Aberrant KIF26B expression promotes bladder cancer progression through driving NSUN2 nuclear localization Authors: Jia-Ming Wang, Hai-Yun Xie, Feng-Hao Zhang, Xuan Shu, Jin-Dan Luo, Li-Ping Xie, Jiang-Feng Li Published: 2025 ABSTRACT Abnormal expression of kinesins has been observed in several types of cancer. The present study provided the first evidence that KIF26B could induce RNA m5C modification in bladder cancer (BCa). Detailly, KIF26B interacted with NSUN2 and recruited ubiquitinase STUB1 to promote K63-linked ubiquitination of NSUN2 at K511 site. K63-linked ubiquitination enabled NSUN2 to bind to KPNA1 and translocate into nucleus where it drove RNA m5C modification. Furthermore, KIF26B induced liquid-liquid phase separation (LLPS) of YBX1 and upregulated IL-6 expression through NSUN2/m5C/YBX1 axis. Secreted IL-6 then activated STAT3 signaling to promote transcription of kif26b through direct binding between STAT3 and kif26b promoter. IL-6 also recruited DLAT to acetylate NSUN2 at K229 site and increased association affinity between NSUN2 and KIF26B. Together, these established a KIF26B/NSUN2/m5C/IL-6 positive feedback loop, suggesting that targeting KIF26B may be a promising therapeutic strategy for BCa.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13045-025-01749-4. To the editor RNA m5C modification accelerates BCa progression [1, 2]. The methyl-transferase NSUN2 promotes disease by installing m5C mark on nuclear RNA [1], yet how it enters the nucleus remains unknown. Kinesins power long-distance cargo transport [3], and we previously showed that KIF26B ferried TRAF2 into nucleus of BCa cells [4]. Because KIF26B also mediated the nuclear import of RNA m6A regulators in breast cancer [5], we asked whether it similarly escorted NSUN2 to drive m5C deposition in BCa. Main KIF26B robustly enhanced global RNA m5C levels (Fig. 1A; Suppl. Figure 1A–C). Mechanistically, KIF26B engaged NSUN2 through tyrosine 1784 (T1784) within its C-terminus (Fig. 1B-C; Suppl. Figure 1D–F). Depletion of KIF26B trapped NSUN2 in the cytoplasm; wild-type KIF26B or its C-terminal fragment restored nuclear accumulation, whereas the T1784A mutant failed to do so (Fig. 1D; Suppl. Figure 2A–E). NSUN2 constitutively bound importin-α5 (KPNA1) (Suppl. Figure 3A–C), and KPNA1 knock-down phenocopied KIF26B loss (Fig. 1E; Suppl. Figure 3D). Over-expressed KIF26B drove NSUN2 into the nucleus only when KPNA1 was present, and co-immunoprecipitation showed that KIF26B enabled NSUN2–KPNA1 complex formation (Fig. 1F). Consequently, KPNA1 was indispensable for KIF26B-induced m5C installation (Fig. 1G). Fig. 1KIF26B induces RNA m5C modification via enabling NSUN2 nuclear localization in BCa. (A) Dot blot was conducted to examine total RNA m5C modification after knockdown of kif26b in UM-UC3. (B) Association between KIF26B and NSUN2 was examined using IP in UM-UC3 and T24. (C) Association between KIF26B and NSUN2 was detected using IF in T24 (Left). White arrow indicated representative overlapped dot. Line-scan analyses of fluorescent intensity of NSUN2 and KIF26B in T24 was shown (Right) (Scale bar = 25 μm) (D) IF detected nuclear localization of NSUN2 after knockdown of kif26b in UM-UC3 (Scale bar = 25 μm). (E) T24 was treated with siRNA targeting kpna1 or KIF26B expressing plasmids or Importazole. IF detected nuclear localization of NSUN2 (Scale bar = 25 μm) (F) Wild-type KIF26B (WT) was fragmented to C-terminal (C) and motor domain (motor). Tyrosine 1784 of KIF26B was mutated to alanine to generate T1784A KIF26B (MT). Related plasmids were transfected into kif26b-deficient UM-UC3 and T24. Association between NSUN2 and KPNA1 was examined (G) Related plasmids or siRNA were transfected into UM-UC3 and T24. Dot blot was used to examine total RNA m5C modification (H) Total ubiquitination, K48 or K63-linked ubiquitination as well as interaction between NSUN2 and STUB1 were examined using IP after knockdown of kif26b in UM-UC3 (I) Related plasmids were transfected into UM-UC3. K63-linked ubiquitination of Flag-NSUN2 was detected using IP. K511 site was predicted by https://thebiogrid.org/ (J) Related plasmids or siRNA were transfected into UM-UC3. Cellular localization of Flag-NSUN2 was examined using cyto/nuc extraction assay. Interaction between NSUN2 and KPNA1 was examined using IP (K) Potential intrinsic-disordered regions were predicted using PONDR (Up, https://pondr.com). Schematic showed 3 functional domains of YBX1 (Down) (L) IF detected puncta of YBX1 in fixed and live UM-UC3 (Scale bar = 15 μm) (M) Time-lapse imaging detected the process regarding two YBX1 puncta forming one YBX1 punctum in UM-UC3 (Scale bar = 15 μm) (N) After bleached, IF detected the process of YBX1 punctum recovery in UM-C3 (Scale bar = 15 μm) (O) UM-UC3 and T24 were treated with DMSO or 1,6-hexanediol. IF detected YBX1 puncta (Scale bar = 25 μm) (P) Related plasmids were transfected into kif26b-deficient T24. IF examined YBX1 puncta (Scale bar = 25 μm) KIF26B induces RNA m5C modification via enabling NSUN2 nuclear localization in BCa. (A) Dot blot was conducted to examine total RNA m5C modification after knockdown of kif26b in UM-UC3. (B) Association between KIF26B and NSUN2 was examined using IP in UM-UC3 and T24. (C) Association between KIF26B and NSUN2 was detected using IF in T24 (Left). White arrow indicated representative overlapped dot. Line-scan analyses of fluorescent intensity of NSUN2 and KIF26B in T24 was shown (Right) (Scale bar = 25 μm) (D) IF detected nuclear localization of NSUN2 after knockdown of kif26b in UM-UC3 (Scale bar = 25 μm). (E) T24 was treated with siRNA targeting kpna1 or KIF26B expressing plasmids or Importazole. IF detected nuclear localization of NSUN2 (Scale bar = 25 μm) (F) Wild-type KIF26B (WT) was fragmented to C-terminal (C) and motor domain (motor). Tyrosine 1784 of KIF26B was mutated to alanine to generate T1784A KIF26B (MT). Related plasmids were transfected into kif26b-deficient UM-UC3 and T24. Association between NSUN2 and KPNA1 was examined (G) Related plasmids or siRNA were transfected into UM-UC3 and T24. Dot blot was used to examine total RNA m5C modification (H) Total ubiquitination, K48 or K63-linked ubiquitination as well as interaction between NSUN2 and STUB1 were examined using IP after knockdown of kif26b in UM-UC3 (I) Related plasmids were transfected into UM-UC3. K63-linked ubiquitination of Flag-NSUN2 was detected using IP. K511 site was predicted by https://thebiogrid.org/ (J) Related plasmids or siRNA were transfected into UM-UC3. Cellular localization of Flag-NSUN2 was examined using cyto/nuc extraction assay. Interaction between NSUN2 and KPNA1 was examined using IP (K) Potential intrinsic-disordered regions were predicted using PONDR (Up, https://pondr.com). Schematic showed 3 functional domains of YBX1 (Down) (L) IF detected puncta of YBX1 in fixed and live UM-UC3 (Scale bar = 15 μm) (M) Time-lapse imaging detected the process regarding two YBX1 puncta forming one YBX1 punctum in UM-UC3 (Scale bar = 15 μm) (N) After bleached, IF detected the process of YBX1 punctum recovery in UM-C3 (Scale bar = 15 μm) (O) UM-UC3 and T24 were treated with DMSO or 1,6-hexanediol. IF detected YBX1 puncta (Scale bar = 25 μm) (P) Related plasmids were transfected into kif26b-deficient T24. IF examined YBX1 puncta (Scale bar = 25 μm) KIF26B did not alter NSUN2 abundance (Suppl. Figure 4A-B), but markedly increased its K63-linked ubiquitination (Fig. 1H; Suppl. Figure 4C). The E3 ligase STUB1 selectively catalyzed K63-ubiquitination of NSUN2 without affecting protein levels or K48-ubiquitination (Suppl. Figure 4D–I). KIF26B loss weakened STUB1–NSUN2 interaction, whereas KIF26B over-expression strengthened it in a dose-dependent manner (Fig. 1H; Suppl. Figure 4 J). Catalytically-dead STUB1 abolished KIF26B-mediated NSUN2 ubiquitination and m5C accumulation (Suppl. Figure 4K-L). Mass-spectrometry mapped K511 within the NSUN2 NLS as the STUB1 target site (Fig. 1I; Suppl. Figure 5A–C). The K511R mutant exhibited reduced K63-ubiquitination, impaired KPNA1 binding, and persistent cytoplasmic retention that could not be rescued by KIF26B or STUB1 over-expression (Fig. 1J; Suppl. Figure 5D–E). In the cytoplasm, the m5C reader YBX1 underwent LLPS in a CTD-dependent manner (Fig. 1K–O; Suppl. Figure 6A). KIF26B depletion dissolved these condensates; wild-type KIF26B reinstated them, whereas the T1784A mutant did not (Fig. 1P; Suppl. Figure 6B). Integrative multi-omics identified IL-6 as a direct KIF26B target (Fig. 2A). KIF26B knock-down decreased IL-6 mRNA and protein abundance by shortening transcript half-life (Fig. 2B–D). NSUN2 bound the il6 mRNA (Suppl. Figure 7A), and KIF26B loss diminished m5C modification at this locus (Suppl. Figure 7B). Catalytically-dead (C271A/C321A) or K511R NSUN2 mutants blocked KIF26B-driven up-regulation of IL-6 (Fig. 2E–G; Suppl. Figure 7C). YBX1 stabilized il6 mRNA in an m5C-dependent manner (Suppl. Figure 7D-E), and this enrichment required both KIF26B and wild-type NSUN2 (Suppl. Figure 7F-G). Disruption of YBX1 LLPS abolished KIF26B-mediated IL-6 induction (Fig. 2H; Suppl. Figure 7H). Fig. 2One IL-6/DLAT/KIF26B/NSUN2 feedback loop exists in BCa. (A) RNA-seq was conducted after knockdown of kif26b or nsun2 in UM-UC3. RNA-Bis-seq was conducted after knockdown of kif26b in UM-UC3. IL-6 was picked after intersecting above three sequence data. (B) Transcription level of il6 was examined after knockdown of kif26b in UM-UC3 and T24. (C) Protein level of IL-6 was detected after knockdown of kif26b in UM-UC3 and T24. (D) RNA degradation speed targeting il6 was recorded after knockdown of kif26b in UM-UC3. (E) Related plasmids were transfected into nsun2-deficient UM-UC3 and T24. Protein level of IL-6 was explored. (F) Related plasmids were transfected into nsun2-deficient T24. RNA degradation speed targeting il6 was recorded. (G) Related plasmids were transfected into nsun2-deficient T24. MeRIP using anti-m5C antibody was conducted to explore RNA m5C modification of il6. (H) Related plasmids or siRNA were transfected into UM-UC3 and T24. Protein level of IL-6 was examined. (I) Protein level of KIF26B and phosphorylated STAT3 were examined after UM-UC3 and T24 were treated with IL-6 in a dose-dependent manner. (J) Binding motif of STAT3 was shown. (K) ChIP was conducted to evaluate enrichment of STAT3 to promoter of kif26b in UM-UC3 and T24 under IL-6 treatment. (L) IP analyzed acetylation of NSUN2 and its association with DLAT after IL-6 treatment in UM-UC3 and T24. (M) Related plasmids were transfected into dlat-deficient T24. IP evaluated K63-linked ubiquitination and acetylation of NSUN2 as well as its association with STUB1, KIF26B and KPNA1 under IL-6 treatment (CD: catalytic dead) (N) Related plasmids were transfected into UM-UC3. IP examined K63-linked ubiquitination and acetylation of Flag-NSUN2 as well as its association with STUB1 and KPNA1 (O) Related plasmids or siRNA were transfected into T24. IP was used to examine K63-linked ubiquitination of Flag-NSUN2 and its association with KPNA1 under IL-6 treatment (P) Related plasmids or siRNA were transfected into UM-UC3. IP was used to examine K63-linked ubiquitination of Flag-NSUN2 and its association with KPNA1 under IL-6 treatment. (Q) Related plasmids were transfected into kif26b-deficient UM-UC3 and T24. Cyto/nuc extraction assay determined cellular localization of Flag-NSUN2 under IL-6 treatment (R) Related plasmids were transfected into kif26b-deficient UM-UC3 and T24. Dot blot examined total RNA m5C modification under IL-6 treatment. (S) Related plasmids were transfected into nsun2-deficient UM-UC3 and T24. Dot blot examined total RNA m5C modification under IL-6 treatment. (T) Subcutaneous tumor model was established. (U) Tumor sizes of indicated groups were recorded. Proposed model of this study was shown One IL-6/DLAT/KIF26B/NSUN2 feedback loop exists in BCa. (A) RNA-seq was conducted after knockdown of kif26b or nsun2 in UM-UC3. RNA-Bis-seq was conducted after knockdown of kif26b in UM-UC3. IL-6 was picked after intersecting above three sequence data. (B) Transcription level of il6 was examined after knockdown of kif26b in UM-UC3 and T24. (C) Protein level of IL-6 was detected after knockdown of kif26b in UM-UC3 and T24. (D) RNA degradation speed targeting il6 was recorded after knockdown of kif26b in UM-UC3. (E) Related plasmids were transfected into nsun2-deficient UM-UC3 and T24. Protein level of IL-6 was explored. (F) Related plasmids were transfected into nsun2-deficient T24. RNA degradation speed targeting il6 was recorded. (G) Related plasmids were transfected into nsun2-deficient T24. MeRIP using anti-m5C antibody was conducted to explore RNA m5C modification of il6. (H) Related plasmids or siRNA were transfected into UM-UC3 and T24. Protein level of IL-6 was examined. (I) Protein level of KIF26B and phosphorylated STAT3 were examined after UM-UC3 and T24 were treated with IL-6 in a dose-dependent manner. (J) Binding motif of STAT3 was shown. (K) ChIP was conducted to evaluate enrichment of STAT3 to promoter of kif26b in UM-UC3 and T24 under IL-6 treatment. (L) IP analyzed acetylation of NSUN2 and its association with DLAT after IL-6 treatment in UM-UC3 and T24. (M) Related plasmids were transfected into dlat-deficient T24. IP evaluated K63-linked ubiquitination and acetylation of NSUN2 as well as its association with STUB1, KIF26B and KPNA1 under IL-6 treatment (CD: catalytic dead) (N) Related plasmids were transfected into UM-UC3. IP examined K63-linked ubiquitination and acetylation of Flag-NSUN2 as well as its association with STUB1 and KPNA1 (O) Related plasmids or siRNA were transfected into T24. IP was used to examine K63-linked ubiquitination of Flag-NSUN2 and its association with KPNA1 under IL-6 treatment (P) Related plasmids or siRNA were transfected into UM-UC3. IP was used to examine K63-linked ubiquitination of Flag-NSUN2 and its association with KPNA1 under IL-6 treatment. (Q) Related plasmids were transfected into kif26b-deficient UM-UC3 and T24. Cyto/nuc extraction assay determined cellular localization of Flag-NSUN2 under IL-6 treatment (R) Related plasmids were transfected into kif26b-deficient UM-UC3 and T24. Dot blot examined total RNA m5C modification under IL-6 treatment. (S) Related plasmids were transfected into nsun2-deficient UM-UC3 and T24. Dot blot examined total RNA m5C modification under IL-6 treatment. (T) Subcutaneous tumor model was established. (U) Tumor sizes of indicated groups were recorded. Proposed model of this study was shown Interestingly, IL-6 activated STAT3 to transcriptionally up-regulate KIF26B. STAT3 bound the kif26b promoter at − 599∽−589, and IL-6 enhanced this occupancy (Fig. 2I–K; Suppl. Figure 8A–F). In turn, IL-6 stimulated DLAT-dependent acetylation of NSUN2 without altering its abundance (Fig. 2L; Suppl. Figure 9A-B). K229-acetylated NSUN2 displayed stronger affinity for KIF26B (Fig. 2L-M; Suppl. Figure 9C–G). The K229R mutant failed to interact with KIF26B (Suppl. Figure 9H) and consequently resisted IL-6-induced K63-ubiquitination, KPNA1 binding, and nuclear import (Fig. 2N-P; Suppl. Figure 9I–K). DLAT required KIF26B and STUB1 to potentiate NSUN2 nuclear translocation and m5C deposition (Suppl. Figure 9L-N). Neither K511R nor K229R NSUN2 responded to IL-6 stimulation (Suppl. Figure 10A–D), and blocking KIF26B–NSUN2 interaction phenocopied these mutations (Fig. 2Q–S; Suppl. Figure 10E). Functionally, KIF26B over-expression enhanced colony formation, cell viability, and xenograft growth only when wild-type NSUN2 was present; K229R or K511R mutants were ineffective (Fig. 2T; Suppl. Figure 11A-B). Although SUMOylation has been reported to escort NSUN2 into the nucleus [6], we uncovered a previously unrecognized K63-ubiquitination cascade that drove NSUN2 nuclear import in BCa (Fig. 2U). IL-6-triggered acetylation primed NSUN2 for KIF26B engagement, enabling STUB1-mediated K63-ubiquitination at K511 and subsequent KPNA1-dependent nuclear entry. Moreover, we revealed cytoplasmic YBX1 LLPS, not nuclear LLPS [7], as a downstream consequence of this transport pathway and identified a positive feedback loop whereby IL-6 transcriptionally induced its own upstream regulator KIF26B. Thus, KIF26B acts as a non-canonical kinesin that licenses NSUN2 nuclear function and fuels BCa progression. Supplementary Information Supplementary Material 1 Supplementary Material 1