Fig 2: Tax-A and Tax-B do not differ in their binding to NF-κB proteins. To test whether one of the two amino acid differences between Tax-A and Tax-B, i.e., at position 221, which are located just beside the Tax1 (225-232) motif is involved in the p100 processing, we tested whether this amino acid difference affects the binding affinity to the NF-кB protein, thereby altering the ability to activate the NF-кB pathway and their downstream target genes. Jurkat cells were transfected with each expression plasmid and cells were harvested after 24 h, then immunoprecipitation and western blot analyses were performed. a Western blot analysis performed on immunoprecipitated Tax protein revealed that each subgroup of Tax, i.e., Tax-A and -B, interacted with each NF-κB components, i.e., c-Rel, RelA, RelB, p50/p105, and p52/p100, with similar efficiencies. b The amounts of p100, p52, RelB, p65, p105, p50, and c-Rel proteins in immunoprecipitated samples were quantified by densitometric scanning of corresponding bands of the western blot using Image J software. Vertical bars indicate mean ± SD of the densitometric analysis from four independent experiments

Fig 3: PAK4 inhibits senescence-like growth arrest via the NF-?B subunit RELB. a Column clustered heatmaps of NF-?B (RELA) target genes that are differentially expressed in Hs 578T cells upon siPAK4. Genes are by column and samples by row. The color intensity represents column Z-score, where red indicates highly and blue lowly expressed. Selected NF-?B target genes are indicated. b RT-qPCR validation of differentially expressed NF-?B target genes in Hs 578T cells. mRNA was prepared 72 h after transfection with the indicated siRNAs (n = 3 per group). c Correlation between PAK4 expression and an NF-?B signature in breast tumors (T) and normal tissue (N) in the METABRIC dataset. d Correlation between the expression of PAK4 and the alternative NF-?B subunit RELB in breast tumors (T) and normal tissue (N) in the METABRIC dataset. e Quantification of EdU+ cells in Hs 578T cells 4 days after transfection with the indicated siRNAs (n = 3 per group). Data are represented as mean ± SEM in b and e. p-values by two-tailed unpaired t test are indicated in b and by one-way ANOVA followed by Tukey's post hoc test in e. The Spearman correlations and p-values by Spearman’s test are indicated in c and d for tumors (T, blue, n = 1992) and normal tissue (N, gray, n = 144)

Fig 4: RELB–Ser151 phosphorylation controls DNA-binding and C/EBPß transcription. a Structure of RELB:?B DNA complex (PDB ID: 3DO7) highlighting RELB–Ser151 in close proximity to DNA, the RELB subunit shown in purple and DNA in greens. Top inset zooms into Ser151 and shows a hydrogen bond with the DNA backbone. Bottom inset models the disturbing impact on DNA binding of Ser151 phosphorylation. b ELISA-based measurements of RELB DNA-binding 24 h upon expression of the indicated constructs in Hs 578T cells (n = 3 per group). c Representative immunoblot for experiments quantified in b using VINCULIN as loading control. d ELISA-based measurements of RELB DNA-binding in MCF7 cells stably expressing the indicated constructs (n = 4 per group). e ELISA-based measurements of RELB DNA-binding 4 days after the indicated siRNA transfections in Hs 578T cells (n = 3 per group). f Differentially expressed genes from an RT-qPCR array performed in Hs 578T cells transfected with FLAG-tagged RELB–S151E and RELB-wt (n = 3 per group). Statistically discernible hits (Supplementary Fig. 7c) are ordered by Fold Change. g Correlation between the expression of C/EBPß and RELB in tumors (T) and normal tissue (N) in METABRIC. h RT-qPCR analyses of C/EBPß mRNA in Hs 578T cells 72 h after the indicated siRNA transfections (n = 3 per group). i Quantification of EdU+ cells in Hs 578T cells 4 days after the indicated siRNA transfections (n = 3 per group). j Schematic summary diagram. In proliferating cancer cells, abundant PAK4 contributes to senescence avoidance. PAK4 inhibits RELB by direct interaction and phosphorylation of RELB–Ser151 thereby impairing binding affinity to DNA. In the absence of PAK4, RELB accumulates and binds DNA, which culminates in the expression of senescence regulators including C/EBPß. Data are represented as mean ± SEM in b, d, e, h, and i. p-values by one-way ANOVA followed by Tukey's post hoc test in b and by two-tailed unpaired t tests in d, e, h, and i are indicated. The Spearman correlation and p-value by Spearman’s test are indicated in g for tumors (T, blue, n = 1992) and normal tissue (N, gray, n = 144)

Fig 5: Enhanced nuclear staining of RelB in OVCAR8-NUAK1KO xenograft tumours. (A) Representative immunohistochemistry DAB channel images for RelA and RelB in xenograft tumours using OVCAR8 and OVCAR8-NUAK1KO cells. Scale bars represent 100 μm. (B, C) Quantification of DAB staining intensity for RelA (B) and RelB (C) as determined by the IHC Profiler ImageJ plugin. Data are shown as the percentage of pixels classified as “Negative”, “Low Positive”, “Positive”, and “High Positive” for the full image (Total) or nuclear regions as defined by hematoxylin staining (Nuclear). OVCAR8-NUAK1KO were compared with OVCAR8 by unpaired, two-tailed Student’s t-test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001; n ≥ 10 tumour nodules per group). Error bars indicate standard error of the mean (S.E.M.).