Biosynthetic CircRNA_001160 induced by PTBP1 regulates the permeability of BTB via the CircRNA_001160/miR-195-5p/ETV1 axis

The presence of the blood-tumor barrier (BTB) severely impedes the transport of anti-neoplasm drugs to the central nervous system, affecting the therapeutic effects of glioma. Glioma endothelial cells (GECs) are the main structural basis of the BTB. Circular RNA is considered to be an important regulator of endothelial cell growth. In this study, we found that polypyrimidine tract binding protein 1 (PTBP1) and circRNA_001160 were remarkably upregulated in GECs. Knockdown of PTBP1 or circRNA_001160 significantly increased BTB permeability, respectively. As a molecular sponge of miR-195-5p, circRNA_001160 attenuated its negative regulation of the target gene ETV1 by adsorbing miR-195-5p. In addition, ETV1 was overexpression in GECs. ETV1 bounded to the promoter regions of tight junction-related proteins and increased the promoter activities, which significantly promoted the expression levels of tight junction-related proteins. The present study showed that the combined application of PTBP1, circRNA_001160, and miR-195-5p with the anti-tumor drug Dox effectively promoted Dox through BTB and extremely induced the apoptosis of glioma cells. Our results demonstrated that the PTBP1/circRNA_001160/miR-195-5p/ETV1 axis was critical in the regulation of BTB permeability and provided new targets for the treatment of glioma.


Introduction
Glioblastoma (GBM) is the most common primary malignant brain tumor in the adult central nervous system with its highly invasiveness and diffusivity 1 . Despite the use of most conventional therapies such as surgical resection and chemoradiotherapy, patients with glioma still gain poor postoperative outcomes, high recurrence rates, and low survival rates 2 . The blood-tumor barrier (BBB) is composed of normal capillary endothelial cells (ECs), preventing the entry of harmful substances, and is an indispensable vascular barrier to protect the brain 3,4 .
Glioma leads to the BBB structure remodeling, forming the blood-tumor barrier (BTB) 5 . The BTB constitutes major obstacles to the transport of therapeutics in brain tumors 6 . Selective opening of the BTB to increase antineoplasm drugs concentration in tumor tissues is crucial for the treatment of malignant glioma.
In addition to the lipid-soluble small molecules with a molecular mass under a 400-600-Da threshold, BTB and BBB limit most compounds, making potentially powerful anti-neoplasm drugs unusable for GBM 7 . Microbubbleenhanced diagnostic ultrasound effectively increases the permeability of BTB in glioma by opening KCa channels 8 . In preclinical studies, it was confirmed that inhibition of BMX kinase by ibrutinib in vivo strongly destroys tumor pericytes, and selectively damages BTB instead of BBB, thereby increasing drug effusion into established tumors and enhancing the chemotherapeutic efficacy of drugs with poor BTB penetration 3 . In addition, in targeted therapy of molecules, researches show that, low doses of EMAP-II induces high permeability of BTB through the cAMP/PKA/Rac1 signaling pathway 9 . Single or combination therapy with KHDRBS3, cDENND4C, and miR-577 is effective in inducing apoptosis of glioma cells by promoting the anti-neoplasm drug doxorubicin across BTB 10 .
RNA-binding proteins (RBPs) are involved in multiple steps of post-transcriptional regulation of cellular biological process, including pre-mRNA splicing, RNA editing, and polyadenylation 11 . Polypyrimidine tract binding protein 1 (PTBP1) is located at 19p13.3 and is one of the most widely studied splicing regulatory factors in the hnRNP family 12 . PTBP1 is a key protein that regulates precursor mRNA splicing and alternative splicing events 13 . PTBP1 promotes the expression conversion of two pyruvate kinases PKM1 and PKM2, and enhances the aerobic glycolysis of cells, thereby promoting the proliferation of glioma cells 14 . In addition, PTBP1 shows high expression level and participates in the malignant biological behaviors of bladder cancer 15 , colon cancer 16 , and breast cancer 17 cells. The study of PTBP1 on the regulation of BTB permeability has not been reported yet.
Circular RNAs (circRNAs) consist of a covalently closed loop without a 5'-terminal cap and 3'-terminal poly A tail, which are more stable than long non-coding RNAs 18 . Cir-cRNAs have a variety of biological functions. For example, circRNAs act as molecular sponges for microRNAs; 19 posttranscriptional regulation of gene expression; 20 and encode polypeptides involved in protein translation 21 . Cir-cRNA_001160 (hsa_circ_0001417, circ-ANKRD17) is located on chromosome 14 and has a full length of 1832 bp. CircRNA_001160 is derived from the cyclization of linear RNA ANKRD17 (ankyrin repeat domain 17). ANKRD17 plays a vital role in liver and cardiovascular development as well as cell cycle regulation 22 . In addition, ANKRD17 is critical for vascular integrity during embryonic development 23 . At present, there is no report on the expression and function of circRNA_001160 in GECs.
MicroRNAs (miRNAs) are highly conserved single stranded non-coding RNAs 24 . MiRNAs regulate target gene expression by inducing mRNA degradation or blocking protein translation 25 . The regulation of miRNAs mediated BBB damage therapy can improve vascular cognitive impairment 26 . Numerous studies demonstrated that miRNAs could regulate BTB permeability 27,28 . Overexpression of miR-195-5p could reduce angiogenesis of ovarian cancer 29 . In addition, miR-195-5p is downregulated in glioma tissues, and upregulated miR-195-5p suppresses the malignant biological behaviors of glioma cells 30 . Bioinformatics software starBase v2.0 was used to predict that miR-195-5p binding sites existed on cir-cRNA_001160. However, the expression of miR-195-5p in GECs and the underlying molecular mechanism of cir-cRNA_001160/miR-195-5p are unclear.
Ets variant gene 1 (ETV1), also known as ER81, is a member of the E-26 (ETS) transcription factor family 31 . ETS factors are transcription factors that bind to promoters and enhancers to recruit other transcriptional components 32 . ETV1 has been shown to regulate the biological processes of tumor cells as an oncogene in Ewing's sarcoma 33 , melanoma 34 , and triple-negative breast cancer 35 .
This study established an in vitro BTB model. Firstly, the endogenous expressions of PTBP1, circRNA_001160, miR-195-5p and ETV1 in astrocyte endothelial cells (AECs) and glioma endothelial cells (GECs) were detected. Furthermore, the above-mentioned molecular interaction mode and the mechanisms of regulating BTB permeability were clarified. This study aimed to demonstrate the role of the PTBP1/circRNA_001160/miR-195-5p/ETV1 pathway in regulating BTB permeability, which can provide a new strategy for the comprehensive treatment for glioma.

Establishment of in vitro BTB and BBB model
In vitro BTB and BBB models were established by coculture of EC and U87 cells. For the BTB model in vitro, the U87 cells were inoculated in six-well plates at 2 × 10 4 cells/well, and culture medium was developed for 2 days. After culturing U87 cells for 2 days, ECs were inoculated in the upper compartment at 2 × 10 5 cells/well on the upper side of inserts pre-coated with Cultrex Rat Collagen I (R&D Systems, Minneapolis, MN, USA). EC and U87 cells were cultured with prepared EBM-2 medium, and the medium was updated every 2 days. A GEC model EC phenotype obtained in vivo gliomas was obtained after 4 days of co-culture. For the BBB model in vitro, ECs were co-cultured with NHA using the same method described above. An AEC model EC phenotype obtained in vivo gliomas was obtained after 4 days of co-culture.

Quantitative real-time PCR (qRT-PCR) assays
Total RNAs were extracted from AECs and GECs with Trizol reagent (Life Technologies, Carlsbad, CA, USA), according to the manufacturer's description. RNA concentration and quality were measured by the 260/280 nm ratio using Nanodrop Spectrophotometer (ND-100; Thermo Fisher Scientific, Waltham, MA, USA). One Step SYBR Prime Script RT-PCR Kit (Takara Biomedical Technology, Dalian, China) was used to identify the expression levels of PTBP1 (NM_002819.5), linear ANKRD17 (NM_032217.5), ETV1 (NM_004956.5). One Step Prime-Script™ RT-PCR Kit (Takara) was used to identify the expression level of circRNA_001160 (NM_198889). In addition, RNase R was used to confirm the relative expression of circRNA _001160 and eliminate the effect of linear ANKRD17. GAPDH was used as endogenous control. The design of divergent primer and convergent primer was preformed by Geneseed Biotechnology company (Guangzhou). The expression level of miR-195-5p (NR_029712.1) was detected by TaqMan MicroRNA Reverse Transcription kit and TaqMan Universal Master Mix II (Applied Biosystems). U6 was used as endogenous control. All qRT-PCR reactions were performed using the 7500 Fast RT-PCR System (Applied Biosystems, USA). The relative quantification 2 −ΔΔCt method was applied to calculate the expressions of genes. The primers and probes used in this study are shown in Supplementary table S1.

Cell transfection
Short-hairpin RNAs (shRNAs) directed against human PTBP1, circRNA_001160 and ETV1 gene were constructed in pGPU6/GFP/Neo vector (Gene Pharma, Shanghai, China) to generate silencing plasmids. The human ETV1 gene coding sequence was ligated into the pIRES2-EGFP vector (GeneChem, Shanghai, China) to construct the ETV1 overexpression plasmid. The pGPU6/GFP/Neo and pIRES2-EGFP empty vectors without targeting sequences were used as negative controls (NCs). ECs were cultured into 24-well plates and transfected with plasmids by Opti-MEMI, Lipofectamine LTX and Plus Reagents (Life Technologies) when they reached about 80% confluence. Stable transfected cells were selected using G418 and puromycin (Sigma-Aldrich, St. Louis, MO, USA). G418-resistant (or purinomycin-resistant) cell clones were established at about 4 weeks. For co-transfection of sh-PTBP1 and sh-cir-cRNA_001160, cells with stably knocked down PTBP1 were transfected with sh-circRNA_001160. G418 and purinomycin dual-resistant clone were selected. Furthermore, agomir miR-195-5p (pre-miR-195-5p), antagomir miR-195-5p (anti-miR-195-5p), and their respective NC (Gene Pharma, Shanghai, China) were transfected into ECs using Lipofectamine 3000 Reagents (Life Technologies). Then qRT-PCR and western blot were performed to measure transfection efficiency. For co-transfection of sh-circRNA_001160 and pre/anti-miR-195-5p or overexpressed ETV1 (ETV1) and pre-miR-195-5p, cells that stably sh-circRNA_001160, ETV1, or their respective NC were transiently transfected with pre/anti-195-5p or its NC. The sequences of all sh-RNA templates are shown in Supplementary Table S2. To avoid shRNA-mediated off-target effects, we transfected silencing plasmids at multiple sites and tested transfection efficiency. Then, the shRNA with the best silencing efficiency was selected for functional study. The transfection efficiencies are shown in Additional file 1: Fig. S2.

Transendothelial electric resistance (TEER) assays
In order to measure the integrity and permeability of BTB, TEER assay was determined by the Millicell-ERS apparatus (Millipore, Billerica, MA, USA) after the successful establishment of an in vitro BTB model. Before each measurement, the TEER assay was recorded at room temperature for 30 min and the culture medium was updated to ensure temperature balance and homogeneity of the culture environment. Through the measurement of resistance barrier property subtracting background, and then multiplied by the effective surface area of membrane to calculate the final resistance (Ω cm 2 ).

Horseradish peroxidase flux (HRP) assays
After the establishment of a normal BBB model for in vitro glioma regulation, 1 mL medium containing 10 mg/mL HRP (0.5 mm, Sigma-Aldrich) was added to the upper system, and 2 mL medium was added to the hole. Then, after incubating for 1 h at 37°C, 5 μL of medium was collected from the lower compartment. The samples using tetramethyl-benzidine to colorimetric analysis. The absorbance was measured at 370 nm with a spectrophotometer. The final HRP flux was expressed as picomole passed per square centimeter surface area per hour (pmol/cm 2 /h).

Western blot assays
The total proteins were lysed on ice with RAPI buffer (Beyotime Institute of Biotechnology) and protease inhibitors (10 mg/mL aprotinin, 10 mg/mL phenyl-methylsulfonyl fluoride [PMSF], and 50 mM sodium orthovanadate) for half an hour and then centrifuged at 17,000g for 45 min at 4°C. The amount of the sample was measured using a BCA protein assay kit (Beyotime Institute of Biotechnology, Jiangsu, China). Proteins were transferred to polyvinylidene fluoride (PVDF) membrane by SDS-PAGE electrophoresis. After SDS-PAGE electrophoresis and transfer, the PVDF membrane was blocked with skim milk powder for 2 h at room temperature. Then membranes were incubated with the primary antibodies as follows: PTBP1 (1:1,000; Proteintech, USA), ETV1 (1:1,000, Affinity Biosciences, USA), GAPDH (1:10,000; Proteintech, USA), ZO-1 (1:300; Life Technologies, Frederick, MD, USA), occludin (1:1,000; Proteintech, USA), and claudin-5 (1:300; Life Technologies, Frederick, MD, USA) at 4°C overnight. The membrane was incubated for 2 h at room temperature with HRP-conjugated secondary antibody. These protein blots were observed with enhanced chemiluminescence kit (ECL) (Santa Cruz Biotechnology, Dallas, TX) and detected by the ECL assay system (Thermo Scientific, Beijing, China). The protein strips were then scanned using Chemi Imager 5500 V2.03 software, and the integrated light density values (IDVs) was calculated by Fluor Chen 2.0 software, with GAPDH as an internal control.

Northern blotting
RNAs were isolated from AECs and GECs samples using Trizol reagent (Life Technologies). CircRNA_001160 for RNA imprinting was implemented using a Biotin RNA marker mixture (Roche). RNA samples were isolated by electrophoresis and transferred to NC membranes, which were then incubated with a hydrated buffer containing probes. Finally, the chemiluminescent nucleic acid detection module (Thermo Scientific) was used to detect RNA signals. The measurement was repeated three times.
CircRNA, miRNA, and transcription factor microarrays CircRNA, miRNA, and transcription factor microarray analyses, sample preparation and microarray hybridization were performed by Kangchen Biotechnology Corporation (Shanghai, China).

Native and nascent RNA immunoprecipitation (RIP) assays
Whole cell lysates from the control group and the anti-miR-195-5p group were incubated with RIP immunoprecipitation buffer containing magnetic beads conjugated with human anti-argonaute2 (Ago2) antibody (Millipore, Billerica, MA, USA) and NC normal mouse IgG (Millipore). The samples were incubated with the protease K buffer and the immunoprecipitated RNA was isolated. In addition, purified RNA was collected and analyzed by qRT-PCR to verify the presence of binding targets.
The wild-type and mutant plasmids were performed by Gene Pharma. The plasmids used with Lipofectamine 3000 were co-transfected into ECs. For nascent RIP, cells were incubated with 100 μM DRB (Sigma, D1916) for 3 h to block Pol II transcription. Transcription was restarted after removal of the DRB and the newly transcribed RNA were labeled with 200 μM 4sU for 30 min (Sigma, T4509) as described above. Native RIP was performed with antihuman PTBP1 (Santa Cruz, USA) and NC normal mouse IgG (Millipore), and the 4su-labeled RNA was purified with RIP products, and the primers for qRT-PCR are performed listed in table S1. The sequences of wild-type and mutant plasmids are listed in Supplementary table S4.

Chromatin immunoprecipitation (ChIP) assays
The ChIP assays were performed using the Simple Chip Enzymatic Chromatin IP kit (Cell Signaling Technology, Danvers, MA, USA) according to the manufacturer's instructions. In brief, GECs were cross-linked with EBM-2 containing 1% formaldehyde for 10 min, then glycine was added and incubated at room temperature for 5 min to quench the cross-linking. The cells were collected in a lysis buffer containing PMSF. In addition, the use of micrococcus nuclease digested chromatin at temperature of 37°C for 20 min and frequently mixed. Immunoprecipitation was incubated with 2 μg anti-ETV1 antibody (Abcam, USA). ProteinG agarose beads were used for immunoprecipitation in each sample and incubated overnight, and they were gently shaken at 4°C while 2% lysates were used as input reference. Deliver DNA ChIP connected with 5 mol/L NaCl and proteinase K reversed 2 h under 65°C, and then purified DNA. In each PCR reaction, the corresponding inputs were obtained in parallel for PCR validation. PCR products were isolated from 3% agarose gel. Primers for ChIP PCR are shown in Supplementary table S3.

Apoptosis analysis
In vitro BTB models were established using transfected sh-PTBP1 cells, sh-circRNA_001160 cells, pre-miR-195-5p cells, sh-PTBP1 + sh-circRNA_001160 cells, sh-circRNA_001160 + pre-miR-195-5p cells. After establishing the in vitro BTB model, 10 μM of Doxorubicin (Beyotime Institute of Biotechnology, Jiangsu, China) was added to the transwell chamber. The normal control group (control) did not join the Dox processing. After 12 h in each group, the U87 cells in the lower chamber were digested to prepare a cell suspension, and the supernatant was discarded after centrifugation. Cell apoptosis was detected by staining with Annexin V-PE/ 7AAD (Southern Biotech, Birmingham, AL, USA) according to the manufacturer's instructions. Add 500 μl of Binding Buffer suspension cells per tube, add 5 μl PI and 5 μl of FITC were mixed, and reacted at room temperature for 15 min protecting from light. Flow cytometry (FACScan, BD Biosciences) was used to detect apoptosis.

Statistical analysis
All statistical analyses were carried out using the GraphPad Prism v5.01 (GraphPad, La Jolla, CA). All data were expressed as the mean ± SD. Significant statistical differences between the two groups were determined by Student's t test (two-tailed). For three or more groups, one-way ANOVA was used for statistical analysis, followed by post-Dunnett testing. If P < 0.05, the difference was considered statistically significant.

PTBP1 was upregulated in GECs, and knockdown of PTBP1 increased BTB permeability
QRT-PCR and western blot were used to detect the expression of PTBP1. As shown in Fig. 1a, b, PTBP1 was markedly upregulated in GECs. To further validate the effect of PTBP1 on BTB permeability, we transfected ECs with knockdown of PTBP1. In order to assess the integrity and permeability of BTB, we measured the TEER value and HRP flux. There was no significant difference between the knockdown of PTBP1 negative control (sh-NC) group and the control group. Compared with the sh-NC group, the TEER value was evidently decreased and the HRP flux was obviously increased in the knockdown of PTBP1 (sh-PTBP1) group (Fig. 1c, d). Meanwhile, the effects of PTBP1 knockdown on the expression levels of ZO-1, occludin and claudin-5 were shown in Fig. 1e. Compared with the sh-NC group, the expression levels of ZO-1, occludin and claudin-5 significantly decreased in the sh-PTBP1 group.
Furthermore, IF assays confirmed that knockdown of PTBP1 reduced the expression levels of ZO-1, occludin and claudin-5, with discontinuous distribution on the boundaries of GECs (Fig. 1f).
CircRNA_001160, but not linear ANKRD17, was upregulated in GECs and regulated the expression levels of tight junction-related proteins By comparing the genomic DNA sequence of ANKRD17 and circRNA_001160, the composition of circRNA_001160 was shown in Fig. 2a. The sanger sequencing experiment was used to validate the sequence on the junction sites of circRNA_001160. The junction sites of circRNA_001160 are GCATCAAATG and GGGTGGCACT. Divergent primer detected the presence of circRNA_001160 in cDNA and no circRNA_001160 was detected in the genomic DNA (gDNA) (Fig. 2b). RNase R, an RNA exonuclease that degrades linear RNA but does not degrade the circular form, was used to confirm circular RNA. As shown in Fig. 2c, d, cir-cRNA_001160 was resistant to RNase R treatment, whereas linear ANKRD17 treating with RNase R was signifificantly reduced. Notably, there was no significant difference in the expression of linear ANKRD17 (cir-cRNA_001160 parental gene) between the AECs and GECs group. To investigate the expression and location of circRNA_001160, FISH and northern blot were used to detect the expression of circRNA_001160. Results confirmed that circRNA_001160 was located in the cytoplasm and was conspicuously elevated in GECs (Fig. 2e, f). We transfected two circRNA_001160 junction shRNAs in order to determine the effect of circRNA_001160 on the permeability of BTB. These plasmids of shRNAs could specifically knockdown the circRNA_001160 expression level, without affecting the expression of linear ANKRD17 (Fig. S2b, Fig. 2g). After knockdown of circRNA_001160, the TEER value was evidently decreased and the HRP flux was obviously increased (Fig. 2h, i). Knockdown of cir-cRNA_001160 reduced the expression levels of ZO-1, occludin, and claudin-5 (Fig. 2j). IF assays also confirmed that knockdown of circRNA_001160 reduced the expression levels of tight junction-related proteins, and destoryed the continuous distribution of tight junctionrelated proteins in GECs (Fig. 2k). e Effects of PTBP1 knockdown on ZO-1, occludin, and claudin-5 expression levels determined by western blot. Data represent mean ± SD (n = 3, each). *P < 0.05, **P < 0.01 versus sh-NC group. f Effects of PTBP1 knockdown on ZO-1, occludin, and claudin-5 expression levels and distribution determined by immunofluorescence staining (n = 3, each). ZO-1, occludin, and claudin-5 (green) were labeled with secondary antibody against anti-ZO-1, anti-occludin, and anti-claudin-5 antibody, respectively, and nuclei (blue) were labeled with DAPI. Scale bar represents 20 μm.

PTBP1 promoted the expression of circRNA_001160
In order to further clarify the mechanism by which PTBP1 regulated BTB permeability, after stable knockdown of PTBP1, we performed a circRNA microarray to assess the expression levels of circRNAs in GECs. As shown in Fig. 3a, circRNA_0008035, circRNA_0034642, circRNA_001160, and circRNA_001569 were the four circRNAs with the most differential expression. QRT-PCR analysis also showed a significant decrease in the expression levels of the four circRNAs (Fig. 3b). In addition, there was no significant change in the expression of linear ANKRD17 between the sh-NC and sh-PTBP group (Fig. 3c). Importantly, the result of Nascent RIP assay was shown in Fig. 3d. Compared with the IgG co-precipitation group, the enrichment of ANKRD17 pre-mRNA in the PTBP1 co-precipitation group was significantly increased. In the PTBP1 co-precipitation group, after the binding sites of PTBP1 and ANKRD17 pre-mRNA were mutated, we found a reduction in the enriched ANKRD17 pre-mRNA. This result further revealed the interaction between PTBP1 and the pre-mRNA of the ANKRD17 gene which formed circRNA_001160. Those results indicated that PTBP1 promoted the expression of cir-cRNA_001160, but had no effect on the expression of linear ANKRD17. To further expore the function of PTBP1 and circRNA_001160, ECs stably expressing sh-PTBP1 and sh-circRNA_001160 were established. Double knockdown of PTBP1 and circRNA_001160 could significantly increase BTB permeability (Fig. 3e-g). These results suggested that PTBP1 regulated BTB permeability by promoting the expression of circRNA_001160.

MiR-195-5p was downregulated in GECs and regulated BTB permeability
FISH and qRT-PCR were conducted to reveal the expression and location of miR-195-5p. MiR-195-5p was predominately located in the cytoplasm and was downregulated in GECs (Fig. 4a, b). To further investigate the function of miR-195-5p, plasmids with overexpressed and knockdown of miR-195-5p were transfected into ECs. In the pre-miR-195-5p group, the TEER value was evidently decreased, and the HRP flux was obviously increased (Fig.  4c, d). Western blot and IF assays results demonstrated that the expression levels of tight junction-related proteins were significantly downregulated in pre-miR-195-5p group (Fig. 4e, f). The above results indicated that overregulated of miR-195-5p increased BTB permeability.

CircRNA_001160 negatively regulated and interacted with miR-195-5p
MiRNA microarray analysis was used to screen the circRNA_001160-related miRNAs. We found that miR-195-5p had the highest relative expression after downregulation of circRNA_001160 (Fig. S3a). We applied dual luciferase reporter assay to confirm the existence of targeted binding sites between circRNA_001160 and miR-195-5p. As shown in Fig. 5a, luciferase activity was decreased in HEK293T cells co-transfected with pre-miR-195-5p and circRNA_001160-Wt, pre-miR-195-5p and circRNA_001160-Mut1, pre-miR-195-5p and circRNA_001160-Mut2. However, luciferase activity in HEK293T cells co-transfected with pre-miR-195-5p and circRNA_001160-Mut3 was restored to the control level, suggesting that miR-195-5p bound to circRNA_001160 in a sequence specific manner. Furthermore, RIP experimental results showed that circRNA_001160 and miR-195-5p in Ago2 co-precipitation group were significantly increased (Fig. 5b, c), suggesting that circRNA_001160 and miR-195-5p existed simultaneously in RNA-induced silencing complex (RISC). After knockdown of circRNA_001160, the expression of miR-195-5p was significantly increased. Overexpression of miR-195-5p resulted a significant decrease in the expression of circRNA_001160 (Additional file 1: Fig.  S2f, g). These results suggested that circRNA_001160 reduced the expression of miR-195-5p in RISC, and there may be a mutual inhibitory feedback loop between cir-cRNA_001160 and miR-195-5p. Having confirmed that cir-cRNA_001160 bound to miR-195-5p, we further investigated whether miR-195-5p was involved in circRNA_001160mediated BTB permeability. As shown in Fig. 5d-f, silencing of miR-195-5p significantly reversed the increase of BTB permeability caused by circRNA_001160 knockdown.

ETV1 was upregulated in GECs and directly bound to the promoter regions of tight junction-related proteins
QRT-PCR and western blot showed that ETV1 was significantly upregulated in the GECs group (Fig. 6a, b). As shown in Fig. 6c, d, in the ETV1 group, the TEER value was evidently increased, and the HRP flux was obviously decreased. The western blot showed that the expression levels of tight junction-related proteins were significantly increased (Fig. S4b). IF assays showed a high fluorescence intensity and cell boundary enrichment of ZO-1, occludin (see figure on previous page) Fig. 2 Knockdown of circRNA_001160 increased BTB permeability in vitro and reduced the expression levels of tight junction-related proteins. a A schematic representation of how circRNA_001160 arosed from the ANKRD17 gene as determined by scanning ANKRD17 genomic DNA and circBase. Sanger sequencing validated the sequence on the junction sites of circRNA_001160. The red arrow indicated the head-to-tail splicing sites of circRNA_001160. bThe existence of circRNA_001160 in GECs. c The expression of circRNA_001160 in GECs with RNase R treatment. Data represent mean ± SD (n = 5, each group). **P < 0.01 versus control group in AECs; ## P < 0.01 versus RNase R group in AECs. d The expression of linear ANKRD17 in GECs with RNase R treatment. Data represent mean ± SD (n = 5, each group). **P < 0.01 versus control group in AECs; ## P < 0.01 versus control group in GECs. e The expression and location of circRNA_001160 in AECs and GECs were detected by FISH. (red, circRNA_001160; blue, DAPI nuclear staining). Scale bar represents 20 μm. f Relative circRNA_001160 expression in AECs and GECs was detected by northern blotting. g The effect of circRNA_001160 knockdown on linear ANKRD17 expression level was detected by qRT-PCR. Data represent mean ± SD (n = 5, each group). (H and I) Effects of circRNA_001160 knockdown on TEER values (h) and HRP flux (i). j Effects of circRNA_001160 knockdown on ZO-1, occludin, and claudin-5 expression levels determined by western blot. Data represent mean ± SD (n = 3, each). *P < 0.05, **P < 0.01 versus sh-NC group. (k) Effects of circRNA_001160 knockdown on ZO-1, occludin, and claudin-5 expression levels and distribution determined by immunofluorescence staining (n = 3, each). ZO-1, occludin, and claudin-5 (green) were labeled with secondary antibody against anti-ZO-1, anti-occludin, and anti-claudin-5 antibody, respectively, and nuclei (blue) were labeled with DAPI. Scale bar represents 20 μm. and claudin-5 in the ETV1 group (Fig. S4c). The results indicated that ETV1 overexpression inhibited BTB permeability by upregulating the expression levels of tight junction-related proteins.
Using the JASPAR CORE database, a total of eight potential binding sites for ETV1 were found in the promoter regions of tight junction-related proteins. The predicted ETV1 binding sites were 177 bp and 992 bp upstream of the ZO-1 transcription initiation site; 104 bp, 553 bp, and 730 bp upstream of the occludin transcription initiation site; and 160 bp, 334 bp and 751 bp upstream of the claudin-5 transcription initiation site. To test whether ETV1 promoted tight promoter activity in tight junctionrelated proteins in GECs, a series of constructs were RNAs isolated from sh-NC and sh-PTBP1 cells. Red indicates high relative expression and green indicates low relative expression. b Relative expression levels of circRNA_0008035, circRNA_0034642, circRNA_001160, and circRNA_001569 were detected by qRT-PCR. Data represent mean ± SD (n = 5, each group). *P < 0.05, **P < 0.01 versus sh-NC group. c The effect of PTBP1 knockdown on linear ANKRD17 expression level was detected by qRT-PCR. Data represent mean ± SD (n = 5, each group). d The relative enrichment of ANKRD17 pre-mRNA in anti-IgG and anti-PTBP1 was detected by a nascent RIP assay. Data represent mean ± SD (n = 3, each group). **P < 0.01 versus anti-IgG group; ## P < 0.01 versus Bio-Wt1-3 in anti-PTBP1 group. e, f The co-effects of PTBP1 and circRNA_001160 knockdown on TEER values (e) and HRP flux (f). g The co-effects of PTBP1 and circRNA_001160 knockdown on ZO-1, occludin, and claudin-5 expression levels determined by western blot. Data represent mean ± SD (n = 3, each). *P < 0.05, **P < 0.01 versus sh-PTBP1-NC + sh-circRNA_001160-NC group; # P < 0.05 versus sh-PTBP1 + sh-circRNA_001160-NC group; & P < 0.05 versus sh-PTBP1-NC + sh-circRNA_001160 group.
designed and the dual-luciferase reporter assays were performed. The results confirmed that ETV1 activated the promoter region activities of tight junction-related proteins ( Fig. 6e-g).
Finally, CHIP assays were used to detect whether ETV1 bound to the promoter regions of ZO-1, occludin and claudin-5. As shown in Fig. 6h-j, ETV1 connected with the predicted binding sites of ZO-1, occludin and claudin-5, but did not interact with the control region. The all results indicated that the ETV1 response elements were located in the promoter region of ZO-1, occludin and claudin-5, respectively, and promoted the activity of the promoter regions.

MiR-195-5p targeted ETV1 3'-UTR and impaired its expression
First of all, the mRNA and protein expression levels of ETV1 were examined after changing the expression of miR-195-5p. The mRNA and protein expression levels of ETV1 were significantly decreased in the pre-miR-195-5p group (Fig. 7a, b). Subsequently, the dual luciferase reporter assay showed that co-transfected ETV1-Wt and pre-miR-195-5p group had significantly reduced luciferase activity, while the luciferase activity was no statistically significant in co-transfected ETV1-Mut and pre-miR-195-5p group (Fig. 7c). These results confirmed that miR-195-5p regulated the expression of ETV1 by binding to ETV1 3'-UTR, and ETV1 was a direct target of miR-195-5p. To further investigate whether ETV1 was involved in miR-195-5p-mediated BTB permeability, we established ECs with pre-miR-195-5p and ETV1 plasmids. As shown in Fig. 7d-f, ETV1 overexpression attenuated the decrease in TEER value, the increase in HRP flux and the increase in the expression of tight junction-related proteins induced by miR-195-5p overexpression. The results corroborated that ETV1 participated in miR-195-5p-mediated BTB permeability and overexpression of miR-195-5p attenuated ETV1's regulation of BTB permeability by targeting ETV1 3'-UTR.

Discussion
This study found that PTBP1 and circRNA_001160 were highly upregulated in GECs. The respective knockdown of PTBP1 and circRNA_001160 decreased the expression levels of tight junction-related proteins and increased the permeability of the BTB. In GECs, miR-195-5p was downregulated and ETV1 was overexpressed.
Overexpression of miR-195-5p negatively regulated the expression of ETV1, which could reduce the expression levels of tight junction-related proteins. The combination of sh-PTBP1, sh-circRNA_001160, pre-miR-195-5p with Dox significantly induced the apoptosis of U87 cells.
Current researches showed that there are two main ways in which chemotherapeutic drugs can be transported through the BTB, including paracellular pathway and transcellular pathway 36 . The paracellular pathway refers to the transport of substances from blood vessels through the tight junctions between microvascular ECs into brain tissue 37 . Selectively opening tightly linked cell-by-cell pathways is an important way to increase BTB permeability. When the expression levels of tight junction-related proteins are decreased, the BTB permeability is increased 38 . The MiR-195-5p targetd circRNA_001160 and reversed the effects of circRNA_001160 on BTB permeability in vitro. a Dual-luciferase reporter assays were performed to determine the binding sites of circRNA_001160 and miR-195-5p in HEK293T cells. Data represent mean ± SD (n = 5, each). **P < 0.01 versus pre-NC group. b, c MiR-195-5p was identified in circRNA_001160-RISC complex. Relative expression of circRNA_001160 and miR-195-5p were measured using qRT-PCR. Data represent mean ± SD (n = 5, each group). *P < 0.05, **P < 0.01 versus anti-normal IgG group; # P < 0.05, ## P < 0.01 versus anti-Ago2 in control group. d, e The co-effects of circRNA_001160 and miR-195-5p on TEER values (d) and HRP flux (e). f The co-effects of circRNA_001160 and miR-195-5p on ZO-1, occludin, and claudin-5 expression levels determined by western blot. Data represent mean ± SD (n = 3, each). *P < 0.05, **P < 0.01 versus sh-NC + pre-NC group.
TEER value and HRP flux involved in this study are important indicators for evaluating the integrity and permeability of BTB. Downregulation of TEER value and upregulation of HRP flux represent the increasing of BTB permeability 39 .
The abnormal expression of RBPs is closely related to the occurrence of various human diseases. Our study first found that knockdown of PTBP1 significantly increased the permeability of BTB, which proved that PTBP1 had a regulatory effect on vascular ECs. Recently, the regulatory effect of PTBP1 on vascular ECs has attracted researchers' attention. Studies have confirmed that PTBP1 promotes proliferation and migration of vascular ECs by forming complexes with cold-shock domain proteins 40 . In patients Fig. 6 Knockdown of ETV1 increased the BTB permeability and inhibited the expression levels of tight junction-related proteins at the transcriptional level. a Relative ETV1 expression in AECs and GECs determined by qRT-PCR. Data represent mean ± SD (n = 5, each). *P < 0.05 versus AECs group. (b) Relative ETV1 expression in AECs and GECs determined by western blot. Data represent mean ± SD (n = 4, each). **P < 0.01 versus AECs group. (C and D) Effects of ETV1 on TEER values (c) and HRP flux (d). e-g Schematic depiction of the different reporter plasmids and relative luciferase activity: ZO-1 (e), occludin (f), and claudin-5 (g) were shown. Data represent mean ± SD (n = 5, each). **P < 0.01. h-j Schematic representation of the human ZO-1 (h), occludin (i), and claudin-5 (j) promoter region 3,000 bp upstream of the transcription start sites (TSSs), which were designated as +1. ChIP PCR products for putative ETV1 binding sites and an upstream region not expected to associate with ETV1 were depicted with bold lines. Dashed arrows represent the primers used for each PCR. GECs were used to conduct ChIP assays. PCR was conducted with the resulting precipitated DNA. Images are representative of independent experiments. with pulmonary hypertension, the upregulation of PTBP1 accelerates glycolysis of endothelial cells (BOECs) in blood growth, leading to abnormal metabolism and proliferation of BOECs 41 .
There is increasing evidence that circRNAs regulate various biological behaviors of ECs [42][43][44] . For the first time, we found that circRNA_001160 knockdown significantly increased BTB permeability. The regulatory effect of cir-cRNAs on vascular ECs can also be seen in cZNF292, which demonstrates the function of promoting angiogenesis in ECs 45 . The knockdown of cZNF609 could increase ECs' migration and tube formation, reduce retinal vascular loss and inhibit pathological angiogenesis in vivo 46 . CircANRIL induces atherosclerosis by promoting apoptosis and proliferation in human vascular cells and tissues 47 .
There were three binding sites upstream of the region of circRNA_001160 formation on PTBP1 and ANKRD17 pre-mRNA predicted by the application of UCSC and RBPmap database. The experimental data demonstrated that PTBP1 specifically bound to ANKRD17 pre-mRNA.
A growing number of studies have indicated that different types of RBPs regulate the expression of circRNAs. The RBP QKI has two potential binding sites upstream and downstream of the circRNA-forming region on the SMARCA5 pre-mRNA. If both upstream or downstream binding sites are mutated, or four sites are simultaneously mutated, the generation of circular RNA is greatly reduced 48 . Eukaryotic initiation factor 4A3 (EIF4A3) binds to the upstream region of the MMP9 mRNA transcript and induces circMMP9 expression 49 . RBP FUS regulates circRNA biogenesis by binding the introns flanking the back-splicing junctions 50 .
CircRNAs adsorb miRNAs and act as molecular sponges to regulate the development of multiple malignancies [54][55][56] . CircRNA-MYLK could directly bind to miR-29a and reduce the inhibition of target VEGFA by miR-29a, thereby promoting the development of bladder cancer 57 . Hsa_circ_0010729 regulates vascular ECs' proliferation and apoptosis by targeting the miR-186/HIF-1α axis 58 . The results of dual luciferase reporter assays and RIP assays confirmed that miR-195-5p bound to circRNA_001160 and they were involved in the RISC complex. Further, qRT-PCR results showed that miR-195-5p expression was negatively correlated with circRNA_001160 expression, while circRNA_001160 knockdown significantly upregulated the expression of miR-195-5p. At the same time, silencing of miR-195-5p reversed the regulation of circRNA_001160 knockdown on BTB permeability. These results indicated that cir-cRNA_001160 regulated BTB permeability by adsorbing miR-195-5p.
Many studies have confirmed the transcriptional regulation of ETV1 on downstream target proteins. Analysis of DNA sequence data revealed that there are many motifs interacting with Etv1 (DA motif)(G/C/A)GGA(A/ T)(G/A) in the promoter region of the mature gene 59 . In prostate cancer, studies have confirmed that ETV1 is capable of binding to the matrix metalloproteinase 7 (MMP-7) gene promoter in vitro and ETV1 stimulates the Fig. 8 PTBP1, circRNA_001160, miR-195-5p enhanced the effect of Dox in promoting apoptosis of U87 cells. a Apoptosis analysis in U87 cells was evaluated by Annexin V-PE/7-AAD staining. Data represent mean ± SD (n = 3, each). **P < 0.01 versus control group; # P < 0.05, ## P < 0.01 versus Dox group. b Schematic diagram of PTBP1/circRNA_001160/miR-195-5p/ETV1 pathway to regulate BTB permeability.
activity of the MMP-7 promoter 60 . Using the bioinformatics database (targetscan, DBTSS and JASPAR CORE), ETV1 was identified as a direct target of miR-195-5p, and ETV1 was bound to the promoter regions of ZO-1, occludin and claudin-5. We hereby confirmed that ETV1 was highly expressed in GECs and was negatively regulated by miR-195-5p. We further utilized dual-luciferase reporter assays and ChIP assays to show that ETV1 directly bound to the promoter regions of ZO-1, occludin and claudin-5, showing transcriptional promotion. Therefore, ETV1 could regulate the permeability of BTB by a paracellular pathway.
MiRNAs can be used as post-transcriptional gene regulators by targeting the 3'-UTR of target mRNA 61 . We confirmed that miR-195-5p targeted the 3'-UTR of ETV1, and overexpression of miR-195-5p reduced ETV1 mRNA and protein expressions. At the same time, double overexpression of miR-195-5p and ETV1 reversed the pre-miR-195-5p alone on the BTB permeability. The above results confirmed that miR-195-5p regulated the expression of ETV1 at the post-transcriptional level by targeting the 3'-UTR of ETV1 mRNA, thereby regulating BTB permeability.
Doxorubicin (Dox) is a classic anti-neoplasm drug that affects the supercoiling in DNA by inhibiting the activity of the enzyme topoisomerase II 62 . Due to the presence of BTB, the therapeutic effect of Dox on glioma is not obvious 63 . The circular RNA DENND4C could effectively promote the Dox across BTB to induce apoptosis of glioma cells 10 . We found that the combination of sh-PTBP1, sh-circRNA_001160, pre-miR-195-5p with Dox significantly increased the apoptosis of U87 cells. The results indicated the potential value of the combination of genes and drugs in the treatment of glioma.
In summary, PTBP1 promoted the function of cir-cRNA_001160 by regulating the expression of cir-cRNA_001160. Double silencing of PTBP1 and circRNA_001160 could significantly increase BTB permeability. CircRNA_001160 acted as a molecular sponge of miR-195-5p, which had a negative regulatory effect on ETV1. ETV1 promoted the expression levels of tight junction-related proteins at the transcriptional level, and overexpression of ETV1 attenuated BTB permeability. The combined use of PTBP1, circRNA_001160 and miR-195-5p enhanced the role of Dox in promoting apoptosis in glioma cells. The results of this study could provide a new direction for the combined treatment of genes and anti-tumor drugs.