The expression of circRACK1 is downregulated in GC cells.
Recent advancements in our understanding of circRNAs, particularly their role in protein translation via CAP-independent mechanisms, have not only expanded our knowledge of these molecules but also highlighted their potential in disease diagnosis and treatment. Current research indicates that circRNA translation is tightly regulated under stress conditions, possibly serving as a compensatory cellular mechanism. Notably, hypoxia, a hallmark of the tumor microenvironment and a key feature of solid tumors, may influence circRNA expression. In this context, we investigated whether circRNA expression in gastric cancer (GC) cells alters under hypoxic conditions.
We cultured HGC27 cells, which were in the logarithmic growth phase, in a hypoxic incubator (5% CO2, <1% O2) for 36 hours. Post-cultivation, we lysed the cells and conducted Ribo-seq detection. Our circular heatmap analysis revealed a notable decrease in the expression of a novel circRNA, designated as novel_circ_000494, under these hypoxic conditions (see Fig. 1A). Further investigation in the Circbase database identified this circRNA as hsa_circ_0075402, located on chromosome 5:181241492-181242345. Its host gene is RACK1 (ENSG00000204628). Subsequent Sanger sequencing analysis showed that this circRNA, which we have named circRACK1, forms a closed-loop structure through the reverse splicing of two exons from the RACK1 gene (exon1: 181242345-181242174, exon2: 181241639-181241492), as illustrated in(Fig. 1B).
To confirm the circular structure of circRACK1, we designed convergent and divergent primers for amplifying circRACK1 and linear RACK1, respectively. Our results, depicted in Fig. 1C, demonstrate that circRACK1 is exclusively amplified from cDNA, confirming its nature as a back-spliced product of the host mRNA. Additionally, circRACK1 exhibited significantly higher resistance to RNase R degradation compared to its linear mRNA counterpart (Fig. 1D-E). Actinomycin D assays further confirmed the enhanced stability of circRACK1, showing a markedly longer half-life than the linear RACK1 transcript (Fig. 1F-G).
RT-qPCR detection in common GC cells revealed a reduction in circRACK1 expression. Particularly, AGS and HGC27 cell lines, showing the lowest circRACK1 levels, were selected for further study (Fig. 1H). Upon re-examination under hypoxic conditions, we observed an additional decrease in circRACK1 expression levels (Fig. 1I-J).
CircRACK1 inhibits malignant phenotypes of gastric cancer cells
To delve into the functional importance of circRACK1 in GC cells, we strategically designed siRNAs targeting circRACK1's specific binding sites and introduced overexpression plasmids into HGC27 and AGS cells. This intervention notably impacted the expression levels of circRACK1, while the mRNA and protein levels of its host gene, RACK1, remained largely unaffected, as illustrated in Fig(S1 A-F).
Our research extended to evaluating the effect of circRACK1 on the proliferation of GC cells. EdU incorporation analysis further corroborated these findings, showing increased EdU assimilation in cells with diminished circRACK1 expression, and the opposite in those with heightened expression (Fig 2A-B, S2A-B). The results from colony formation assays were in line with these observations. Cells overexpressing circRACK1 formed fewer and smaller colonies, whereas inhibition of circRACK1 resulted in a significant increase in colony numbers, underscoring its suppressive role in GC cell proliferation (Figure S3A).The role of circRACK1 in apoptosis was elucidated through flow cytometry, employing ANEXIN V-PI. The data revealed that circRACK1 knockdown markedly diminished apoptosis in GC cells, while its overexpression promoted apoptotic processes (Fig S2E-F).Our investigation further spanned the influence of circRACK1 on GC cell migration and angiogenesis. Transwell assays demonstrated an increased migration of cells in the circRACK1 knockdown groups and a decrease in the overexpressed groups (Fig 2C, S2C-D). Scratch assays reinforced these findings, indicating that high levels of circRACK1 slow down wound healing, whereas its knockdown accelerates this process (Figure S3B-C). These outcomes suggest a significant inhibitory effect of circRACK1 on the migratory capabilities of GC cells. Moreover, the angiogenic potential of GC cells, assessed through their impact on HUVEC-induced migration and tube formation, was enhanced in circRACK1 knockdown cells, while overexpression of circRACK1 exhibited the opposite effect (Figure 2D).
CircR-127aa encoded by circRACK1 inhibits gastric cancer cell malignancy
In exploring the coding potential of circRACK1, our search in the CIRCRNADb and TransCirc databases revealed multiple internal ribosome entry sites (IRES) and open reading frames (ORFs) within circRACK1, hinting at its possible coding function. To test the translatability of circRACK1, we analyzed its putative ORF using ORF Finder. We identified a potential overlapping ORF (384nt) within circRACK1, capable of encoding a peptide of 127 amino acids in length (Fig 3A). We cloned the circRACK1 sequence into a circRNA expression vector (circRACK1-3xFLAG), with a FLAG tag positioned before the stop codon to detect successful circularization. Immunoblotting suggested that circRACK1 might encode a new protein of approximately 15kDa. To verify this protein's origin from circRACK1, we engineered plasmids with FLAG-tagged start codon mutations (ATG-ATT) and empty vector controls. The unique peptide fragment formed at the junction site of circRACK1 is illustrated in Fig 3B. Post-transfection of these plasmids into HGC27 and AGS cells, followed by Western Blot analysis after 48 hours, FLAG was detected in cells overexpressing circRACK1 but not in those with the vector or circRACK1-mut plasmid (Fig 3C). Given that part of circRACK1-encoded circR-127aa sequence overlaps with the host gene RACK1, we considered the possibility of RACK1 polyclonal antibody binding to circR-127aa. Using an anti-RACK1 antibody, we detected circR-127aa, observing increased expression levels following overexpression of the OE plasmid, while the expression of host gene RACK1 remained relatively unchanged (Fig 3D). Immunofluorescence experiments further confirmed the presence of circR-127aa in HGC27 and AGS cells, predominantly localized in the cytoplasm and enriched on the cell membrane (Fig 3E).To ascertain whether the biological impact of circRACK1 on gastric cancer cells is mediated through itself or its translation products, we re-examined common tumor phenotypes using circRACK1 overexpression plasmids, circRACK1-mut, and control plasmids. Gastric cancer cells transfected with circRACK1-mut plasmids showed no significant changes in proliferation, migration, and scratch healing capabilities compared to control group plasmids. However, cells transfected with circRACK1 overexpression plasmids exhibited an inhibitory effect on these phenotypes (Fig S4 A-D).
CircR-127aa promotes ubiquitination and degradation of vimentin
To elucidate the biological mechanism behind circR-127aa's influence on gastric cancer behavior, we conducted immunoprecipitation experiments to identify proteins potentially interacting with circR-127aa. The immunoprecipitated products were subjected to gel electrophoresis and stained with Coomassie Brilliant Blue to accentuate the distinct bands between the Vector and OE groups. These bands were then analyzed through mass spectrometry, indicating potential interactions of circR-127aa with Vimentin and ACTIN (Fig 3F, S5A). Further immunoprecipitation assays validated these interactions (Fig 3G, H, S6B).
We also performed immunofluorescence co-localization experiments, observing that circR-127aa co-localizes with Vimentin and ACTIN, particularly on the cell membrane (Fig S5C-D). To explore the regulatory mechanism of circR-127aa on Vimentin expression in gastric cancer cells, we conducted RT-qPCR and Western Blotting experiments. These revealed that Vimentin expression did not significantly change in the gastric cancer cell line with overexpressed or knocked-down circRACK1 compared to the control group (Fig S6A,-B). However, Western Blotting showed a decrease in Vimentin expression following circRACK1 overexpression and an increase following circRACK1 knockdown (Fig4A-B).
We further treated OE and shRNA group cells with cycloheximide (CHX) to inhibit new protein synthesis in gastric cancer cells. Results indicated a significant reduction in Vimentin expression in the OE group compared to the ShRNA group (Fig 4C). After treatment with the proteasome inhibitor MG132, Vimentin expression levels in the OE group's two cell lines recovered, suggesting that circR-127aa might influence Vimentin expression through ubiquitin-mediated proteasomal degradation (Fig 4D). In HEK293T cells transfected with HA-ubiquitin and Vimentin overexpression plasmids, circRACK1, or Vector plasmids for 48 hours, exogenous ubiquitin detection revealed increased ubiquitination modifications in the OE group compared to the Vector group (Fig 4E). Lastly, immunoprecipitation experiments on two stable cell lines showed that circRACK1 knockdown decreased Vimentin ubiquitination, while its overexpression increased Vimentin ubiquitination (Fig 4F)
Modulatory Effects of the HIF-1α/circRACK1/vimentin Axis on Gastric Cancer Cell Behavior
In the HGC27 gastric cancer cell line, circRACK1 overexpression (OE) led to an increase in E-cadherin levels and a decrease in N-cadherin, β-catenin, MMP9, and Snail, indicative of reduced invasive capacity. This modulation was reversed by concurrent HIF-1α overexpression (HIF-1α OE), which suggests a regulatory mechanism where HIF-1α enhances cancer cell invasion and migration by suppressing circRACK1 (Fig5A,S6A). Colony formation and invasion assays further supported the role of circRACK1 in altering cellular behaviors. The cells with circRACK1 OE displayed a diminished ability to form colonies and invade through the matrix. This was notably counteracted by co-overexpression with vimentin (OE+Vim), placing vimentin as a key effector in circRACK1-driven functions (Fig5B-C,S6B-C). Immunofluorescence assays demonstrated that circRACK1 silencing (sh-circR) results in significant cytoskeletal reorganization, characterized by increased pseudopodia, suggesting a promotion of cellular motility, as opposed to the more stable architecture observed in the OE group (Fig5D,S6D). Western blot analysis across different genetic backgrounds established that circRACK1 knockdown and vimentin co-overexpression increased the levels of mesenchymal markers while decreasing the epithelial marker E-cadherin, aligning with the molecular signature of EMT (Fig5E,S6E). Collectively, these results illuminate the comprehensive role of circRACK1 and its peptide product, CircR-127aa, in modulating gastric cancer cell behavior through EMT and cytoskeletal dynamics, potentially orchestrated by vimentin interactions.
Knockdown of circRACK1 promotes gastric cancer growth and metastasis in vivo
To further validate the influence of circRACK1 on the proliferation of gastric cancer cells in vivo, we utilized stable cell lines with circRACK1 knockdown (experimental group) and corresponding control cell lines (control group) to establish subcutaneous tumors in nude mice. Regular tumor volume measurements indicated that circRACK1 knockdown enhances subcutaneous tumor growth (Fig6A,C, S7C). live animal imaging showed larger tumors in the circRACK1 knockdown group (Fig6B). Subsequent H&E staining and immunohistochemical analyses of the tumors indicated significantly increased expression of Ki-67, Vimentin, MMP9, and CD31 in the experimental group compared to the control group. This suggests enhanced proliferation and invasion capacities of tumor cells and increased microvascular density when circRACK1 is knocked down. Lower expression of Caspase 3 in the experimental group implied reduced apoptosis in tumor cells (FigS7F). Additionally, F-ACTIN staining within the tissues showed increased fluorescence intensity in mouse tumors with circRACK1 knockdown (Fig 6D).During the live imaging and pathological examination of the subcutaneously implanted tumors in the nude mice, we observed more instances of intrathoracic and abdominal metastases in the experimental group than in the control group (S7D,E). We injected the stable cell lines with circRACK1 knockdown (experimental group) and the corresponding control cell lines (control group) into the tail vein of mice to create a distant metastasis model. After four weeks, we conducted regular small animal live imaging and CT scans for distant metastases monitoring. Starting from the fifth week post-model establishment, more distant metastases were observed in the experimental group compared to the control group (Fig6E, S7G). Upon sacrificing and dissecting the mice at eight weeks post-modeling, we found that the experimental group exhibited more and larger lung metastases (Fig6F-G).
Additionally, we examined the livers of the mice and noted metastatic lesions in the liver of some mice in the experimental group. We selected typical hepatic metastatic lesions for observation and documentation (Fig S7H).
In the concluding section of our study, we present a detailed schematic diagram illustrating the role of the peptide circR-127aa, encoded by circRACK1, in gastric cancer (GC). This diagram emphasizes the tumor-suppressing function of circR-127aa, particularly highlighting its involvement in the ubiquitination process of Vimentin, thereby elucidating its significance in the pathophysiology of GC(Fig6H).