PABPC1-induced stabilization of IFI27 mRNA promotes angiogenesis and malignant progression in esophageal squamous cell carcinoma through exosomal miRNA-21-5p

A total of 190 ESCC tissue samples were obtained from patients who underwent cholecystectomy without prior radiotherapy or chemotherapy between 2015 and 2020 at Shantou University Medical College. Informed consent was obtained from all patients participating in the study. This study was approved by the Institutional Ethics Board of the Cancer Hospital of Shantou University Medical College (2021117). ESCC cell lines TE1, 81 T, KYSE410, KYSE180, KYSE450, KYSE150, KYSE510, and KYSE520 were cultured in high-glucose Dulbecco’s modified Eagle’s medium (DMEM, Gibco, Shanghai, China) or Roswell Park Memorial Institute (RPMI, Gibco) 1640 medium supplemented with 10% fetal bovine serum (FBS, Gibco). All cells were maintained in a humidified incubator (5% CO2) at 37°C. Immortalized esophageal epithelial NE1 cells were cultured in Eagle’s minimum essential medium (Gibco) supplemented with 10% FBS (Gibco). Cell lines were authenticated by short tandem repeat (STR) profiling and confirmed to be mycoplasma negative.

Cells were lysed in RIPA buffer (Beyotime, Jiangsu, China) containing a protease inhibitor cocktail (Sigma, St. Louis, CA, USA). Total protein was assessed using a BCA Protein Assay Kit (Beyotime). Western blot analysis was performed as previously described [18]. Antibodies against the following were used in this study, PABPC1 (1:1000, Abcam, SF, USA), IFI27, HA tag, Flag tag, TSG101, CD63, CD9, PCNA, Alix, Ki67, caspase 3, CXCL10, CD34, cleaved-PARP, PARP, eIF4G, cleaved caspase 9, caspase 9, ERK, p-ERK, IFI27, STAT3, p-STAT3, STAT1, p-STAT1, NF-kB, p-NF-kB, β-actin and GAPDH (all 1:1000, Cell Signaling Technology, MA, USA), EXOSC2 and EXOSC4 (both 1:1000, Santa Cruz, MA, USA).

Using an RNeasy Mini Kit (Qiagen, Valencia, CA, USA), total cellular RNA was extracted from cells. qRT-PCR was performed according to a previously published procedure using the SYBR Green system (Takara, Dalian, China) [19]. The primer sequences are shown in Table S1. For RNA-seq, the mRNA was amplified by PCR and sequenced using an Illumina NovaSeq 6000 (Gene Denovo Biotechnology Co., Guangzhou, China).

IHC staining was performed using the Envision Labeled Peroxidase System (Dako, Carpinteria, CA, USA) as described previously [19]. PABPC1 and IFI27 expression were analyzed based on the proportion and intensity of positively-stained tumor cells. The scores for intensity and fraction of positive cells were multiplied; scores of 0–4 were defined as low expression, and scores of 6–12 were defined as high expression. Immunofluorescence was performed as previously described [19]. Briefly, cells were first fixed in 3% paraformaldehyde, and then permeabilized with 0.5% Triton X-100. The cells were blocked with 5% FBS in PBS, followed by incubation with primary antibody, fluorescent secondary antibody (Invitrogen) and counterstained with DAPI (Beyotime) were then incubated with cells to visualize the targeted proteins and nuclei. The data were then analyzed by fluorescence microscopy.

Carbobenzoxy-Leu-Leu-leucinal (MG132), actinomycin D, and cycloheximide (CHX) were purchased from MCE (St. Louis, MO, USA). 5-Ethynyl-uridine (5-EU) and HDAC inhibitor sodium butyrate (NaBu) were purchased from Selleck (St. Louis, MO, USA). The siRNAs were designed and synthesized by GenePharma Company (Shanghai, China). Full-length PABPC1 plasmid, five truncated PABPC1 RRM1-RRM4 and MLLE plasmids, mutant PABPC1 ^M161A/D165K and RRM1 deletion PABPC1 plasmids, eIF4G, and IFI27 plasmids were purchased from Vigene (Shanghai, China). ESCC cell transfection was carried out using Lipofectamine 3000, Lipofectamine IMAX, and Opti-MEM (Invitrogen, Shanghai, China). For stable expression of PABPC1 or IFI27 in OS cell lines, lentivirus or short hairpin(sh) -RNA transfected cells were selected with puromycin (5 μg/mL, MCE) for 2 weeks.

Co-IP was performed using the relevant antibodies and protein A/G-conjugated Dynabeads (Beyotime) according to the manufacturer's instructions. In brief, cell lysates were incubated with antibodies overnight at 4 °C. Protein A/G-conjugated beads were added into the lysate at 4 °C for 2 h. Then, the beads were washed with lysis buffer or PBS and boiled in SDS loading buffer. Western blotting was used to detect the immunoprecipitated proteins.

Cell proliferation was examined with a Cell Counting Kit-8 (CCK-8) according to the manufacturer’s instructions (Beyotime). Apoptosis was determined by flow cytometry using Annexin V/PI staining according to the manufacturer’s protocol, and data analysis was performed using BD Accuri C6 software. Cell migration was evaluated by scratch wound healing assay where wounds were scratched on the monolayer of cells using a 200 μL pipette tip. After the cells had been cultured for 48 h, plates were washed once with fresh medium to remove non-adherent cells, and the plates were then photographed. Cell invasion and migration was tested using a transwell assay. Briefly, 100 μL Matrigel (Corning, CA, USA) was first added onto the bottom of the transwell chamber (Corning), and then 1 × 10⁵ cells in serum free medium were placed on the coated membrane in the top chamber and the bottom chamber was filled with DMEM with 10% FBS, then incubated for 24 h. The membrane was then fixed and stained with crystal violet. The cell imagines were taken and counted in 3 random fields with microscope.

After co-cultured with different treatment of ESCC cells, 10⁴ HUVECs were cultured in μ-Slide Angiogenesis plate (ibidi, Germany) coated with 10 μL Matrigel (R&D Systems, Minneapolis, MN) for 6 h at 37 °C. The formation of capillary-like structures was captured under a light microscope. The degree of in vitro angiogenesis was represented the formed tube numbers scanned and quantitated in five low-power fields (100x).

Briefly, HEK293T or tumor cells (3 × 10⁴ cells per well) grown in a 24-well plate were co-transfected, with a luciferase reporter (200 ng per well), miR-21-5p mimic or inhibitor and 10 ng Renilla luciferase vector (pRL-CMV; Genomeditech, China), using Lipofectamine™ 3000 (Invitrogen). After 48 h, a dual reporter luciferase assay was performed according to the manufacturer’s instructions (Promega, Madison, MD, USA). The relative luciferase activity was expressed as the ratio of firefly luciferase activity to Renilla luciferase activity.

To measure the stability of IFI27 mRNA, cells were treated with 5 μg/mL actinomycin D (MCE) to block transcription. Cells were collected at 0–24 h after addition of actinomycin D, and the total cellular RNA was isolated. qRT-PCR was performed to measure the half-life of RNA, and GAPDH mRNA was used as an internal control.

To capture the newly synthesized RNA transcripts of IFI27 mRNA, a nascent RNA capture assay was carried out using the Click-IT Nascent RNA Capture Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol.

A poly (A) tail-length assay was performed using a Poly(A) Tail-Length Assay Kit (Invitrogen) according to the manufacturer’ s instructions. Briefly, total RNAs were first added to poly (G/I) and reverse transcribed. Then, the poly (A) length was detected by PCR using both gene-specific primers and the Universal PCR Reverse Primer (Table S1).

ChIP assays were performed using an EZ-ChIP kit (Millipore, Beijing, China). DNA was analyzed by qPCR with SYBR Green (Bio-Rad, Shanghai, China) on an ABI-7500 (Applied Biosystems, Shanghai, China) using the primers specified in Table S1. RIP assays were performed using an EZ-Magna RIP RNA-Binding Protein Immunoprecipitation kit (Millipore). Briefly, cells were collected and lysed in lysis buffer with protease inhibitors. The protein extract (500 μg) was incubated with 3 μg of PABPC1 antibody or IgG overnight at 4 °C. Approximately 30 μL of A/G protein magnetic beads were then added and the mixture was incubated at 4 °C for 4 h. After washing, co-immunoprecipitated RNAs were extracted. RNA fold enrichment is presented as percent input and compared with the IgG control.

Biotin RNA pull-down experiments were performed using RNA pulldown kits (Bersin Bio, Guangzhou, China) according to the manufacturer’s instructions. Briefly, the cell lysates were incubated with 100 pmol of synthetic 5’-end biotin-modified IFI27 oligonucleotides overnight at 4 °C. After adding streptavidin agarose beads and incubating at 4 °C for 4 h, precipitates were washed five times and boiled in SDS buffer, followed by western blotting analysis.

Exosome isolation was performed as previously described [18]. In brief, cancer cells were cultured with exosome-free FBS-containing media and grown to 70% confluence. Then, cells washed 3 times with PBS, and incubated for 24 h in serum-free media and the supernatant was collected. The supernatant was centrifuged at 3000 × g for 15 min to remove cells and cell debris and it was then mixed with ExoQuick exosome precipitation solution (SBI, Palo Alto, CA) and incubated overnight according to the manufacturer’s protocol. Then, the mixture was centrifuged at 1500 × g for 30 min at 4 °C. The pelleted exosomes were dissolved in PBS and were subsequently divided and transferred to RNase-free tubes to be stored or undergo electron microscopy, protein assays, RNA extraction, and use in in vitro or in vivo treatment. For exosome uptake experiments, exosomes were labeled with a PKH67 Green Fluorescent Cell Linker Kit (Sigma) following the manufacturer’s protocol. Then, 10 µg of exosomes was resuspended in 100 µl PBS and were added to 1 × 10⁵ HUVECs. HUVECs were harvested at 24 h for qRT-PCR and immunofluorescence analysis.

For in vivo assays, each experimental group consisted of six 5-week-old male BALB/c-nu/nu mice. Briefly, 1.0 \(\times\) 10⁷ cells stably overexpressing PABPC1 were suspended in 50 μL serum-free DMEM/Matrigel (1:1) and injected into the oxter of each mouse. The mice were euthanized at 5 weeks after injection, and the weight and size of the tumors were measured. For lung metastasis assays, 1.0 \(\times\) 10⁶ PABPC1-overexpressing cells were suspended in 50 μL serum-free DMEM and injected into the tail veins of mice (n = 6/group). After 6 to 8 weeks, mice were euthanized, lung tissue was excised, and tumor nodules formed in the lung were counted and analyzed by H&E staining. The experimental protocol was approved by the Ethics Committee of Animal Experiments of the Cancer Hospital of Shantou University Medical College (SUMC2021-035).

Statistical analyses were performed with SPSS version 22.0. Data are presented as the mean \(\pm\) standard deviation (SD), and statistical significance was determined using unpaired Student’s t-tests. Survival curves were generated by the Kaplan–Meier method. P-values < 0.05 were considered to indicate statistical significance.

Analyses of the expression levels of all RBPs in the esophageal cancer database from The Cancer Genome Atlas (TCGA) revealed higher PABPC1 expression in ESCC tissue than in normal esophageal epithelial tissue (Fig. 1A). We also examined PABPC1 expression levels in pan-cancers in the Gene Expression Profiling Interactive Analysis (GEPIA) database (http://​gepia.​cancer-pku.​cn/​index.​html) and found that PABPC1 was expressed at higher levels in tumor tissue (Fig. 1B). In our ESCC cohort, PABPC1 expression was higher in ESCC tissue than in normal esophageal mucosa in the majority of cases (15/24), as determined by western blotting (Fig. 1C and S1A). Based on our data (n = 48) and TCGA, the RNA expression of PABPC1 was higher in ESCC compared with normal ESCC epithelia (Fig. 1D). As shown in Fig. 1E and F, the IHC staining score of PABPC1 was higher in ESCC tissue than in case-matched normal epithelial tissue (n = 190), and positively correlated with the microvessel density (MVD) detected by CD34 antibody. In addition, PABPC1 expression was higher in eight ESCC cell lines (KYSE150, KYSE520, KYSE510, KYSE410, KYSE450, KYSE180, 81 T, and TE1) compared with an immortalized normal esophageal cell line (NE1), as detected by western blotting and qRT-PCR (Fig. 1G). We assessed the subcellular localization of PABPC1 using immunofluorescence and observed detectable but weak nuclear PABPC1 and strong PABPC1 staining in the cytoplasm of KYSE150 and KYSE520 cells (Fig. 1H and S1B). According to the IHC scores, PABPC1 expression in ESCC tissue was positively and significantly correlated with tumor differentiation (Fig. 1I and Table 1). We also investigated the correlation between PABPC1 expression and survival time in ESCC patients, with a mean follow-up period of 12.5 months after surgery. Patients with high PABPC1 expression (n = 117) had a shorter median survival time (13.46 \(\pm\) 1.12 months) after surgery than those with low PABPC1 expression (n = 73; 17.32 \(\pm\) 1.38 months) (p = 0.043; log-rank test) (Fig. 1J). Similar results were found by analyzing ESCC data from TCGA (Figure S1C). Univariate and multivariate Cox regression analyses revealed that low PABPC1 expression was correlated with improved overall survival of ESCC patients (Figure S1D).

We explored the probable mechanisms of high PABPC1 expression in ESCC by analyzing data from the Encyclopedia of DNA Elements (ENCODE) database (http://​genome.​ucsc.​edu/​) (Figure S2A). To determine whether histone acetylation is involved in human PABPC1 transcriptional regulation, we treated KYSE150 and KYSE520 cells with the histone deacetylase (HDAC) inhibitor NaBu. PABPC1 mRNA levels and H3K27ac enrichment in the promoter region of PABPC1 were remarkably increased in NaBu-treated ESCC cells compared with that in untreated controls (Fig. 2A, B). As shown in Fig. 2C, higher H3K27Ac levels were observed in the PABPC1 promoter region in ESCC tissues (n = 3) and cells (KYSE150 and KYSE520) than in normal tissues and normal NE1 cells. These results suggest that histone acetylation might regulate PABPC1 expression in transcriptional regulation.

Therefore, we hypothesized that certain histone acetylation-modifying enzymes or co-activators might be involved in PABPC1 regulation. As shown in Figs. 2D and S2B, two histone acetyltransferase p300 inhibitors (C646 and A485) decreased, while p300 transfection increased the expression of PABPC1 in a dose-dependent manner. p300 overexpression also increased the levels of H3K27ac in the promoter region of the PABPC1 gene, as identified by chromatin immunoprecipitation (ChIP) (Fig. 2E). These results indicate that p300 might serve as an epigenetic regulator of PABPC1 transcription. We further investigated the transcription factor that combined with p300 to regulate H3K27 acetylation at the PABPC1 promoter. There are two putative Sp1 binding sites close to each other within the PABPC1 promoter region (-226/-217 and -56/-45 bp). Western blot analysis showed that Sp1 increased PABPC1 expression in a dose-dependent manner (Figure S2C). As expected, western blotting revealed strong PABPC1 expression in the p300 and Sp1 co-transfected cells, which was substantially stronger than the expression in cells transfected with the p300 or Sp1 plasmids alone (Fig. 2F). To evaluate whether the two Sp1-binding sites in the PABPC1 promoter region contribute to Sp1-mediated p300 activity, we constructed three reporter plasmids with mutations in the -226/-217 Sp1 site, the -56/-45-Sp1 site, and in both sites. Relative to wildtype promoter, all mutations reduced the transcription activity of p300 and Sp1 on the PABPC1 promoter, and the double-mutated plasmid was associated with substantially decreased activity of p300 and Sp1 (Fig. 2G). Furthermore, TCGA database analysis revealed a positive correlation between p300 and PABPC1 and between Sp1 and PABPC1 expression in ESCC tissues (Fig. 2H). Taken together, these results indicate that the binding of Sp1 with p300 leads to H3K27ac enrichment at the PABPC1 promoter to result in activation of PABPC1 transcription.

We then investigated the biological effects of PABPC1 in ESCC. The efficiency of PABPC1 plasmid and siRNA transfection was confirmed by qRT-PCR and western blot (Fig. 3A). By using CCK-8 assays, we observed that the growth rate of KYSE520 and KYSE150 cells overexpressing PABPC1 was increased compared to that of the negative control, while the growth rates of PABPC1 knockdown cell lines was decreased (Figure S3A). Furthermore, colony numbers were increased in PABPC1-overexpressing cell lines, but decreased in knockdown cell lines (Figure S3B). To further explore the effect of PABPC1 on apoptosis, western blot and flow cytometry were performed in ESCC cells. The results showed a significant decrease in cleaved PARP and caspase 9 expression, and in the percentage of apoptotic cells when cells overexpressed PABPC1 (Figure S3C, D). Next, transwell and wound healing assays results showed that PABPC1 overexpression promoted ESCC cell invasion and migration, whereas PABPC1 knockdown inhibited these functions (Fig. 3B, C). We further investigated whether PABPC1 could promote tumor angiogenesis. We observed that PABPC1-overexpressing ESCC cells causeda substantial increase in the tube formation of HUVECs when compared with control cells, while knockdown PABPC1 decreased the tube formation (Fig. 3D). Nude mice were subcutaneously injected with KYSE150 cells stably overexpressing PABPC1, and tumor growth was larger compared with control cells. As shown in Fig. 3E, tumors derived from the PABPC1 groups were larger and weighed more than those in the KYSE150 control groups. Moreover, the KYSE150-PABPC1 groups had higher expression of Ki-67 and CD34, and decreased levels of cleaved caspase-3 compared with the control group (Fig. 3F). In vivo studies showed that ectopic overexpression of PABPC1 largely inhibited ESCC cell migration (Fig. 3G). Taken together; these data suggest that PABPC1 increases ESCC cell proliferation, invasion, and migration in vitro, and tumor formation in vivo.

We performed RNA-sequencing (RNA-seq) to explore the downstream genes potentially involved in PABPC1-induced ESCC proliferation and invasion. We identified 440 differentially expressed genes (DEGs) that were upregulated and 359 that were downregulated between the control and PABPC1 overexpressing groups. Selected DEGs from the RNA-seq results were confirmed by qRT-PCR (Figure S4A). Analysis of DEGs via Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) indicated that PABPC1 regulates the interferon (IFN) signaling pathway (Fig. 4A). We detected the expression of key mediators of IFN pathways, and found that PABPC1 increased STAT3, ERK, and NF-kB phosphorylation, while it decreased STAT1 phosphorylation (Fig. 4B). Upregulation of IFI27, a key regulator in the IFN-α signaling pathway, was among the most significantly altered DEGs upregulated by PABPC1 (Fig. 4C). The mRNA and protein expression level of IFI27 was also regulated by PABPC1, as detected by qRT-PCR, western blotting, and immunofluorescence (Fig. 4D-F). Then, we sought to determine whether PABPC1 promotes ESCC tumorigenesis via IFI27. We knocked down IFI27 in PABPC1-overexpressing cells, or knocked down PABPC1 in IFI27-overexpressing KYSE150 cells, and detected cell apoptosis, proliferation, invasion, migration and angiogenesis (Figs. 4G, H and S4B-F). IFI27 knockdown dampened PABPC1-mediated increases in STAT3, ERK, and NF-kB phosphorylation (Figure S4G). To further evaluate the effects of PABPC1 and IFI27 on tumor growth, we established xenograft mouse models of ESCC (6 mice per group). The size and weight of xenograft tumors expressing PABPC1 were significantly larger than those of the negative control (NC), and were further reduced when PABPC1 expression was combined with shIFI27 transfection (Fig. 4I). Furthermore, the most significant reduction in tumor size and weight was observed with shIFI27 alone. Tumors with high PABPC1 expression had markedly increased Ki-67 levels but low caspase 3 (Figure S4H). The increase in Ki-67 was attenuated in the xenografts treated with shIFI27. Taken together, these data suggest that PABPC1-induced tumor progression is associated with IFI27 expression.

To evaluate the clinical relevance of our findings in vivo and in vitro, we explored IFI27 expression in public databases and our ESCC patient tissue cohort. IFI27 expression was higher in cancer tissues than in normal tissues in most cancer types in GEPIA database (Figure S5A). IFI27 expression was significantly higher in ESCC tissues compared to paired non-tumor tissues in the GSE20347 (n = 34) and GSE23400 cohorts (n = 106), as well as in our ESCC cohort detected by western blotting (n = 24) and immunohistochemistry (n = 190) (Figure S5B-D). Survival analysis revealed that patients in the high IFI27 group had shorter overall survival than those in the low IFI27 group in both our cohort and TCGA (Figure S5E). We then used IHC staining of 190 ESCC tissues to explore whether a clinical correlation exists between PABPC1 and IFI27 expression. We found that the IFI27 immunohistochemistry score was positively correlated with PABPC1, with approximately 78.3% of tumor tissues (83/106) with high IFI27 expression showing high PABPC1 staining, and 59.5% (50/84) of those with low IFI27 expression displaying low PABPC1 staining (Fig. 4J). Furthermore, survival analysis illustrated that high expression of both PABPC1/IFI27 predicted the poorest overall survival, while patients with low PABPC1/IFI27 expression had best overall survival (p < 0.01) (Fig. 4K).

We next sought to gain mechanistic insight into how PABPC1 affects IFI27 expression. First, PABPC1 did not affect IFI27 protein stability, gene transcription activity, poly (A) tail length, or mRNA synthesis, as detected by luciferase reporter gene assay, poly(A) tail length, and nascent RNA capture assays (Figure S6A-D). Since PABPC1 mediated mRNA stability, we hypothesized that PABPC1 increases IFI27 mRNA expression by increasing its stability. We treated ESCC cells with the RNA synthesis inhibitor actinomycin D (5 μg/mL) and found that the half-lives of IFI27 mRNA were significantly shorter in PABPC1-depleted cells compared to in control cells. Meanwhile, PABPC1 expression could increase IFI27 mRNA stability compared to in control cells (Fig. 5A and S6E). To exclude the possibility that actinomycin D itself could be affecting IFI27 mRNA stability, we also measured RNA stability using the 5-ethynyluridine (EU) labeling (0.1 mM) and release method. We consistently observed a significant decrease in 5-EU-labeled IFI27 mRNA after PABPC knockdown (Fig. 5B). These results indicate that PABPC1 regulates IFI27 expression at the posttranscriptional level.

We confirmed the binding of IFI27 mRNA and PABPC1 by biotin-RNA pull-down assays (Figures S6F). IFI27 mRNA enrichment was significantly increased in the PABPC1 group compared to the control group by RIP assay (Figs. 5C). To detect the possible RNA binding domain via which PABPC1 interacts with IFI27 mRNA, we established five Flag-tagged plasmids containing N-terminal RNA recognition motifs 1–4 (RRMs) and the C-terminal MLLE of PABPC1. As shown in Fig. 5D, the RRM1 domain had strong binding affinity for IFI27 mRNA, whereas the RRM2 domain had weak binding affinity for IFI27 mRNA. PABPC1 interacts with eukaryotic initiation factor 4G (eIF4G) to trigger mRNA circularization and protect mRNA from degradation [20]. To determine whether eIF4G is involved in PABPC1-mediated IFI27 mRNA degradation, we induced eIF4G knockdown and found that the half-life of IFI27 RNA was significantly reduced compared to control (Fig. 5E). Then, we confirmed the colocalization of eIF4G and PABPC1 by immunofluorescence (Fig. 5F). The binding between eIF4G and PABPC1, as well as between IFI27 mRNA and PABPC1, was weaker in response to eIF4G knockdown (Figure S6G-K). To confirm the function of the PABPC1-eIF4G-RNA ternary complex in mRNA decay, we introduced two-point mutations into PABPC1 (PABPC1^M161A/D165K) that abolish binding to eIF4G (Fig. 5G). Notably, PABPC1^M161A/D165K transfection no longer suppressed mRNA degradation, and IFI27 mRNA levels remained unchanged in response to eIF4G knockdown (Fig. 5H). To further confirm our findings, we constructed a PABPC1 plasmid with an RRM1 domain deletion, ΔRRM1 PABPC1. As shown in Figure S6L and M, ΔRRM1 PABPC1 abolished PABPC1 binding to eIF4G, and did not affect the mRNA expression and half-life of IFI27. Interestingly, we found that IFI27 knockdown decreased the binding of PABPC1 and eIF4G (Figure S6M). These results indicate that the interaction between PABPC1 and eIF4G is critical for protecting IFI27 mRNA from degradation.

EXOSC2 (exosome component 2), an important catalytic component of the RNA exosome complex, plays a vital role in regulating RNA degradation [21]. As expected, the decreased stability of IFI27 mRNA could also be rescued by EXOSC2 knockdown (Figure S7A-B). Using a RIP assay, we confirmed that EXOSC2 interacts with IFI27 RNA, and EXOSC2 knockdown increased the binding of PABPC1 with IFI27 RNA (Fig. 5I). This is similar to the manner in which PABPC1 knockdown increased the binding affinity of EXOSC2 with IFI27 mRNA (Figures S7C). In addition, EXOSC2 knockdown reversed the decrease in IFI27 expression levels and RNA half-life caused by PABPC1 knockdown (Fig. 5J-L). Similar results were found for EXOSC4, a core component of the RNA exosome (Figure S7D-F). These findings demonstrate that PABPC1 competed with the RNA exosome to prevent IFI27 mRNA degradation.

We then asked how PABPC1/IFI27 regulated angiogenesis in ESCC. We did not observe any significant changes in the expression levels of certain important angiogenesis-associated factors, including VEGFA, EGF, PDGFA, ANG1, and ANG2, after PABPC1 or IFI27 transfection (Figure S8A). Recently, it has been suggested that exosomes released from cancer cells can act as key factors to affect tumor angiogenesis and metastasis [22]. Thus, we isolated exosomes from ESCC cell culture medium and confirmed exosomes by western blotting, electron microscopy and nanoparticle tracking analysis (Fig. 6A). We further investigated whether PABPC1 could promote tumor angiogenesis through exosomes. We observed that exosomes derived from PABPC1-overexpressing ESCC cells conferred a substantial increase in the migration and tube formation of HUVECs when compared with exosomes from control cells, while exosomes from PABPC1-overexpressing ESCC cells, treated with the exosome inhibitor SW4869, largely restrained the migration and tube formation of HUVECs (Fig. 6B). In addition, our immunofluorescence assay demonstrated that PHK-67-stained exosomes were directly taken up by HUVECs in vitro (Fig. 6C). Collectively, these results suggest that PABPC1 exerts an angiogenic effect via ESCC cell exosomes.

Exosomal miRNA plays a key role in tumor angiogenesis [22]. Therefore, we used miRNA microarray analysis to explore the potential of exosomal miRNA in PABPC1-induced angiogenesis and identified miR-21-5p as the top candidate upregulated in exosomes from PABPC1-overexpressing cells, following comparison of exosomes from PABPC1-transfected cells and control cells (Fig. 6D and S8B, C). We initially confirmed that PABPC1 and IFI27 could mediate the expression of miR-21-5p in ESCC cells (Fig. 6E), and then we extracted exosomes from ESCC cells transfected with Cy3-labeled miR-21-5p mimics, and added the exosomes to HUVECs. Exosomal miR-21-5p could be taken by HUVECs and promote tube formation of HUVECs, indicating that ESCC cell-derived exosomal miR-21-5p could induce angiogenesis (Fig. 6F and S8D). Further investigation revealed that knockdown of PABPC1 decreased the level of miR-21-5p in ESCC cell-derived exosomes, as detected by qPCR (Fig. 6G). Additionally, knockdown of PABPC1 inhibited the process of exosomal miR-21-5p transfer from ESCC cells to HUVEC (Fig. 6H). To explore how PABPC1 mediated exosomal miR-21-5p, we analyzed the specific interaction between the miR-21-5p sequence and PABPC1 motifs (ACUGAUG) by using the database of RBP specificities (RBPDB, http://​rbpdb.​ccbr.​utoronto.​ca/​; threshold 0.6). MiRNA pull-down assays revealed that there was significant binding between miR-21-5p and PABPC1 in the cytoplasm and in exosomes. However, this binding could be impaired by mutating the binding sequence of miR-21-5p (Fig. 6I). Thus, we speculated that PABPC1 may mediate the process of miR-21-5p packaging into exosomes. RIP assays further demonstrated that miR-21-5p was more enriched in the PABPC1 antibody group than in the IgG group, both in ESCC cells and their exosomal lysates (Fig. 6J). Therefore, we demonstrated that PABPC1 could stimulate angiogenesis by mediating miR-21-5p expression and promoting exosomal miR-21-5p packaging for release and targeting of endothelial cells.

To analyze the mechanisms of regulated exosomes in controlling tumor angiogenesis, we performed RNA-seq to identify DEG profile and signaling pathways in HUVECs incubated with exosomes derived from PABPC1-overexpressing or control KYSE150 cells. KEGG and GO analyses of genes potentially regulated by exosomes revealed significant alternations in the cytokine-cytokine receptor signaling pathway (Fig. 7A). The top DEGs are shown in Fig. 7B. Numerous studies have shown that chemokine (C-X-C motif) ligand 10 (CXCL10), a target of miR-21-5p, is a key mediator in tumor angiostasis [23]. Western blotting and ELISA for CXCL10 showed a significant decrease in the expression of CXCL10 in HUVECs treated with exosomes derived from PABPC1-overexpressing cells (Fig. 7C, D). Bioinformatics predicted that CXCL10 was a putative direct miR-21-5p target. Using 3′ UTR luciferase reporter assays, we found that miR-21-5p overexpression inhibited luciferase activity in KYSE150 cells expressing wild-type CXCL10 3′ UTR reporters, whereas mutated miR-21-5p specifically abolished this suppression (Fig. 7E). Real-time PCR analysis revealed that a miR-21-5p mimic also inhibited CXCL10 expression in HUVEC cells, whereas a miR-21-5p inhibitor exhibited the opposite effect (Fig. 7F). Furthermore, the tube formation and cell migration of HUVECs treated with exosomal-miR-21-5p could be successfully halted by overexpressing CXCL10 (Fig. 7G).