N6‑methyladenosine upregulates miR‑181d‑5p in exosomes derived from cancer‑associated fibroblasts to inhibit 5‑FU sensitivity by targeting NCALD in colorectal cancer.

Resistance to 5‑Fluorouracil (5‑FU) is a frequent occurrence in patients with colorectal cancer (CRC). MicroRNAs (miRNAs) from cancer‑associated fibroblasts (CAFs)‑secreted exosomes have been associated with 5‑FU sensitivity. The potential molecular mechanism of CAFs‑exosomal miRNAs in CRC remains unclear. The aim of the present study was to elucidate the role of exosomal miRNAs in 5‑FU sensitivity in CRC. Exosomes derived from CAFs were extracted. Exosomal miR‑181d‑5p was identified as a miRNA associated with 5‑FU sensitivity. The putative function of exosomal miR‑181d‑5p was evaluated by ethynyl‑2‑deoxyuridine staining, flow cytometry, RNA immunoprecipitation, luciferase reporter assay, tumor xenograft formation, reverse transcription‑quantitative PCR and western blot analysis. Modification of miR‑181d‑5p by the RNA N6‑methyladenosine (m6A) methyltransferase like (METTL)3 was examined by m6A methylation analysis. The results indicated that m6A modification and METTL3 expression were upregulated in CRC patients. METTL3‑dependent m6A methylation promoted the miR‑181b‑5p process by DiGeorge Syndrome Critical Region 8 (DGCR8) in CAFs. CAFs‑derived exosomes inhibited 5‑FU sensitivity in CRC cells through the METTL3/miR‑181d‑5p axis. A mechanistic study revealed that miR‑181d‑5p directly targeted neurocalcin δ (NCALD) to inhibit the 5‑FU sensitivity of CRC cells. Patients with higher NCALD levels exhibited a higher survival rate. Taken together, METTL3‑dependent m6A methylation was upregulated in CRC to promote the processing of miR‑181d‑5p by DGCR8. This led to increased miR‑181d‑5p expression, which inhibited the 5‑FU sensitivity of CRC cells by targeting NCALD. The results of the present study provided novel insight into exosomal microRNAs in 5‑FU sensitivity in CRC cells. Furthermore, exosomal miR‑181d‑5p may represent a potential prognostic marker for CRC.

Introduction

Colorectal cancer (CRC) is a leading cause of cancer-related deaths (1). Early CRC can be resected endoscopically (2). Unfortunately, most CRCs are diagnosed after local or regional spread, therefore, laparoscopic surgery or partial colectomy must be performed (3,4). Furthermore, recurrence is common following surgery (5). Adjuvant chemotherapy may prevent recurrence and improve survival rate (6). 5-Fluorouracil (5-FU)-based chemotherapy is widely used for CRC treatment (7); however, the development of resistance to 5-FU is a limitation to CRC treatment (8). The tumor microenvironment (TME) or stroma interacts closely with tumor cells (9). Cancer-associated fibroblasts (CAFs), a component of the tumor stroma, builds up and remodels the extracellular matrix structure (10). Studies have indicated that CAFs are implicated in tumorigenesis and tumor progression (11,12). Keller et al (13) demonstrated that co-culture of CAFs and the HT-29 human colon adenocarcinoma cell line, enhances proliferation and metabolism of this CRC cell line. Co-culture of CAFs and CRC cells has also been shown to promote metastasis and chemo-resistance (14). The accumulation of CAFs in the TME is correlated with poor prognosis and recurrence of CRC (15).

Exosomes are a type of extracellular vesicle (EVs) that regulates various biological processes (16). Exosomes contain proteins, lipids, DNA and RNA from the cells that secrete them (17). Increasing evidence suggests that exosomes also contain microRNAs (miRNAs) (18,19). miRNA-carrying exosomes can be taken up by recipient cells and exosomal miRNAs have been shown to affect tumor proliferation and metastasis (16,20). All prokaryotic and eukaryotic cells release EVs as part of their normal physiology and during abnormal conditions (17). A previous study also indicates that miR-181d-5p-containing exosomes derived from CAFs promote epithelial-mesenchymal transition (EMT) by regulating CDX2/HOXA5 in breast cancer (21). Hepatoma cell-derived EVs promote the differentiation of bone marrow mesenchymal stem cells and promote liver cancer metastasis through the delivery of miR-181d-5p (22). Although it is clear that exosomal miRNAs from CAFs play a role in various cancers, the regulatory mechanism of exosomal miRNAs from CAFs is largely unknown.

N6-methyladenosine (m6A) methylation plays an important role in the regulation of gene expression and cell fate (23). m6A methylation also regulates the generation and function of noncoding RNAs, including miRNAs (24). m6A methylation is initiated by a methyltransferase complex which consists of Wilms tumor 1-associating protein (WTAP), methyltransferase like (METTL) 3 and METTL14 (25). Liu et al (26) found that METTL3 and WTAP are markedly upregulated in tumor tissues, whereas METTL14 is downregulated in CRC. m6A modification results in increased miR-25-3p to promote pancreatic cancer growth (27). Microprocessor DiGeorge Syndrome Critical Region 8 (DGCR8) processes primary microRNAs (pri-miRNAs) during miRNA biogenesis by interacting with the apical UGU motif of pri-miRNAs (28). m6A modification could mark pri-miRNAs for processing by recognizing DGCR8 in a manner dependent on METTL3 or METTL14/m6A, suggesting that altered METTL3 or METTL14/m6A may contribute to the aberrant expression of miRNAs in a number of biological processes, including cancer (29,30). However, the regulatory mechanism of m6A methylation in CAFs and its effect on tumorigenesis and development remain to be elucidated.

The present study determined the function of m6A methylation in CRC and revealed the mechanisms through which m6A regulates CRC. It first demonstrated that the level of m6A methylation was markedly enhanced in CRC tissues and METTL3 was a key factor for aberrant m6A methylation. Next, the present study showed that METTL3 upregulation suppressed 5-FU sensitivity by regulating CAFs-derived exosomal miR-181d-5p through modulation of DGCR8, which binds to pri-miR-181d-5p in an m6A-dependent manner. Overexpression of miR-181d-5p inhibited 5-FU sensitivity by targeting NCALD. Based on these data, the present study proposed that m6A modification upregulated miR-181d-5p in CAFs-derived exosomes to inhibit chemosensitivity by targeting NCALD in CRC.

Materials and methods

Human specimens and cell culture

A total of two cohorts of patients (age ranging from 23-86 years) with CRC treated between June 2016 and June 2019 at the Shanghai Eighth People Hospital were included. Cohort 1 consisted of 30 fresh CRC tissues and cohort 2 included 111 formalin-fixed paraffin-embedded CRC tissues, which were determined to be intratumoral by a pathologist. Tumor-node-metastasis (TNM) staging was performed based on pathology reports and histologic sections. The present study was carried out according to the provisions of the Declaration of Helsinki of 1975. Fibroblasts were isolated from surgically resected CRC (named cancer-associated fibroblasts, CAFs) and adjacent-normal tissues (named adjacent-normal fibroblasts; NFs) in cohort 1 as previously described (14,31). Briefly, CRC tissues were minced with a sterile blade and resuspended in a solution of DMEM with 20% fetal bovine serum (FBS, Gibco; Thermo Fisher Scientific, Inc.), penicillin/streptomycin, amphotericin B and 3% collagenase type I (Sigma-Aldrich; Merck KGaA) for 2 h at 37°C. The samples were then filtered through an 8 µm mesh to remove undigested debris. The single cell suspension with viable fibroblasts was cultured in DMEM (with 10% FBS) at 37°C for 2 to 3 weeks in a 24-well plate and then transferred to a T75 flask for continuous maintenance. The counterpart fibroblasts were collected ≥5 cm away from the tumor. Primary fibroblasts used in the present study were between passages 2 and 5. Fibroblasts (passage 3) identified by immunocytochemistry staining using anti-vimentin (Abcam; cat. no. ab92547; 1:1,000), anti-α-smooth muscle actin (Abcam; cat. no. ab124964; 1:500), anti-FAP (eBioscience; cat. no. BMS168; 1:500), anti-CK19 (Abcam; cat. no. ab52625; 1:500), or anti-CD31 (Abcam; cat. no. ab9498; 1:500), which showed strong vimentin, α-smooth muscle actin and fibroblast activation (FAP) staining and negative CK19 and CD31 staining (Fig. S1) were transfected with plasmid PBabe-SV40-T (neo) by electroporation (280V, 960 mF) for cell immortalization as previous described (32). Primary fibroblasts were free of cross-contamination and authenticated using the short tandem repeat profiling method. In cohort 2, 90 patients had received appropriate chemotherapy or radiotherapy. Of these, 78 patients received 5-FU-based chemotherapy. Using the proportion of changes in tumor volume after treatment, patients were divided into four subgroups: i) Complete response (CR; no tumor), ii) partial response (PR; tumor shrinkage by more than 50%), iii) stable disease (tumor shrinkage by less than 50% or tumor enlargement by more than 25%) and iv) progressive disease (PD; tumor enlargement by >25%). CR plus PR was defined as chemotherapy-sensitive patients and stable disease plus PD as chemotherapy-resistant patients. Written informed consents were waived because of the retrospective nature of the present study, which was approved by the Ethics Committee of Shanghai Eighth People Hospital (approval no. 2021-YS-067).

Immunohistochemical staining

Tissues were fixation with 4% paraformaldehyde, embedded in paraffin and then subjected to standard dewaxing and rehydration. The sections were incubated in citric acid buffer (pH 6.0) for 15 min for antigen retrieval, followed by incubation for 10 min with 3% H2O2 solution to inactivate endogenous enzymatic activity. The sections were then incubated with anti-NCALD antibody (ProteinTech Group, Inc.; cat. no. 12925-1-AP; 1:200) or anti-METTL3 antibody (Abcam; cat. no. ab195352; 1:500) for 1 h at 25°C followed by Elivision plus Polyer HRP (mouse/rabbit) immunohistochemistry kit (Maxim Biotech, Inc.; cat. no. KIT-9903; 1:1,000) for 30 min at 20°C. The results were evaluated by two pathologists who were blinded to the patients. Immunoreactivity was scored using the H-score system (33) ranging from 0 to 4. The staining intensity ranged from 0-3, thus giving a range from 0-12. Patients were divided into a low-expression group (score <4) or a high-expression group (score ≥4).

m6A methylation analysis

TRIzol reagent (Thermo Fisher Scientific, Inc.) was used to extract total RNA. Poly(A)+ RNA was purified using GenElute mRNA Miniprep kit (Sigma-Aldrich; Merck KGaA). m6A levels were measured by m6A RNA Methylation Assay kit (Abcam; cat. no. ab185912). Briefly, 80 µl of binding solution and 200 ng of sample RNA were added to each designated well and then incubated at 37°C for 90 min for RNA binding. Each well was washed three times with wash buffer from the kit. Then, 50 µl of the diluted capture antibody was added to each well and incubated at room temperature for 60 min. Each well was subsequently incubated with detection antibody and enhancer solution at room temperature for 30 min. Finally, the wells were incubated with developer solution in the dark for 1-10 min at 25°C. The reactions were stopped with stop solution and measured using a microplate reader at 450 nm wavelength within 2-10 min.

Cell culture

CRC cells (HCT116 and HT29) were obtained from the Shanghai Biology Institute and cultured in 95% humidity, 5% CO2 and 37°C in DMEM containing 10% FBS and 1.0% penicillin-streptomycin (Beijing Solarbio Science & Technology Co., Ltd.).

Isolation and characterization of exosomes

CAFs were cultured in DMEM complete medium with exosome-free FBS (Thermo Fisher Scientific, Inc.; cat. no. A2720801) for 2 days. The medium was centrifuged at 2,200 × g for 15 min and 11,000 × g for 35 min at 4°C, then filtered through a 0.22-mm filter. The medium was then centrifuged at 110,000 × g for 100 min at 4°C. The resulting pellets were resuspended and centrifuged at 110,000 × g for 100 min at 4°C and resuspended in 50 µl of PBS. The morphology of exosome was observed by transmission electron microscopy (FEI Tecnai 12, Philips Medical Systems B.V.). Briefly, exosomes were fixed with 4% paraformaldehyde for 20 min at 4°C and spotted onto glow-discharged copper grids. The copper grids were dried for 10 min at room temperature. Samples were stained with 2% uranyl acetate and dried for 10 min. Then samples were observed at 100 kV.

Size distribution of exosomes was analyzed using the Flow Nano Analyzer (NanoFCM Co., Ltd.) according to manufacturer. s protocol and the data was processed using NanoFCM software (NanoFCM Profession V1.0; NanoFCM Co., Ltd.). To monitor the interaction between exosomes and CRC cell lines, the exosomes were labeled with the PKH67 green fluorescent cell linker kit (Sigma-Aldrich; Merck KGaA) and co-cultured with CRC cells for 24 h at 37°C. The uptake of exosomes by HCT116 and HT29 cells was assessed by confocal microscopy (Olympus FV1200; Olympus Corporation) ≥5 fields of view randomly selected in each section and counted at ×400 magnification.

Co-culture assay

Using a Transwell chamber (Thermo Fisher Scientific, Inc.; 0.4 mm pore size), CRC cells (5×104) were cultured in DMEM containing 10% exosome-free FBS in the lower compartment and the same number of fibroblasts was seeded on the Transwell at 37°C. Fibroblasts were treated with or without 20 µM GW4869 for 24 h before seeding. After 48 h of co-culture, HCT116 and HT29 cells were collected for further cytological experiments.

In vitro exosome supplementation

HCT116 and HT29 cells were provided with fresh DMEM media plus 100 µg/ml exosomes isolated from CAFs for 48 h.

Lentiviral packaging and infection

Short hairpin (sh) RNAs targeting METTL3 were designed and synthesized by Obio Technology Company (shMETTL3-1, GCT GCA CTT CAG ACG AAT T; shMETTL3-2, GGA TAC CTG CAA GTA TGT T) and inserted into the pLKO.1 lentivirus plasmid (Addgene, Inc.). A non-targeting sequence (GTC ACG CCA TAC CTA ATA C) was inserted into the pLKO.1 lentivirus plasmid as a sh negative control (NC). Human METTL3- or NCALD-encoding DNA was cloned into pLVX-Puro lentivirus plasmid (Clontech; Takara Bio USA) between EcoRI and BamHI restriction sites to form the overexpression vector pLVX-Puro-METTL3 or pLVX-Puro-NCALD. Empty pLVX-Puro lentivirus plasmid was regarded as vector control. A lentiviral package was generated by introducing lentiviral plasmids and helper virus packaging plasmids (pMD2G and psPAX2; 10:9:1) into 293T cells using Lipofectamine 2000® (Invitrogen; Thermo Fisher Scientific, Inc.) for 6 h at 37°C. After 48 h of transfection, the recombinant lentivirus in the cell supernatant was collected by centrifugation at 5,000 × g for 5 min at 25°C and the purification and titration of recombinant lentivirus was performed as previously described (34). Cells were infected with recombinant lentivirus-transducing units at an Multiplicity of infection of 20 in the presence of 8 µg/ml polybrene (Sigma-Aldrich; Merck KGaA) for 24 h at 37°C. Stable cells were selected with puromycin (3 µg/ml; Thermo Fisher Scientific, Inc.) for four more days and then used for subsequent experiments.

miRNA inhibitor and mimic

miR-181d-5p mimic (5′-AAC AUU CAU UGU UGU CGG UGG GU-3′), miR-181d-5p inhibitor (5′-ACC CAC CGA CAA CAA UGA AUG UU-3′) and corresponding NC (5′-CAG UAC UUU UGU GUA GUA CAA-3′) were synthesized by GeneChem, Inc. and introduced into CRC cells using Lipofectamine 2000® (Invitrogen; Thermo Fisher Scientific, Inc.) at a concentration of 100 nM for 6 h at 37°C.

Ethynyl-2-deoxyuridine (EdU) assay

Cell proliferation was determined by EdU staining using a BeyoClick EdU Cell Proliferation kit (Beyotime Institute of Biotechnology). Briefly, cells were incubated with 20 µM EdU for 2 h at room temperature. After washing with PBS, cells were fixed in 4% polyformaldehyde for 15 min at room temperature, permeabilized with 0.3% Triton X-100 and treated with Click Additive Solution for 30 min at room temperature. After incubating with DAPI solution for 5 min at room temperature, the cells were observed by fluorescence microscopy ≥5 fields of view randomly selected in each section and counted at ×200 magnification.

Cell apoptosis assay

Cells (50% confluence) with METTL3 knockdown or overexpression were treated with 1 µg/ml 5-FU. After 48 h of treatment, the cells were collected and incubated sequentially at 4°C with Annexin V-FITC for 15 min and propidium iodide (Beyotime Institute of Biotechnology) for 5 min. Apoptosis (percentage of early plus late apoptotic cells) was analyzed using an Accuri C6 flow cytometer (BD Biosciences). The results of flow cytometry were analyzed using FlowJo 7.6 software (FlowJo LLC).

Co-immunoprecipitation

Cells were lysed using immunoprecipitation buffer (containing 1 mM DTT, 100 mmol/l NaCl and 1 mM MgCl2) and protease inhibitor cocktails (Bimake.com). Lysates were centrifuged at 12,000 × g for 10 min at 4°C. The total cell lysates were used for IP with anti-METTL3 (Abcam; cat. no. ab195352; 1:50) or normal IgG and then incubated with Protein A/G PLUS-Agarose beads (Santa Cruz Biotechnology, Inc.; cat. no. sc-2003) at 4°C overnight according to manufacturer.s protocol. The immunoprecipitates were collected and analyzed by western blotting as described below using anti-METTL3 (Abcam; cat. no. ab98009; 1:1,000) and anti-DGCR8 (Abcam; cat. no. ab191875; 1:1,000) antibodies.

RNA immunoprecipitation (RIP) assays

RIP was performed using the Magna RIP RNA-Binding Protein Immunoprecipitation kit (MilliporeSigma). Cell extracts were prepared using RIP lysis buffer. The RNA-protein complexes were conjugated with anti-m6A (Abcam; cat. no. ab208577), anti-DGCR8 (ab191875), or anti-IgG antibody (ab172730) overnight at 4°C and washed successively with RIP-wash buffer for 10 min and 5 min at 4°C. The co-precipitated RNAs were purified using phenol:chloroform:isoamyl alcohol and subjected to reverse transcription-quantitative PCR.

RNA extraction and reverse transcription-quantitative (RT-q) PCR

RNA was extracted from 1×107 fibroblasts, CRC tissues and cell lines using TRIzol (Thermo Fisher Scientific, Inc.) according to the manufacturer.s instructions. Reverse transcription was performed with Prime-Script™ RT Master Mix (Takara Bio, Inc.) according to the manufacturer.s instructions. RT-qPCR was performed using SYBR-Green Master mix (Invitrogen; Thermo Fisher Scientific, Inc.) using an ABI PRISM 7500 real time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.). The amplification reactions for qPCR were as follows: 94°C for 10 min, 29 cycles of 94°C for 30 sec, 55°C for 30 sec and 72°C for 30 sec. GAPDH or U6 was used to normalize expression. RT-qPCR results were analyzed using the 2−ΔΔCq method (35). Primers are listed in Table I. Experiments were replicated three times.

Table I

Primes sequences used in the present study.

Gene	Sequences (5′-3′)
METTL3-forward	CCAGATGCTCCTGCCACTC
METTL3-reverse	ACAGTCCCTGCTACCTCCC
METTL14-forward	CCCATGTACTTACAAGCC
METTL14-reverse	CAGTGATGCCAGTTTCTC
WTAP-forward	GTAATGCGACTAGCAACC
WTAP-reverse	TATCAGGCGTAAACTTCC
NCALD-forward	AATGGTATAAAGGCTTCTTG
NCALD-reverse	TCCAGGTCGTACATGCTG
Pri-miR-181d-forward	CGGTGACTCTGACCTTCC
Pri-miR-181d-reverse	CCACAGTGACATTCATCCC
GAPDH-forward	ACAACTTTGGTATCGTGGAAGG
GAPDH-reverse	GCCATCACGCCACAGTTTC
miR-92a-3p-forward	AACAGATATTGCACTTGTC
miR-92a-3p-reverse	CAGTGCAGGGTCCGAGGT
miR-181d-5p-forward	AACATTCATTGTTGTCGGTGG
miR-181d-5p-reverse	ACCCACCGACAACAATGAAT
miR-221-3p-forward	CGCGAGCTACATTGTCTGCTG
miR-221-3p-reverse	CAGTGCAGGGTCCGAGGT
miR-125b-5p-forward	CGCGTCCCTGAGACCCTAAC
miR-125b-5p-reverse	AACAAGTCCCTGAGACCCT
miR-185-5p-forward	CGCTGGAGAGAAAGGCAGT
miR-185-5p-reverse	CAGTGCAGGGTCCGAGGT
miR-625-3p-forward	GCAGGACTATAGAACTTTC
miR-625-3p-reverse	CAGTGCAGGGTCCGAGGT
Pre-miR-181d-forward	CAGCCGAGGTCACAATCAAC
Pre-miR-181d-reverse	ATTCCCCTTAAGCCGTGTCT
U6-forward	CTCGCTTCGGCAGCACA
U6-reverse	AACGCTTCACGAATTTGCGT

METTL, methyltransferase like; WTAP, Wilms tumor 1-associating protein; NACLD, neurocalcin δ; miR/miRNA, microRNA; pre, precursor.

Western blot analysis

RIPA buffer (cat. no. 9806; Cell Signaling Technology, Inc.) was used to extract the proteins from the whole cell lysate and exosomes and the mixtures were then centrifuged at 10,000 × g for 5 min at 4°C. Protein concentrations were determined by the bicinchoninic acid assay. Proteins (25 mg) were resolved by 10% SDS-PAGE, transferred to PVDF membranes, blocked with 5% skimmed milk overnight at 4°C and incubated with primary antibodies (METTL3; cat. no. ab98009; 1:1,000; METTL14; cat. no. ab220030, 1:500; WTAP; cat. no. ab195380; 1:1,000; CD9; cat. no. ab263019; 1:1,000; CD63; cat. no. ab216130; 1:2,000; CD81; cat. no. ab79559; 1:1,000; all from Abcam; NCALD, ProteinTech Group, Inc.; cat. no. 12925-1-AP; 1:2,000; GAPDH; Cell Signaling Technology, Inc.; cat. no. 5174; 1:2,000) overnight at 4°C. After washing, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (Beyotime Institute of Biotechnology; cat. nos. A0208 and A0216; 1:1,000). Membranes were then washed and incubated with ECL (MilliporeSigma) for imaging and the bands were analyzed using ImageJ software (v4.6.2, National Institutes of Health).

Luciferase reporter assay

The NCALD 3′-UTR region carrying a putative miR-181d-5p binding site was inserted into the luciferase reporter plasmid pGL3-Promoter (Promega Corporation). For reporter assay, cells were transfected with miR-181d-5p mimic/inhibitor and pGL3-Promoter-NCALD-wild type (WT) (NCALD-WT) or pGL3-Promoter-NCALD-mutant (MUT) plasmid (NCALD-MUT) and the pRL-TK vector (Promega Corporation) expressing the Renilla luciferase for normalization at a ratio of 2:2:1 using Lipofectamine® 2000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.). The Dual-luciferase reporter assay system (Promega Corporation) was used to measure luciferase activity at 48 h post-transfection.

In vivo model

A total of 24 male BALB/c nude mice (5-week-old; body weight, 15-20 g) were purchased from the (Shanghai SLAC Laboratory Animal Co., Ltd.). All rats were housed at 23-25°C with 50-60% humidity, 12-h light/dark cycle and food and water ad libitum. The study was approved by the Ethics Committee of Shanghai Eighth People Hospital (approval number 2021-0462). BABL/c nude mice were housed in individually ventilated cages under specific pathogen-free conditions including a 12-h light/dark cycle, 25°C temperature and 80% humidity. Mice were provided access to sterilized water and food ad libitum. A tumor-bearing model was established by subcutaneously injecting 100 µl HT29 cells (5×106) followed by an intravenous injection of CAFs-derived exosomes (50 µg/mouse every three days) into the tail vein of the mice. An intraperitoneal injection of 5-FU (50 mg/kg, every week) was administered on day 12. Finally, the mice were anesthetized by inhalation with 3% isoflurane and sacrificed by cervical dislocation. The mice were euthanized on day 33 and tumors were collected for terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining. The largest tumor diameter was <13 mm and the largest tumor volume was <600 mm3.

TUNEL staining

TUNEL staining was performed using the Roche In Situ Cell Death Detection kit for the detection of programmed cell death (Roche Applied Science). The tissue sections were then examined by microscopy (CX41RF; Olympus Corporation) and the number of TUNEL-positive cells was counted using ImageJ software version 1.61 (National Institutes of Health).

Statistical analysis

Data are reported as the mean ± standard deviation (SD). Statistical analysis was performed using GraphPad Prism 8.0.2 software (GraphPad Software, Inc.). A two-tailed unpaired or paired Student.s t-test was used to compare differences between two groups. A one-way ANOVA followed by Tukey.s post-multiple test was used to compare differences between multiple groups. Kaplan-Meier method and Cox.s proportional hazards regression models were used to calculate overall survival and the differences were analyzed by a log-rank test. P<0.05 was considered to indicate a statistically significant difference.

Results

m6A modification is increased in CRC patients

The present study first measured the levels of m6A modification in adjacent-normal colorectal tissues and CRC tissues. The patient information for cohort 1 is shown in Table II. The results indicated that m6A modification was markedly upregulated in CRC tissues in cohort 1 (Fig. 1A). In addition, CAFs exhibited a spindle-shaped morphology, were adherent in culture and compared with NFs, expressed higher levels of the specific fibroblast markers, α-SMA, FAP and vimentin (Fig. S1). The levels of m6A modification in primary NFs and CAFs from patients with CRC in cohort 1 were determined. The results indicated that m6A modification was also significantly increased in CAFs (Fig. 1B). Next, the expression levels of METTL3, METTL14 and WTAP were measured and the results indicated that METTL3 was significantly increased in CAFs at both the mRNA and protein level, whereas WATP mRNA was slightly increased in CAFs and no significant change was observed for WATP protein and METTL14 mRNA and protein levels (Fig. 1C and D). To further investigate the role of METTL3, the expression levels of METTL3 in adjacent-normal colorectal tissues and CRC tissues were measured by immunohistochemistry in cohort 2 (Fig. 1E). A survival curve revealed that patients with lower METTL3 protein levels exhibited a higher survival rate (Fig. 1F). Furthermore, METTL3 was notably correlated with three of the clinicopathologic characteristics in the patients with CRC: tumor size (P=0.012), TNM classification (P=0.018) and distant metastasis (P=0.026; Table III).

Table II

Clinicopathological parameters in patients with CRC in cohort 1.

Characteristic	Cases	%
Sex
Male	18	60.0
Female	12	40.0
Age (years)
<60	15	50.0
≥60	15	50.0
Tumor size (cm)
≤4	11	36.7
>4	19	63.3
TNM classification
I	3	10
II	15	50.0
III	10.0	33.3
IV	2	6.7
Distant metastasis
Yes	9	30.0
No	21	70.0

CRC, colorectal cancer.

Figure 1

m6A modification in CRC patients. m6A levels in (A) adjacent-normal colorectal tissues (n=30) and CRC tissues (n=30) and (B) in primary NFs (n=30) and CAFs (n=30) from patients with CRC in cohort 1. (C) mRNA expression of METTL3, METTL14 and WTAP in primary NFs (n=10) and CAFs (n=10) from patients with CRC in cohort 1. (D) Protein expression of METTL3, METTL14 and WTAP in primary NFs (n=5) and CAFs (n=5) from patients with CRC in cohort 1. (E) Expression of METTL3 in cohort 2. Scale bar, 100 µm. (F) Survival curves for patients in cohort 2. All experiments were performed in triplicate. *P<0.05, ***P<0.001. m6A, RNA N6-methyladenosine; CRC, colorectal cancer; NFs, named adjacent-normal fibroblasts; CAFs, cancer-associated fibroblasts; METTL, methyltransferase like; WTAP, Wilms tumor 1-associating protein; NACLD, neurocalcin δ; IHC, immunohistochemistry.

Table III

Correlation between the METTL3 protein expression and clinicopathological parameters in patients with CRC in cohort 2.

Clinicopathological parameter	Protein expression of METTL3		P-value
	High (n=58)	Low (n=53)
Sex			0.312
Male	24	27
Female	34	26
Age (years)			0.150
<60	26	31
≥60	32	22
Tumor size (cm)			0.012
≤4	18	29
>4	40	24
TNM classification			0.018
I	13	5
II	17	31
III	20	12
IV	8	5
Distant metastasis			0.026
Yes	33	19
No	25	34

Differences between groups were determined by the Chi-square test. METTL, methyltransferase like; CRC, colorectal cancer.

METTL3 regulates CAFs-induced cell proliferation and 5-FU sensitivity of CRC cells

To study the effect of METTL3, METTL3 was either silenced or overexpressed in primary CAFs (Fig. S2A and B). Then, the CAFs were used to treat CRC cell lines HT29 and HCT116. Immunostaining results indicated that the CAFs significantly increased HT29/HCT116 cell proliferation, which was significantly attenuated by silencing METTL3, but significantly increased by overexpressing METTL3 (Figs. 2A and S3A and B). Flow cytometry indicated that in the presence of 5-FU, CAFs significantly decreased HT29/HCT116 cell apoptosis, which was significantly attenuated by silencing METTL3, but further decreased by overexpressing METTL3 (Fig. 2B and C). These findings indicated that the regulation of cell proliferation and 5-FU sensitivity by CAFs was METTL3-dependent.

Figure 2

METTL3 regulates CAFs-induced cell proliferation and 5-FU sensitivity of CRC cell lines. (A) Proliferation of HT29 and HCT116 cells co-cultured with METTL3-overexpressing or METTL3-silencing lentivirus-transduced CAFs. (B and C) Apoptosis of HT29 and HCT116 cells co-cultured with METTL3-overexpressing or METTL3-silencing lentivirus-transduced CAFs in the presence of 1 µg/ml 5-FU. Scale bar, 100 µm. All experiments were performed in triplicate. **P<0.01, ***P<0.001. METTL, methyltransferase like; CAFs, cancer-associated fibroblasts; 5-FU, 5-Fluorouracil; CRC, colorectal cancer; CAFs, cancer-associated fibroblasts; EdU, Ethynyl-2-deoxyuridine; sh, short hairpin; NC, negative control.

CAFs-derived exosomes inhibit the 5-FU sensitivity of CRC cells

To determine the manner in which CAFs affect 5-FU sensitivity, exosomes were isolated from primary CAFs (CAF-exo) and characterized by transmission electron microscopy (Fig. S4A). The size range of particles measured by NanoFCM is ~42 nm to ~197 nm, with an average size of ~66.7 nm (Fig. S4B) and exosomes markers CD63, CD9 and CD81 were measured by immunoblot assay (Fig. 3A). Internalization of exosomes by HT29/HCT116 cells was examined by the PKH-67 assay and a laser scanning confocal microscope (Fig. 3B). The CAFs were treated with DMSO or an exosome inhibitor, GW4869 and co-cultured with HT29/HCT115 cells in the presence of 5-FU. Flow cytometry revealed that the control CAFs caused a significant decrease in apoptosis, which was not shown in CAFs treated with GW4869 group (Fig. 3C and D). Next, exosomes from control CAFs, METTL3-silencing CAFs, or METTL3-overexpressing CAFs were used to treat HT29/HCT115 cells in the presence of 5-FU. Flow cytometry indicated that control exosomes caused a significant decrease in apoptosis, which was significantly attenuated by exosomes from METTL3-silencing CAFs, but inhibited by exosomes from METTL3-overexpressing CAFs (Fig. 3E and F). These findings confirmed that CAF-derived exosomes inhibit the 5-FU sensitivity of CRC cells.

Figure 3

CAFs-derived exosomes inhibit 5-FU sensitivity of CRC cell lines. (A) Immunoblot analysis of exosome markers in CAF cell lysate and CAFs-exo. (B) Laser scanning confocal microscope analysis of exosome uptake. Scale bar, 50 µm. (C and D) Effect of exosome inhibitor, GW4869, on apoptosis of CRC cells in the presence of 1 µg/ml 5-FU. (E and F) Apoptosis of CRC cells treated with exosomes derived from METTL3-overexpressing or METTL3-silencing CAFs in the presence of 1 µg/ml 5-FU. All experiments were performed in triplicate. *P<0.05, ***P<0.001. CAFs, cancer-associated fibroblasts; 5-FU, 5-Fluorouracil; CRC, colorectal cancer; METTL, methyltransferase like; CAFs-exo, primary CAFs; sh, short hairpin; NC, negative control.

METTL3-dependent m6A modification regulates the processing of miR-181d-5p by DGCR8

Considering the important roles of miRNAs in tumorigenesis, it was hypothesized that METTL3 influences 5-FU sensitivity by targeting miRNAs in an m6A-dependent, pri-miRNA-processing manner. Initially, the present study assessed whether METTL3 was required for the engagement of pri-miRNAs by the microprocessor protein, DGCR8, in CRC cells in a manner similar to a previous study (30). Co-immunoprecipitation was performed and a METTL3-interacting protein, DGCR8, was identified (Fig. 4A). In addition, a significant increase in binding between METTL3 and DGCR8 was also observed in METTL3-overexpressing cells (Fig. 4B). Next, the miRNA levels in exosomes derived from METTL3-overexpressing or METTL3-silencing CAFs were measured by RT-qPCR. The results indicated that overexpressing METTL3 upregulated miR-181d-5p, whereas silencing METTL3 significantly decreased miR-181d-5p levels. No significant change was observed for the expression of other CAF-derived exosomal miRNAs including miR-92a-3p, miR-221-3p, miR-185-5p, miR-125b-5p, or miR-625-3p (Fig. 4C) (14). In addition, some miRNAs contain m6A tags in their pri-miRNA and are involved in DGCR8-dependent pri-miRNA processing (29). CAFs were then lysed, RNA were extracted and the levels of pri-miR-181d, precursor (pre-)miR-181d and miR-181d-5p were determined. RT-qPCR analysis suggested that pre-miR-181d and miR-181d-5p were significantly increased by METTL3 overexpression and decreased by METTL3 silencing. By contrast, the levels of pri-miR-181d were significantly decreased by METTL3 overexpression, but significantly increased by METTL3 silencing (Fig. 4D). RIP results indicated that overexpressing METTL3 enhanced m6A modification of pri-miR-181d, which was decreased by silencing METTL3 (Fig. 4E). Furthermore, the results indicated that silencing METTL3 markedly decreased the binding of DGCR8 and pri-miR-181d and m6A modification of pri-miR-181d (Fig. 4F and G). Co-culture of CRC cells with exosomes from METTL3-manipulated CAFs demonstrated that CAF-exo significantly increased the levels of miR-181d-5p, which were suppressed by METTL3 silencing, but enhanced by METTL3 overexpression (Fig. 4H and I). These findings suggested that m6A modifications enhanced the recognition of pri-miR-181d by DGCR8 and its subsequent processing to mature miRNA.

Figure 4

METTL3-dependent m6A modification regulates DGCR8 processing of miR-181d-5p. (A) Co-immunoprecipitation analysis of DGCR8. (B) Immunoprecipitation of DGCR8 in METTL3-overexpressing cells. (C) Expression of miRNAs in exosomes derived from METTL3-overexpressing or METTL3-silencing CAFs. (D) Expression of pri-miR-181d, pre-miR-181d and miR-181d-5p and (E) the levels of pri-miR-181d-5p m6A in METTL3-overexpressing or METTL3-silencing lentivirus-transduced CAFs. (F) Analysis of pri-miRNA binding to DGCR8. (G) Analysis of pri-miRNAs m6A modification. (H and I) miR-181d-5p levels in CRC cells treated with exosomes derived from METTL3-overexpressing or METTL3-silencing lentivirus-transduced CAFs. All experiments were performed in triplicate. *P<0.05, **P<0.01, ***P<0.001. METTL, methyltransferase like; m6A, RNA N6-methyladenosine; DGCR8, DiGeorge Syndrome Critical Region 8; miR/miRNA, microRNA; CAFs, cancer-associated fibroblasts; pri, primary; pre, precursor; sh, short hairpin; NC, negative control; CAFs-exo, primary CAFs.

CAFs-derived exosomes inhibit 5-FU sensitivity through the METTL3/miR-181d-5p axis

Next, miR-181d-5p was silenced or overexpressed in CRC cells (Fig. S2C) to further study its role in chemosensitivity. Apoptosis was measured in HT29 and HCT116 cells transfected with miR-181d-5p inhibitor/mimic in the absence or presence of 5-FU. The results indicated that overexpressing miR-181d-5p inhibited the 5-FU sensitivity of CRC cells, whereas suppressing miR-181d-5p promoted 5-FU sensitivity (Fig. 5A and B). Treatment of miR-181d-5p-silencing or miR-181d-5p-overexpressing CRC cells with exosomes from METTL3-silencing or METTL3-overexpressing CAFs revealed that silencing METTL3 significantly increased 5-FU sensitivity, whereas overexpressing METTL3 significantly decreased 5-FU sensitivity in HT29 cells (Fig. 5C and D). In addition, silencing miR-181d-5p significantly increased 5-FU sensitivity, whereas overexpressing miR-181d-5p significantly decreased 5-FU sensitivity in HT29 cells. These findings suggested that the inhibition of 5-FU sensitivity in CRC cells was METTL3/miR-181d-5p dependent.

Figure 5

CAFs-derived exosomes inhibit 5-FU sensitivity via the METTL3/miR-181d-5p axis. (A and B) Apoptosis of HT29 and HCT116 cells transfected with miR-181d-5p inhibitor/mimic in the presence of 1 µg/ml 5-FU. (C and D) Apoptosis of HT29 cells transfected with miR-181d-5p inhibitor/mimic and treated with exosomes derived from METTL3-overexpressing or METTL3-silencing lentivirus-transduced CAFs in the presence of 1 µg/ml 5-FU. All experiments were performed in triplicate. ***P<0.001. CAFs, cancer-associated fibroblasts; 5-FU, 5-Fluorouracil; METTL, methyltransferase like; miR/miRNA, microRNA; 5-FU, 5-Fluorouracil; sh, short hairpin; NC, negative control; exo, primary.

miR-181d-5p mimic inhibits 5-FU sensitivity of CRC cell lines by targeting NCALD

To further understand how miR-181d-5p regulates 5-FU sensitivity, a bioinformatic analysis was conducted and a potential miR-181d-5p binding site in 3′-UTR of NCALD mRNA was identified (Fig. 6A). WT or MUT 3′-UTR of NCALD was used for a luciferase reporter assay. The results indicated that overexpressing miR-181d-5p significantly suppressed the luciferase activity of NCALD 3′-UTR, which was significantly enhanced by silencing miR-181d-5p (Fig. 6B). NCALD expression was markedly increased by silencing miR-181d-5p, but suppressed following miR-181d-5p overexpression in CRC cell lines (Fig. 6C and D). NCALD overexpression significantly increased 5-FU sensitivity, which was attenuated by miR-181d-5p overexpression (Figs. S2D and 6E-F). Furthermore, the results also indicated that overexpressing miR-181d-5p markedly decreased NCALD levels (Fig. 6G). Overall, the results indicated that miR-181d-5p targets NCALD to inhibit 5-FU sensitivity.

Figure 6

miR-181d-5p mimic inhibits 5-FU sensitivity by targeting NCALD. (A) Predictive miR-181d-5p binding site. (B) Luciferase reporter assay of the binding of NCALD and miR-181d-5p. (C and D) Expression of NCALD in HT29 and HCT116 cells transfected with miR-181d-5p inhibitor, mimic, or NC. (E and F) Cell apoptosis and (G) NCALD expression in HT29 and HCT116 cells transfected with miR-181d-5p inhibitor/mimic and transduced with NCALD-overexpressing lentivirus in the absence or presence of 1 µg/ml 5-FU. All experiments were performed in triplicate. *P<0.05, **P<0.01, ***P<0.001. miR/miRNA, microRNA; 5-FU, 5-Fluorouracil; NACLD, neurocalcin δ; NC, negative control.

CAFs-derived exosomes inhibit 5-FU sensitivity in vivo

The effect of CAFs-derived exosomes on 5-FU sensitivity was further evaluated in an animal model. Administration of 5-FU significantly inhibited HT29 tumor growth, which was sharply attenuated by administration of CAF-exo (Fig. 7A). 5-FU also significantly decreased tumor weight, which was significantly inhibited by the administration of CAF-exo (Fig. 7B). TUNEL staining indicated that 5-FU markedly increased tumor cell apoptosis, which was attenuated by administration of CAF-exo (Fig. 7C and D). These data suggested that CAF-exo decreases 5-FU sensitivity of CRC cells in a CRC animal model.

Figure 7

CAF-derived exosomes inhibit the 5-FU sensitivity of CRC cell lines in vivo. HT29 cells and CAF-derived exosomes were injected into nude mice. Mice were administered 5-FU (50 mg/kg) every week on day 12. Tumor (A) volume and (B) weight. (C and D) TUNEL staining. Scale bar, 100 µm. All experiments were performed in triplicate. ***P<0.001. CAFs, cancer-associated fibroblasts; 5-FU, 5-Fluorouracil; CRC, colorectal cancer.

NCALD expression is decreased and associated with treatment outcomes of 5-FU-treated CRC patients

To further elucidate the role of NCALD, expression levels of miR-181d-5p and NCALD in adjacent-normal colorectal tissues and CRC tissues were measured by RT-qPCR. The results showed that CRC patients exhibited increased miR-181d-5p levels (Fig. 8A) and decreased NCALD levels in cohort 1 (Fig. 8B). A Pearson.s correlation analysis showed that miR-181d-5p was negatively correlated with NCALD (Fig. 8C). It was also observed that NCALD protein expression was significantly inhibited in CRC tissues from cohort 2 (Fig. 8D). A survival curve showed that patients with higher NCALD protein levels had a higher survival rate (Fig. 8E). Furthermore, NCALD was notably correlated with three of clinicopathologic characteristics, tumor size (P=0.004), TNM classification (P=0.033) and distant metastasis (P=0.009), in CRC patients (Table IV). Multivariate regression analysis indicated that tumor size, TNM classification, distant metastasis and NCALD expression were all risk factors for CRC (Fig. 8F). Next, NCALD expression in CRC tissues from patients with or without 5-FU-resistance was measured. The results indicated that NCALD was markedly higher in CRC tissues from patients without 5-FU-resistance (Fig. 8G). A correlation analysis of NCALD and 5-FU sensitivity in CRC tissues was performed using a Chi-square test and the results indicated that NCALD was negatively correlated with 5-FU sensitivity (Fig. 8H). A Kaplan-Meier survival analysis indicated that patients with higher NCALD had an improved prognosis (Fig. 8I). These results suggest that NCALD was decreased in CRC patients and NCALD was correlated to outcomes following 5-FU treatment of CRC patients.

Figure 8

NCALD expression is decreased and associated with treatment outcome of 5-FU-treated CRC patients. (A and B) Expression of miR-181d-5p and NCALD in adjacent-normal colorectal tissues (n=30) and CRC tissues (n=30) in cohort 1. (C) Pearson.s correlation scatter plots for CRC tissues (n=30). (D) Expression of NCALD in cohort 2. Scale bar, 100 µm. (E) Survival curves for patients in cohort 2. (F) Multivariate regression analysis of the tissues from cohort 2. (G) NCALD expression in CRC tissues with or without 5-FU-resistance. (H) Correlation analysis (Chi-square test) of NCALD and 5-FU sensitivity in CRC tissues (n=78). (I) Survival curves for overall survival according to NCALD expression in cohort 2 with 5-FU treatment. All experiments were performed in triplicate. ***P<0.001. NACLD, neurocalcin δ; 5-FU, 5-Fluorouracil; CRC, colorectal cancer.

Table IV

Correlation between the NCALD protein expression and clinicopathological parameters in patients with CRC in-cohort 2.

Clinicopathological parameter	Protein expression of NCALD		P-value
	High (n=42)	Low (n=69)
Sex			0.288
Male	22	29
Female	20	40
Age (years)			0.162
<60	18	39
≥60	24	30
Tumor size (cm)			0.004
≤4	25	22
>4	17	47
TNM classification			0.033
I	10	8
II	22	26
III	8	24
IV	2	11
Distant metastasis			0.009
Yes	13	39
No	29	30

Differences between groups were determined by the Chi-square test. NACLD, neurocalcin δ; CRC, colorectal cancer.

Discussion

Exosome transport is believed to be an effective means to modulate cell signaling and the function of recipient cells. Some onco-proteins, oncomiRs and DNA fragments harboring oncogenic dysregulation are found in cancer-derived exosomes (36). For instance, exosomes from CAFs containing Annexin A6 induce FAK-YAP activation by stabilizing β1 integrin and enhancing drug resistance (37). Studies have indicated that miRNAs can be loaded into exosomes or high-density lipoprotein to protect miRNAs from degradation and maintain their stability in cell-to-cell communication (38,39). Exosomal miRNAs have garnered attention in recent years because evidence suggests that exosomal miRNAs influence the properties of cancer cells (40). CAFs interact with cancer cells through CAF-secreted factors including exosomal miRNAs to influence tumor progression (16). The present study found that exosomes released from CAFs inhibited the 5-FU sensitivity of CRC cells. In addition, CAFs-derived exosomes inhibited 5-FU sensitivity through the METTL3/miR-181d-5p axis. These results indicated a new role for CAFs-derived exosomes in CRC, which is to increase miR-181d-5p levels to inhibit 5-FU sensitivity.

m6A methylation is a form of posttranscriptional RNA modification (41). Studies suggest that m6A RNA methylation regulates proliferation, metabolism and tumorigenesis (42,43). For example, Shen et al (44) demonstrate that increased m6A modification stabilizes HK2 and GLUT1 expression to enhance glycolysis, which promotes CRC progression. A study by Liu et al (45) indicates that m6A modification is suppressed in endometrial tumors. In mammals, m6A is created by a methyltransferase complex that contains an enzymatic subunit METTL3 and its co-factors, METTL14 and WTAP (46). High METTL3 levels are implicated in poor prognosis (47). METTL3 is also upregulated in breast cancer, whereas METTL3 knockdown suppresses breast tumor growth (48). The results of the present study indicated that METTL3 expression was sharply increased in tumor tissues and CAFs of patients with CRC compared with normal tissues and NFs, respectively, leading to enhanced m6A modification in CRC patients. These data suggested that CAFs, but not NFs, contributed to m6A modification in CRC. The present study also demonstrated that METTL3-dependent m6A methylation inhibited the 5-FU sensitivity of CRC cells. These results not only increase our knowledge of METTL3/m6A modification in CRC, but also provided putative targets for CRC therapy.

miRNAs regulate gene expression (49). Dysregulated miRNAs are capable of inhibiting apoptosis, activating invasion and inducing angiogenesis (50). For example, miR-21 is identified as overexpressed in six different cancer types including CRC (51). miR-708 is decreased in CRC and functions as a tumor suppressor by targeting ZEB1 (52). miRNA genes are transcribed into a pri-miRNA and cleaved to form a pre-miRNA, which is then cleaved again to form a miRNA (53). Pri-miRNA is converted to pre-miRNA by a microprocessor complex consisting of DGCR8, which recognizes an N6-methyladenylated GGAC sequence (30,54). Guo et al (55) demonstrate that DGCR8 knockdown suppresses the maturation of miR-27b and downregulates miR-27b to promote cell proliferation in ovarian cancer cells. Zhang et al (27) suggest that excessive miR-25-3p maturation via m6A promotes pancreatic cancer progression. A study by Ma et al (29) revealed that decreased m6A modification negatively modulates DGCR8 processing of pri-miRNA-126 resulting in metastasis of hepatocellular carcinoma. The data from the present study demonstrated that METTL3-dependent m6A methylation served an important role in the processing of miR-181d-5p by DGCR8. The results indicated the importance of METTL3-dependent m6A methylation/DGCR8 in miR-181d-5p maturation and improved our understanding of the role of miR-181d-5p dysregulation in human cancer.

miR-181d-5p has been shown to play a role in cancer through various targets. For example, Wang et al (21) show that miRNA-181d-5p promotes EMT via CDX2/HOXA5 in mammary cancer. Gao et al (56) demonstrate that miR-181d-5p regulates proliferation and invasion of lung cancer cells via CDKN3. Overexpressing miR-181d inhibits metastasis of CRC cells through PEAK1 (57). In addition, previous studies have shown that NCALD may regulate chemo-resistance by the ERK1/2, NF-κB and immune response pathways (58) and prognosis via CX3CL1 in ovarian cancer (59). The results of the present study demonstrated that exosomal miR-181d-5p binds directly with NCALD to regulate 5-FU sensitivity. This is the first report, to the best of the authors. knowledge, showing that miR-181d-5p targets NCALD to regulate the chemosensitivity of CRC cells. These results not only increase our knowledge of miR-181d-5p/NCALD in CRC, but also broaden our understanding of the chemosensitivity of CRC cells.

The role of NCALD was further studied using clinical samples. The results indicated that miR-181d-5p was increased and NCALD was decreased in CRC patients. miR-181d-5p levels were negatively correlated with NCALD levels. Higher NCALD levels were associated with a higher survival rate. These findings indicated an important role for miR-181d-5p/NCALD in CRC and increased our understanding of the roles of miR-181d-5p and NCALD in the pathogenesis of CRC. Future studies will be focused on the regulatory effect of other molecules in CAF exosomes, such as lncRNA, miRNA and protein and the molecular mechanisms underlying chemo-resistance in CRC cells. The role of the METTL3/miR-181d-5p forced-expression NFs in regulating the 5-FU sensitivity of CRC cells will also be further examined. According to the consensus statement on CAFs (60), whether there are tumor-specific mutations in the isolated CAFs should be determined. Although further studies are needed, the present study identified a novel molecular mechanism underlying chemosensitivity in CRC cells which may provide a new targets and strategies for treating CRC.

In conclusion, the findings of the present study revealed a new role for CAF-secreted exosomal miR-181d-5p. METTL3-dependent m6A methylation was upregulated in CRC to promote the processing of miR-181d-5p by DGCR8, leading to increased miR-181d-5p, which inhibits 5-FU sensitivity by targeting NCALD (Fig. S5). These findings highlighted the importance of m6A/DGCR8/miR-181d-5p/NCALD signaling, which may facilitate the development of new drugs for CRC treatment.