HNRNPL induced circFAM13B increased bladder cancer immunotherapy sensitivity via inhibiting glycolysis through IGF2BP1/PKM2 pathway

Abstract

Background: The response rate to immunotherapy in patients with bladder cancer (BCa) remains relatively low. Considering the stable existence and important functions in tumour metabolism, the role of circRNAs in regulating immune escape and immunotherapy sensitivity is receiving increasing attention.

Methods: Circular RNA (circRNA) sequencing was performed on five pairs of BCa samples, and circFAM13B (hsa_circ_0001535) was screened out because of its remarkably low expression in BCa. Further mRNA sequencing was conducted, and the association of circFAM13B with glycolysis process and CD8⁺ T cell activation was confirmed. The functions of circFAM13B were verified by proliferation assays, glycolysis assays, BCa cells-CD8⁺ T cell co-culture assays and tumorigenesis experiment among human immune reconstitution NOG mice. Bioinformatic analysis, RNA-protein pull down, mass spectrometry, RNA immunoprecipitation, luciferase reporter assay and fluorescence in situ hybridization were performed to validate the HNRNPL/circFAM13B/IGF2BP1/PKM2 cascade.

Results: Low expression of circFAM13B was observed in BCa, and it was positively associated with lower tumour stage and better prognosis among patients with BCa. The function of CD8⁺ T cells was promoted by circFAM13B, and it could attenuate the glycolysis of BCa cells and reverse the acidic tumour microenvironment (TME). The production of granzyme B and IFN-γ was improved, and the immunotherapy (PD-1 antibodies) sensitivity was facilitated by the inhibition of acidic TME. Mechanistically, circFAM13B was competitively bound to the KH3-4 domains of IGF2BP1 and subsequently reduced the binding of IGF2BP1 and PKM2 3'UTR. Thus, the stability of the PKM2 mRNA decreased, and glycolysis-induced acidic TME was inhibited. The generation of circFAM13B was explored by confirming whether heterogeneous nuclear ribonucleoprotein L (HNRNPL) could promote circFAM13B formation via pre-mRNA back-splicing.

Conclusions: HNRNPL-induced circFAM13B could repress immune evasion and enhance immunotherapy sensitivity by inhibiting glycolysis and acidic TME in BCa through the novel circFAM13B/IGF2BP1/PKM2 cascade. Therefore, circFAM13B can be used as a biomarker for guiding the immunotherapy among patients with BCa.

Keywords: Bladder cancer; Glycolysis; IGF2BP1; Immunosensitivity; PKM2; circFAM13B.

Fig. 1

Identification and characterisation of circFAM13B in BCa. A Heat map of circRNA sequencing in five pairs of BCa tissues. B Schematic illustration that shows that circFAM13B was composed of FAM13B exon 8, exon 9 and exon 10, and the back splicing junction was confirmed by sanger sequencing. C The expression of circFAM13B and FAM13B mRNA in T24 and UMUC3 cells treated with or without RNase R were determined by qRT-PCR (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). D The remaining RNA levels of circFAM13B and FAM13B mRNA in T24 cells treated with actinomycin D at different time points were determined by qRT-PCR (*P < 0.05, Student’s t-test). E The expression of circFAM13B in cDNA and gDNA of T24 cells were confirmed by qRT-PCR with the divergent and convergent primers. F The subcellular location of circFAM13B was validated by FISH experiment. G The expression of circFAM13B in seven BCa cell lines and SV-HUC cell line were investigated by qRT-PCR (**P < 0.01, ***P < 0.001, Student’s t-test). H The expression of circFAM13B in 72 pairs of BCa tissues was confirmed by qRT-PCR (**P < 0.01, Student’s t-test). I The relationship of circFAM13B and the overall survival of patients with BCa was confirmed by Kaplan–Meier analysis. Data are expressed as mean ± SD, n = 3

Fig. 2

CircFAM13B promoted the effect of CD8⁺ T cells in vitro. A The efficiency of CD8 + cell screening was validated by flow cytometry by using CD3 and CD8 antibodies. B The morphology of CD8 + T cells before and after activation was observed under a microscope. C Schematic diagram of the co-culture model. D-E ELISA assays showed that CD8 + T cells co-cultured with circFAM13B knockdown T24 or UMUC3 cells secreted less granzyme B. CD8 + T cells co-cultured with circFAM13B overexpression T24 or UMUC3 cells secreted more granzyme B (***P < 0.001, Student’s t-test). F–G ELISA assays showed that CD8 + T cells co-cultured with circFAM13B knockdown T24 or UMUC3 cells secreted less IFN-γ. CD8 + T cells co-cultured with circFAM13B overexpression T24 or UMUC3 cells secreted more IFN-γ (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). H The killing ability of CD8⁺ T cells and the immunotherapy sensitivity of BCa were inhibited when co-cultured with circFAM13B knockdown T24 cells. The killing ability of CD8⁺ T cells and the immunotherapy sensitivity of BCa were increased when co-cultured with circFAM13B overexpressed T24 cells. I The killing ability of CD8⁺ T cells and the immunotherapy sensitivity of BCa were inhibited when co-cultured with circFAM13B knockdown UMUC3 cells. The killing ability of CD8.⁺ T cells and the immunotherapy sensitivity of BCa were increased when co-cultured with circFAM13B overexpressed UMUC3 cells. Data are expressed as mean ± SD, n = 3

Fig. 3

CircFAM13B inhibited the glycolysis of BCa cells by attenuating PKM2 expression. A Heat map of mRNA sequencing in three pairs of circFAM13B overexpression and relative control T24 cells. B Volcano plot of mRNA sequencing that indicates the differentially expressed genes, including PKM2. C GSEA analysis showed that differentially expressed genes were enriched in the glycolysis and immune response pathway. D-E QRT-PCR results indicated that circFAM13B inhibited the mRNA level of PKM2 in T24 and UMUC3 cells (**P < 0.01, ***P < 0.001, Student’s t-test). F Western blot results confirmed that circFAM13B inhibited the protein level of PKM2 in T24 and UMUC3 cells. G. Pearson’s correlation analysis was conducted to validate the correlation of circFAM13B and PKM2. H-I Glucose detection assays indicated that circFAM13B inhibited the glucose intake of T24 and UMUC3 cells (**P < 0.01, ***P < 0.001, Student’s t-test). J-K Lactic acid detection assays confirmed that circFAM13B inhibited the lactic acid production of T24 and UMUC3 cells (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). L-M ATP detection assays indicated that circFAM13B inhibited the ATP production of T24 and UMUC3 cells (**P < 0.01, ***P < 0.001, Student’s t-test). Data are expressed as mean ± SD, n = 3

Fig. 4

CircFAM13B interacts with IGF2BP1 protein via the KH3–4 domain. A RNA–protein pulldown experiment and silver staining assay were conducted to investigate the possible proteins, which could bind circFAM13B. B Mass spectrometry assay indicated that IGF2BP1 was pulled down by the circFAM13B probe. C GO enrichment analysis showed that the proteins pulled down by circFAM13B probe were enriched in the mRNA stability regulation pathway. D The results of StarBase database predictions and mass spectrometry analysis were intersected, and IGF2BP1 binding to both circFAM13B and PKM2 was found. E The binding site of circFAM13B and IGF2BP1 was predicted using the catPAPID algorithm. F RNA–protein pulldown and Western blot assays were performed to confirm that circFAM13B could bind to IGF2BP1. G RIP assays in circFAM13B overexpression and relative control T24 cells were conducted to validate the binding of IGF2BP1 and circFAM13B (**P < 0.01, ***P < 0.001, Student’s t-test). H Full length or truncations of flag-tagged recombinant IGF2BP1 protein vectors were designed and constructed. I The vectors containing full length or truncations of flag-tagged recombinant IGF2BP1 were transfected into T24 cells successfully. J RIP assays and qRT-PCR were carried out by using flag antibodies to determine the region of IGF2BP1 that could bind to circFAM13B. K Nuclear-cytoplasmic fractionation and qRT-PCR were performed to validate whether circFAM13B and IGF2BP1 are located in the nucleus or cytoplasm. L IF-FISH assays were carried out to indicate the co-localisation of circFAM13B and IGF2BP1 in T24 cells. M The expression of IGF2BP1 in BCa tissues and adjacent normal tissues were detected by qRT-PCR (**P < 0.01, Student’s t-test). Pearson’s correlation analysis was conducted to validate the correlation of circFAM13B and IGF2BP1. Data are expressed as mean ± SD, n = 3

Fig. 5

CircFAM13B inhibited the stability of PKM2 by attenuating the binding of IGF2BP1 to PKM2. A The binding sites of IGF2BP1 in the 3’ UTR region of PKM2 were predicted using RBPsuite. B RIP and qRT-PCR assays were performed to indicate the binding of IGF2BP1 protein and PKM2 3’ UTR in circFAM13B overexpression and relative control T24 cells (*P < 0.05, **P < 0.01, Student’s t-test). C RIP assays and qRT-PCR were carried out using flag antibodies to determine the region of IGF2BP1 that could bind to PKM2. D Relative luciferase activity of PKM2 3’ UTR-WT or PKM2 3’ UTR-MT that was detected in IGF2BP1 overexpression or relative control 293 T cells with or without ectopic expression of circFAM13B (***P < 0.001, Student’s t-test). E–F QRT-PCR assays were performed to validate the efficiency of IGF2BP1 siRNAs transfection and expression level of PKM2 in T24 and UMUC3 cells (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). G Western blot analysis was performed to validate the efficiency of IGF2BP1 siRNAs transfection and expression level of PKM2 in T24 and UMUC3 cells. H The remaining PKM2 mRNA levels in IGF2BP1 siRNAs transfected or control T24 and UMUC3 cells treated with actinomycin D at different time points were determined by qRT-PCR (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). I-J The remaining PKM2 mRNA levels in circFAM13B knockdown or overexpression T24 and UMUC3 cells treated with actinomycin D at different time points were determined by qRT-PCR (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). Data are expressed as mean ± SD, n = 3

Fig. 6

IGF2BP1 promoted the glycolysis and immune escape of BCa cells. A-B Glucose, lactic acid and ATP detection assays were performed in IGF2BP1 siRNAs transfected or control T24 and UMUC3 cells (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). C–D ELISA assays were carried out to detect the granzyme B and IFN-γ produced by CD8⁺ T cells, which were co-cultured with IGF2BP1 siRNA transfected or control T24 and UMUC3 cells (**P < 0.01, ***P < 0.001, Student’s t-test). E–F The killing ability of CD8.⁺ T cells and the immunotherapy sensitivity of BCa were increased when co-cultured with IGF2BP1 knockdown T24 or UMUC3 cells. Data are expressed as mean ± SD, n = 3

Fig. 7

IGF2BP1 rescued the decreased PKM2 stability, repressed glycolysis and inhibited immune escape induced by circFAM13B. A-B QRT-PCR assays showed that the overexpression of IGF2BP1 rescued the attenuation of PKM2 mRNA expression caused by circFAM13B in T24 and UMUC3 cells. (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). C Western blot assays showed that the overexpression of IGF2BP1 rescued the attenuation of PKM2 protein expression caused by circFAM13B in T24 and UMUC3 cells. D Actinomycin D treatment assays showed that the overexpression of IGF2BP1 rescued the inhibition of mRNA stability caused by circFAM13B in T24 and UMUC3 cells (**P < 0.01, ***P < 0.001, Student’s t-test). E–F Glucose, lactic acid, and ATP detection assays showed that the overexpression of IGF2BP1 rescued the inhibition of glycolysis caused by circFAM13B in T24 and UMUC3 cells (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). G–H ELISA assays showed that the overexpression of IGF2BP1 weakened the enhancement of granzyme B and IFN-γ production of CD8 + T cells caused by co-culturing with circFAM13B overexpressed T24 and UMUC3 cells (*P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). I–J Overexpression of IGF2BP1 weakened the enhanced CD8.⁺ T cells killing ability and increased the immunotherapy sensitivity of BCa cells caused by co-culturing with circFAM13B overexpressed T24 and UMUC3 cells. Data are expressed as mean ± SD, n = 3

Fig. 8

CircFAM13B overexpression augments the efficacy of anti PD-1 therapy in HuNOG mice model. A Schematic of the huNOG mice model establishment and experimental design to investigate the role of circFAM13B in regulating immune escape and anti-PD-1 therapy efficacy in vivo. B Flow cytometry was performed to detect the positive rate of human CD45 in the HuNOG mice model. C Representative image of HuNOG mice injected with circFAM13B or relative control T24 cells, which were treated with PD-1 or IgG antibodies (n = 5). D Representative image of the tumour formation of HuNOG mice injected with circFAM13B or relative control T24 cells, which were treated with PD-1 antibodies or (n = 5). E The weights of the tumours removed from HuNOG mice were measured using electronic scales (**P < 0.01, ***P < 0.001, Student’s t-test). F The tumour volumes of HuNOG mice were measured every 3 days (***P < 0.001, Student’s t-test). G Representative images of IHC staining of Ki-67, PKM2, CD8 and CD3 in tumour samples removed from HuNOG mice. Data are expressed as mean ± SD, n = 3

Fig. 9

HNRNPL induced the back-splicing of circFAM13B. A RNA–protein pulldown experiment and silver staining assay were conducted to investigate the possible proteins, which could bind the flanking introns of circFAM13B. B Mass spectrometry assay results indicated that HNRNPL was pulled down by the probes of flanking introns of circFAM13B. C The binding sites of HNRNPL in the flanking introns sequences of the pre-FAM13B mRNA were predicted via catRAPID. D RNA–protein pulldown and Western blot assays were performed to confirm that the flanking introns of circFAM13B could bind to HNRNPL. E RIP assay in T24 cells was conducted to validate the binding of HNRNPL and flanking introns of circFAM13B (**P < 0.01, Student’s t-test). F IF-FISH assays were carried out to indicate the co-localisation of circFAM13B and HNRNPL in T24 cells. G QRT-PCR assays were performed to validate the expression level of circFAM13B in HNRNPL knockdown T24 and UMUC3 cells (**P < 0.01, ***P < 0.001, Student’s t-test). H Pearson’s correlation analysis was conducted to validate the correlation of circFAM13B and HNRNPL. Data are expressed as mean ± SD, n = 3

Fig. 10

Mode pattern of the HNRNPL-induced circFAM13B/IGF2BP1/PKM2 regulatory and function network