Extended Data Fig. 3: circFAM53B-transfected DCs were capable of eliciting anti-tumour immune response in vitro.

(a) Heatmap of Z-score normalized log₂(count+1) expression of the selected differential expressed genes in untransfected (-), mock-, linFAM53B- and circFAM53B-transfected DCs (n = 3 independent experiments). (b) Relative quantitation of indicated genes in DCs, as evaluated by RT-qPCR. **P = 0.0088 (IL12A), 0.0012 (CCL3); CD86: ***P = 0.0005 ((-) vs linFAM53B), 0.0006 (linFAM53B vs circFAM53B). (c) The migration of DCs towards PBS, CCL21 and CXCL7 was determined by the Transwell assay. ***P = 0.0001. (d, e) Representative histograms and quantitation of CD86, CD80, HLA-DR (d) and IL-12 expression (e) in DCs, determined by flow cytometry. HLA-DR: **P = 0.0063. CD86: **P = 0.0022. CD80: **P = 0.0098. IL-12: *P = 0.0416, **P = 0.0066, ***P = 0.0002. MFI: mean fluorescence intensity. (f) HLA-I immunoprecipitation followed by MS analysis identified circFAM53B(192-200) as the HLA-I binding peptide from DCs transfected with circFAM53B (n = 3 independent experiments). (g) Percentages of the in vitro primed T cells stained for CD45RO and CCR7 expression, evaluated by flow cytometry. ***P = 0.0002. (h) Quantification of the spot count per 5 × 10⁴ T cells determined by IFNγ ELISpot. (i-k) Flow cytometric analysis for indicated intracellular cytokines (i, j), as well as perforin or GZMB (k) immunostaining, in the in vitro primed T cells of HLA-A*02⁺ (i, k) or HLA-A*11⁺ (j, k) patients. Percentages of the stained CD8⁺ or CD4⁺ T cells are shown. (i) ***P = 0.0008 ((-) vs circFAM53B), 0.0005 ((-) vs tumour lysates). (j) TNF: ***P = 0.0008, **P = 0.0022. IL-2: ***P = 0.0003 ((-) vs circFAM53B), 0.0003 ((-) vs tumour lysates). (k) Perforin: ***P = 0.0002 ((-) vs circFAM53B), 0.0005 ((-) vs tumour lysates). (l) The autologous breast cancer cells were transduced with polybrene only (mock), empty vector (shvec), shGFP, circFAM53B shRNA-1 and shRNA−2 (shcirc-1 and shcirc-2, respectively) and then co-cultured with in vitro primed CTLs. Percentages of CTLs stained for intracellular IFNγ, perforin and GZMB are shown, analysed by flow cytometry. (m) Tumour cell death induced by the in vitro primed T cells of HLA-A*11⁺ patients was examined by PI uptake through flow cytometry. Percentages of the dead tumour cells are shown. (n, o) The CTLs primed by circFAM53B-transfected DCs were rechallenged by circFAM53B WT, circFAM53B KD and “rescued” breast tumour cells MCF-7. The “rescued” breast tumour cells were established by transfecting circFAM53B KD cells with empty vector (vec), full length of circFAM53B (circFAM53B^fl)and truncated circFAM53B^Δ1-117 RNAs, respectively. (n) Percentages of IFNγ stained CTLs are shown, evaluated by flow cytometry. ***P = 0.0003 (mock vs circFAM53B^fl), 0.0005 (circFAM53B^fl vs circFAM53B^Δ1-117). (o) Percentages of the PI⁺ dead tumour cells induced by in vitro primed CTLs are shown. ***P = 0.0001 (circFAM53B WT vs mock), 0.0002 (mock vs circFAM53B^fl), 0.0006 (circFAM53B^fl vs circFAM53B^Δ1-117). Results are mean ± s.d. of n = 3 (b-e, g-o) independent experiments producing similar results. ****P <  0.0001. P values, compared with DCs with indicated treatment (b-e), unprimed T cells (-) (g-k, m), mock-transduced tumour cells (mock) (l), tumour cells with indicated transfection (n, o), were determined by two-tailed one-way ANOVA with Dunnett’s multiple-comparisons test (g-m) and Tukey’s multiple-comparisons test (b-e, n, o).

Source Data