Introduction

Carcinoma-associated fibroblasts (CAFs) are major components of the tumor microenvironment and play pivotal roles in tumor progression1,2,3. Despite having originated from activated tissue fibroblasts, characteristics of CAFs vary significantly from fibroblasts in normal tissues4. The expressions of various fibrotic markers, growth factors, chemokines and cytokines are increased in CAFs than that in normal tissue fibroblasts; these factors may have tumorigenic effects2. Autophagy is a catabolic process that mediates the degradation of unnecessary or dysfunctional cellular components5,6. Different studies have reported the role of autophagy in cancer cells7,8,9. Recent reports have focused on the function of autophagy in both cancer and stromal cells10,11,12. Autophagy status of fibroblasts in the tumor stroma was reported to differ from that in extra-tumoral fibroblasts12. However, the significance of autophagy in the transition of normal tissue fibroblasts to CAFs during tumor progression and the mechanism of autophagy in CAFs remain unclear.

TP53 is a tumor suppressor gene, and its mutational inactivation is frequently observed in many human cancers13. In colorectal cancer, TP53 mutation is a common genetic abnormality that develops with the cancerization of the colon and is associated with chromosomal instability. p53 is a transcription factor that regulates the expression of genes associated with cell cycle arrest, apoptosis, and senescence. In addition, p53 has recently been suggested to modulate autophagy in cancer cells14,15. The TP53 gene has a non-cell-autonomous function and can affect the surrounding cells in the tumor microenvironment, such as CAFs, by changing the secretion of a large number of proteins, production of reactive oxygen species, or profiles of miRNAs sequestered in the exosomes16. We have previously reported that the functional loss of p53 in colon cancer cells promoted the modification of tumor stroma and subsequent tumor growth through the above mechanisms17,18. However, to the best of our knowledge, the impact of p53 deficiency in cancer cells on the autophagy of co-existing fibroblasts has not been reported. Therefore, we aimed to clarify the significance of autophagy in the activation of fibroblasts, focusing on the TP53 status of co-existing cancer cells in the tumor microenvironment.

Results

Inactivation of TP53 in cancer cells suppresses autophagy in fibroblasts

The importance of autophagy function of fibroblasts has been reported previously10,11,12. Therefore, this study aimed to examine whether autophagy might play a role in the activation of fibroblasts. We intended to clarify the significance of autophagy in the activation of fibroblasts, while focusing on the TP53 status. Thus, we examined activation and autophagy of CCD-18Co cells cocultured with HCT116sh control cells and HCT116sh p53 cells in Fig. 1.

Figure 1
figure 1figure 1

Inactivation of p53 in cancer cells activates fibroblasts and suppresses autophagy. (a) TP53 mRNA expression analysis in HCT116sh p53 and HCT116sh control cells using quantitative real-time PCR (qRT-PCR) (left; mean ± SD, *P < 0.05). The expression of p53 in HCT116sh p53 and HCT116sh control cells using western blotting (right). (b) Relative expressions of FGF2, PDGFRA, ACTA2, CXCL12, TGFβ1, and VEGFA mRNAs in CCD-18Co cells co-cultured with or without HCT116sh p53 or HCT116sh control cells analysed using qRT-PCR (n = 3, mean ± SD, *P < 0.05). (c) Cell numbers of CCD-18Co cells co-cultured with　HCT116sh p53 or HCT116sh control cells for 24, 48, and 72 h. (n = 3, mean ± SD, *P < 0.05). (d) Western blotting and autophagic flux assays of CCD-18Co cells co-cultured with or without HCT116sh p53 or HCT116sh control cells (n = 4, mean ± SD, *P < 0.05). (e) Western blotting and autophagic flux assays of CCD-18Co cells co-cultured with or without HT29 cells (n = 5, mean ± SD, *P < 0.05). Protein expression levels were measured using imageJ software 1.8.0_172 (https://imagej.nih.gov/).

Full size image

We used HCT116 cells which are human colon cancer cell lines with wild-type TP53 expression.

To explore the significance of p53 functional deficiency in cancer cells, we generated TP53-deficient HCT116 cells (HCT116sh p53 cells) using shRNA for p53. We confirmed the suppression of TP53 in HCT116 cells by qRT-PCR and western blotting analysis (Fig. 1a). The relative mRNA expression of FGF2, PDGFRA, CXCL12, TGFβ1, and VEGFA—the phenotypic markers of fibroblast activation—was higher in CCD-18Co cells co-cultured with HCT116sh p53 cells in Transwell than that in CCD-18Co cells co-cultured with HCT116sh control cells and in those cultured alone (Fig. 1b). CCD-18Co cells co-cultured with HCT116sh p53 cells were found to have proliferated more than those co-cultured with HCT116sh control cells (Fig. 1c). To clarify the effect of p53 deficiency in cancer cells on the autophagy of co-existing fibroblasts We evaluated autophagy in CCD-18Co cells co-cultured with HCT116sh control and HCT116sh p53 cells by western blotting and autophagic flux assays. The autophagic flux was significantly suppressed in CCD-18Co cells co-cultured with HCT116sh p53 cells than that in CCD-18Co cells co-cultured with HCT116sh control cells or cultured alone; p62(SQSTM1), which is a selective substrate for autophagy, accumulated in CCD-18Co cells co-cultured with HCT116sh p53 cells (Fig. 1d). These results suggest that autophagy in CCD-18Co cells co-cultured with HCT116sh p53 cells was suppressed than that in CCD-18Co cells co-cultured with HCT116sh control cells or cultured alone. To confirm that the suppression of autophagy in fibroblasts was induced by the inactivation of p53 in cancer cells, we used HT29 cells that are colon cancer cells with mutations in TP53. Autophagy in CCD-18Co cells was significantly suppressed even when co-cultured with HT29 cells than that in CCD-18Co cells cultured alone, similar to the condition in CCD-18Co cells co-cultured with HCT116sh p53 cells (Fig. 1e). Therefore, autophagy in CCD-18Co cells was inhibited by some humoural factors secreted by cancer cells along with functional deficiency of p53.

Suppression of autophagy induces activation of fibroblasts and accelerates growth in co-cultured cancer cells

Next, we examined whether the suppression of autophagy in fibroblasts can affect their activation. We suppressed autophagy in CCD-18Co cells using siRNA for ATG7 (a major autophagy-related protein) and pepstatin (an autophagy inhibitor). We confirmed the inhibition of ATG7 in CCD-18Co cells transfected with ATG7 siRNA using qRT-PCR and western blotting (Fig. 2a). Transfection of ATG7 siRNA significantly suppressed the autophagic flux in CCD-18Co cells than that of control siRNA (Fig. 2b). The relative expression levels of ACTA2, CXCL12, FGF2, TGFβ1, and VEGFA mRNA were high in CCD-18Co cells transfected with ATG7 siRNA than in those with control siRNA (Fig. 2c). HCT116 cells co-cultured with CCD-18Co cells transfected with ATG7 siRNA proliferated more than those co-cultured with CCD-18Co cells transfected with control siRNA (Fig. 2d). Moreover, to inhibit autophagy in fibroblasts in another way and observe the activation, we used the autophagy inhibitor pepstatinA. when pepstatin A was added to CCD-18Co cells, the relative expression levels of ACTA2 and CXCL12 mRNA increased than that in control cells (Fig. 2e). These results suggest that suppression of autophagy in fibroblasts may lead to their activation and promote fibroblast-mediated cancer cell proliferation.

Figure 2
figure 2figure 2

Suppression of autophagy activates fibroblasts. (a) ATG7 mRNA expression in CCD-18Co cells transfected with ATG7 or control siRNA analysed using quantitative real-time PCR (qRT-PCR) (left; n = 3, mean ± SD, *P < 0.05). The expression of ATG7 protein in CCD-18Co cells transfected with ATG7 or control siRNA analysed using western blotting (right). (b) Western blotting and autophagic flux assays in CCD-18Co cells transfected with ATG7 or control siRNA (n = 4, mean ± SD, *P < 0.05). (c) Relative expressions of ACTA2, CXCL12, TGFβ1, FGF2, PDGFRA and VEGFA mRNA in CCD-18Co cells transfected with ATG7 or control siRNA analysed using qRT-PCR (n = 3, mean ± SD, *P < 0.05). (d) Cell numbers of HCT116 cells co-cultured with CCD-18Co cells transfected with ATG7 or control siRNA for 24, 48 and 72 h. (n = 3, mean ± SD, *P < 0.05). (e) Relative expressions of ACTA2, CXCL12, TGFβ1, FGF2, PDGFRA and VEGFA mRNA in CCD-18Co cells treated with pepstatin or control (n = 3, mean ± SD, *P < 0.05). Protein expression levels were measured using imageJ software 1.8.0_172 (https://imagej.nih.gov/).

Full size image

Exosomes derived from TP53-inactivated cancer cells suppress autophagy in fibroblasts

Recently, p53 inactivation in cancer cells was reported to affect the surrounding stromal cells by modifying the secretion of miRNAs sequestered in exosomes18. We focused on the role of exosomes derived from cancer cells in the autophagy of fibroblasts in the tumor microenvironment. We isolated exosomes from the culture media of HCT116sh control and HCT116sh p53 cells (Fig. 3a). Protein expression in the isolated exosomes was confirmed using western blotting with representative exosome markers, such as CD9, CD63, and CD81 (Fig. 3b). To examine whether exosomes derived from cancer cells were involved in the autophagy of fibroblasts, we observed the autophagic flux in CCD-18Co cells after the addition of exosomes derived from HCT116sh control or HCT116sh p53 cells. The autophagic flux in CCD-18Co cells was significantly suppressed and accumulation of p62 was enhanced with the addition of exosomes from HCT116sh p53 cells than that of exosomes from HCT116sh control cells and in control CCD-18Co cells (Fig. 3c). In addition, we stimulated CCD-18Co cells with exosomes derived from HT29 cells and examined their autophagic flux. The autophagic flux was significantly suppressed and accumulation of p62 was enhanced in CCD-18Co cells cultured with exosomes from HT29 cells than in CCD-18Co cells cultured in control PBS (Fig. 3d). These results suggest that exosomes derived from p53-deficient cancer cells may be involved in the suppression of autophagy in fibroblasts.

Figure 3
figure 3

Exosomes derived from TP53-inactivated cancer cells suppress autophagy in fibroblasts. (a) Representative transmission electron microscopy images of exosomes isolated from culture media of HCT116sh control or HCT116sh p53 cells. Scale bar: 100 nm. (b) Western blotting analysis of CD9, CD63, and CD81 in exosomes from HCT116sh control or HCT116sh p53 cells. (c) Western blotting and autophagic flux assays of CCD-18Co cells cultured with or without exosomes derived from HCT116sh p53 (+ Exo-HCT116sh p53) or HCT116sh control cells (+ Exo-HCT116sh control) (n = 5, mean ± SD, *P < 0.05). (d) Western blotting and autophagic flux assays of CCD-18Co cells cultured with (+ Exo-HT29) or without (PBS) exosomes derived from HT29 cells (n = 5, mean ± SD, *P < 0.05). Protein expression levels were measured using imageJ software 1.8.0_172 (https://imagej.nih.gov/).

Full size image

Suppression of ATG2B activates fibroblasts via inhibition of autophagy

To explore the mechanism of suppression of autophagy in fibroblasts associated with p53 inactivation in cancer cells, we examined the expression of representative autophagy-related proteins in CCD-18Co cells co-cultured with HCT116sh control or HCT116sh p53 cells using western blotting. Among these autophagy-related proteins, the expression of ATG2B in CCD-18Co cells co-cultured with HCT116sh p53 cells was significantly suppressed than that in CCD-18Co cells co-cultured with HCT116sh control cells or CCD-18Co cells cultured alone (Fig. 4a). In addition, the expression level of ATG2B protein in CCD-18Co cells cultured with exosomes derived from HCT116sh p53 cells was suppressed than that in CCD-18Co cells cultured with exosomes derived from HCT116sh control cells and control CCD-18Co cells (Fig. 4b). Therefore, we hypothesized that the suppression of autophagy and subsequent activation of fibroblasts by HCT116sh p53 cells may be induced through ATG2B suppression. To determine the function of ATG2B in the activation of fibroblasts, we suppressed the expression of ATG2B in CCD-18Co cells using siRNA. We confirmed the inhibition of ATG2B in CCD-18Co cells by qRT-PCR and western blotting analysis (Fig. 4c). Inhibition of ATG2B suppressed autophagic flux in CCD-18Co cells than expression in control cells (Fig. 4d). The relative expression levels of ACTA2, CXCL12, TGFβ1, FGF2 and VEGFA mRNA significantly increased in CCD-18Co cells with ATG2B suppression than in control CCD-18Co cells (Fig. 4e). We further investigated the effect of ATG2B suppression in CCD-18Co cells on the proliferation of co-existing cancer cells. The proliferation of HCT116 cells co-cultured with CCD-18Co cells with ATG2B suppression was significantly higher than that of HCT116 cells co-cultured with control CCD-18Co cells (Fig. 4f). These results suggest that the suppression of ATG2B in p53-deficient cancer cells can activate fibroblasts and accelerate fibroblast-mediated cancer cell proliferation.

Figure 4
figure 4figure 4

Suppression of ATG2B activates fibroblasts. (a) Western blotting analysis of autophagy-related proteins in CCD-18Co cells co-cultured with HCT116sh control or HCT116sh p53 cells. Quantification of ATG2B signal intensity (n = 5, mean ± SD, *P < 0.05). (b) Western blotting analysis of autophagy-related proteins in CCD-18Co cells cultured with or without exosomes derived from HCT116sh p53 or HCT116sh control cells. Quantification of ATG2B signal intensity (n = 5, mean ± SD, *P < 0.05). (c) ATG2B mRNA expression in CCD-18Co cells transfected with ATG2B or control siRNA analysed using quantitative real-time PCR (qRT-PCR) (left; n = 3, mean ± SD, *P < 0.05). The expression of ATG2B in CCD-18Co cells transfected with ATG2B or control siRNA analysed using western blotting (right). (d) Western blotting and autophagic flux assay in CCD-18Co cells transfected with ATG2B or control siRNA (n = 4, mean ± SD, *P < 0.05). (e) Relative expressions of ACTA2, CXCL12, TGFβ1, FGF2, RDGFRA and VEGFA mRNA in CCD-18Co cells transfected with ATG2B or control siRNA analysed using qRT-PCR (n = 3, mean ± SD, *P < 0.05). (f) Cell numbers of HCT116 cells co-cultured with CCD-18Co cells transfected with ATG2B or control siRNA for 24, 48, and 72 h. (n = 3, mean ± SD, *P < 0.05). Protein expression levels were measured using imageJ software 1.8.0_172 (https://imagej.nih.gov/).

Full size image

miR-4534 suppresses ATG2B expression in fibroblasts

Finally, we attempted to identify exosomal miRNAs that regulate ATG2B expression in fibroblasts. Our previous study compared the miRNA profiles in HCT116sh p53 and HCT116sh control exosomes using miRNA microarrays, as detailed in the Gene Expression Omnibus database of the National Center for Biotechnology Information (GSE120013; https://www.ncbi.nlm.nih.gov/)18. We determined the miRNAs targeting ATG2B using the microarray dataset, based on their expression levels in HCT116sh p53 and HCT116sh control exosomes and their probability of targeting ATG2B based on their sequences. miR-4534 was identified as the top de-regulated microRNA with a signal intensity of > 100 in HCT116sh p53 exosomes, an expression ratio (expression in HCT116sh p53 exosomes / expression in HCT116sh control exosomes) > 2, and a target score > 80 as calculated in the miRDB (http://mirdb.org/)19,20 (Fig. 5a).

Figure 5
figure 5figure 5

miR-4534 inhibits ATG2B expression in fibroblasts. (a) Strategy for identifying miR-4534 using microarray analysis and miRDB. (b) The relative expression levels of miR-4534 in exosomes derived from HCT116sh control and HCT116sh p53 cells analysed using quantitative real-time PCR (qRT-PCR) (n = 4, mean ± SD, *P < 0.05). (c) Relative expressions of miR-4534 in CCD-18Co cells transfected with or without miR-4534 mimics analysed using qRT-PCR (n = 5, mean ± SD, *P < 0.05). (d) Western blotting analysis of ATG2B in CCD-18Co cells transfected with or without miR-4534 mimics. (e) Western blotting and autophagic flux assays in CCD-18Co cells transfected with or without miR-4534 mimics (n = 4, mean ± SD, *P < 0.05). (f) Relative expressions of FGF2, PDGFA, ACTA2, CXCL12, TGFβ1, and VEGFA mRNAs in CCD-18Co cells transfected with or without miR-4534 mimics analysed using qRT-PCR (n = 3, mean ± SD, *P < 0.05). Protein expression levels were measured using imageJ software 1.8.0_172 (https://imagej.nih.gov/).

Full size image

We compared the levels of miR-4534 expression in HCT116sh p53 and HCT116sh control exosomes using qRT-PCR with syn-cel-miR-39 as the external control. miR-4534 expression in HCT116sh p53 exosomes was significantly higher than that in HCT116sh control exosomes (Fig. 5b). We then over-expressed miR-4534 in CCD-18Co cells using the miRNA mimic (Fig. 5c). ATG2B expression in CCD-18Co cells was suppressed by the over-expression of miR-4534 than that by control miRNA (Fig. 5d). We examined the autophagy and activation of CCD-18co cells transfected with miR4534 mimic. We found that the autophagy of CCD-18co cells transfected with miR4534 mimic was more suppressed compared to that of CCD-18co cells transfected with miR control (Fig. 5e). The relative expression levels of ACTA2 and CXCL12 mRNA in CCD-18Co cells transfected with miR4534 mimic were higher than with that of CCD-18co cells transfected miR control (Fig. 5f).

Discussion

In the present study, we showed that p53 deficiency in colon cancer cells suppressed autophagy and promoted subsequent activation of fibroblasts. Suppression of the autophagy-related protein, ATG2B, by exosomes derived from TP53-deficient colon cancer cells suppressed autophagy in fibroblasts. Furthermore, we also identified specific miRNAs in exosomes derived from TP53-deficient cancer cells that can suppress expression of ATG2B in fibroblasts.

The role of autophagy in tumor progression is still controversial21,22,23,24,25, particularly in the tumor stromal cells. It has been reported that autophagy in tumor stroma promotes tumor growth by supplying nutrients, including amino acids11,26. Another report also described the pro-tumorigenic effect of autophagy in CAFs by protecting against oxidative stress10; therefore, autophagy in CAFs promotes tumor growth. However, some studies have reported that the suppression of autophagy in fibroblasts can induce their activation27. For instance, myofibroblast differentiation in the pulmonary fibroblasts was promoted by suppression of autophagy and contributed to the pathogenesis of chronic obstructive pulmonary disease28. The present study showed that suppression of autophagy in colon fibroblasts co-existing with TP53-deficient colon cancer cells can induce activation of fibroblasts. Autophagy in tumor stromal fibroblasts may play diverse roles in different phases of tumor progression, such as in intratumoral nutrient supply or myofibroblast differentiation during a desmoplastic reaction. We believe that the suppression of autophagy in fibroblasts induced by cancer cells may be involved in their activation; fibroblast activation plays a pivotal role in the transition of normal tissue fibroblasts into CAFs and formation of tumor stroma. CAFs have higher expression of activation markers and tumor growth factors than normal fibroblasts29; our study showed that normal fibroblasts with suppressed autophagy had acquired properties similar to that of CAFs. This is the first report validating the association between the suppression of autophagy in colon fibroblasts and activation and transition of normal tissue fibroblasts into CAFs.

Normal tissue fibroblasts can acquire CAF-like phenotypes in response to soluble secretions from cancer cells, including exosomes18,30,31. Exosomes communicate information on myriad proteins, mRNAs and miRNAs between secreting and recipient cells, and hence, have an impact on the expression of various genes in recipient cells that governs the conversion of normal fibroblasts into CAFs18,31. Many target genes and related factors, such as TP53, sphingosine kinase 1, E-cadherin, TGFβ1, and senescence are involved in the activation of fibroblasts induced by cancer cell-derived exosomes18,31,32,33,34,35. CAFs have a heterogeneous cell population with varying origins, phenotypes, and functions in the tumor stroma29. Different processes are associated with the transition of normal tissue fibroblasts into CAFs; herein, we focused on the significance of autophagy in activated fibroblasts and identified a specific miRNA in exosomes that can suppress autophagy and induce fibroblast activation with CAF-like phenotypes. Although the detailed mechanisms of miRNA sorting into exosomes have not been elucidated, several reports have indicated that TP53 expression in donor cancer cells can modify their inherent exosomal miRNA profiles and affect gene expression in surrounding cells18,36. After analysis of the miRNAs sorted into exosomes in donor cancer cells with p53 deficiency, we discovered that miR-4534 suppressed ATG2B and activated fibroblasts via suppression of autophagy.

A previous report indicated that miR-4534 was over-expressed in prostate cancer tissues and showed oncogenic effects by downregulating the tumor suppressor gene, PTEN37; however, most of its functions remain unclear. ATG2B is an autophagy-related protein that can regulate the transfer of lipids for autophagosome formation38,39. The present report demonstrates a novel and important function of miR-4534 as a regulator of autophagy via suppression of ATG2B in the tumor stroma during colorectal tumorigenesis. To reveal the clinical usefulness of miR-4534, a better understanding of its roles and further research are required.

In conclusion, loss of p53 function in colon cancer cells may promote the activation of surrounding fibroblasts through the suppression of autophagy. miR-4534 in the exosomes of TP53-deficient cancer cells activated fibroblasts by suppressing ATG2B. Therefore, we propose that the suppression of autophagy by specific miRNAs in exosomes can play an important role in the transition of normal tissue fibroblasts into CAFs and subsequent stroma-mediated tumor growth.

Materials and methods

Cell culture

Human colon cancer cell lines—HCT116 with wild-type TP53 expression, and HT29 with TP53 mutation—were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum. Non-transformed human colon fibroblasts (CCD-18Co) were cultured in Eagle’s minimum essential medium (EMEM) (ATCC, Manassas, VA, USA) supplemented with 10% fetal bovine serum. All cell lines were obtained from the American Type Culture Collection (ATCC); the cell lines were used within < 6 months after purchase, and were authenticated to verify their identity and the absence of cross-contamination (National Institute of Biomedical Innovation, Osaka, Japan). HCT116 and CCD-18Co cell lines were last authenticated on 23 February 2015, and HT29 cell line was last authenticated on 9 August 2018. All cell lines were used between 3 and 9 passages in this study.

RNA interference

Lentiviral GFP-IRES-shRNA vectors against TP53 (RHS4430-101161166; 101162286; 101168779; 99365289) were obtained from Thermo Fisher Scientific (Waltham, MA, USA), and HCT116sh control and HCT116sh p53 cells were generated. Cells were cultured with 2 µg/ml puromycin after colony selection to maintain stable shRNA expression. The cultured fibroblast cell lines were transfected with small interfering RNAs (siRNAs) against the autophagy genes, ATG2B and ATG7 (Invitrogen, Carlsbad, CA, USA), using the transfection reagent, Lipofectamine RNAiMAX (Invitrogen, 13778150), as described in the manufacturer’s instructions.

Exosome isolation

Cancer cells (HCT116 or HT29) were first washed with phosphate-buffered saline (PBS), and then fresh serum-free DMEM was added. After incubation for 48 h, the conditioned medium was collected and centrifuged. Centrifugation was performed in three steps: first at 300×g for 10 min, then at 2000×g for 10 min to remove the cells, and subsequently at 10,000×g for 30 min. The extracellular vesicles (EVs) were separated by ultracentrifugation at 100,000×g for 70 min, and were washed by suspending in PBS. EVs were then ultracentrifuged again at 100,000×g for 70 min. The final pellet was re-suspended in 100 μL of PBS and collected as exosomes. Protein concentration of the purified exosomes was quantified with bicinchoninic acid protein assay kit (Thermo Fisher Scientific). Morphology of the exosomes was observed using transmission electron microscopy (Hitachi High-Technologies Corporation, Tokyo, Japan, H-7650) after preparation using the exosome-TEM-easy kit (101Bio, Palo Alto, CA, USA), according to the manufacturer’s instructions. Exosomes derived from cancer cells were added to fibroblasts at a concentration of 100 µg/mL.

Co-culture experiments

We co-cultured the colon fibroblast (CCD-18Co cells) and cancer cells (HCT116sh control, HCT116sh p53, and HT29 cells) using Transwell inserts with 0.4 μm pore size (Corning, NY, USA). Fibroblasts or cancer cells were seeded in 6- or 12-well plates, and an equal amount of the other cell type was seeded in Transwell inserts. The culture media in the plates and Transwell inserts were changed to EMEM after 24 h, and the Transwell inserts were placed on companion plates. Further assays were performed after co-culturing the cells for 48–72 h.

Cell proliferation and viability

We analyzed cell proliferation in 12-well plates (5 × 104 cells per well). Co-cultured cells were counted using c-chips (NanoEntek, Seoul, Korea) at 24, 48 and 72 h respectively.

Quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was extracted from cell lines using the RNeasy kit (QIAGEN, Tokyo, Japan). Complementary DNA was synthesized from 1 µg of total RNA using the ReverTra Ace qPCR RT Kit (Toyobo, Osaka, Japan). The mRNA expression of genes was quantified using the TaqMan gene expression assay (Applied Biosystems, Foster City, CA). The expression of genes was normalized to that of B2M (beta-2-microglobulin).

We used miRNA-specific primers (TaqMan MicroRNA Assays, Applied Biosystems) for miRNA reverse transcription, and TaqMan gene expression assay for miRNA expression quantification. The expression data were normalized to RNU6B expression for cellular miRNAs and syn-cel-miR-39 spike-in (QIAGEN, Tokyo, Japan) expression for exosomal miRNAs.

Western blotting analysis

Cultured cells were lysed with a lysis buffer (1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate and 1× protein inhibitor cocktail (Nacalai Tesque, Kyoto, Japan), phosphate-buffered saline; pH 7.4). After incubation on ice for 15 min, the lysate was centrifuged at 10,000g for 15 min at 4 °C. The protein content of the supernatants was determined using a bicinchoninic acid protein assay kit (Pierce, Rockford, IL). Equal amounts of protein were separated by sodium dodecyl sulphate–polyacrylamide (SDS) gel electrophoresis and transferred to polyvinylidene difluoride membranes (Bio-Rad, Hercules, CA, USA). We cut the membrane before hybridisation with antibodies. Protein expression levels were measured using imageJ software 1.8.0_172 (https://imagej.nih.gov/). All full length data of western blots are in Supplementary Information.

Autophagic flux assay

We used bafilomycin A1 (BioViotica, Dransfeld, Germany), an autophagic flux inhibitor, prepared at 125 nM concentration in ethanol. Cells were treated with bafilomycin A1 for 2 h before protein extraction. The autophagic flux index was calculated using the following method: autophagy flux index = (LC3-II expression levels with bafilomycin A1)/(LC3-II expression levels without bafilomycin A1). LC3-II expression levels were normalized to beta-actin expression levels. In each experiment, the autophagic flux index of the control group was normalized to 1.

MicroRNA expression analysis

We used miRNA mimics (TaqMan MicroRNA Assays, Applied Biosystems) to over-express miRNAs in CCD-18Co cells. We then evaluated the effect of this over-expression on ATG2B expression using qRT-PCR, as described above.

Statistical analysis

Comparisons between groups were performed using student’s t-test. P < 0.05 was considered statistically significant.