Exosomal transfer of miR-15b-3p enhances tumorigenesis and malignant transformation through the DYNLT1/Caspase-3/Caspase-9 signaling pathway in gastric cancer

Histologically confirmed GC tissue and paired adjacent noncancerous tissue were obtained from 108 patients undergoing surgical procedures at Nanjing Medical University’s First Affiliated Hospital in China. The 108 patients mentioned above, were gender, age and disease history matched with 108 non-GC volunteers, who provided human serum samples. All clinical specimens were collected under the guidance of the Health Insurance Portability and Accountability Act (HIPAA) protocol and were stored at − 80 °C after being frozen in liquid nitrogen, once collected. First Affiliated Hospital of Nanjing Medical University Ethics Committee Approval was obtained to conduct this study, while written consent was obtained from all participants.

The following three GC cell lines: normal GES-1 gastric mucosa epithelium cell line; moderately differentiated adenocarcinoma SGC-7901 cell line and the poorly differentiated adenocarcinoma BGC-823 cell line, were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences. The cells were cultured at 37 °C, in RPMI 1640 medium supplemented with 1% penicillin/streptomycin, 10% fetal bovine serum (FBS) and 5% CO₂. All culture medium reagents were obtained from Gibco, USA.

After the cells had reached a confluency of 70–80%, the medium was changed to a RPMI 1640 medium with 10% exosome-depleted FBS (obtained through ultracentrifugation at 120,000×g at 4 °C for 6 h [18]). After 48 h, 50 ml of the conditioned medium (CM) was collected from each cell line, and ultracentrifugation was used to extract exosomes from the medium, following previously described standard procedures [19]. In order to collect blood samples for serum exosome isolation, ethylenediaminetetraacetic acid (EDTA) containing collection tubes were used. Within an hour, the tubes were centrifuged at 1900×g at 4 °C for 10 min, using a swinging bucket rotor. A new tube was used to collect the upper (yellow) serum phase, and 16,000×g centrifugation at 4 °C for 10 min was conducted to eliminate additional cellular fragments, as well as cell debris. Then, an exoEasy Maxi Kit (Qiagen, Hilden, Germany; Cat. Number: 76064) was used, as instructed by the manufacturer, to isolate serum exosomes. As described in a previous study [20], a FEI Tecnai T20 transmission electron microscope (TEM) (FEI Company, USA) was used to observe the exosomes, while a Nano Sight NS 300 system (Nano Sight Technology, Malvern, UK) was used to determine exosome quantity and size.

TRIzol reagent (Invitrogen, USA) was used to extract total RNA from tissues, cells and CM derived-exosomes, which were purified using a miRNeasy Serum/Plasma Kit (Qiagen, Germany; Cat. Number: 217184), as instructed by the manufacturer. In addition, exosomal RNA was isolated directly from serum, using an exoRNeasy Serum/Plasma MidiKit (Qiagen, Hilden, Germany; Cat. Number: 77044). The miRNeasy Serum/Plasma Spike-In Control (cel-miR-39, Qiagen, Hilden, Germany; Cat. Number: 219610) was used as the serum miRNA expression profiling internal control, as instructed by the manufacturer. The cDNA of the RNAs were created with the aid of a PrimeScript™ RT Reagent Kit (TaKaRa, Japan; Code No. RR037A (miRNAs)/RR036A (mRNAs)). TB Green® Premix Ex Taq™ (TaKaRa, Japan, Code No. RR420A) was used to conduct the qRT -PCR, with the results recorded using ABI StepOne™ Software v2.3 (Applied Biosystems, USA). GAPDH functioned as an internal control for DYNLT1 mRNA levels and the relative miR-15b-3p expression of serum-exosomes were normalized to cel-miR-39, which was normalized to U6 in CM-exosomes, cells and tissues. The 2^−ΔCT formula was used to determine gene expression fold change. Additional file 7: Table S1 lists all primary sequences used.

Lipofectamine2000 Reagent (Invitrogen, USA) and Opti-MEM (Gibco, USA) were used, as instructed by the manufacturer, in 6-well plates to transfect the GenePharma Corporation (SGC, China) synthesized miR-15b-3p mimics/scrambled negative control RNA (NC) or miR-15b-3p inhibitor/scrambled negative control RNA (inhibitor-NC) into cells. After 48 h and 24 h of oligonucleotide transfection, cells were harvested to isolate total cell lysates and total RNA for western blotting and qRT-PCR analyses, in order to determine DYNLT1 and miR-15b-3p levels, respectively. The miR15b-3p mimics and inhibitor sequences mentioned above are listed in Additional file 7: Table S2.

Genechem Inc. (China) constructed luciferase-labelled lentivirus vectors carrying miR-15b-3p (Lv-miR-15b-3p)/negative control (Lv-NC), miR-15b-3p inhibitor (Lv-inhibitor)/negative control (Lv-inNC) and GFP-labelled lentivirus vectors containing CD63 (GFP-Lv-CD63) were used. BGC-823 cells were infected in 6-well plates, using 10 μl of the aforementioned lentiviral vectors for 3 days at 37 °C. Then, selection of successful lentiviral transfected cells was done using 1.0 μg/ml puromycin (Sigma Aldrich, USA). The primers used for the amplification of miR-15b-3p were: 5′-.

AGGTATGCACGCGTGAATTGTTACTTTTTTTTCTATAAAGCTAGGTTGG - 3′ (sense) and 5′-GCCGACACGGGTTAGGATCAAAAAACACTACGCCAATATTTA-CGTGC-3′(antisense). Sequences used for the Lv-miR-15b-3p inhibitor were: 5′-AATTCAAAAACGAATCATTATTTGCTGCTCTA-3′ (sense) and 5′-CCGGTAGAGCAGCAAATAATGATTCGTTTTTG-3′(antisense). qRT-PCR was performed to validate infection efficiency.

Into 6-well plates, the harvested cells were added at a concentration of 1 × 10³ cells/well, for 10–15 days, to be used for the colony formation assay. Fixation of the colonies were done using 2 ml of paraformaldehyde for 30 min, while 0.1% crystal violet was used for 30 min at room temperature for cell staining. In addition, a Cell-Light EdU Apollo567 In Vitro Kit (RiboBio, China) and a Cell Counting Kit-8 (CCK-8) kit (Dojindo Laboratories, Japan) were utilized to evaluate the proliferation of the cells. For the CCK-8 assay, into each well of a 96-well plate containing 2 × 10³ transfected cells, 10 μL of CCK-8 reagent was added at the same time every day for further incubation (2 h). A Microplate reader (ELX-800; Bio-Tek, USA) was used to measure absorption at 450 nm, at a series of time points (0, 24, 36, 48, 72 and 96 h). For the 5-ethynyl-2′-deoxyuridine (EdU) assay, rigorous processing was carried out on the cells in 96-well plates with the cells at a concentration of 2 × 10⁴ cells/well, as instructed by the manufacturer [21]. Finally, a Nikon ECLIPSE E800 fluorescence microscope was used to examine the cell samples.

An Annexin V-PI apoptosis detection kit (Vazyme Biotech Co. Ltd., China) was used in a manner similar to that of a previous description [22, 23] to detect apoptosis. Thereafter, fluorescence-activated cell sorting (FACS) was utilized to count the stained cells using CellQuest software (BD Biosciences, USA) connected to a Calibur flow cytometer. A TUNEL FITC Apoptosis Detection Kit (Vazyme Biotech Co. Ltd., China) was used, as instructed by the manufacturer, to conduct TUNEL staining. Immunofluorescence was observed using a Nikon ECLIPSE E800 fluorescence microscope.

First, into a 24-well plate, transwell assay inserts (Millipore, USA) were added. A Matrigel-coated membrane (50 μL/well, BD Biosciences, Franklin Lakes, NJ) was used for the invasion assay, while a normal membrane was used for the migration assay, as the apical chamber membrane. Then, 600 μL of 10% FBS containing medium was seeded into the basolateral chamber, and 100 μL of FBS-free RPMI 1640 medium (Gibco, USA) was added into the apical chamber containing 2 × 10⁵ cells in each well to re-suspend the cells. After incubation for 24 h at 37 °C, PBS was used to rinse the Transwell plates twice, fix with 4% paraformaldehyde for 30 min, while 0.1% crystal was used for 30 min at room temperature, for staining. Subsequently, utilizing an inverted light microscope the cells were observed, photographed and counted.

The pmirGLO dual-luciferase miRNA target expression vector (Promega, USA) was transfected with the PCR amplified 3′ untranslated regions (3′-UTR) of DYNLT1 mRNA. In 24-well plates, the luciferase construct containing wild-type (WT) or mutated binding site of DYNLT1 (constructed by Genechem Inc., China) were transfected into target cells. This was followed by co-transfection with miR-15b-3p mimics, inhibitor, NC or inhibitor-NC using Lipofectamine2000, to identify the binding site between DYNLT1 and miR-15b-3p. Determination of luciferase activity after 48 h of transfection and normalization with Renilla luciferase was done utilizing a Dual-Luciferase Reporter System Kit (E1910, Promega, USA), as previously reported [24].

Protein extraction from cells, tissues, and exosomes were performed using a radioimmunoprecipitation assay (RIPA) kit (Sigma-Aldrich, USA), as instructed by the manufacturer. After determination of protein concentration using a bicinchoninic acid (BCA) kit (Pierce, USA), SDS-containing polyacrylamide gel (SDS-PAGE) was used for the separation of equal amounts (35 μg for cells and tissues, and 10 μg for exosome pellets) of protein samples. Thereafter, the samples were moved onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad, USA). Then, for 1 h, 5% non-fat milk in TBSTween (TBST) (0.1 M, pH 7.4) was used to block the membranes, followed by hybridization with primary antibodies against CD9 (ab92726, 1:1000 dilution), CD63 (ab217345, 1:1000 dilution), DYNLT1 (ab202583, 1:2000 dilution), BAX (ab32503, 1:1000 dilution), BCL-2 (ab32124, 1:1000 dilution) and TSG101 (ab125011, 1:1000 dilution), from Abcam (USA); Cleaved caspase-3 (9664, 1:1000 dilution) and Cleaved caspase-9 (7237, 1:1000 dilution), from Cell Signaling Technology (USA), overnight at 4 °C. The antibodies for GAPDH (QYA03819B, 1:2000 dilution) and β-Actin (sc-47,778, 1:1000 dilution) from Santa Cruz Biotechnology (USA) served as reference proteins. The immunocomplexes were incubated with corresponding horseradish peroxidase conjugated secondary antibodies (Applygen, China; 1:2000 dilution), for 2 h at room temperature. Thereafter, an enhanced chemiluminescence assay was conducted on a SuperSignal™ West Femto Maximum Sensitivity Substrate (34,095, Thermo Fisher, USA) to visualize the blots.

The cells cultured on four-well chamber slides were washed with PBS thrice, fixed using 4% paraformaldehyde for 15 min, once again washed with PBS, and permeabilized using 0.5% Triton-X 100 (dissolved in PBS) for 20 min. For exosome tracking, exosomes secreted by the BGC-823 cells were labeled using PKH26 red fluorescent dye (Sigma-Aldrich, USA) or exosomal marker, CD63 (green; Genechem Inc., China), while F-actin was stained using phalloidin-FITC (green), and DAPI (blue) was used to label nuclei. Cy3-(miR-15b-3p inhibitor/inhibitor-NC/mimics/NC) were synthesized, as well as purified by RiboBio Co. (China). A Nikon ECLIPSE E800 fluorescence microscope was used to capture images. The uptake capacity of SGC-7901 and GES-1 into exosomes containing different miRNA sequences (mimics/NC/inhibitor/inhibitor-NC) was determined using immunofluorescence assays and qRT-PCR.

6–8 week old BALB/c-nu male nude mice were kept in an animal facility that was pathogen-free and were randomly divided into five groups (n = 5). The groups received subcutaneous injections of miR-15b-3p enriched/Lv-NC exosomes, miR-15b-3p inhibited/Lv-inNC exosomes (1 × 10⁹ exosomes/ml) or PBS treated SGC-7901 cells (2 × 10⁶ cells in 200 μl PBS). The anesthetization of the mice was done using xylazine (10 mg/kg) or ketamine (100 mg/kg), while bioluminescence signals were observed using an IVIS 100 Imaging System (Xenogen, USA) 15 min after D-luciferin (100 mg/kg, Xenogen, USA) was injected into the mice. Once in 4 days, a digital caliper was used to measure the tumors and the following formula was used to calculate tumor volume: (width² × length)/2, until euthanasia, 28 days after cell inoculation. Finally, the subcutaneous tumors of the mice were excised and at room temperature were frozen in liquid nitrogen or fixed in 4% paraformaldehyde for subsequent studies. Approved protocols provided by Nanjing Medical University’s Institutional Animal Care and Research Advisory Committee were followed for all animal experiments.

The tumor masses of both mice and clinical samples were 4% paraformaldehyde fixed, paraffin embedded at 58 °C and cut into 4 μm sections, followed by staining with anti-DYNLT1 antibodies (1:50 dilution, ab202583, Abcam, USA). Aperio Scan-Scope AT Turbo (Aperio, USA) was used to capture images of the tumors, while image-scope software (Media Cybernetics Inc.) was used to conduct the quantitative analysis.

GraphPad Prism 7.00 Software (USA) and SPSS version 22.0 (SPSS, USA) were used to conduct the statistical analyses. Expression is presented as mean ± SEM of at least three independent experiments for all results. One-way analysis of variance (ANOVA) or student’s t test was performed to determine statistical differences among two or more groups. Sensitivity, specificity, and area under the curve (AUC), including 95% confidence interval (CI), were computed with the aid of the constructed receiver-operating characteristic (ROC) curves, using the Youden index (J) [25] to calculate the optimum cut-off values. The survival analysis included log-rank tests and Kaplan–Meier analyses. A P value of < 0.05 was used to indicate a statistically significant result. For all figures: *, P < 0.05; **, P < 0.01; &**, P < 0.001; and &&**, P < 0.0001.

Gene Expression Omnibus (GEO) database microarray data (accession number: GSE86226) were analyzed and the top 61 highly expressed miRNAs (fold change > 1.5, FDR < 0.01), in comparison with 3 pooled peripheral serum samples from 30 GC patients and 1 pooled sample from 10 controls (Fig. 1a). 281 miRNAs with highly significant expression were identified after the same criteria (fold change > 1.5, FDR < 0.01) for GC tissues was applied to the TCGA database (Fig. 1b). 29 miRNAs were found to fall into the intersection between the two sets of data (Fig. 1c). Among them, miR-15b-3p was the most prominent in the tumor tissues of 108 GC patients (Additional file 1: Figure S1a-l and Fig. 1d), which is consistent with the expression trend in the TCGA database (Fig. 1e). Table 1 shows the baseline characteristics of the 108 GC patients. Next, significantly higher miR-15b-3p levels were found in GC serum, compared with that of normal serum using qRT-PCR assay (n = 30, Fig. 1f). In addition, compared with the GES-1 cell line (Fig. 1g), distinctively elevated expression of miR-15b-3p was found in the moderately differentiated adenocarcinoma SGC-7901 cell line and the poorly differentiated adenocarcinoma BGC-823 cell line. Considering the significantly higher miR-15b-3p expression in BGC-823 cells, above that of SGC-7901 cells, we hypothesized that the miR-15b-3p expression is higher in GC cell lines with high malignancy. Collectively, these results show that GC development may involve miR-15b-3p. Hence, we focused on the functional role of miR-15b-3p.

In order to determine whether miR-15b-3p plays a role in GC progression, we first investigated its effect on GC cell proliferation. Additional file 2 Figure S2 shows miR-15b-3p expression after transfection into SGC-7901 and BGC-823 cells. Results of the colony formation, CCK-8 and 5-ethynyl-2′-deoxy-uridine (EdU) assays reveal that in comparison with the respective control groups, treatment with miR-15b-3p mimics accelerate the proliferation of SGC-7901 and BGC-823 cells, while the miR-15b-3p inhibitor significantly inhibits their proliferation (Fig. 2a-c). Figure 2d shows that compared with that of the control group, the invasion and migration rates of the miR-15b-3p mimics-transfected SGC-7901 and BGC-823 cells were significantly higher, whereas the miR-15b-3p inhibitor alone transfected cells could only migrate or invade a short distance. TUNEL assay and flow cytometry analysis were used to explore whether the regulation of apoptosis is a potential factor for miR-15b-3p-induced cell growth progression. Thus, the apoptotic percentage of miR-15b-3p-silenced GC cells were found to be obviously elevated, and the cells overexpressing miR-15b-3p were found to show lower apoptosis levels (Fig. 2e and f). In addition, apoptosis-related protein expression levels were ascertained using western blotting analysis, as shown in Fig. 2g. Significant upregulation of the anti-apoptotic protein, BCL-2, expression was detected in the miR-15b-3p mimics group, which is contrary to that of BAX, Cleaved caspase-9 and Cleaved caspase-3 levels. The above mentioned changes were found to be opposite to the changes seen in the miR-15b-3p inhibition group. Accordingly, we speculate that miR-15b-3p functions as an oncogene in GC.

miRNAs play a key role as negative gene expression regulators at post-transcription level by fusing target mRNA complementary 3’UTR sequences [13, 16]. In order to explore miR-15b-3p regulation at mRNA level, prediction of potential miR-15b-3p target genes was done by simultaneously using four bioinformatics tools (miRDB, RNA22, TarBase and TargetScan) (Fig. 3a). A significant decrease in DYNLT1 at mRNA level was found in GC tissues, as shown by qRT-PCR assays (Additional file 3: Figure S3a-c and Fig. 3b). A similar trend was seen with DYNLT1 expression in TCGA database GC tissues and normal tissues (Fig. 3c). Determination of DYNLT1 expression was done using western blotting and immunohistochemistry (IHC) analyses. As predicted, the results were similar, indicating that in GC tissues, DYNLT1 is downregulated, compared with matched normal tissues (n = 30, Fig. 3d-f). The results of the qRT-PCR analysis performed on 108 paired GC tumor tissues (R² = 0.3655, P<0.0001), SGC-7901 cells (R² = 0.7726, P = 0.0008) and BGC-823 cells (R² = 0.8703, P<0.0001) found a negative correlation between the expressions of miR-15b-3p and DYNLT1 (Fig. 3g and Additional file 4: Figure S4). In addition, miR-15b-3p was confirmed to downregulate DYNLT1 expression in BGC-823 and SGC-7901 cells at both mRNA, as well as protein level, while DYNLT1 expression upregulation could be achieved by silencing miR-15b-3p (Fig. 3h and i). The direct interaction between miR-15b-3p and DYNLT1 was demonstrated using either a wild-type (WT) or mutant 3′-UTR DYNLT1 mRNA containing a luciferase reporter plasmid. As shown in Fig. 3j, miR-15b-3p harbors a complementary binding sequence of DYNLT1. Subsequently, miR-15b-3p overexpressing BGC-823 and SGC-7901 cells showed a significant decrease in luciferase activity, whereas the luciferase activity was obviously enhanced by miR-15b-3p inhibition (Fig. 3k). However, the loss of binding sites eliminated the miR-15b-3p inhibitory effect on luciferase activity, as shown in Fig. 3k. Thus, DYNLT1 was confirmed as a direct downstream miR-15b-3p target.

Considering that exosomes are highly stable disease biomarkers, exo-miRNAs may be potential GC diagnostic or prognostic biomarkers that are more accurate and stable than miRNAs [26‐28]. In order to explore whether exo-miR-15b-3p performs the above functions, we first extracted and purified exosomes from the conditioned media of three cell lines (BGC-823, SGC-7901 and GES-1) and the serum of GC patients and non-GC volunteers (n = 108, patients and volunteers as listed in Table 1). Exosome quantity and number, as well as their cup-shaped morphology were determined through TEM analysis and Nano Sight particle tracking analysis (Fig. 4a and b). Moreover, the exosomal markers, TSG101, CD63 and CD9, were identified using western blotting analysis, further confirming that exosomes were the particles that were isolated (Fig. 4c). exo-miR-15b-3p was found to be enriched in the CM of SGC-7901 cells, and in particular, BGC-823 cells, rather than GES-1 cells, as shown by the qRT-PCR assay (Fig. 4d). Similarly, miR-15b-3p expression in the 108 GC patient serum exosomes was found to be significantly higher, compared with that of corresponding non-GC controls (Fig. 4e). exo-miR-15b-3p diagnostic efficacy for GC was determined using the ROC curve. The results show that the AUC of 0.820 (95%CI, 0.763–0.876), with a specificity of 80.6% and a sensitivity of 74.1% was obtained for exo-miR-15b-3p, as shown in Fig. 4f. The diagnostic effect of serum exo-miR-15b-3p was found to be better than that of miR-15b-3p in tissues (AUC = 0.674 [0.600–0.748]) and serum (AUC = 0.642 [0.499–0.786], Additional file 5: Figure S5a and b). The correlation between clinicopathological features and serum exo-miR-15b-3p levels of expression were determined by dividing patients into a high-expression group and a low-expression group, with 54 patients assigned to each, based on the median miR-15b-3p expression level. A statistically significant correlation was observed between high serum exo-miR-15b-3p expression and alcohol abuse, tumor size (≥3.5 cm in diameter), poorly differentiated histological type, TNM stage (III and IV) and lymph vascular invasion (Table 2). Furthermore, Kaplan-Meier analysis was performed to judge whether serum exo-miR-15b-3p expression is correlated with GC patient cancer-specific survival. As shown in Fig. 4g, poor overall survival (P = 0.019) can be accurately predicted by high exo-miR-15b-3p expression levels. These results indicate that serum-secreted exo-miR-15b-3p can function as a sensitive and specific predictive and prognostic liquid biomarker for GC and may be related with the malignant transformation of GC.

Since miR-15b-3p expression of BGC-823 cells (poorly differentiated adenocarcinoma) is higher than that of SGC-7901 and GES-1 cells, we hypothesized that exosomes can mediate a novel mechanism of cell-to-cell communication in GC, by transmitting miR-15b-3p between cells of varying degrees of differentiation and malignancy, and then take part in the malignant transformation of GC. In order to confirm our assumption and the manner of miR-15b-3p intercellular delivery, we performed co-culture experiments to determine if exosomes and their contents could be internalized by target cells. First, 50 mg of BGC-823 cell derived PKH26-labeled exosomes were incubated with 5 × 10⁵ GES-1 or SGC-7901 cells, and the uptake of exosomes was observed after co-culture for 0, 6, 12 and 24 h. It was found that in a time-dependent manner, GES-1 and SGC-7901 cells gradually engulfed the exosomes (Fig. 5a). After 24 h of co-cultivation, many exosomes were found to have entered recipient cells and accumulated around the nucleus (Fig. 5a). Moreover, in order to visualize the exosome-mediated intercellular miRNA transfer, after confirming that GFP-Lv-CD63 had been successfully transfected into BGC-823 cells (Additional file 6: Figure S6a), the fluorescence labeled miR-15b-3p mimics (Cy3-miR-15b-3p mimics) was transiently transfected into BGC-823 cells (Additional file 6: Figure S6b) and then, the medium was refreshed (Additional file 6: Figure S6c). Next, exosomes in the CM of transfected BGC-823 cells were further isolated and added to the untreated GES-1 and SGC-7901 cells for 24 h. The apparent green and red fluorescence seen in Fig. 5b confirms the successful shuttling of Cy3-miR-15b-3p mimics through exosomes into recipient cells. Moreover, Cy3-miR-15b-3p mimics and CD63-labeled exosomes were found to be co-localized in the cytoplasm (Fig. 5b). In addition, the qRT-PCR assay results show that oligonucleotide sequences (miR-15b-3p mimics/NC/inhibitor/inhibitor-NC) can be taken-up by exosomes and transported into the extracellular medium, where recipient cell uptake regulates miR-15b-3p expression (Fig. 5c).

For further investigating BGC-823 cell-derived exo-miR-15b-3p function in recipient cells, we isolated exosomes from the CM of miR-15b-3p inhibitor/inhibitor-NC /mimics/NC transfected BGC-823 cells. Next, 50 mg of the purified exosomes or PBS was co-cultured for 24 h with 5 × 10⁵ GES-1 or SGC-7901 cells. As expected, when exo-miR-15b-3p mimics were incubated with either SGC-7901 or GES-1 cell lines, increased cell proliferation (tested using colony formation, CCK-8 and EdU assays) (Fig. 6a-c), cell invasion and migration (tested using Transwell chamber migration assay) (Fig. 6d) were observed. In contrast, significant repression of these biological functions were found in cells co-cultured with exosomes, in which miR-15b-3p was knocked down (Fig. 6a-d). However, recipient cells treated with control exosomes (Exo-NC and Exo-inhibitor-NC) showed greater proliferation, migration and invasion capacities than those treated with PBS (Fig. 6a-d). Furthermore, we found that the apoptosis of GES-1 and SGC-7901 cells treated with exosomes carrying miR-15b-3p mimics was significantly reduced, while knockdown of exo-miR-15b-3p reversed the apoptosis of these cells (Fig. 6e-g). At the protein level, the expression of DYNLT1 and BAX were found to be inhibited in GES-1 and SGC-7901 cells treated with exo-miR-15b-3p mimics, compared with Exo-NC and PBS groups. Conversely, transfection with exosomes containing a miR-15b-3p inhibitor had an opposite effect on their expression (Fig. 6g). In addition, anti-apoptotic protein BCL-2 levels increased in GES-1 and SGC-7901 cells co-incubated with exosomes packed with miR-15b-3p mimics and decreased in cells treated with the exo-miR-15b-3p inhibitor (Fig. 6g). It is known that DYNLT1 is involved in the regulation of apoptosis [29, 30]. In order to further explore its mechanism of action, we analyzed levels of major proteins in the Caspase-3 classical apoptosis signaling pathway and found increased cleavage of Caspase-9 and Caspase-3 in the high DYNLT1 expression group, which was inhibited in the low DYNLT1 expression group (Fig. 6g). Collectively, these results indicate that BGC-823 cell-derived exo-miR-15b-3p is effectively involved in the malignant transformation of recipient cells.

Subsequently, the in vitro changes seen in GES-1 and GC cells in the presence of exo-miR-15b-3p were confirmed in vivo. The nude mice were subcutaneously injected with SGC-7901 cells, following BGC-823 cell-derived Exo-Lv-NC, Exo-Lv-miR-15b-3p, Exo-Lv-inNC and Exo-Lv-inhibitor or PBS pre-incubation. The stable transfection efficiency of the luciferase-labelled lentivirus into BGC-823 cells and the miR-15b-3p expression of exosomes derived from cells of different treatment groups are shown in Fig. 7a. We observed that tumor growth increased significantly in mice receiving exosomes enriched with miR-15b-3p, compared with those injected with PBS or exosomes containing Lv-NC (Fig. 7b and c), at the same time luciferase intensities were also detected (Fig. 7d). However, visibly smaller tumors were formed in the Exo-Lv-inhibitor group (Fig. 7b-d). Next, tumor tissues harvested for TUNEL staining were found to show a decrease in GC cell apoptotic rate, after treatment with Exo-Lv-miR-15b-3p, compared with those treated with Exo-Lv-NC, whereas the opposite result was detected in the Exo-Lv-inhibitor group (Fig. 7e and f). Furthermore, we found the results to be consistent with the in vitro results, indicating that both exosome-delivered NC and inhibitor-NC significantly repressed cell apoptosis, compared with treatment with PBS alone, suggesting that BGC-823 cell-derived exosomes can inhibit target cell apoptosis (Fig. 7e and f). Compared with the control group, qRT-PCR quantified tumor tissue miR-15b-3p levels were significantly higher in the Exo-Lv-miR-15b-3p group and decreased in the Exo-Lv-inhibitor group, while the results of both qRT-PCR and IHC assays indicate that DYNLT1 expression in tissues is inhibited in the first group and enhanced in the second group (Fig. 7g and h). However, as shown in Fig. 7g and h, no obvious difference was found in the PBS group. Exo-Lv-miR-15b-3p treated tumors with high miR-15b-3p levels tended to express lower protein levels of DYNLT1, BAX, Cleaved caspase-9 and Cleaved caspase-3, but higher protein levels of BCL-2 (Fig. 7i). Conversely, higher protein levels of DYNLT1, BAX, Cleaved caspase-9 and Cleaved caspase-3, but lower BCL-2 levels were observed in the low miR-15b-3p groups (Exo-Lv-inhibitor treated) (Fig. 7i). Our results suggest that the exo-miR-15b-3p/DYNLT1 axis inhibits apoptosis via modulating the Caspase-3/Caspase-9 signaling pathway, maintaining high levels of SGC-7901 cell proliferation in vivo.