In order to determine whether MSCs can promote stemness and chemoresistance in GC cells, GC cell lines AGS and MKN45 were co-cultured with MSCs via transwell co-culture system. After co-culturing, stemness genes CD133, OCT4, SOX2 and LIN28 expression levels were enhanced in both mRNA and protein levels (Supplementary Fig. 1a-b). The population of CD44 positive (CD44+) GC cells, recognized as cancer stem cell (CSC) typical markers, was significantly increased as detected by flow cytometric analysis (Supplementary Fig. 1c). Moreover, sphere-formation ability of GC cells and the expression levels of stemness genes in sphere-forming GC cells were obviously increased when co-cultured with MSCs indicating the enhanced self-renewal capability of GC cells (Fig. 1a, b). Colony-formation assay showed that growth inhibition effects of 1 μg/mL 5-FU and 3 μg/mL oxaliplatin on GC cells were remarkably decreased in the presence of MSCs (Fig. 1c). These findings suggested that MSCs can promote stemness and chemoresistance in GC cells.
To further investigate the effect of MSCs on tumor initiation, we subcutaneously injected limiting dilution of MKN45 cells at three dosages, 5 × 103, 5 × 104, and 5 × 105, with or without 5 × 106 admixed MSCs. Compared to injection of GC cells alone, the presence of MSCs increased the tumor-initiating frequencies from 0/6, 0/6, and 3/6 to 4/6, 6/6, and 6/6. Notably, injection of GC cells alone was not able to induce tumor formation even with the cell count of 5 × 103 or 5 × 104 cells after 6 weeks. However, when they were mixed with 5 × 106 MSCs, tumor formation occurred within 3 weeks or 1 week, which indicates that MSCs induce de novo tumor formation of GC cells (Fig. 1d, e). As measured by the extreme limiting dilution analysis (ELDA) , the frequency of tumor initiation in GC cells alone was from 3.88 × 10–7 to 3.70 × 10–6; in the presence of admixed MSCs the frequency was increased, from 7.86 × 10−5 to 6.15 × 10−4 (Fig. 1f). These results indicate that MSCs enhance tumor initiation and are responsible for facilitating de novo tumor formation.
To determine whether MSCs were recruited to tumor tissues and accounted for poor outcome of GC patients, we analyzed the expression levels of MSC surface antigen markers, CD29 and CD90, in The Cancer Genome Atlas (TCGA) database. CD29 and CD90 double-positive MSCs were more abundant in GC tissue as compared to non-cancerous gastric tissue and also associated with shortened recurrence-free survival (RFS) in patients (Fig. 1g, h). Additionally, we analyzed the expression levels of CD29 and CD90 in primary GC samples and matched adjacent nontumorous gastric tissues from 151 patients by immunohistochemistry (IHC). The results showed that the expression levels of CD29 and CD90 were expression correlated with more advanced tumor, node metastasis stages, tumor invasion, lymph node, and distant metastases and recurrence (Fig. 1i, j and Table 1). Kaplan–Meier analysis showed that higher co-expression of CD29 and CD90 was correlated with worse disease-free survival (DFS) (stage I–III GC patients) and overall survival (OS) (stage IV GC patients) (Fig. 1k, l). Additionally, the score of stemness gene OCT4 was higher in CD29 and CD90 double-positive GC tissues in comparison with CD29 and CD90 double-negative GC tissues (Supplementary Fig. 1d-e). These results suggest that MSCs promote stemness and chemoresistance and predict a poor outcome in GC patients.
FAO plays an important role in MSC induced-stemness and chemoresistance
During the past decade, researchers have gained better understanding in the relationship between cancer metabolism and cancer stemness [23,24,25]. However, the contribution of lipid catabolism to cancer stemness remains unclear. Recent studies have demonstrated that stemness maintenance benefits from FAO by mitochondrial metabolism of FAs [13, 26]. Hence, we wondered whether FAO played an important role in MSC-inducing stemness and chemotherapy resistance. We found that the expression levels of FAO-associated enzymes, carnitine palmitoyltransferase 1 (CPT1) and acetyl-coenzyme A synthetase (ACS), were increased in both GC cells and sphere-forming GC cells when co-cultured with MSCs (Fig. 2a, b). Consistently, FA uptake, CPT1 activity, FAO rate, and ATP level of GC cells were enhanced in the presence of MSCs (Supplementary Fig. 2a and Fig. 2c–e).
To investigate the role of FAO in stemness and chemotherapy resistance, AGS and MKN45 cells were transfected with siCPT1 sequences, which decreased CPT1 mRNA and protein expression (Supplementary Fig. 2b-c). Expression levels of stemness genes were remarkably reduced with silencing of CPT1 (Fig. 2f). MTT (3-(4,5-dimethyl-2-thiazolyl)−2,5-diphenyltetrazolium bromide) assay showed that inhibition of CPT1 expression increased the growth inhibition effect in GC subjected to 5-FU and oxaliplatin, suggesting that silence of CPT1 enhances the sensitivity of GC cells to chemotherapy (Supplementary Fig. 2d).
Next FAO inhibitor etomoxir (ETX), which specifically suppresses CPT1, was employed to further investigate the contribution of FAO to MSC-induced stemness and chemoresistance. We treated GC cells with or without 100 μmol/L ETX for 48 h cultured either alone or with MSCs. As expected, the expression levels of stemness genes and the ability of sphere formation were decreased when GC cells co-cultured with MSCs were exposed to ETX, compared to their control group without ETX (Fig. 2g–i). Colony formation showed that sensitivity to 5-FU and oxaliplatin were increased in GC cells with MSCs when ETX was added (Fig. 2j). Besides, ETX impaired ATP production in GC cells co-cultured with MSCs (Fig. 2k). These results suggest that inhibition of FAO can abrogate the influence of MSCs in promoting stemness and chemoresistance in GC cells, implicating that FAO is required for maintaining MSC-regulated stemness and chemotherapy resistance.
MACC1-AS1 is induced by MSCs and contributes to stemness and chemoresistance
MACC1-AS1 is the antisense RNA of metastasis-associated in colon cancer-1 (MACC1). We previously reported that MACC1-AS1 is overexpressed in GC tissues and facilitates metabolic plasticity through antioxidant production, resulting in a lower ROS load under metabolic stress . Lower ROS levels are critical for maintaining stemness and drug resistance [19,20,21]. Thus we hypothesized MACC1-AS1 may contribute to stemness and chemoresistance. We found that the expression of MACC1-AS1 was obviously increased both in GC cells and sphere-forming GC cells after co-culture with MSCs (Fig. 3a, b). Additionally, compared to injection of GC cells alone, the expression of MACC1-AS1 in subcutaneous tumor of nude mice was significantly increased in GC cells admixed with MSCs (Fig. 3c). We also performed in situ hybridization (ISH) to evaluate the correlation between MACC1-AS1 expression and co-expression of CD29 and CD90 in GC specimens (Supplementary Fig. 3a). The score of MACC1-AS1 was higher in CD29 and CD90 double-positive GC tissues in comparison to CD29 and CD90 double-negative GC tissues (Fig. 3d). These results indicate that MSCs promote the expression of MACC1-AS1 in GC cells.
Given the promotion of MSCs on MACC1-AS1 expression in GC cells, we next investigated the role of MACC1-AS1 in stemness and chemoresistance. To evaluate whether MACC1-AS1 is functionally involved in GC progression, MACC1-AS1 was overexpressed in stable lentivirus transfected way and the overexpression efficiency was examined (Supplementary Fig. 3b). qRT-PCR and western blotting showed that stemness genes were enhanced after stable transfection with MACC1-AS1 (Fig. 3e, f). The population of CD44+ GC cells evaluated by flow cytometric analysis was significantly increased (Supplementary Fig. 3c) and sphere-formation assay showed that sphere-formation capability of GC cells was obviously increased after overexpression of MACC1-AS1 (Fig. 3g). Moreover, overexpression of MACC1-AS1 facilitated the ability of clone formation and reduced growth inhibition when treated with 5-FU and oxaliplatin (Fig. 3h and Supplementary Fig. 3d). To further investigate the role of MACC1-AS1 in chemotherapy, 5FU and oxaliplatin-resistant cell lines were established (Supplementary Fig. 3e). We found that the expression of MACC1-AS1 was gradually increased with augment of 5-FU and oxaliplatin concentration during the establishment of the chemotherapy-resistant cells (Supplementary Fig. 3f). Additionally, the expression levels of stemness genes and FAO relative genes and CPT1 activity were increased in 5-FU and oxaliplatin-resistant cells, compared to 5-FU and oxaliplatin-sensitive cells (Supplementary Fig. 3g–h). Collectively, the results suggest a possible role of MACC1-AS1 in the cross-talk of MSCs and GC cells, participating in MSC-mediated stemness and chemoresistance of GC cells.
MSC-derived TGF-β1 contributes to MACC1-AS1 upregulation in GC cells
It is reported that MSCs communicate with cancer cells mostly by secreting soluble factors. We collected and fractioned MSC-conditioned medium (CM) based on molecular weight into <3-kD and >3-kD fractions via selective filters and centrifugation. We found that only >3 kD MSC-CM induced MACC1-AS1 upregulation in GC cells (Fig. 3i). These results strongly suggest that the >3-kD fraction of the MSC-CM contains at least one soluble factor that contributes to MSC-mediated stemness and chemoresistance through MACC1-AS1. TGF-β1, a key cytokine that can be secreted by MSCs, has been considered as a critical factor involved in epithelial–mesenchymal transition, metastasis, and stemness [27,28,29]. Based on the fact that >3 kD MSC-CM induced MACC1-AS1 upregulation, it is likely that TGF-β1 secreted by MSCs contributes to MACC1-AS1 induction.
To verify this possibility, TGF-β1 level from MSC-CM of <3-kD and >3-kD fractions was evaluated by enzyme-linked immunosorbent assays. Compared to MSC-CM of <3 kD and the GC medium collected from AGS and MKN45, the level of TGF-β1 in >3-kD fractions was significantly higher (Fig. 3j). As we have known, TGF-β exerts cellular effects by binding to heterotetrameric complexes of type I and type II serine/threonine kinase receptors (TGFβR-I and TGFβR-II) and activates the downstream SMAD family [30,31,32]. We found that the expression levels of p-TGFβR-I, p-TGFβR-II, p-SMAD2, and p-SMAD3 were increased when GC cells were co-cultured with MSCs (Fig. 3k), whereas that of SMAD7 were decreased (Supplementary Fig. 3i). These results suggest that TGF-β1 secreted by MSCs contributes to stemness and chemoresistance through binding to TGF-β receptors and activating SMAD2 and SMAD3. To test the effect of TGF-β1 on MACC1-AS1, TGF-β1 (20 μg/mL) was supplemented to GC medium for 24 h. We found that MACC1-AS1 was obviously increased with TGF-β1 treatment (Fig. 3l). Furthermore, TGF-β1 inhibitor disitertide and TGFβR-I inhibitor LY-364947 abrogated the MACC1-AS1 expression in GC cells treated with MSC-CM of >3-kD fraction (Fig. 3m, n). These findings suggest that TGF-β1, as one of the soluble factors secreted by MSCs, is essential for MACC1-AS1 upregulation.
The role of MACC1-AS1 on stemness and chemoresistance is dependent on FAO
Based on the existing results, we hypothesized that MACC1-AS1 may be a key participant in FAO-regulated stemness and chemoresistance. We performed RNA sequencing technique (RNA-seq) to profile gene expression after MACC1-AS1 overexpression. BioCyc pathway enrichment revealed that FAO pathway was activated in cancer cells with MACC1-AS1 overexpression (Fig. 4a). Then we found that overexpression of MACC1-AS1 gradually upregulated the levels of CPT1 and ACS as the concentration of MACC1-AS1 transient plasmid increased (Fig. 4b, c and Supplementary Fig. 4a). Consistent results were obtained from detection of FAO rate, ATP production, and FA uptake (Fig. 4d, e and Supplementary Fig. 4b). Next we further assessed whether the promoting effect of MACC1-AS1 in stemness and chemoresistance depended on FAO. As expected, the expression levels of stemness genes were decreased in MACC1-AS1-overexpressing GC cells with 100 μmol/L ETX for 48 h in comparison to MACC1-AS1-overexpressing GC cells without ETX (Fig. 4f and Supplementary Fig. 4c). The same result was obtained by suppression of CPT1 with siRNA#1 (Fig. 4g). Moreover, ETX abrogated the promoting effect of MACC1-AS1 on the ability of self-renewal and reduced mitochondria ATP production accelerated by MACC1-AS1 (Fig. 4h, i). MTT assay showed that ETX treatment partially reversed MACC1-AS1-mediated resistance to 5-FU and oxaliplatin (Fig. 4j). Taken together, FAO is greatly responsible for MACC1-AS1-mediated stemness and chemoresistance of GC cells.
miR-145-5p is downstream of MACC1-AS1 to promote stemness and chemoresistance through FAO
Emerging evidences suggest that lncRNAs are likely to function through interaction with microRNAs (miRNAs). To elucidate the mechanism underlying the contribution of MACC1-AS1 to FAO-dependent stemness and chemoresistance, we found that miR-145-5p was one of the target miRNAs of MACC1-AS1 predicted by LncRNASNP database (Fig. 5a). Previous studies have reported that miR-145-5p exerts an inhibitory effect on drug resistance [33, 34] and is correlated with lipid metabolism [35, 36]. Therefore, we wondered whether miR-145-5p played a role in MACC1-AS1-induced stemness and chemoresistance.
First, miR-145-5p expression was significantly decreased after overexpression of MACC1-AS1 and it was decreased as the concentration of MACC1-AS1 transient plasmid increased (Fig. 5b and Supplementary Fig. 5a). Thus dual luciferase reporter assay was employed to further confirm their relationship. The result showed that overexpression of miR-145-5p decreased the reporter activity of MACC1-AS1 wild type (WT) in 293T cell and MKN45 cell but not that of MACC1-AS1 mutant type (MUT), confirming the direct binding sites predicted by LncRNASNP database (Fig. 5c and Supplementary Fig. 5b). For validation, we performed RNA pull down assay to capture RNA complexes associated with miR-145-5p probes. Compared to the control probe, miR-145–5p positive probe successfully pulled down MACC1-AS1 in MKN45 cells (Fig. 5d). Taken together, these data suggest that MACC1-AS1 is able to directly bind to miR-145-5p and regulate its activity.
To determine the role of miR-145-5p in FAO-dependent stemness and chemoresistance, we found that miR-145-5p significantly repressed stemness genes and FAO enzyme expression levels (Supplementary Fig. 5c–d). Moreover, miR-145-5p reduced FAO rate and ATP production and promoted ROS production and sensitivity to 5-FU and oxaliplatin (Supplementary Fig. 5e–h). When MACC1-AS1 and miR-145-5p were co-transfected into GC cells, miR-145–5p remarkably repressed the promoting effect of MACC1-AS1 on the expression levels of stemness genes and FAO enzymes (Fig. 5e, f). In addition, overexpression miR-145-5p significantly reduced the FA uptake, FAO rate, and ATP production induced by MACC1-AS1 overexpression (Supplementary Fig. 5i and Fig. 5g, h). MTT assay showed that miR-145-5p partially rescued the inhibitory impact of MACC1-AS1 on sensitivity of GC cells to 5-FU and oxaliplatin (Fig. 5i). These data suggest that MACC1-AS1/miR-145-5p signaling axis contributes to the FAO-dependent stemness and chemoresistance.
Inhibition of FAO attenuates MSC-induced chemoresistance in vivo
We found that higher CPT1 expression in GC patients was associated with shorter OS and progression-free survival (PFS) analyzed by the online Kaplan Meier Plotter database (Fig. 6a, b). Then we analyzed the expression of CPT1 in primary GC samples and matched adjacent nontumorous gastric tissues by IHC. Compared to the matched adjacent nontumorous gastric tissues, CPT1 score was higher in GC (Fig. 6c, d). It was also higher in CD29 and CD90 double-positive GC tissues in comparison to CD29 and CD90 double-negative GC tissues (Fig. 6e). Moreover, higher CPT1 expression was linked to worse DFS in stage I–III patients and worse OS in stage IV patients (Fig. 6f, g).
We further investigated whether inhibition of FAO attenuated MSC-induced chemoresistance in vivo. FOLFOX regiment, composed of 5-FU, oxaliplatin, and calcium folinate, is the first-line therapy of metastatic GC . The results showed that the tumor volumes of FOLFOX-treated mice were remarkably decreased compared to mice without FOLFOX treatment. MSCs remarkably abolished the antitumor effects of FOLFOX, but intraperitoneal (i.p.) injection with ETX significantly reduced the tumor volumes (Fig. 6h, i). Furthermore, IHC staining of tumors showed that the expression levels of OCT4, CPT1, and proliferation-related nuclear antigen ki-67 were decreased after ETX treatment, while marker of apoptosis cleaved caspase 3 was increased (Fig. 6j). No significant difference was observed in tissues and structures between groups treated with or without ETX by hematoxylin–eosin staining of the lungs, livers, spleens, kidneys, and hearts of xenograft nude mice (Fig. 6k). Taken together, these results suggest that inhibition of FAO attenuates MSC-induced chemoresistance in vivo.
Source : https://www.nature.com/articles/s41388-019-0747-0