Introduction
Hepatocellular carcinoma (HCC) is the most common histological subtype of primary liver cancer with high incidence and poor prognosis [
1]. By the year 2025, liver cancer is projected to impact over 1 million individuals annually [
2]. Hepatitis B virus (HBV) infection is the leading risk factor for the development of HCC, contributing to approximately 50% of cases [
3]. Due to the achievement of sustained virological response (SVR) through the use of antiviral drugs, the risk associated with hepatitis C virus (HCV) infection has markedly declined [
4]. Nevertheless, patients with cirrhosis remain at a heightened risk for the development of HCC even after HCV clearance. Additionally, non-alcoholic steatohepatitis (NASH), which is linked to metabolic syndrome or diabetes mellitus, is emerging as the rapidly growing cause of HCC, particularly in Western countries [
5]. Metastasis is a defining characteristic of cancer [
6]. HCC tends to invade the vascular system to form microvascular invasion (MVI) or portal vein tumor thrombosis (PVTT) [
7]. Vascular invasion is a major route for intrahepatic and distant metastasis in HCC and is a strong negative prognostic factor [
8]. PVTT is present in 10–60% of patients at the time of initial diagnosis of HCC [
9]. Most treatment guidelines classify patients with PVTT as advanced stage with limited treatment options [
2]. The median survival time of patients with PVTT treated with supportive care only is just 2–4 months [
10]. Except for a small number of patients with simultaneously resectable HCC and PVTT who may benefit from surgery, the outcomes of non-surgical treatment for PVTT are unsatisfactory [
11]. Hence, better understanding of the molecular mechanisms of vascular invasion is needed to develop more effective therapeutic approaches for PVTT.
Circular RNA (circRNA) is a new class of RNA that is generated by an alternative splicing method called back-splicing [
12]. Due to the loop-like structure and because they lack poly(A) tails, circRNAs are resistant to RNase R degradation and are more stable than linear RNAs [
13]. circRNAs exert their biological functions through various mechanisms, such as acting as microRNA (miRNA) sponges, interacting with proteins, and translating novel polypeptides [
14]. circRNAs broadly exist in various eukaryotes and play important roles in organ development and disease progression, especially tumorigenesis and metastasis [
15]. We previously showed that ciRS-7 is an independent predictor of MVI and circSETD3 inhibits the proliferation of HCC cells by sponging miR-421 [
16,
17]. Hu et al. [
18] identified circASAP1 as a key regulator of metastasis that may serve as a prognostic biomarker based on circRNA sequencing (circRNA-seq) of patients with HCC and postoperative pulmonary metastases. However, circRNAs that regulate vascular invasion in HCC are unknown.
Given that MVI is only observed under a microscope during pathological diagnosis, studies on MVI tissues are difficult [
19]. Unlike MVI, PVTT can be detected by radiological examination and staged according to the site of thrombosis [
20]. As patients with simultaneously resectable HCC and PVTT can derive a survival benefit from surgery [
8], paired HCC and PVTT tissues harvested during surgery provide a valuable opportunity to study the mechanisms involved in the vascular invasion and metastasis of HCC. RNA sequencing (RNA-seq) has identified numerous dysregulated messenger RNAs (mRNAs) [
21], miRNAs [
22], and long non-coding RNAs (lncRNAs) [
23] in PVTT, some of which have been shown to play an important role in HCC metastasis. In this study, using high throughput circRNA-seq, we showed that circRNA pleckstrin and Sect. 7 domain containing 3 (circPSD3) (derived from exons 13 and 14 of
PSD3; circBase ID: hsa_circ_0136098) was significantly downregulated in PVTT tissues. Mechanistically, circPSD3 inhibited the migration and invasion of HCC cells in a urokinase-type plasminogen activator (uPA) system-dependent manner.
Materials and methods
HCC samples
In total, 159 HCC samples were obtained from the West China Biobanks, Department of Clinical Research Management, West China Hospital, Sichuan University. Nineteen patients had paired PVTT tissues and 48 patients had paired non-cancerous liver tissues. All patients underwent radical resection between 2016 and 2018. None of the patients received other preoperative anticancer treatments. The study was approved by the Biomedical Ethics Committee of West China Hospital (Ethic approval ID: 2022(1685)). Written informed consent was obtained from each patient.
circRNA-seq
Five PVTT and matched HCC tissues were selected for RNase R-treated circRNA-seq (Novogene, Beijing, China). Briefly, 5 µg of high-quality RNA per sample was used as input material. Ribosomal and linear RNAs were removed before library preparation. circRNAs were identified using find_circ and CIRI. The R package “DESeq2” (version 2.15.13) was employed to screen differentially expressed circRNAs between PVTT and matched HCC tissues. circRNAs with |log2(fold-change)| >2.5 and FDR < 0.05 were considered to be differentially expressed.
RNA-seq with unique molecular identifiers (UMIs)
To determine downstream targets and pathways of circPSD3, RNA-seq was performed using UMIs (Seqhealth Technology Co. Ltd., Wuhan, China). Briefly, 2 µg of total RNA extracted from HCC-LM9 cells stably transfected with circPSD3 overexpression or control lentivirus was used for stranded RNA-seq library preparation (Catalogue #DR08502; KC-Digital™ Stranded mRNA Library Prep Kit for Illumina, Seqhealth Technology Co. Ltd.), following the manufacturer’s instructions. The kit eliminates duplication bias in PCR and sequencing steps by using the UMIs of eight random bases to label the pre-amplified cDNA molecules. Library products corresponding to 200–500 bps were enriched, quantified, and sequenced on a NovaSeq 6000 sequencer (PE150 model) (Illumina, San Diego, CA, USA). Raw sequencing data were filtered using Trimmomatic (version 0.36). Clean reads were further treated with in-house scripts to eliminate duplication bias introduced in library preparation and sequencing. TopHat2 (version 2.0.13) was used to align de-duplicated reads to the reference genome (GRCh38/hg38). The R package “DESeq2” (version 2.15.13) was used to identify differentially expressed genes between groups. Genes with |log2(fold-change)| >1 and FDR < 0.05 were considered to be differentially expressed. The molecular functions and pathways of dysregulated genes were analysed in KOBAS 2.0.
Cell culture and transfection
Human HCC cell lines (HCC-LM9 and SK-Hep-1) cells were maintained in Dulbecco’s modified Eagle medium/high-glucose medium (HyClone, Logan, UT, USA) supplemented with 10% foetal bovine serum (FBS) (PAN-Biotek, Aidenbach, Bavaria) and antibiotics (1% penicillin/streptomycin; HyClone) in a humidified incubator with 5% CO2 at 37ºC.
Small interfering RNAs (siRNAs) targeting circPSD3, TAR DNA-binding protein 43 (TDP43), serpin family B member 2 (SERPINB2), and histone deacetylase 1 (HDAC1) were designed and synthesised by RiboBio (Guangzhou, China). The siRNA sequences are listed in Table
S1 (Additional file 1). siRNAs were transfected using GeneMute (SignaGen Laboratories, Rockville, MD, USA), according to the manufacturer’s instructions. The lentiviruses used to express circPSD3 and uPA receptor (uPAR) were constructed by GeneChem (Shanghai, China) and infected following the manufacturer’s protocol.
Quantitative real-time (qRT)-PCR
Genomic DNA (gDNA) was isolated using the PureLink™ Genomic DNA Mini Kit (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions. Total RNA was extracted using TRIzol (Invitrogen Life Technologies Inc., Germany). Reverse transcription was performed using HiScript III RT SuperMix for qPCR (+ gDNA wiper) Kit (Vazyme Biotech Co. Ltd., Nanjing, China). Nuclear and cytoplasmic fractions were isolated using the PARIS™ Kit (Thermo Fisher Scientific). qRT-PCR was performed in triplicate using 2× ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech Co. Ltd.) and the CFX Connect Real-Time System (Bio-Rad, Hercules, CA, USA). The primers used in this study are listed in Table
S2 (Additional file 1).
Fluorescence in situ hybridisation (FISH)
Cy3-labelled probes were synthesised by RiboBio. FISH was performed using the FISH Kit (RiboBio), according to the manufacturer’s instructions. Images were captured using the A1RþMP Confocal Laser Microscope System (Nikon, Tokyo, Japan). U6 and 18 S ribosomal RNAs (rRNAs) were used as positive controls in the nucleus and cytoplasm, respectively.
Actinomycin D and RNase R treatment assays
circRNA stability was assessed using actinomycin D and RNase R treatment assays. For actinomycin D assay, HCC cells were cultured in medium containing 2 mg/mL actinomycin D. Total RNA was isolated at the indicated time points and subjected to qRT-PCR. For RNase R treatment, 3 µg of total RNA was treated with 10 U RNase R (20 U/µL; Epicenter, Madison, WI, USA) at 37 °C for 45 min, followed by 70 °C for 10 min to deactivate RNase R. Linear and circular RNA degradation was determined by PCR followed by agarose gel electrophoresis.
Western blotting and immunohistochemistry
Western blotting was performed as described previously [
17,
24]. The following primary antibodies were used: anti-SERPINB2 (1:1,000) (Proteintech, Wuhan, China), anti-uPAR (1:1,000) (Abcam, Cambridge, UK), anti-uPA (1:1,000) (Abcam), anti-HDAC1 (1:1,000) (Abclonal, Wuhan, China), anti-AGO2 (1:1,000) (Abclonal), anti-GST (1:1,000) (Abcam), anti-FLAG (1:1,000) (Abclonal), and anti-TDP43 (1:1,000) (Proteintech). Immunohistochemistry was performed using antibodies against TDP43 (1:200) (Proteintech), SERPINB2 (1:200) (Proteintech), and uPAR (1:200) (Abcam), as described previously [
17,
24].
Immunofluorescence staining
Cells were seeded into a 24-well plate with a coverslip at the bottom of each well. After incubation for 24 h at 37 °C, the cells were fixed with 4% paraformaldehyde for 30 min, permeabilised with 0.2% Triton X-100 for 10 min, and blocked with 3% bovine serum albumin for 1 h. Cells were incubated with specific antibodies at 4 °C overnight and with fluorescence-conjugated secondary antibodies at 37 °C for 1 h. Nuclei were stained with 4’6-diamidino-2-phenylindole (DAPI) for 10 min. After sealing, images were acquired using a confocal laser scanning microscope (Nikon).
Cell viability assay
Cell viability was assessed using the Cell Counting Kit-8 (CCK-8) and colony formation assays. For CCK-8 assay, HCC cells were suspended in 100 µL of medium and seeded into 96-well plates at a density of 2,000 cells/well. After incubation with 10 µL of CCK-8 solution for 1.5 h, absorbance at 450 nm was measured at indicated time points using the Eon™ Microplate Reader (BioTek, Whiting, VT, USA). For colony formation assay, 1,000 cells were seeded into each well of a 6-well plate. After incubation for 14 days, the colonies were fixed with 4% paraformaldehyde (Solarbio, Beijing, China) and stained with 0.05% crystal violet (Beyotime Biotechnology, Nantong, China). Colonies were photographed and counted. All experiments were performed in triplicate.
Wound healing assay
Confluent monolayer cells in 6-well plates were wounded using a 200 µL pipette tip. After washing with PBS twice, cells were cultured in medium containing 3% FBS. Images were acquired using an inverted microscope (Carl Zeiss, Jena, Germany) at 0 and 48 h after wounding. At least three separate fields were photographed. The relative healed area was calculated using ImageJ (National Institutes of Health, Bethesda, MD, USA) and normalised to 0 h control.
Transwell migration and Matrigel invasion assays
For Transwell migration assay, 3 × 104 HCC cells were suspended in 300 µL of serum-free medium and seeded into the upper chamber (pore size, 8 μm) (Millipore, Billerica, MA, USA). The bottom chamber contained 600 µL of medium containing 10% FBS as a chemoattractant. After 24 h, cells on the lower surface of the upper chamber were fixed with 4% paraformaldehyde, stained with 0.05% crystal violet, and imaged at 100× magnification. At least three random fields were photographed. Migrated cell numbers were counted using Image J. A similar protocol was performed for Matrigel invasion assay, except 30 µL of diluted Matrigel (BD Bioscience, Bedford, MA, USA) was added to the upper chamber before cell seeding.
3D spheroid-based Matrigel invasion assay
Complete growth medium containing 1 × 10
4 SK-Hep-1 cells (200 µL) was seeded into ultralow attachment 96-well round-bottom plates (Corning, USA) for 3 days, as described previously [
25‐
27]. Images were acquired when tumor spheroids were formed (Day 0). Then, 100 µL of plating medium was removed and replaced with the same volume of Matrigel. After Matrigel solidification, 100 µL of complete growth medium was added to each well. After 3 days, images were photographed and analysed using the Celigo
™ cytometer (Nexcelom Bioscience, Lawrence, MA, USA).
RNA immunoprecipitation (RIP) assay
RIP assay was performed using the Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (Millipore), according to the manufacturer’s instructions. Briefly, magnetic beads were sequentially incubated with primary antibodies and prepared cell lysates. Enriched RNA was isolated using TRIzol reagent and quantified by qRT-PCR. Immunoglobulin G antibody served as the negative control.
MS2-tagged RNA affinity purification (TRAP) assay
A 2× MS2 stem–loop sequence was inserted at the back-splicing site of circPSD3 (Biosense, Guangzhou, China). The MS2-circPSD3 expression plasmid or negative control plasmid was co-transfected with the GST-MS2 expression plasmid into HCC cells. The cell lysates were incubated with anti-GST-coated magnetic beads (Beyotime Biotechnology) at 4 °C overnight. Enriched RNAs and proteins were collected. RNA was quantified by qRT-PCR. Proteins were quantified by liquid chromatography–tandem mass spectrometry (LC-MS/MS) (OEBiotech, Shanghai, China) and validated by western blotting.
Dual-luciferase reporter assay
A dual-luciferase reporter plasmid was constructed as a vector containing Renilla luciferase (Rluc) and firefly luciferase (Luc) (VectorBuilder, Guangzhou, China). The predicted internal ribosome entry site (IRES) of circPSD3 or mutant sequences was subcloned into the promoter region of Luc. The plasmid was transfected into HEK-293T cells using Lipo6000™ Transfection Reagent (Beyotime Biotechnology), following the manufacturer’s instructions. After incubation for 48 h, Rluc and Luc activity was determined using the Duo-Lite Luciferase Assay System (Vazyme Biotech Co. Ltd.). Results are presented as the ratio of Luc to Rluc.
Male BALB/c nude mice (5–6 weeks old) were purchased from HFK Bioscience (Beijing, China) and maintained under specific pathogen-free conditions. All animal experiments were approved by the Animal Care Committee of Sichuan University (Ethic approval ID: 20,220,224,052). To establish a lung metastasis model, HCC cells were injected into the tail vein of nude mice. Five weeks later, the mice were anaesthetised. Resected lung specimens were stained with haematoxylin and eosin (H&E). To establish an intrahepatic metastasis model, 2 × 106 stably transfected cells were injected into the liver of nude mice. Six weeks later, liver lobes containing tumors were harvested and subjected to H&E staining. Serial sectioning of liver tissues was performed to confirm vascular invasion. Pathological images were acquired using a Whole Slide Image Scanner (Unic Technologies Inc., Beijing, China). The number of metastatic nodules and vascular invasion were determined by two independent investigators. At least five mice were included in each group. All animal experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals.
Statistical analysis
Statistical analyses were conducted using SPSS (version 21.0) (IBM Corp., Armonk, NY, USA), GraphPad Prism (version 8.0) (GraphPad Software, La Jolla, CA, USA), and R (version 3.5.0) (The R Foundation, Vienna, Austria;
http://www.r-project.org/). Continuous variables were expressed as mean ± SD and compared using Student’s
t-test. Categorical variables were expressed as numbers and percentages and compared using the chi-square test or Fisher’s exact test, as appropriate. Correlations were determined using Pearson’s correlation coefficients. The optimal cut-off value for circPSD3 expression in HCC tissues was determined using X-tile software. Survival curves were plotted using the Kaplan–Meier method and compared using the log-rank test. All statistical tests were two-tailed. A p-value < 0.05 was considered statistically significant.
Discussion
Vascular invasion is a well-known prognostic factor in HCC that is used to guide classification and treatment. Until now, the mechanisms of vascular invasion have remained largely unknown, and treatment options are limited. In this study, using circRNA-seq and qRT-PCR validation, we showed that circPSD3 is downregulated in PVTT tissues. Decreased circPSD3 expression in HCC tissues is an indicator of MVI and is a predictor of poor prognosis in patients undergoing partial hepatectomy. Artificial regulation of circPSD3 affects the migration and invasion of HCC cells, which was demonstrated by a series of in vitro and in vivo experiments. Mechanistic investigations revealed that SERPINB2 is the downstream target of circPSD3. SERPINB2, an endogenous inhibitor of the uPA system, mediates the inhibitory effect of circPSD3 on the invasion and metastasis of HCC cells. To our knowledge, this is the first study to show the relationship between circRNA and vascular invasion, as well as the uPA system.
Theoretically, a single host gene can produce a panel of circRNAs that have different expression patterns and biological functions, due to disparate compositions. The circRNAs produced by PSD3 pre-mRNA have attracted attention in cancer and other diseases. Consistent with our study, a PSD3-derived circRNA was found to be significantly downregulated in ccRCC tissues and was associated with metastasis in patients with ccRCC. circPSD3 overexpression suppressed cell migration and invasion and epithelial–mesenchymal transition in vitro, and inhibited pulmonary metastasis in vivo, in a miR-25-3p/FBXW7-dependent manner [
47]. Another PSD3-derived circRNA was found to be significantly upregulated in papillary thyroid cancer tissues and cell lines and was positively associated with a larger tumor size, TNM stage, and lymph node metastasis. circPSD3 knockdown suppressed papillary thyroid cancer cell proliferation and invasion by enhancing the inhibitory effect of miR-7-5p on the expression of METTL7B [
48]. The pathological functions of different PSD3-derived circRNAs have also been investigated in hepatic fibrosis [
49] and hepatitis C virus (HCV) infection [
50]. These researches suggest that PSD3-derived circRNAs play important roles in multiple disease contexts. In-depth investigations of these circRNAs will be meaningful in identifying novel therapeutic targets.
Emerging publications have reported that the biogenesis of circRNAs can be regulated by multiple RBPs. Such as QKI regulates the formation of cSMARCA5 [
51], FUS regulates the biogenesis of circCNOT6L [
52], and EIF4A3 regulates circMMP9 [
53] and circTOLLIP [
54] expression. In current study, we demonstrated that TDP43 is an essential regulator on the biogenesis of circPSD3. To our knowledge, this is the first study to reveal the regulating effect of TDP43 on the biogenesis of circRNAs. TDP43 encoded by the
TARDBP gene mediates multiple aspects of RNA metabolism, including RNA transcription, alternative splicing, and mRNA stabilisation [
31]. Previous studies have mainly focused on neurodegenerative diseases, such as Alzheimer’s disease, Parkinson’s disease, and Huntington’s disease [
55]. However, studies have also demonstrated a relationship between TDP43 and a variety of malignancies, such as triple-negative breast cancer [
56,
57], lung cancer [
58], and melanoma [
59]. In HCC, several studies have reported that TDP43 is upregulated in HCC tissues. The interaction between TDP43 and
GSK3β mRNA inhibits the translation of GSK3β, thereby activating the Wnt/β-catenin pathway to enhance HCC cell proliferation and metastasis [
33]. TDP43 enhances the stability of ABHD2 mRNA by binding to its 3’UTR. Upregulated ABHD2 enhances lipid metabolism and suppresses apoptosis in HCC cells [
34]. TDP43 also inhibits the expression of the miR-520 family by interacting with the miR-520 family promoter. The miR-520/PFKP axis mediates TDP43 regulation of glycolysis in HCC cells [
32]. Consistent with these findings, our study showed that TDP43 mRNA levels were significantly higher in HCC tissues than in non-cancerous tissues. This result was strengthened by western blotting in six paired HCC tissues and immunohistochemical staining in a patient with HCC and PVTT. TDP43 expression was inversely correlated with the expression of circPSD3 in HCC tissues, and TDP43 knockdown significantly increased the expression of circPSD3 in HCC cells. Considering the inhibitory effect of circPSD3 on invasion and metastasis, we believe that circPSD3 may be another mediator of TDP43 regulation of vascular invasion and distant metastasis in HCC.
Studies have reported that PSD3-derived circRNAs can sequester miRNAs. However, due to differences in sequences among PSD3-derived circRNAs and the undetectable interaction between circPSD3 and AGO2, sponging miRNAs may not be the primary mechanism by which PSD3-derived circRNAs exert their function. Bioinformatics analysis predicted that circPSD3 may encode novel peptides. The function of the IRES in mediating protein translation and the ORF as a template was also confirmed using appropriate vectors. However, upregulated protein level in circPSD3-overexpressing cells was not detected using specific antibodies against two different amino acid sequences of the predicted encoded peptides. circRNAs tend to form short imperfect intramolecular double-stranded RNAs [
60,
61] and interact with diverse molecules [
13], these all may largely restrict the recognition of ribosome and the extension of peptide chain. Therefore, the research on translation of circRNAs is still initially, and the encoding notion of circNRAs should be strengthened by more high-quality explorations.
TRAP assay revealed numerous proteins that may potentially interact with circPSD3. Among them, HDAC1, is involved in transcriptional repression through histone deacetylation, allowing histones to wrap DNA tightly. To target specific genomic regions, HDAC1 must interact with DNA binding factors (e.g., transcription factors, nuclear receptors, and DNA methyltransferases). In human corneal and conjunctival epithelial cells, HDAC1 binds to the SERPINB2 promoter, together with pRb2/ p130, E2F5, DNMT1, and SUV39H1 [
44]. In MCF-7 breast cancer cells, HDAC1 interacts with CPT1A variant 2 (CPT1AV2). CPT1AV2 knockdown upregulates the levels of HDAC1 and alters the expression of multiple cancer-related genes, including the downregulation of SERPINB2 [
45]. HDAC1 is significantly upregulated in HCC tissues compared to matched non-cancerous tissues. Elevated HDAC1 expression in HCC tissues is associated with a higher incidence of portal vein invasion, poorer histological differentiation, and a shorter survival time after hepatectomy [
62,
63]. HDAC1 plays important roles in the proliferation, differentiation, apoptosis, and metastasis of HCC cells. In a recent study, HDAC1 interacted with HIF-1α to downregulate the expression of FAM99A, thereby inhibiting HCC metastasis and epithelial–mesenchymal transition by negatively regulating miR-92a during hypoxia [
64]. Consequently, we selected HDAC1 as a candidate mediator of the regulatory effect of circPSD3 on SERPINB2 expression and HCC metastasis. The interaction between circPSD3 and HDAC1 was demonstrated in a series of experiments. We found that HDAC1 knockdown enhanced the expression of SERPINB2 in HCC cells. Rescue experiments showed that HDAC1 knockdown attenuated the promoting effects of circPSD3 knockdown on the migration and invasion of HCC-LM9 and Sk-Hep-1 cells. These results suggest that HDAC1 is a key mediator of the inhibitory effect of circPSD3 on HCC metastasis. However, circPSD3 did not have any impact on the protein levels of HDAC1. No protein modifying proteins were identified by LC-MS/MS. HDAC1 is localised to both the nucleus and cytoplasm of HCC cells. The nuclear localising signal of HDAC1 is partially overlaps with the putative binding site of circPSD3. Therefore, we speculated that preferential localisation of circPSD3 in the cytoplasm may retain HDAC1 in the cytoplasm, thereby reducing the transcriptional repression effect of HDAC1. This hypothesis was tested by immunofluorescence and nuclear and cytoplasmic fractionation assays. However, the binding sequence on circPSD3 that mediates its interaction with HDAC1, and the detailed mechanism involved in the cytoplasmic retention of HDAC1, require further investigation.
Degradation of the basement membrane and ECM is primarily mediated by plasmin [
65] and is essential for cell migration, invasion, and metastasis [
6]. Plasmin is produced by the inactive zymogen, plasminogen [
66], which is activated by two types of plasminogen activators: the tissue-type plasminogen activator (tPA) and the uPA [
66]. tPA preferentially mediates the degradation of intravascular fibrin deposition [
67], whereas uPA, uPAR, and two endogenous inhibitors, plasminogen activator inhibitor-1 (PAI-1; also known as SERPINE1) and PAI-2 (also known as SERPINB2), constitute the uPA system [
46]. Once bound to uPAR, uPA catalyses the conversion of plasminogen to plasmin, forming an effective proteolytic enzyme system at the cell surface [
68,
69]. Plasmin not only degrades components of the ECM, but also promotes ECM degradation by activating latent MMPs and growth factors [
70,
71]. The proteolytic activity of uPA can be suppressed by SERPINE1 or SERPINB2 [
72,
73]. After uPA inhibition, uPA–PAI complexes facilitate the interaction between uPAR and low-density lipoprotein receptor, thereby stimulating endocytosis of uPA–PAI complexes and partial recycling of uPAR to the cell surface [
74]. Generally, SERPINB2 expression in tumor tissues is negatively associated with cancer growth and metastasis. Using immunohistochemistry, Zhou et al. [
38] revealed that the proportion of SERPINB2-positive cells in HCC tissues was significantly lower than that in non-cancerous tissues (26.9% vs. 71.8%, respectively). Negative intratumoral staining of SERPINB2 was independently associated with the presence of PVTT and predicted a poor prognosis. Comparable results were achieved in our study. The expression of SERPINB2 was significantly lower in HCC tissues than in non-cancerous tissues, whereas the reverse was true for the expression of uPA and uPAR. In two patients with HCC and PVTT, the staining intensity of SERPINB2 reduced from normal to HCC and to PVTT tissues. However, the opposite trend was observed for uPAR. These findings further strengthen the ability of the uPA system to regulate HCC cell migration and invasion.
Given the potency of the uPA system in promoting invasion and metastasis in a wide variety of malignancies, targeting the uPA system is a potential strategy for the treatment of cancer [
37,
46,
75‐
77]. In the past, therapeutic agents and approaches have focused on regulating constituents of the uPA system or blocking their biological activity [
78‐
80]. In a previous study, recombinant SERPINB2 and SERPINB2 cDNAs were used to upregulate SERPINB2 expression [
72,
81‐
83]. In this study, we introduced a novel circPSD3 to enhance SERPINB2 expression, thereby inhibiting the uPA system. circRNAs are gradually being recognised as promising treatment agents by virtue of their high stability, low molecular weight, and low immunogenicity [
13]. The manageable ectopic expression and in vitro synthesis of stable circRNAs have made circRNA-based therapies possible. For example, a circRNA packaged aptamer exhibited greater suppression of the NF-κB pathway than a linear packaged aptamer [
84]. miRNA-122 is indispensable for the HCV life cycle by binding to the 5’UTR of HCV RNA. A synthesised circRNA sponge containing four miRNA-122 binding sites sequestered miRNA-122, thereby inhibiting protein synthesis with greater efficacy than Miravirsen [
85]. Heterogeneous nuclear ribonucleoprotein L (hnRNPL), an RNA-binding protein, regulates alternative splicing by binding to short CA repeats of nuclear pre-mRNAs. An artificial circRNA containing 20–100 CA dinucleotides showed high affinity for hnRNPL and sequestered hnRNPL in the cytoplasm, resulted in alternative splicing events similar to those caused by siRNA-mediated hnRNPL depletion [
86]. It is noteworthy that the aforementioned circRNA-based treatments have only been studied in cell lines or mouse models. Further researches are needed before these findings can be translated into the clinic.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.