Fig 2: Depletion of TREX/TAP/SRSF1/SRSF7 blocks export of the NKILA reporter. (A) Western blot showing the relative level of proteins in MCF-7 cells after transfection of siRNA targeting UAP56 (and its homolog URH49), TAP, THOC2, and ALYREF. Tubulin was used as the loading control. (B) RNA-FISH showing the localization of NKILA in knockdown cells. Scale bar = 5 µm. C) Percentage of cells with predominant cytoplasmic signal after transfection of appropriate siRNA (n = 3, 200–300 cells were counted for each transfection). ***P < 0.001. (D) Silver staining of RNP from CAR-N and control sequence after separation on SDS-PAGE. (E) Western blot showing the relative level of SRSF1 and SRSF7 in MCF-7 cells after treatment with appropriate siRNA. (F) Quantification of SRSF1 and SRSF7 protein level on the western blot in E (n = 3), ***P < 0.001. (G) RNA-FISH for NKILA localization in cells treated with siRNA. Scale bar = 10 µm. (H) Percentage of cells with apparent cytoplasmic signal in RNA-FISH (n = 3, 200–300 cells were counted in each transfection). ***P < 0.001.

Fig 3: SRSF1/7 interacts with NKILA and components of the TREX complex. (A) RT-PCR analysis of NKILA in RIP with SRSF1/7 antibodies after full-length NKILA and NKILA lacking SRSF1/7 binding sites were transfected into MCF-7 cells. IgG was used as the control. (B) Quantification of RT-PCR bands in A from two independent experiments. ***P < 0.001, **P < 0.01. (C) Western blot showing the relative enrichment after protein IP. IgG was used as the control antibody. Nuclear extract from MCF-7 cells was used for IP with ALYREF and SRSF1/7 antibodies and UAP56, ALYREF, SRSF1 and SRSF7 were checked by western blotting analysis. (D) Total peptide numbers identified by mass spectrometry in IP samples. (E) RT-PCR analysis of NKILA in RIP performed with the ALYREF antibody after treatment of MCF-7 cells with siNC and siSRSF1/7. Right panel showing the mean of NKILA detected from two independent experiments. *P < 0.05. (F) As for E, except RIP was performed with the TAP antibody. *P < 0.05.

Fig 4: SRSF1 regulated the proliferation of GSCs by binding to circATP5B and upregulating circATP5B expression. a The relative expression of circATP5B after SRSF1 knockdown or overexpression was detected by qRT-PCR. (GSC406: p < 0.01; GSC201: p < 0.05; One-Way ANOVA). b, c The RIP assay was performed after SRSF1 knockdown (b) or overexpression (c), followed by qRT-PCR to detect the enrichment of circATP5B in GSCs. (GSC406: p < 0.001; GSC201: p < 0.01; One-Way ANOVA). d, e The RNA pull-down assays showed the SRSF1 protein immunoprecipitation with circATP5B as detected by western blotting. f MTS assays showed that SRSF1 knockdown or overexpression affected the cell viability of GSCs and was reversed by circATP5B overexpression or knockdown, respectively. (GSC406: p < 0.001; GSC201: p < 0.001; One-Way ANOVA). g The relative expression correlation between HOXB5 and SRSF1 in 70 cases of glioma patients were detected by qRT-PCR. (Total: r = 0.6606, p < 0.0001; Grade II: r = 0.4726, p = 0.0354; Grade III: r = 0.5013, p = 0.0107; Grade IV: r = 0.5526, p = 0.0042; Pearson’s correlation analyses). h The EDU assays showed that SRSF1 knockdown or overexpression affected the proliferation of GSCs and was reversed by circATP5B overexpression or knockdown, respectively. Scale bar = 100 µm. (GSC406: p < 0.01; GSC201: p < 0.01; One-Way ANOVA). j The neurospheres formation assays showed that SRSF1 knockdown or overexpression affected the relative size of the neurospheres of GSCs and was reversed by circATP5B overexpression or knockdown, respectively. Scale bar = 20 µm. (GSC406: p < 0.01; GSC201: p < 0.01; One-Way ANOVA). i and k Limiting dilution assays showed that SRSF1 knockdown or overexpression affected the neurosphere-forming capacity of GSCs and was reversed by circATP5B overexpression or knockdown, respectively (GSC406: p < 0.01; GSC201: p < 0.05; ELDA analysis; circles represent corresponding points, triangles mean the point is outside of the log fraction number wells). l Schematic diagram of the putative HOXB5 binding site in the 3'-UTR of SRSF1. m and n The luciferase reporter assays showed that HOXB5 knockdown or overexpression affected the luciferase activities of SRSF1 in GSCs. (GSC406: p < 0.001; GSC201: p < 0.001; One-Way ANOVA). o The ChIP qRT-PCR showed that HOXB5 bound to the promoter of SRSF1. (GSC406: p < 0.01; GSC201: p < 0.001; One-Way ANOVA). p, q, r qRT-PCR (p) and western blotting (q, r) showed the SRSF1 expression was affected after HOXB5 knockdown or overexpression in GSCs. (GSC406: p < 0.01; GSC201: p < 0.001; One-Way ANOVA). EV: empty vector, OE: overexpression, NC: negative control, KD: knockdown. All data were expressed as the mean ± SD (three independent experiments). *p < 0.05; **p < 0.01; ***p < 0.001

Fig 5: Model for the export of the intronless lncRNA NKILA. CAR-N with cluster of SRSF1/7 binding sites in NKILA recruits SRSF1 and SRSF7, which facilitate TREX recruitment by interaction with UAP56 and ALYREF and mediate the export of NKILA via the TREX-TAP pathway. (Created with Biorender.com)