HNRNPL Circularizes ARHGAP35 to Produce an Oncogenic Protein

Abstract

Circular RNAs (circRNAs) are an intriguing class of widely prevalent endogenous RNAs, the vast majority of which have not been characterized functionally. Here, we identified a novel oncogenic circRNA originating from the back-splicing of Exon2 and Exon3 of a tumor suppressor gene, ARHGAP35 (also known as P190-A), termed as circARHGAP35. have observe that circARHGAP35 and linear ARHGAP35 have antithetical expression and functions. Interestingly, circARHGAP35 contains a 3867 nt long ORF with an m⁶A-modified start codon and encodes a truncated protein comprising four FF domains and lacking the Rho GAP domain. Mechanistically, circARHGAP35 protein promotes cancer cell progression by interacting with TFII-I protein in the nucleus. The RNA binding protein, HNRNPL, facilitates the formation of circARHGAP35. Clinically, circARHGAP35 is associated with poor survival in cancer patients. Our findings characterize an oncogenic circRNA and demonstrate a novel mechanism of oncogene activation in cancer by circRNA through the production of a truncated protein.

Keywords: ARHGAP35; HNRNPL; cancer progression; circular RNAs; oncogenic proteins; translation.

Figure 1

Identification of a circRNA derived from ARHGAP35 gene. A) Hierarchically clustered heatmap of circRNAs differentially expressed in 12 paired HCC and adjacent non‐cancerous liver tissue samples. Rows represent circRNAs and columns represent tissues. B) The expression of 43 circRNAs and the corresponding linear transcripts from the same gene locus detected by qRT‐PCR in 12 paired HCC and adjacent non‐tumor tissues normalized to β‐actin. Data were presented as the log₂ fold change. p values were from paired Student's t‐test (n = 12) and adjusted with Benjamini–Hochberg method. C) The genomic loci of circular ARHGAP35 isoforms. The expression of circARHGAP35 was validated by qRT‐PCR followed by Sanger sequencing. The horizontal arrows refer to the divergent primers used to identify circARHGAP35. The junction site of circARHGAP35 is marked with vertical arrow. D) The expression of six circular ARHGAP35 isoforms in SK‐Hep‐1 and HuH‐7 cells. E) qRT‐PCR analysis of circARHGAP35 and linear ARHGAP35 RNA expression in the cytoplasm or nucleus of SK‐Hep‐1 and HuH‐7 cells. F) Identification of circARHGAP35 by fluorescence in situ hybridization (FISH) with negative control (NC) or the siRNA specifically targeting the back‐splice junction of circARHGAP35 in HuH‐7 cells. Red: circARHGAP35 probes were labeled with Cy3; Blue: nuclei were stained with DAPI. Scale bars, 10 µm. 18S was used as the cytoplasmic control. G) Northern blot for circARHGAP35 and linear ARHGAP35 without or with RNase R treatment using specific probes in HuH‐7 cells. H) qRT‐PCR analysis of circARHGAP35 and linear ARHGAP35 RNA following RNase R treatment in HuH‐7 cells. I) qRT‐PCR analysis of circARHGAP35 and linear ARHGAP35 RNA following actinomycin D treatment at the indicated time points in SK‐Hep‐1 cells. These data were represented as mean ± SEM. Results were performed in at least three independent experiments.

Figure 2

circARHGAP35 and linear ARHGAP35 have antithetical functions in cancer cell lines. A) Schematic illustration of three siRNAs specifically targeting circARHGAP35, linear ARHGAP35, and both, respectively. B) CCK‐8 proliferation assay of HuH‐7 and HCT‐116 cells transfected with the control or indicated siRNAs. C,D) Transwell migration and invasion assays of HuH‐7 (C) and HCT‐116 (D) cells performed following transfection with control or indicated siRNAs. Scale bars, 10 µm. E) Western blot validation of ARHGAP35 knockdown using CRISPR/Cas9 technology in HuH‐7 cells. F) Transwell migration and invasion assays in ARHGAP35‐knockdown HuH‐7 cells following transfection with the indicated siRNAs. G) CCK‐8 proliferation assay of HuH‐7 and HCT‐116 cells following stable overexpression of circARHGAP35. H) Transwell migration and invasion assays of HuH‐7 and HCT‐116 cells following stable overexpression of circARHGAP35. Scale bars, 10 µm. I) The effect of circARHGAP35 on tumor formation in a nude mouse xenograft model. Cells infected with either circARHGAP35 shRNA expressing lentivirus or vector control lentivirus were injected subcutaneously into the flank of each nude mouse. Scale bar, 10 mm. J) The tumor weight of the two groups. K) The effect of circARHGAP35 on tumor metastasis in a mouse tail vein injection model. Cells infected with either shRNA expressing lentivirus or vector control lentivirus were injected into the tail vein of each nude mouse (n = 11). Statistical analysis of the differences between the two groups was performed using the χ2 test. L) Hematoxylin‐eosin‐stained sections of lung metastatic nodules formed in the two groups. Slides were examined by an expert pathologist. Black arrows indicate the nodules formed in the lung. The number of metastatic nodules in the lungs of the two groups were counted and analyzed. Scale bars, 100 µm. Shown were representative images. Data were represented as mean ± SEM. Results were performed in at least three independent experiments; two‐way ANOVA and Tukey post hoc test were performed for (B) and (G); one‐way ANOVA and Dunnett post hoc test were performed for (C,D); unpaired Student's t‐tests were performed for (H), (J) and (L). *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 3

circARHGAP35 encodes a protein. A) Schematic representation of a putative open reading frame (ORF) in circARHGAP35. The junction is present inside the ORF. Start and stop codons are indicated in green and red, respectively. B) Schematic representation of the expression constructs. circARHGAP35 sequence was inserted into a circular RNA expression vector, which contains Alu elements to form the vector ‘p‐circARHGAP35’; Flag tag was added directly to upstream of the stop codon (TGA) to establish the construct ‘p‐circARHGAP35‐Flag’; the circARHGAP35‐Flag sequence was cloned to a linear vector to form a negative control vector ‘p‐circARHGAP35‐Flag‐NC’. The start codon and stop codon are shown in green and red, respectively. The Flag tag is shown in pale yellow. C) The indicated plasmids were transfected into HEK‐293T cells and potential proteins were detected using Western blot analysis. D) Immunoprecipitation assay was performed using ARHGAP35 antibody in SK‐Hep‐1 cells. The immunoprecipitated protein sample was subject to SDS‐PAGE and mass spectrometry analysis to identify specific sequences of the circARHGAP35 protein (red letters). E–H) Polysome profiling was performed using a linear 15% to 50% sucrose gradient. The polysomes of HCT‐116 cell cytoplasmic extracts without (HCT‐116) or with EDTA treatment (HCT‐116 + EDTA) were fractionated using sucrose density gradient centrifugation. Absorbance at 254 nm was measured. The relative levels of circARHGAP35 (E), linear ARHGAP35 mRNA (F), circPMS1 (G), and Actin mRNA (H) were analyzed by qRT‐PCR in gradient fractions in HCT‐116 cell lysates with or without EDTA treatment. circPMS1 and actin served as negative and positive controls, respectively. Relative distribution of each RNA in individual fraction represented as a percentage of total RNA. The sum of all fractions was considered as a total of one RNA. I) Methylated RNA immunoprecipitation (MeRIP) assay was performed using total RNA from SK‐Hep‐1 cells. Purified RNA was subsequently analyzed by qRT‐PCR. Nonspecific IgG was used as an isotype negative control. Statistical analysis was performed using unpaired Student's t‐test. *p < 0.05; **p < 0.01; ***p < 0.001. J) MeRIP was performed using total RNA from HEK‐293T cells ectopically expressing either vector or FTO. Purified RNA was subsequently analyzed by qRT‐PCR. Statistical analysis was performed using two‐way ANOVA and Tukey post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001. K) FTO reduces circARHGAP35 translation. Protein was analyzed by Western blot using HEK‐293T cells co‐transfected with circARHGAP35 expression vector and FTO (or vector control).

Figure 4

The circARHGAP35 protein has oncogenic functions. A) Schematic representation of the p‐lin‐cORF‐Flag and p‐lin‐cORF‐Flag‐mut constructs. The circARHGAP35 ORF sequence with Flag tag was inserted into a linear expression vector to form p‐lin‐cORF‐Flag. In the mutant construct, the stop codon TGA was deleted. The start codon and stop codon are shown in green and red, respectively. The Flag tag is shown in pale yellow. B) The circARHGAP35 and ARHGAP35 proteins were tested in SK‐Hep‐1 cells following the indicated lentiviral transduction. The linear ARHGAP35 full length ORF (FL) were inserted into linear vector to form p‐lin‐FL. C–E) Colony formation assay (C), CCK‐8 proliferation assay (D), and transwell migration and invasion assay (E) following the overexpression of circARHGAP35 protein, mutant protein, or ARHGAP35 protein using a linear expression vector. Scale bars, 10 µm. Data were represented as mean ± SEM. One‐way ANOVA and Dunnett post hoc test were performed for (C) and (E). Two‐way ANOVA and Tukey post hoc test were performed for (D). *p < 0.05; **p <0.01; ***p < 0.001.

Figure 5

circARHGAP35 protein interacts with TFII‐I in the nucleus. A) Schematic illustrations of the protein domains of ARHGAP35 and circARHGAP35 proteins. The distinct C‐terminus of circARHGAP35 is shown in red. B) Subcellular localizations of circARHGAP35 protein and ARHGAP35 protein in SK‐Hep‐1 cells infected with circARHGAP35 expressing lentivirus or linear ARHGAP35 expressing lentivirus. circARHGAP35 protein (secondary antibody, Alexa 633, red), ARHGAP35 protein (secondary antibody, Alexa 488, green), and DAPI (blue). Scale bars, 10 µm. C) Western blot analysis of active and total RhoA in SK‐Hep‐1 cells with stable overexpression of circARHGAP35 protein, or ARHGAP35, or vector control. D) F‐actin was stained using phalloidin‐488 in SK‐Hep‐1 cells with stable overexpression of either circARHGAP35 protein or ARHGAP35 protein. Nuclei were stained with DAPI. Scale bars, 25µm. E) Immunofluorescence of circARHGAP35 protein and TFII‐I using Flag or TFII‐I antibody in SK‐Hep‐1 cells infected with circARHGAP35 expressing lentivirus. circARHGAP35 protein (secondary antibody, Alexa 488, green), TFII‐I (secondary antibody, Rhodamine, red), and DAPI (blue). Scale bars, 7.5 µm. F) Immunoprecipitation (IP) assay in SK‐Hep‐1 cells with stable overexpression of circARHGAP35 protein using either Flag or control IgG antibody, followed by immunoblotting using the TFII‐I antibody. G) IP assay in SK‐Hep‐1 cells with stable overexpression of circARHGAP35 protein using either TFII‐I or control IgG antibody, followed by immunoblotting using indicated antibodies. H) Western blot validation of circARHGAP35 protein and TFII‐I protein in circARHGAP35 protein overexpressing SK‐Hep‐1 cells following transfection with siRNA targeting TFII‐I. I,J) CCK‐8 proliferation (I) and transwell (J) assays following transfection with siRNA targeting TFII‐I in circARHGAP35 protein overexpressing SK‐Hep‐1 cells. Scale bars, 10 µm. Data were represented as mean ± SEM. Two‐way ANOVA and Tukey post hoc test were performed for (I); one‐way ANOVA and Dunnett post hoc test were performed (J), ***p < 0.001.

Figure 6

HNRNPL regulates circARHGAP35 formation. A) The expression fold change (siRNA/NC) of circARHGAP35 and linear ARHGAP35 following treatment with the siRNA library targeting 63 RBPs. The siRNA pool targeting HNRPL is highlighted in red. B) The expression of HNRNPL, circARHGAP35, and linear circARHGAP35 in SK‐Hep‐1 cells transfected with three siRNAs targeting HNRNPL. Data represent the mean ± SEM. One‐way ANOVA and Dunnett post hoc test were performed. *p < 0.05; **p < 0.01; ***p < 0.001. C) A schematic illustration of putative binding sites of HNRNPL upstream and downstream of the circARHGAP35 genomic site. D) RNA immunoprecipitation (RIP) was performed in SK‐Hep‐1 cells. qRT‐PCR was performed to quantify the RIP enriched RNA. U2 was used as a negative control. p Values were from unpaired Student's t‐tests. *p < 0.05; **p < 0.01; ***p < 0.001. E) Schematic diagram of circGFP expression vectors without or with HNRNPL binding sites (Up‐1 and Down‐2) in upstream and downstream of the circARHGAP35 genome location (#1, #2). F) qRT‐PCR analysis of the expression of circGFP using specific primer in HEK‐293T cells transfected with the indicated constructs, with siRNA targeting HNRNPL (or negative control, NC). G) The expression of HNRNPL in 110 paired HCC and adjacent non‐tumor (NT) liver tissues. Data were analyzed by paired Student's t‐test. H) Correlation between circARHGAP35 and HNRNPL in 110 HCC samples was determined by qRT‐PCR with β‐actin serving as an internal control. Statistical analysis was performed with Pearson's correlation analysis. I) Correlation between circARHGAP35 and HNRNPL in cancer cell lines. Statistical analysis was performed with Pearson's correlation analysis.

Figure 7

The upregulation of circARHGAP35 is associated with poor survival in cancer patients. A) The expression of circARHGAP35 in 110 paired HCC and adjacent non‐tumor (NT) liver tissues. Data were analyzed by paired Student's t‐test, n = 110. B) The fold change of circARHGAP35 expression in 110 paired HCC samples (downregulated, green; unchanged, gray; upregulated, red). C) The expression of linear ARHGAP35 in 110 paired HCC and adjacent non‐tumor (NT) liver tissues. Data were analyzed by paired Student's t‐test, n = 110. D) The fold change of linear ARHGAP35 expression in 110 paired HCC samples (downregulated, green; unchanged, gray; upregulated, red). E) Kaplan–Meier analysis of the correlation between circARHGAP35 expression and overall survival (OS), disease free survival (DFS), and recurrence in 110 HCC patients. F) Kaplan–Meier analysis of the correlation between linear ARHGAP35 RNA expression and OS, DFS, and recurrence in 110 HCC patients. G) The 110 HCC patients were divided into four groups according to the expression levels of circARHGAP35 and linear ARHGAP35 RNA. Kaplan–Meier analysis of the correlation between circARHGAP35/linear ARHGAP35 RNA expression and OS, DFS, and recurrence in 110 HCC patients. Log‐rank tests were used to determine the statistical significance for (E), (F), and (G).

Figure 8

Integrated model depicting circARHGAP35 driving cancer progression. circARHGAP35 is derived from the locus of a tumor suppressor gene, ARHGAP35, and its biogenesis is regulated by HNRNPL. circARHGAP35 is translated into a large protein in an m⁶A‐dependent manner. The circARHGAP35 protein exerts its oncogenic roles by interacting with TFII‐I, while ARHGAP35 suppresses cancer cell motility by decreasing RhoA activity.