A circular RNA activated by TGFβ promotes tumor metastasis through enhancing IGF2BP3-mediated PDPN mRNA stability

Abstract

Metastasis is the leading cause of cancer-related death, where TGFβ-induced epithelial-mesenchymal transition (EMT) process confers on cancer cells increased metastatic potential. However, the involvement of circRNAs in this process is still obscure. Here, we identify a TGFβ-induced circRNA called circITGB6 as an indispensable factor during the TGFβ-mediated EMT process. circITGB6 is significantly upregulated in metastatic cancer samples and its higher abundance is closely correlated to worse prognosis of colorectal cancer (CRC) patients. Through gain- and loss-of-function assays, circITGB6 is found to potently promote EMT process and tumor metastasis in various models in vitro and in vivo. Mechanistically, circITGB6 enhances the mRNA stability of PDPN, an EMT-promoting gene, by directly interacting with IGF2BP3. Notably, interfering circITGB6 with PEI-coated specific siRNA effectively represses liver metastasis. Therefore, our study reveals the function of a TGFβ-regulated circRNA in tumor metastasis and suggests that targeting circITGB6 is a promising strategy for cancer therapy.

Fig. 1. Characterization of the TGFβ-induced circRNA circITGB6.

a Heat map of TGFβ-responsive circRNAs (annotated in circBase) with high abundance (average RPM > 0.1) and significance (P < 0.05). b Northern blotting assays for circITGB6 expression in MCF10A and A549 cells with or without 5 ng/ml TGFβ treatment for 72 h. The probe used for Northern blotting crosses the back-splicing junction site of circITGB6. c, d TGFβ increased circITGB6 expression in both time-dependent and dose-dependent manners. A549 cells were untreated (-) or treated with the indicated dose of TGFβ for 72 h (c) or with 5 ng/ml TGFβ for the indicated time (d). e The genomic loci of circITGB6. f Sanger sequencing validation of head-to-tail splicing (indicated by red line) between exon 11 and exon 12 of ITGB6 gene. g RT-PCR assays were performed from cDNA or gDNA to detect ITGB6 mRNA or circITGB6 using divergent and convergent primers (indicated by black and white triangle in (e)), respectively. h, i RT-PCR (h) and Northern blotting (i) assays of circITGB6, ITGB6 or Actin mRNA. Total RNAs from A549 cells were digested with or without RNase R. k Subcellular distribution of circITGB6 determined by qPCR assays. The cytoplasmic/nucleus fractionations were prepared in A549 and Capan-2 cells. Actin and U6 transcripts were used as the cytoplasmic and nucleus marker, respectively. Immunoblot assays with anti-GAPDH and anti-Lamin A/C confirmed good cytoplasmic/nucleus fractionation. Cyt cytoplasmic; Nuc nuclear. j RNA fluorescence in situ hybridization for circITGB6. 18 S rRNA and U6 were used as the cytoplasmic and nuclear marker, respectively. Nuclei were stained with DAPI. Scale bar, 10 µm. l–n qPCR assays for exploring the clinical relevance of circITGB6 expression in human CRC cancer tissues (n = 80 samples, cohort 1). Cancer grade (l), tumor size (m) and lymph node metastasis status (n). The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data. o Kaplan–Meier curve for the overall survival of CRC patients, which were divided into two groups according to the mean expression of circITGB6 (n = 80 samples, cohort 1). Data represent mean ± sd from three independent experiments (c, d). The P value was calculated by unpaired, two-tailed Student’s t test (a, c, d), one-way ANOVA test (l–n) and two-sided log-rank test (o). Source data are provided as a Source Data file.

Fig. 2. circITGB6 facilitated EMT process and tumor metastasis.

a Northern blotting assays for circITGB6 expression in A549 and MCF7 cells with circITGB6 overexpression. b, c Transwell migration assays for circITGB6 overexpression (b) and knockdown (c) cells. Scale bar, 20 μm. Vec, vector; Scr, scramble shRNA; sh1 and sh2 represent two independent shRNAs targeting the back-splicing region of circITGB6. d, e Effects of circITGB6 overexpression (d) and knockdown (e) on the protein levels of EMT markers. GAPDH as loading control. f–i qPCR assays for circITGB6 expression (f), immunoblotting assay for the expression of EMT markers and ITGB6 (g), Transwell migration assay (h) and representative images of cell morphology (i) in A549 stable cells with or without circITGB6 knockdown after 5 ng/mL TGFβ treatment for 72 h. j, k Stably expressing vector or circITGB6 SW620 cells (1 × 10⁶) were injected into the spleen to generate liver metastases. j Representative bioluminescence images of liver metastasis and quantitation of bioluminescence were presented. k Representative images for H&E staining of mouse liver lesions. The metastatic nodules indicated by the black arrows. Scale bar, 4 mm. n = 7 mice (vec) or 9 (OE) mice per group (j, k). l, m Capan-2 cells (1 × 10⁵) stably expressing scramble or circITGB6 shRNAs (sh1 and sh2) were intracardially injected for metastasis assays, n = 7 mice (Scr, sh1 or sh2) per group. Representative bioluminescence imaging of tumor metastasis and quantitation of bioluminescence were presented (l). Representative images for H&E staining of mouse brains or right femur tissues (m). The metastatic nodules indicated by the black arrows. Scale bar: 100 μm. The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data (j, i). Data represent mean ± sd from three independent experiments (f), mean ± s.e.m (j, i). Significance of differences was determined by one-way ANOVA test (j, l) and two-tailed Student’s t test (f). Source data are provided as a Source Data file.

Fig. 3. The metastatic role of circITGB6 is independent of its linear transcripts.

a Schematic representation of circularization-dead (CD) circITGB6 plasmids. CMV, cytomegalovirus promoter; Alu, Alu sequences. b RT-PCR assays showed the failure of exogenous circITGB6 formation in circITGB6-CD SW620 stable cells. c, d Transwell migration assays (c) and Immunoblotting assays of EMT markers (d) in circITGB6 wild-type (WT) or CD stable cells. GAPDH was used as the loading control. Scale bar: 20 μm. e Bioluminescence imaging of liver metastasis and quantitation of bioluminescence of vec, circITGB6-WT and -CD groups. n = 6 mice (vec or CD) or 7 mice (WT) per group. f Representative images of liver metastatic nodules and H&E staining of mouse liver tissues (left) and the numbers of liver metastatic nodules (right). The metastatic nodules indicated by the black arrows or dotted lines. Scale bar: 20 μm. The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data (e, f). Data represent mean ± s.e.m (e, f). Significance of differences was determined by one-way ANOVA test (e, f). Source data are provided as a Source Data file.

Fig. 4. circITGB6 interacts with IGF2BP3 to promote tumor metastasis.

a Coomassie blue staining of circITGB6-interacting proteins pulled down by the antisense (AS) DNA probes from A549 cell lysate. (S) and AS represent the DNA probes identical and complementary to circITGB6 sequences crossing back-splicing junction site, respectively. The band shown by red arrow was subjected to mass spectrometry. b Representative IGF2BP3 peptide detected by mass spectrometry in circITGB6 pull-down fractions. c, d Immunoblot assays of IGF2BP3 pulled down by DNA probes (c) or in vitro transcribed and circularized circITGB6 (d) from A549 or SW620 cell lysates. SRSF1 served as a negative control. Enrichment efficiency was determined by detecting circITGB6 abundance in the precipitation (c). e RIP assays showed the association of IGF2BP3 with circITGB6 in A549 or SW620 cells using anti-IGF2BP3 antibody or the control IgG. CD44 served as the positive control. TUBA1B served as the negative control. f Immunoblotting assays of IGF2BP3 pulled down by DNA probes in A549 or HCT116 cells with or without TGFβ treatment. SRSF1 served as the negative control. Enrichment efficiency was determined by detection of circITGB6 abundance in the precipitation. g, h Immunoblot assays of EMT markers (g) and Transwell migration assays (h) for circITGB6 overexpression A549 or SW620 stable cells with or without IGF2BP3 knockdown. Scale bar: 20 μm. Source data are provided as a Source Data file.

Fig. 5. Identification of the IGF2BP3-binding site within circITGB6 sequence.

a Schematic representation of two putative IGF2BP3-binding regions within circITGB6 that contain stretches of CAUH (H = A, C, or U) element (colored in red). The 3′-end biotin-labeled probes (M1-WT, M1-mut (mutated nucleotides shown in blue) and M2) were synthesized. b RNA pull-down assays using different RNA probes with A549 cell lysates. The amounts of probes were monitored by the HRP-streptavidin. c, d In vitro RNA pull-down assays (c) and RNA-EMSA assays (d) showing the direct binding of purified recombinant IGF2BP3 protein with biotin-labeled M1 probe. e (Upper) in vitro RNA pull-down assays showing the enrichment of IGF2BP3 by biotin-labeled circITGB6 in the presence of non-biotin probes (NC, M1, M2 or M1-mut). (bottom) Relative binding intensity derived from signal quantification of three independent experiments (mean ± sd), Band intensities were normalized to control NC samples for each experiment (set at a value of 1.0). f RNA pull-down assays using the purified recombinant IGF2BP3 protein and in vitro circularized wild-type (WT) or mutant circITGB6. g RIP assays were performed with anti-IGF2BP3 antibody or the normal IgG in A549 cells stably expressing circITGB6 WT or IGF2BP3-interaction deficiency mutant (mut), and RT-PCR assays for the enrichment of WT or mutant circITGB6. h RNA pull-down assays from different A549 stable cells using DNA probes. S and AS represent the DNA probe identical and complementary to circITGB6 sequence crossing back-splicing junction site, respectively. i, j Immunoblot assays of EMT markers (i) and Transwell migration assays (j) in SW620 cells stably expressing circITGB6 WT or mutant. k Representative bioluminescence images and bioluminescence quantitation of tumor metastasis in liver. n = 8 mice (vec or mut) or 9 mice (WT) per group. The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data. Data represent mean ± s.e.m. Significance of differences was determined by one-way ANOVA test. l Representative images for H&E staining and the foci of circITGB6 (red point) of mouse liver tissues. The metastatic nodules indicated by the black arrows. Scale bar: 20 μm. Source data are provided as a Source Data file.

Fig. 6. circITGB6 regulates PDPN mRNA stability through its interaction with IGF2BP3.

a qPCR assays for CD164, CD44s, MMP9 and PDPN expression in Vec, circITGB6-WT and circITGB6-CD SW620 stable cells. b Representative IHC staining images using anti-PDPN antibody of the metastatic lesions in the lung from mice tail-vein injected with A549 stable cells. Scale bar: 20 μm. c qPCR assays for PDPN mRNA expression in A549 and MCF10A with or without TGFβ treatment. d Measurement the half-life of PDPN mRNA in Vec, circITGB6-WT and circITGB6–mut A549 cells treated with 5 μM of actinomycin D (ActD) for indicated times. e RIP assays with anti-IGF2BP3 or the normal IgG antibody, and the resultant precipitates were subjected to RT-PCR assays. TUBA1B and GAPDH served as the negative controls. f Top: Schematic diagrams showing RNA-binding domains within IGF2BP3 and the list of different IGF2BP3 truncation mutants. Bottom: Immunoblot assays with anti-Flag antibody after performing RNA pull-down experiments using DNA probes specifically targeting circITGB6 or PDPN mRNA in A549 cells transfected with plasmids encoding Flag-tagged full-length (FL) or truncated IGF2BP3 mutants. g–i Stably expressing vector or circITGB6 A549 cells (1 × 10⁶) with or without PDPN knockdown were tail-vein injected into mice for metastasis assays. n = 8 mice per group. Representative bioluminescence images (g), bioluminescence quantitation (h) and H&E staining images of metastatic lung tissues (i) were presented. The metastatic nodules indicated by the black arrows. The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data. Data represent mean ± s.e.m. j Representative IHC staining images of TGFB1, E-cadherin, N-cadherin, PDPN, IGF2BP3 and the foci of circITGB6 (red point) in the primary within the spleen or metastatic lesions within the liver in mice intra-splenically injected with SW620 cells. Scale bar: 20 μm. Data represent mean ± sd from three independent experiments (a, c, d). Significance of differences was calculated by two-tailed Student’s t test (a, c, d) and one-way ANOVA test (h). Source data are provided as a Source Data file.

Fig. 7. Targeting circITGB6 effectively inhibits tumor metastasis.

a, b qPCR assays (a) and Pearson Correlation (b) of circITGB6 and PDPN expression in non-metastatic (NM) and metastatic (M) CRC tumor tissues (n = 53 samples, cohort 2). c, d qPCR assays (c) and Pearson Correlation (d) of circITGB6 and PDPN expression in non-metastatic and metastatic lung tumor tissues (n = 118 samples). The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data (a, c). e–h SW620 cells were treated with PEI-coated control siRNA (siNC) or circITGB6 siRNA (siRNA). qPCR assays for measuring the expression of circITGB6, ITGB6 and PDPN in SW620 cells (e), immunoblot assays of EMT markers (f), Transwell migration assays (g) and measurement of PDPN mRNA half-life (h) in SW620 cells. Scale bar: 20 μm. i–I Mice were intra-splenically injected with (2 × 10⁶) SW620 cells stably expressing luciferase to establish liver metastases. Two weeks later, mice were randomly divided into two groups and intravenously injected twice a week for additional 3 weeks with PEI-coated siNC or siRNA targeting circITGB6 (6 mg/kg body weight). Representative bioluminescence images of liver metastasis (i, left) and bioluminescence quantitation in vivo (i, right). The horizontal lines represent the median; the bottom and top of the boxes represent the 25th and 75th percentiles, respectively; and the vertical bars represent the range of the data. n = 8 mice per group. Representative images of the foci of circITGB6 (red point) and the IHC staining of PDPN, E-cadherin, N-cadherin, IGF2BP3 and ITGB6 in metastatic lesions (j). Photos of the livers and representative images of H&E staining metastatic liver tissues (k). the metastatic nodules were indicated by the black arrows (k). The metastatic nodules indicated by the black arrows. Scale bar: 20 μm (j); 4 mm (k). Kaplan–Meier curve for the overall survival of mice (l). Data represent mean ± sd (e, h) from three independent experiments, mean± s.e.m. (i). Significance of differences was determined by one-way ANOVA test (a, c, i), two-sided Pearson Correlation test (b, d), two-tailed Student’s t test (e) and the log-rank test (l). m Schematic illustration of circITGB6 molecular mechanism. Upon TGFβ stimulation, more circITGB6 are expressed to stabilize PDPN mRNA molecules through their interactions with IGF2BP3, thus facilitating EMT process and promoting tumor metastasis. Source data are provided as a Source Data file.