Fig 2: PRPF4B, BUD31, and BPTF depletion modulates ECM-receptor signaling. a Over-representation analysis of genes with decreased expression levels (log2FC change < -1) after PRPF4B knockdown in Hs578T and MDA-MB-231 cells using pathways annotated in the KEGG database. b Gene Set Enrichment Analysis (GSEA) identifies the ECM-receptor interaction pathway as significantly enriched in downregulated genes after PRPF4B knockdown in Hs578T and MDA-MB-231 cell lines. c Hierarchical clustering (Euclidean distance, complete linkage) of log2FC in expression levels of genes involved in the KEGG ECM-receptor interaction pathway after knockdown of candidates demonstrates that many genes involved in this pathway are downregulated after candidate knockdown. d LAMA5, ITGB1, and ITGA3 expression upon depletion of PRPF4B, BUD31 and BPTF (mean + s.d of three biological replicates, one-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001). e Effect of PRPF4B, BUD31 and BPTF depletion on levels of different ECM and focal adhesion components in MDA-MB-231 cells analyzed with western blot. f Effect of indicated gene depletion on focal adhesion and actin cytoskeleton organization. Hs578T cells were fixed and stained against the actin cytoskeleton, p-FAK (Y397). Scale bar is 50 µm

Fig 3: Candidate genes directly affect TNBC cell migratory behavior. a Quantification of single cell migration speed of Hs578T-GFP cells after knockdown of validated hits. Hs578T-GFP cells were transfected with siRNAs and cell migration was assessed by live microscopy. Hits that showed significant and consistent effects in both replicates were considered as candidate genes (right panel). Median ± 95% confidence interval is shown and cell populations were compared by Kruskal–Wallis test with Dunn’s post correction test. b Same as in a, with MDA-MB-231-GFP cells. c Single cell trajectories of cell migration upon knockdown of PRPF4B, MXD1, BUD31, or BPTF in (i) Hs578T and (ii) MDA-MB-231 cell lines

Fig 4: PRPF4B is essential for breast cancer metastasis formation in vivo. a mRNA and b protein expression of PRPF4B in shPRPF4B MDA-MB-417.5 cell line compared to shCtrl#1 and shCrtl#2 MDA-MB-417.5 cell lines. Mean + s.d. of three biological replicates. Statistical significance was determined using one-way ANOVA correcting for multiple testing. *p < 0.05, **p < 0.01. c Immunofluorescent staining for PRPF4B (green) and DNA (DAPI; blue) in shCtrl#1, shCtrl#2 and shPRPF4B MDA-MB-417.5 cell lines. Scale bar = 50 µm. d Tumor growth of shPRPF4B, shCtrl#1, and shCtrl#2 MDA-MB-417.5 cells engrafted in the mammary fat pad of Rag2–/– Il2rg–/– mice (n = 7–8 animals per group). e Dissemination of shPRPF4B, shCtrl#1 and shCtrl#2 MDA-MB-417.5 cells in the lung, liver, spleen, kidney, heart, uterus, and axillar lymph node as determined by ex vivo bioluminescent analysis (total flux (p/s)) of different target tissues (n = 7–8 animals per group). f Number of lung metastases for shPRPF4B, shCtrl#1, and shCtrl#2 injected mice as determined by macroscopic evaluation of lungs injected with ink (n = 7–8 animals per group). g Images of ink injected lungs (left) and bioluminescent signal in the lungs (right) of all individual mice. Blanc indicates mice that did not receive MDA-MD-417.5 cells. Groups were compared by Kruskal–Wallis test with Dunn’s post correction test (luminescence) or ANOVA (metastasis count). *p < 0.05, **p < 0.01

Fig 5: Model for how PRP4K regulates Hpo signaling. PRP4K phosphorylates Yki/Yap in the nucleus to promote its cytoplasmic translocation whereas the Hpo/Mst1/2-Wts/Lats1/2 kinase cassette phosphorylates Yki/Yap to promote its binding to 14-3-3, resulting in sequestration of Yki/Yap in the cytoplasm