Fig 2: Requirement of the non-canonical STAT5 site in the Csn1s2b distal enhancer.a Diagram of the enhancer deletions introduced using CRISPR/Cas9 genome editing. The non-canonical STAT5 motif was deleted in combination with other enhancer motifs (?N/S3, ?N/S1/3 and ?S2/3). The canonical GAS motifs S1 and S2 are shown as burgundy circles, and the non-canonical GAS motif S3 is shown in yellow. NFIB binding site is shown in orange. b Csn1s2b mRNA levels were measured by qRT-PCR in day 10 lactating (L10) mammary tissue of WT and mutant mice and normalized to Gapdh levels. Results are shown as the means ± s.e.m. of independent biological replicates (WT, n = 5; ?N/S3, ?N/S1/3 and ?S2/3, n = 4). One-way ANOVA followed by Dunnett’s multiple comparisons test was used to evaluate the statistical significance of differences between WT and each mutant mouse. p-value = 0.0001, 0.0001 and 0.0001, respectively. c The consequences of enhancer deletions were confirmed by STAT5A, NFIB, H3K27ac and Pol II ChIP-seq analysis in WT and mutant tissues at L10. The Cish locus served as control (Supplementary Fig. 3). d STAT5, NFIB, H3K27ac and Pol II coverage was calculated after variation between data set was normalized by Cish promoter coverage. Results are shown as the means ± s.e.m. of independent biological replicates (n = 2 to 4).

Fig 3: Redundant and non-redundant functions of STAT5 and NFIB sites in the Csn1s2b distal enhancer.a Genomic feature of the Csn1s2b distal enhancer (DE) and diagram of the deletions introduced in the mouse genome using CRISPR/Cas9 genome editing. TF binding sites were mutated individually (?S2 and ?N) or in combination (?N/S2). While S1 and S2 display canonical GAS motifs (burgundy circles), S3 (yellow circle) is a non-canonical sequence with a 4bb spacer. N (orange circle) is a NFIB binding site. b Csn1s2b mRNA levels in day 10 lactating (L10) mammary tissues from WT and mutant mice were measured by qRT–PCR and normalized to Gapdh levels. Results are shown as the means ± s.e.m. of independent biological replicates (WT and ?N, n = 5; ?S2 and ?N/S2, n = 6). One-way ANOVA followed by Dunnett’s multiple comparisons test was used to evaluate the statistical significance of differences between WT and each mutant mouse line. p-value = 0.21, 0.63 and 0.0006, respectively. c The Csn1s2b locus was profiled using ChIP-seq data of WT and mutant tissue. d STAT5, NFIB and H3K27ac coverage was calculated after variation between data set was normalized with Cish promoter coverage. Results are shown as the means ± s.e.m. of independent biological replicates (STAT5, H3K27ac of WT and H3K27 of ?S2, n = 4; NFIB of WT, H3K27ac of ?N and NFIB of ?N/S2, n = 3; STAT5, NFIB of ?S2 and ?N, STAT5 and H3K27ac of ?N/S2, n = 2).

Fig 4: Nfib overexpression accelerates tumour growth Limiting dilution assay: (left panel) quantification of tumour initiation upon orthotopic injection of mice with LM1, LM9, LM1 Nfib KO or LM9 Nfib KO cells. (right panel) Tumour-initiating cell (TIC) frequencies and 95% confidence intervals were calculated using ELDA (Hu & Smyth, 2009) as described in the experimental procedures.Representative images of immunostaining for NFIB of tumours from mice injected with LM or LM Nfib KO cancer cells. Scale bar 100 µm.Bar graph showing tumoursphere numbers over four passages of LM1, LM9, LM1 Nfib KO, and LM9 Nfib KO, n = 3 biological replicates and n = 3 technical replicates, means ± s.d., two-tailed Student’s t-test.Representative images of tumourspheres of LM1, LM9, LM1 Nfib KO, and LM9 Nfib KO cancer cell lines at the first passage. Scale bar 100 µm. Source data are available online for this figure.

Fig 5: Nfib is sufficient to induce metastasis Kaplan–Meier plot depicting metastasis onset after tumour removal in mice injected orthotopically with HR1 gCtrl (n = 13) or HR1 g4Nfib (n = 14), two-tailed log-rank test. Nfib overexpression in HR1 parental cells enhances metastasis in the orthotopic model. Representative bioluminescence images (left panel) and bar plot quantification (right panel) of lung metastases at day 20 (HR1 gCtrl) and day 2 (HR1 g4Nfib) after primary tumour removal. n = 4 mice, means ± s.d., two-tailed Student’s t-test. Nfib overexpression in the HR1 parental cells enhances experimental metastases. Kaplan–Meier showing metastatic incidence of animals inoculated i.v. with HR1 gCtrl (n = 11) or HR1 g4Nfib (n = 11) cells, two-tailed log-rank test.Representative bioluminescence images (left panel) and bar plot quantification (right panel) of lung metastases 16 days after i.v. injection of HR1 gCtrl or HR1 g4Nfib cells. n = 4, means ± s.d., two tailed Student’s t-test. NFIB overexpression enhances experimental metastases formation in the SUM159PT model. Kaplan–Meier survival analysis of animals inoculated i.v. with SUM159PT gCTRL (n = 5) or SUM159PT gNFIB (n = 6) cells, two-tailed log-rank test.Representative bioluminescence images (left panel) and bar plot quantification (right panel) of lung metastases 20 days after i.v. injection of SUM159PT gCTRL or SUM159PT gNFIB (n = 5), means ± s.d., two-tailed Student’s t-test. Source data are available online for this figure.