Fig 2: LIN28A is SUMOylated in vivo and in vitro. (A) LIN28A is modified by SUMO1. HA-LIN28A with or without His-SUMO1 and Flag-UBC9 were co-transfected into 293T cells. 48 h later, cells were lysed for precipitation with Ni2+-NTA resin. Western blot analysis was performed with the indicated antibodies. The SUMO1-LIN28A bands by Ni2+-NTA pull down were quantified by imagej software. (B-C) SENP1 removes the SUMOylation of LIN28A. (B) HA-LIN28A and His-SUMO1 were transfected with Flag-SENPs into 293T cells. (C) 293T cells and 293T SENP1 -/- cells were transfected with plasmids encoding HA-LIN28A alone or together with His-SUMO1 for 48 h. Cells were lysed for precipitation with Ni2+-NTA resin. Western blot analysis was performed with the indicated antibodies. The SUMO1-LIN28A bands were quantified by imagej software. (D) LIN28A is SUMOylated in vitro. Plasmid pGEX-4T1-LIN28A with or without pE1E2S1 plasmid was co-transformed into E. coli BL21. Western blot analysis was conducted with anti-SUMO1 antibody after GST-pull down, and the same membrane was also detected with anti-LIN28A and anti-GST antibodies after stripping. (E, F) SUMOylation of endogenous LIN28A occurs naturally in SK-BR-3 cells (E) and T47D cells (F). SK-BR-3 cells or T47D cells were lysed for IP with anti-LIN28A antibody or normal IgG, followed by western blotting with anti-SUMO1 and anti-LIN28A antibodies. (G) SUMOylation of LIN28A is detected in DU145 stably expressing LIN28A cells. DU145-LIN28A cells were lysed for IP and western blot analysis as performed before. (H) SUMOylation of endogenous LIN28A occurs naturally in human LUAD tissue. LUAD tissue was lysed for IP and western blot analysis as performed before. (I) Hypoxia upregulates SUMOylation of LIN28A. 293T cells transfected with HA-LIN28A and His-SUMO1 were cultured in 1% of oxygen condition (hypoxia) for indicated time before being harvested. Ni2+-NTA resin pull down was performed to detect SUMOylated LIN28A. Samples were immunoblotted using the indicated antibodies. The SUMO1-LIN28A bands were quantified by imagej software. (J) Chemotherapy drugs suppress SUMOylation of LIN28A. 293T cells transfected Myc-LIN28A and His-SUMO1 were treated with Hydroxyurea (3 mm), Camptothecin (10 µm), PTX (30 µm), Etoposide (10 µm), Cisplatin (10 µm), and Doxorubicin (1 µm) for 6 h before being harvested. Ni2+-NTA resin pull down was performed to detect SUMOylated LIN28A. Western blot was performed with the indicated antibodies. The SUMO1-LIN28A bands were quantified by imagej software. (K) Cisplatin down-regulates SUMOylation of LIN28A. 293T transfected Myc-LIN28A and His-SUMO1 plasmids were treated with Cisplatin (10 µm) for indicated time. Cells were lysed for precipitation with Ni2+-NTA resin. Western blot analysis was performed with the indicated antibodies. The SUMO1-LIN28A bands were quantified by imagej software. (L) Cisplatin reduces endogenous LIN28A SUMOylation. T47D cells treated with Cisplatin (10 µm) for 12 h were lysed for IP with anti-LIN28A antibody, followed by western blotting with anti-SUMO1 and anti-LIN28A antibodies. The SUMO1-LIN28A bands were quantified by imagej software.

Fig 3: SUMOylation of LIN28A increases its binding to pre-let-7. (A) Knockout of SENP1 increases the interaction of LIN28A with pre-let-7a-1. 293T cells and 293T SENP1 -/- cells transfected with HA-LIN28A and pre-let-7a-1 were lysed for RIP assay. (B) SUMO1 modification enhances the association of LIN28A with pre-let-7a-1. 293T cells transfected with indicated plasmids were used for RIP assays. (C) DeSUMOylation by SENP1 suppresses the interaction of LIN28A with pre-let-7g. 293T cells transfected with indicated plasmids were used for RIP assays. (D) Cisplatin suppresses the interaction of LIN28A with pre-let-7g. DU145 stably expressing HA-LIN28A cells or T47D-shLIN28A re-expressing HA-LIN28A stable cells were treated with Cisplatin (10 µm) for 12 h, and then, RIP assays were performed. (E, F) Pre-let-7s bound to LIN28A-WT were much more than that to SUMO-site mutant LIN28A-K15R. 293T cells transfected with HA-LIN28A or HA-LIN28A-K15R together with indicated pre-let-7 plasmids were used for RIP assays (E). RIP analysis of RNAs associated with HA-LIN28A or HA-LIN28A-K15R from DU145 and T47D stable cell lines (F). RIP assays were carried out with anti-HA (E) or anti-LIN28A (F) antibody. (G) RNA pull-down assay using biotinylated pre-let-7g and lysates from 293T cells transiently expressing HA-LIN28A or HA-LIN28A-K15R. RNA-bound fraction (beads) and unbound fraction (supernatant) were detected by western blotting with anti-HA antibody. LIN28A bands were quantified by imagej software. The schematic diagram of RNA pull-down assay is presented. RNAs extracted from IP complexes were analyzed by qRT–PCR, all RNA signals of RIP were normalized to those of input, then presented by relative binding fold. All data for qRT–PCR are presented as the mean ± SD with triplicates or quadruplicate sets, differences between individual groups as indicated were analyzed using the t-test (two-tailed and unpaired), and P values of < 0.05 (*), < 0.01 (**), <0.001 (***) are considered significant.

Fig 4: Mutation K15R of LIN28A abolishes its oncogenic roles. (A, D) Mutation K15R of LIN28A inhibits cancer cell proliferation. CCK8 cell proliferation assays were performed with DU145 (A) and T47D (D) stable cells. Error bars ± SEM represent three independent experiments with at least triplicate repeats. Differences between individual groups as indicated were analyzed using the two-way ANOVA, and P values of < 0.05 (*) or < 0.01 (**) or < 0.001 (***) are considered significant. (B, E) SUMOylation of LIN28A promotes cancer cell migration. Wound-healing assays were performed on DU145 (B) and T47D (E) stable cell lines. Images were captured at indicated time. The areas of the wound were calculated by imagej software, and the percentage of wound closure was presented by histogram. Error bars ± SD represent three independent experiments. Differences between individual groups as indicated were analyzed using the t-test (two-tailed and unpaired), and P values of < 0.05 (*), < 0.01 (**), or < 0.001 (***) are considered significant. (C, F) LIN28A-K15R loses its ability in promoting cancer cell invasion. The 3D cell culture growth assay was performed to detect the invasive ability of DU145 (C) and T47D (F) stable cell lines. Images were captured at day 7 and shown in the large field of view (upper panel), and the representative images of 3D culture were shown (lower panel).

Fig 5: LIN28A is mainly SUMOylated at K15. (A) Mutation of K15R decreases SUMOylation of LIN28A. HA-LIN28A-WT or different HA-LIN28A mutations and His-SUMO1 were co-transfected into 293T cells. Cells were lysed for Ni2+-NTA pull down, followed by western blotting by indicated antibodies. The SUMO1-LIN28A bands were quantified by imagej software. (B) SUMOylation at K15 of LIN28A is verified in stable DU145 cell lines by the IP method. DU145 cells stably expressing HA-LIN28A-WT or HA-LIN28A-K15R were lysed for IP with anti-LIN28A antibody or normal IgG, followed by western blotting with anti-SUMO1 and anti-LIN28A antibodies. The SUMO1-LIN28A and LIN28A (IP panels) bands were quantified by imagej software. The ratio of SUMO1-LIN28A/LIN28A presents the intensity of SUMOylated LIN28A. (C) Mutation of K15R reduces LIN28A SUMOylation in an E. coli SUMOylation system. Plasmid pGEX-4T1-LIN28A-WT or pGEX-4T1-LIN28A-K15R with pE1E2S1 plasmid was co-transformed into E. coli BL21. Western blot analysis was conducted with anti-SUMO1 antibody after GST-pull down, and the same membrane after stripping was also detected with anti-LIN28A and anti-GST antibodies. The SUMO1-LIN28A bands were quantified by imagej software. (D) Truncated forms show K15 is a major SUMO site of LIN28A. HA-LIN28A-WT or different HA-LIN28A-truncated and His-SUMO1 were co-transfected into 293T cells. Cells were lysed for Ni2+-NTA pull down, followed by western blotting with indicated antibodies. The SUMO1-LIN28A bands were quantified by imagej software.