\times$ 200, Olympus, Japan).

2.6.2 Cell invasion

Serum-free culture medium was firstly blended with Matrigel (BD, USA) at a ratio of 8:1 (v/v), and 100 $\mu$ l of the mixture was paved onto the polycarbonate film of Transwell chamber. After quiescent standing for $>$ 30 min, the Matrigel was solidified. The remaining procedures were in conformity with steps of migration assay, and SW480 and LoVo cells that penetrated through the microporous membrane were counted manually under the inverted microscope ( $\times$ 200, Olympus, Japan).

2.7 Appraisal of cell metabolism

2.7.1 Energy metabolism of CRC cells

SW480 and LoVo cell lines, digested to a density of 4.5 $\times$ 10 ${}^4$ /well, were seeded in XF-24 culture plates (Seahorse Bioscience, USA), and were then placed in an incubator of 37 ${}^{\circ }$ C and 5% CO ${}_2$ for 24 h. Around 1 hour before detection, cells were shifted into an incubator without CO ${}_2$ , and culture medium was replaced by bicarbonate-exclusive XF Base Medium (Seahorse Bioscience, USA) which was made up of 25 mM glucose, 1 mM pyruvic acid and 1 mM glutamine. Subsequently, 1.0 $\mu$ M oligomycin was added into “A” well of Seahorse gauging plate, and 1.5 $\mu$ M carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) was supplemented into “B” well. Then mixture of rotenone and antimycin A (Rot/AA, 0.5 $\mu$ M) was instilled into “C” well, and oxygen consumption rate (OCR) of the CRC cells was determined [27]. In addition, detection of extracellular acidification rate (ECAR) necessitated basal glycolysis rate and acidification rate of mitochondrial aspiration after addition of 11 mM glucose. Maximal glycolytic capacity of CRC cells was recorded after supplementation of 0.5 $\mu$ M Rot/AA, and glycolytic reserve was determined after supplementation of 2-deoxy-D-glucose (2-DG), based on which ECAR was drawn [28].

2.7.2 Glycolysis of CRC cells

SW480 and LoVo cell lines were seeded into 12-well plates at the concentration of 2 $\times$ 10 ${}^4$ /well, and they became adherent to the plate wall after overnight cultivation. Twenty-four hours later, glucose and lactate contents within cell medium were determined with usage of glucose detection kit and lactate detection kit (both from BioAssay Systems, USA), and number of CRC cells was counted utilizing hemocytometer (model: 1492, Hausser Scientific, USA).

2.8 Dual-luciferase reporter gene assay

Wide-type pmirGLO-XIST (i.e. pmirGLO-XIST Wt) was constructed by binding pmirGLO (Promega, USA) to XIST fragments which contained binding sites with miR-137 (Genechem, China). The mutant type of pmirGLO-XIST (i.e. pmirGLO-XIST Mut) was produced almost similarly, except that XIST fragments were mutated in their binding sites with miR-137. Afterwards, SW480 and LoVo cells at the concentration of 1 $\times$ 10 ${}^5$ /well were cultured until they anchored to the plate wall. Then pmirGLO-XIST Wt and pmirGLO-XIST Mut were, respectively, co-transfected with miR-137 mimic or miR-NC into CRC cell lines for 48 h, as guided by the Lipofectamine2000 ${}^{\text{TM}}$ transfection kit (TaKaRa, Japan). After rinse by PBS, the CRC cells were dissociated by 200 $\mu$ l lysis buffer for 15 min, and the products were monitored using GloMax luminometer (Promega, USA).

2.9 Statistical analyses

Data of this investigation were analyzed with the support of SPSS v.18.0 software (IBM Corporation, USA). The measurement data [mean $\pm$ standard deviation (SD)] were compared through student’s t test or one-way analysis of variance (ANOVA), and enumeration data (n, %) were contrasted utilizing chi-square test. The comparisons were statistically significant when $P$ value was less than 0.05.

3. Results

3.1 Association of lncRNA XIST expression with clinico-pathological features of CRC patients

XIST expression was up-regulated in CRC tissues than in para-carcinoma tissues ( $P<$ 0.05) (Supplementary Fig. 1A), and CRC cell lines, including SW480, HT-29, SW620 and LoVo, revealed higher XIST expression than normal colon epithelial cell line (i.e. NCM460) ( $P<$ 0.05)(Supplementary Fig. 1B). Since that XIST expression was augmented more significantly in SW480 and LoVo cell lines than in other CRC cell lines ( $P<$ 0.05), the couple of cell lines were arranged for subsequent experiments. Furthermore, CRC patients with highly-expressed ( $⩾$ 3.86) XIST were associated with higher odds of large tumor size ( $>$ 5 cm), deep (T3 $+$ T4) infiltration, distant metastasis and advanced TNM stage (i.e. III $+$ IV) than patients carrying lowly-expressed ( $<$ 3.86) XIST (all $P<$ 0.05)(Supplementary Table 2). And highly-expressed XIST, large tumor size, poor differentiation, deep infiltration and distant metastasis were independently predictive of poor 3-year survival of the Chinese CRC cohort (all $P<$ 0.05)(Supplementary Table 3, Supplementary Fig. 1C).

3.2 Silencing of lncRNA XIST undermined proliferation, migration, invasion and chemo-resistance of CRC cells

XIST expression in SW480 and LoVo cell lines was markedly reduced after transfection of si-XIST ( $P<$ 0.05) (Fig. 1A), and CRC cells of si-XIST group became less capable of proliferating (Fig. 1B), migrating (Fig. 1C) and invading (Fig. 1D) than those of si-NC group (all $P<$ 0.05). Furthermore, SW480 (IC50 $=$ 1.503 $\mu$ g/ml) and LoVo cell lines (IC50 $=$ 0.1337 $\mu$ g/ml) in the si-XIST group were vulnerable to the lethal effect of 5-FU as relative to CRC cells in the si-NC group (SW480: IC50 $=$ 4.291 $\mu$ g/ml; LoVo: IC50 $=$ 0.4999 $\mu$ g/ml) ( $P<$ 0.05) (Fig. 1E). Cisplatin-tolerance of SW480 (IC50 $=$ 0.1119 $\mu$ g/ml) and LoVo (IC50 $=$ 0.3032 $\mu$ g/ml) cell lines was also abated in the si-XIST group, when compared with si-NC group (SW480: IC50 $=$ 0.2646 $\mu$ g/ml; LoVo: IC50 $=$ 0.6425 $\mu$ g/ml) ( $P<$ 0.05) (Fig. 1F).

Figure 1.

Activity of colorectal cancer (CRC) cells was affected by silencing of lncRNA XIST. (A) XIST expression was determined within SW480 and LoVo cell lines that were transfected by si-XIST and si-NC. *: $P<$ 0.05 as relative to si-NC group. (B-D) Silencing of XIST undermined the proliferative (B), migratory (C) and invasive (D) capacity of SW480 and LoVo cell lines. *: $P<$ 0.05 as relative to si-NC group. (E-F) Resistance of SW480 and LoVo cell lines against 5-FU (E) and cisplatin (F) were compared between si-XIST group and si-NC group. *: $P<$ 0.05 as relative to si-NC group.

3.3 Silencing of XIST weakened glycolysis of CRC cells by increasing PKM1/PKM2 ratio

SW480 and LoVo cell lines in the si-XIST group consumed smaller amounts of glucose (Fig. 2A) and generated less lactic acid (Fig. 2B) than CRC cells of si-NC group ( $P<$ 0.05). Meanwhile, mRNA level of PKM1 and PKM1/PKM2 ratio both went up significantly in the si-XIST group as compared with si-NC group ( $P<$ 0.05), whereas mRNA level of PKM2 was remarkably declined after transfection of si-XIST ( $P<$ 0.05) (Fig. 2C).

Figure 2.

Glycolysis of colorectal cancer (CRC) cells was regulated by lncRNA XIST. (A-E) Glucose consumption (A), lactate production (B), PKM1/PKM2 ratio (C), oxygen consumption rate (OCR) (D) and extracellular acidification rate (ECAR) (E) of SW480 and LoVo cell lines were measured under treatments of si-XIST and si-NC. *: $P<$ 0.05 as relative to si-NC group. (F) Lactate produced by SW480 and LoVo cell lines was assessed after treatment of 2-deoxy-D-glucose (2-DG). *: $P<$ 0.05 as relative to NC group. (G-H) Tolerance of si-XIST-transfected SW480 and LoVo cell lines was appraised about their response to 5-FU (G) and cisplatin (H). *: $P<$ 0.05 as relative to si-NC group. (I) Inhibitory effect of si-XIST on growth of SW480 and LoVo cell lines was enhanced by addition of 2-DG. *: $P<$ 0.05 as relative to si-NC group.

According to Supplementary Fig. 2A, basal OCR, indicating energy need of cells under normal conditions, was comprised of ATP-linked OCR, which reflected ATP synthesis of mitochondria, and proton leak OCR, which symbolized structural damage of mitochondria [29, 30]. We noticed that basal OCR and ATP-linked OCR of SW480 and LoVo cells were lower in the si-XIST group than in the si-NC group ( $P<$ 0.05) (Fig. 2D). It was implied that si-XIST could reduce basal energy consumption of CRC cells mainly by impairing ATP-synthesis capability, rather than by injuring electron transport chain (ETC) and mitochondrial membrane. More than that, maximal respiration and spare respiratory capacity also descended significantly in SW480 and LoVo cells transfected by si-XIST as relative to those transfected by si-NC ( $P<$ 0.05) (Supplementary Fig. 2A, D), which indicated that si-XIST not only deactivated mitochondria of CRC cells, but also made the cells vulnerable to oxidative stress [31]. In regard to ECAR (Supplementary Fig. 2B), we found that basal glycolysis, maximal glycolytic capacity and glycolytic reserve [29] of SW480 and LoVo cells were restrained after transfection of si-XIST ( $P<$ 0.05) (Fig. 2E), which insinuated that glycolytic potency of CRC cells might be attenuated by si-XIST. Additionally, lactate release of SW480 and LoVo cells was remarkably depressed in the wake of 2-deoxy-D-glucose (2-DG) treatment ( $P<$ 0.05) (Fig. 2F), which was aimed to block glycolysis [32]. Treatment of 2-DG also dramatically enhanced 5-FU/cisplatin-sensitivity of SW480 and LoVo cells in both si-NC group and si-XIST group ( $P<$ 0.05) (Fig. 2G and H). It was noteworthy that 5-FU/cisplatin-resistance of SW480 and LoVo cells in the si-XIST group was diminished less significantly by 2-DG treatment than cells in the si-NC group, since that the difference of IC50 value between 5-FU/cisplatin treatment and 5-FU/cisplatin $+$ 2-DG treatment was smaller in the si-XIST group than in the si-NC group(Supplementary Table 4). Similarly, survival rate of SW480 and LoVo cells in the si-XIST group was decreased less pronouncedly by 2-DG treatment than cells in the si-NC group ( $P<$ 0.05) (Fig. 2I). Taken together, SW480 and LoVo cells of si-XIST group were less dependent on glycolysis than those of si-NC group.

Figure 3.

MiR-137 reversed the promoting effect of lncRNA XIST on proliferation, migration and invasion of colorectal cancer (CRC) cells. (A) Expression of miR-137 was determined among NCM460, SW480, HT-29, SW620 and LoVo cell lines. *: $P<$ 0.05 as relative to NCM460 cell line. (B) MiR-137 expression within SW480 and LoVo cell lines was up-regulated after transfection of miR-137 mimic. *: $P<$ 0.05 as relative to miR-NC group. (C) XIST sponged miR-137 in several sites, and luciferase activity of SW480 and LoVo cell lines was compared among XIST-WT $+$ miR-137 mimic group, XIST-MUT $+$ miR-137 mimic group and XIST-WT $+$ miR-NC group. *: $P<$ 0.05 as relative to XIST-MUT $+$ miR-137 mimic group and XIST-WT $+$ miR-NC group. (D) XIST expression was promoted after transfection of pcDNA3.1-XIST. *: $P<$ 0.05 as relative to pcDNA3.1 group; #: *: $P<$ 0.05 as relative to si-NC group. (E) MiR-137 expression in SW480 and LoVo cell lines was altered after transfection of si-XIST and pcDNA3.1-XIST. *: $P<$ 0.05 as relative to si-NC/pcDNA3.1 group. (F) XIST expression was determined when miR-137 mimic was transfected into SW480 and LoVo cell lines. *: $P<$ 0.05 as relative to miR-NC group. (G-I) Proliferation (G), migration (H) and invasion (I) of SW480 and LoVo cell lines were compared among pcDNA3.1-XIST $+$ miR-137 mimic group, pcDNA3.1-XIST group and NC group. *: $P<$ 0.05 as relative to NC group. #: P $<$ 0.05 as relative to pcDNA3.1-XIST group.

3.4 LncRNA XIST sponged miR-137 and depressed its expression in CRC cells

MiR-137 expression fell significantly in SW480, HT-29, SW620 and LoVo cell lines as relative to NCM460 cell line ( $P<$ 0.05) (Fig. 3A), and miR-137 expression ascended significantly in SW480 and LoVo cells transfected by miR-137 mimic in comparison to miR-NC group ( $P<$ 0.05) (Fig. 3B). Moreover, luciferase activity of SW480 and LoVo cells in the pmirGLO-XIST Wt $+$ miR-137 mimic group was reduced as relative to pmirGLO-XIST Mut $+$ miR-137 mimic group and pmirGLO-XIST Wt $+$ miR-NC group ( $P<$ 0.05) (Fig. 3C), which verified a sponging relationship between XIST and miR-137. In addition, XIST expression in SW480 and LoVo cells was heightened after transfection of pcDNA3.1-XIST ( $P<$ 0.05) (Fig. 3D). MiR-137 expression was decreased after transfection of pcDNA3.1-XIST ( $P<$ 0.05), and was up-regulated when XIST was silenced ( $P<$ 0.05) (Fig. 3E). Nonetheless, no statistical difference was discerned in XIST expression of SW480 and LoVo cells between miR-137 mimic group and miR-NC group ( $P>$ 0.05) (Fig. 3F).

3.5 MiR-137 disturbed contribution of lncRNA XIST to proliferation, migration, invasion and chemo-resistance of CRC cells

Proliferation of SW480 and LoVo cells was restrained in the pcDNA3.1-XIST $+$ miR-137 mimic group, when compared with pcDNA3.1-XIST group ( $P<$ 0.05) (Fig. 3G). Analogously, migration and invasion of SW480 and LoVo cells were impaired after co-transfection of pcDNA3.1-XIST and miR-137 mimic, as relative to pcDNA3.1-XIST group ( $P<$ 0.05) (Fig. 3H and I). Furthermore, SW480 (5-FU: IC50 $=$ 10.18 $\mu$ g/ml; cisplatin: 0.4567 $\mu$ g/ml) and LoVo (5-FU: IC50 $=$ 0.8162 $\mu$ g/ml; cisplatin: 1.015 $\mu$ g/ml) cells in the pcDNA3.1-XIST $+$ miR-137 mimic group were less tolerant to 5-FU (Fig. 4A) and cisplatin (Fig. 4B) than SW480 (5-FU: IC50 $=$ 117.5 $\mu$ g/ml; cisplatin: 1.136 $\mu$ g/ml) and LoVo (5-FU: IC50 $=$ 4.463 $\mu$ g/ml; cisplatin: 3.093 $\mu$ g/ml) cells of pcDNA3.1-XIST group ( $P<$ 0.05).

Figure 4.

MiR-137 obstructed XIST-motivated chemo-resistance and glycolysis in colorectal cancer (CRC) cells. (A-B) Resistance of SW480 and LoVo cell lines against 5-FU (A) and cisplatin (B) were compared among pcDNA3.1-XIST $+$ miR-137 mimic group, pcDNA3.1-XIST group and NC group. *: $P<$ 0.05 as relative to NC group. #: $P<$ 0.05 as relative to pcDNA3.1-XIST group. (C-G) Glucose consumption (C), lactate production (D), PKM1/PKM2 ratio (E), oxygen consumption rate (OCR) (F) and extracellular acidification rate (ECAR) (G) were monitored among SW480 and LoVo cell lines transfected by pcDNA3.1-XIST $+$ miR-137 mimic, pcDNA3.1-XIST and none. *: $P<$ 0.05 as relative to NC group. #: $P<$ 0.05 as relative to pcDNA3.1-XIST $+$ miR-137 mimic group. (H-J) Impacts of cisplatin $+$ 2-deoxyglucose (DG) (H), 5-FU $+$ 2-DG (I) and 2-deoxy-D-glucose (2-DG) (J) on growth of SW480 and LoVo cell lines were weighed among pcDNA3.1-XIST $+$ miR-137 mimic group, pcDNA3.1-XIST group and NC group. *: $P<$ 0.05 as relative to NC group. #: $P<$ 0.05 as relative to pcDNA3.1-XIST group.

3.6 MiR-137 inhibited XIST-activated glycolysis in CRC cells by decreasing PKM2/PKM1 ratio

SW480 and LoVo cells in the pcDNA3.1-XIST+miR-137 mimic group were refrained from intaking glucose (Fig. 4C) and releasing lactate (Fig. 4D), as compared with pcDNA3.1-XIST group ( $P<$ 0.05). Besides, CRC cells transfected by pcDNA3.1-XIST demonstrated higher mRNA level of PKM2, lower mRNA level of PKM1 and a larger ratio of PKM2 mRNA and PKM1 mRNA than untreated CRC cells ( $P<$ 0.05) (Fig. 4E). However, mRNA level of PKM2 and PKM2/PKM1 ratio were reduced, and mRNA level of PKM1 was raised obviously in the pcDNA3.1-XIST group $+$ miR-137 mimic group, when compared with pcDNA3.1-XIST group ( $P<$ 0.05).

With respect to OCR and ECAR, maximal respiration and maximal glycolytic capacity of SW480 and LoVo cells were strengthened in the pcDNA3.1-XIST group as relative to NC group ( $P<$ 0.05) (Fig. 4F and G, Supplementary Fig. 2). On the contrary, SW480 and LoVo cells in the pcDNA3.1-XIST $+$ miR-137 mimic group were associated with weaker maximal respiration capability and lower glycolysis-dependent ECAR than CRC cells in the pcDNA3.1-XIST group ( $P<$ 0.05). Moreover, 5-FU/cisplatin-tolerance of SW480 and LoVo cells dropped significantly after 2-DG treatment (Fig. 4H and I) in comparison to without 2-DG treatment (Fig. 4A and B), and decreases in the IC50 value were less prominent in the pcDNA3.1-XIST $+$ miR-137 mimic group than in the pcDNA3.1-XIST group ( $P<$ 0.05)(Supplementary Table 4). Not only that, after treatment of 2-DG, survival of SW480 and LoVo cells was depressed less significantly in the pcDNA3.1-XIST $+$ miR-137 mimic group than in the pcDNA3.1-XIST group ( $P<$ 0.05) (Fig. 4J). Collectively, pcDNA3.1-XIST-enhanced reliant on glycolysis and 5-FU/cisplatin-resistance in CRC cells could be ameliorated by miR-137 mimic.

4. Discussion

Even if early-stage CRC was curable by surgery, massive CRC patients already exacerbated to the advanced stage at diagnosis. Huge endeavors have been made to develop chemotherapies for late-stage CRC, such as 5-FU/leucovorin combined with oxaliplatin/irinotecan [33], nevertheless, growing drug-resistance readily gave rise to failure in CRC treatment [34]. It was noteworthy that tumor cells were equipped with incremental glycolytic power [35], so elucidation of energy metabolism in CRC might be conducive to improving chemotherapeutic regimens.

LncRNAs played irreplaceable roles in a plethora of biological activities, including dynamic change of chromosome structure, genomic imprinting and dosage compensation, and their abnormal expression imposed sizable effects on malignant transformation of normal cells [36, 37]. LncRNA XIST, formed during transcriptional silencing of X chromosome, not only served as an oncogene in CRC [38, 39], but also spurred development of gastric cancer [40], nasopharyngeal carcinoma [41], non-small cell lung cancer [42] and pancreatic cancer [43]. Consistent with former documentations, we noticed that XIST encouraged proliferation, migration and invasion of CRC cells (Fig. 1B–D), which molecularly explained why high XIST expression was suggestive of CRC patients’ undesirable prognosis(Supplementary Fig. 1C). Furthermore, glycolysis of CRC cells was also explored under the influence of si-XIST (Fig. 2B–E), which was a novelty of this investigation. It was reported that glycolysis produced great deals of energy [44], and created an acidic environment made up of high-concentration lactate, which altogether geared up metastasis of tumor cells [45]. Our investigation revealed that maximal respiration (Fig. 2D), maximal glycolytic capacity (Fig. 2E), glucose uptake (Fig. 2A) and lactate release (Fig. 2B) of CRC cells were all decreased by si-XIST, which suggested that si-XIST depressed CRC progression partly by obstructing glycolysis of tumor cells [46]. Moreover, 5-FU/cisplatin-tolerance of CRC cells was also relevant to glycolysis (Fig. 2G and H), owing to that CRC cells with incremental 5-FU/cisplatin-sensitivity, triggered by si-XIST (Fig. 1E and F), depended less on glycolysis to maintain survival. Summing up the above, glycolysis blocked by si-XIST might reverse malignant phenotype and promote drug-sensitivity of CRC cells.

In addition, a sponging relationship was found to exist between XIST and miR-137 in CRC (Fig. 3C), and miR-137 could antagonize the oncogenic impact of XIST in CRC (Fig. 3H and I), which was consistent with the finding of Liu et al. [47]. Apart from CRC, miR-137 also prevented development of glioma [48], oral squamous cell carcinoma [49], gastric cancer [50] and ovarian cancer [51], implying that miR-137 might impose identical effects in diverse neoplasms. Furthermore, PKM2/PKM1 ratio, a pivotal index of glycolysis in CRC [52], was subjected to regulation of miR-137 (Fig. 4E) and XIST (Fig. 2C). There were multiple proofs that measuring PKM2 level in blood and feces was a promising approach to diagnose cancer and to track tumor progression [25]. For example, PKM2 expression was dramatically elevated with increasing severity of glioma patients [53], and patients with esophageal squamous cell carcinoma were inclined to enjoy longer life span if they carried lower PKM2 expression [54]. Not only that, phosphorylation of PKM2 could activate $\beta$ -catenin signaling [55] and drive expression of oncogenes in CRC (e.g. cyclin D1) [56], which persuasively expounded the CRC-promoting role of PKM2. Hence, it seemed tenable that XIST motivated glycolysis of CRC cells by suppressing the inhibitory effect of miR-137 on PKM2/PKM1 ratio (Fig. 4F–G). Last but not the least, miR-137 could diminish XIST-enhanced 5-FU/cisplatin-resistance of CRC cells (Fig. 4A and B), and miR-137 also ameliorated glycolysis-dependence of 5-FU/cisplatin-resistant CRC cells which was intensified by XIST (Fig. 4H and I). Allowing for that PKM2 was a contributor to CRC chemo-resistance [57], XIST/miR-137 axis was likely to implicate in energy metabolism and chemo-resistance of CRC by reshaping PKM2/PKM1 ratio.

In conclusion, XIST undermined the function of miR-137 and then led to biased ratio of PKM1 and PKM2, which eventually drove glycolysis and reinforced chemo-tolerance of CRC. However, several pitfalls were present in the experimental design of this investigation. Firstly, we failed to incorporate populations of different ethnicities or with distinct living habits, so whether conclusions of this study could be applied to other populations was uncertain. Secondly, rat models were not established because of technical obstacles, so that the in-vivo connection of XIST/miR-137/PKM axis with CRC glycolysis and chemo-resistance could not be validated. Finally, other glycolysis-relevant non-coding RNAs, including lncRNAs and circRNAs, should also be probed into, through which the molecular network that explained CRC etiology could be enriched.