Fig 2: Regulation of TGFß signaling and levels, fibronectin protein, and mRNA A, BGSEA enrichment maps from tumors overexpressing hepsin (DOX+) and control tumors (DOX-).CGraph depicting quantification of Tgfß1 mRNA levels (mean ± SD mRNA read counts), normalized to the housekeeping gene Pum1, in mammary tissue of 6-week-old virgin female Wt and Hpn -/- mice. Student’s t-test was used for statistical analyses (n.s. = not significant, N = 3 indicates the number of biological replicates).DImmunoblot analysis of PAI-1, hepsin, and ß-tubulin (loading control) in MCF10A-pIND20-HPN cells treated with doxycycline, galunisertib, and latent-TGFß (small latent complex) as indicated. Numbers below the PAI-1 blot indicate PAI-1 levels, normalized to ß-tubulin.EImmunoblot analysis of TGFß-V5 (antibody detecting V5), pSmad2/3, PAI-1, and GAPDH (loading control) in cell lysates from MCF10A-pIND20-HPN cells with or without TGFß-V5 overexpression.F–ISilver-stained protein gels with samples from in vitro protease activity assays with recombinant hepsin and fragments of fibronectin. The respective fragments (30 kDa (F), 45 kDa (G), 40 kDa (H), and 120 kDa (I); marked by arrowheads) were incubated (1 µg/reaction) with increasing concentrations of recombinant hepsin (nM).JSchematic mapping of the fibronectin fragments used in (G–J) onto full-length fibronectin. RGD indicates the RGD-binding domain in fibronectin.KGraph depicting quantification of Fn1 mRNA levels (mean ± SD mRNA read counts), normalized to the housekeeping gene Pum1, in mammary tissue of 6-week-old virgin female Wt and Hpn -/- mice. Student’s t-test was used for statistical analyses (n.s. = not significant, N = 3 indicates the number of biological replicates).

Fig 3: Overexpression of Hpn in Wap-Myc-driven mammary tumors induces TGFß signaling Schematic representation of the mouse experiment.Immunoblot analysis of Wap-Myc mammary tumor lysates for the indicated TGFß signaling markers and hepsin (T# denotes individual tumors). GAPDH was used as the loading control. Lysates were derived from Wap-Myc mammary tumors from six mice with and six mice without DOX-induced hepsin overexpression (see (A)).Gene Set Enrichment Analysis (GSEA) enrichment map of pathways upregulated in hepsin overexpressing Wap-Myc tumors (DOX+) compared with control tumors (DOX-) (N = 5 tumors in each group). Node size correlates with the number of genes in the signature; node color red correlates with enrichment in hepsin overexpressing Wap-Myc tumors.GSEA graph comparing the expression of the HALLMARK_TGF_BETA_SIGNALING gene set in hepsin overexpressing (DOX+) to control (DOX-) Wap-Myc tumors (N = 5 tumors per group; FDRp—P-value; FWERq—false discovery rate; NES—normalized enrichment score).Heatmap showing changes in expression of all genes in the HALLMARK_TGF_BETA_SIGNALING gene set in hepsin overexpressing (DOX+) compared with control (DOX-) Wap-Myc tumors. Red color indicates upregulation, and blue color indicates the downregulation of the indicated genes.

Fig 4: Hepsin-mediated TGFß1 signaling through the downregulation of fibronectin Immunoblot analysis of fibronectin, pSmad2/3, total Smad2/3, and GAPDH (loading control) in cell lysates from MCF10A-pIND20-HPN cells without hepsin overexpression. The graph depicts quantification of pSmad2/3 normalized to total Smad2/3 levels, compared to siCtrl (N = 3 biological repeats).Immunoblot analysis of cell lysates (pSmad2/3, total Smad2/3, and vinculin as loading control) and concentrated cell culture supernatant (anti-V5 for TGFß) from MCF10A-pIND20-HPN pL6-TGFß1-V5 cells without hepsin overexpression. Ponceau staining is shown as a loading control for the Western blot with concentrated culture supernatant. The graph depicts quantification of pSmad2/3 normalized to total Smad2/3 levels, compared to siCtrl (N = 3 biological repeats).TGFß luciferase reporter assay of control or fibronectin silenced MCF10A-pIND20-HPN pL6-TGFß1 (SRE) TGFß1 reporter cells with (DOX+) or without (DOX-) hepsin overexpression (N = 7 biological repeats; Y-axis shows fold change in relative light units (RLU)).Immunoblot analysis of cell lysates (fibronectin and ß-tubulin (loading control)) and concentrated cell culture supernatant (TGFß) from Hs578T cells with knockdown of fibronectin. Ponceau staining is shown as the loading control for the Western blot with concentrated culture supernatant. A representative blot from three biological repeats is shown.Immunoblot analysis of fibronectin, pSmad2/3, total Smad2/3, hepsin, and GAPDH (loading control) in cell lysates from MCF10A-pIND20-HPN cells with (DOX+) and without (DOX-) hepsin overexpression, and with or without fibronectin silencing (siFN1). The graph depicts quantification of pSmad2/3 normalized to total Smad2/3 levels (N = 3 biological repeats).TGFß luciferase reporter assay of control or fibronectin silenced MCF10A-pIND20-HPN pL6-TGFß1 (SRE) TGFß1 reporter cells, with (DOX+) and without (DOX-) hepsin overexpression. (N = 6 biological repeats; Y-axis shows fold change in relative light units (RLU)).Model figure depicting how hepsin regulates TGFß signaling. Under Wt conditions, hepsin promotes degradation of fibronectin, which releases latent-TGFß from ECM stores, thus resulting in the induction of TGFß signaling. In hepsin knockout mammary glands, TGFß signaling is compromised as latent-TGFß cannot be released from ECM stores (Created with BioRender.com). Statistical analyses in (A, B, C, E, F) were done using Student’s t-test. Data in (A, B, C, E, F) are presented as dot plots where black lines represent the mean. Source data are available online for this figure.

Fig 5: The knockout of hepsin with CRISPR/Cas9 inhibits TGFß signaling in the mouse mammary gland Hepsin knockout mice harbor a 50 bp frameshift deletion in the 4th exon of the Hpn gene (TM = transmembrane domain, SRCR = scavenger receptor cysteine-rich domain, SPD = serine protease domain; red bar indicates gRNA-binding site). Immunoblot from whole mammary lysates against hepsin protein (Hpn +/+ = Wt, Hpn +/- = heterozygous deletion, Hpn -/- = homozygous deletion) (representative of 3 mice per group).Weight of indicated organs isolated from Wt (N = 3) and Hpn -/- (N = 4) mice.Acoustic startle reflex test to compare hearing ability between Wt (N = 7) and Hpn -/- (N = 4) mice.Cytoscape enrichment map of pathways affected in Hpn -/- whole mammary glands compared with Wt controls (N = 3 per group). Node size correlates with the number of genes in the signature; node color correlates with either gene set enrichment (red) or reduction (blue) in Hpn -/- mammary glands. A full list of gene signatures affected in Hpn -/- mammary glands is shown in Table EV1.Gene Set Enrichment Analysis (GSEA) graphs showing enrichment of indicated TGFß1 signaling gene sets in Hpn -/- mammary glands compared with Wt mammary glands (FDRp—P-value; FWERq—false discovery rate; NES—normalized enrichment score).Immunoblot analysis of phospho-Smad2/3 (TGFß pathway signaling marker) and total Smad2/3 in lysates from indicated tissues isolated from Wt, Hpn +/-, and Hpn -/- mice. GAPDH was used as the loading control. The histogram depicts quantification of pSmad2/3 compared to Wt, normalized to total Smad2/3.Representative Carmine alum stained mammary gland whole mounts from Wt and Hpn -/- mice. The histogram depicts quantification of duct length normalized to duct length in Wt mammary glands. The scale bar represents 1 mm. Data in (B, C, F, G) are represented as mean ± SD, and Student’s t-test was used for statistical analyses. Source data are available online for this figure.