Uncovering mutation-specific morphogenic phenotypes and paracrine-mediated vessel dysfunction in a biomimetic vascularized mammary duct platform

Abstract

The mammary gland is a highly vascularized tissue capable of expansion and regression during development and disease. To enable mechanistic insight into the coordinated morphogenic crosstalk between the epithelium and vasculature, we introduce a 3D microfluidic platform that juxtaposes a human mammary duct in proximity to a perfused endothelial vessel. Both compartments recapitulate stable architectural features of native tissue and the ability to undergo distinct forms of branching morphogenesis. Modeling HER2/ERBB2 amplification or activating PIK3CA(H1047R) mutation each produces ductal changes observed in invasive progression, yet with striking morphogenic and behavioral differences. Interestingly, PI3Kα^H1047R ducts also elicit increased permeability and structural disorganization of the endothelium, and we identify the distinct secretion of IL-6 as the paracrine cause of PI3Kα^H1047R-associated vascular dysfunction. These results demonstrate the functionality of a model system that facilitates the dissection of 3D morphogenic behaviors and bidirectional signaling between mammary epithelium and endothelium during homeostasis and pathogenesis.

Fig. 1. A vascularized mammary duct platform models native 3D tissue architectures and morphogenesis.

Tissues are at one week of co-culture unless otherwise indicated. a Organotypic microfluidic device consisting of an engineered 3D endothelial vessel (green) and mammary epithelial duct (cyan) in physiologic ECM (magenta), enabling long-term culture and paracrine signaling. Inset: 3D reconstruction of endothelial vessel (green—VE-cadherin) and epithelial duct (cyan—phalloidin). b Left: Composite stitched maximum intensity projection micrograph of tissue base to medial section of a mammary epithelial duct (cyan, phalloidin) and proximal vasculature (green, phalloidin); Alexa Fluor-labeled collagen I (magenta), DAPI (blue). Top right: Representative maximum intensity micrographs of mature endothelial vessels immunostained for VE-cadherin, ZO-1, actin, and Ki67, as indicated. Bottom right: Individual confocal slice micrographs from 3D epithelial duct midsections immunostained for GM130, Ki67, α6 integrin, E-cadherin, actin, and DAPI, as is indicated in each panel. c Quantification of golgi localization as measured by the nuclear-GM130 axis (n = 52 cells examined from three independent ducts). 0 corresponds to a nuclear-GM130 axis apically oriented toward the lumen, while 180 corresponds to a basal orientation. d Cartoon depicting morphogen gradients established by perfusing growth factors through an acellular vascular channel. Immunographs of quiescent MCF10A ducts exposed to vehicle (DMSO), FGF2 (3 nM), or TGFβ1 (5 ng/ml) gradients for three days. Inset: high magnification slice micrograph of duct terminus immunostained with phalloidin (cyan), DAPI (blue) and Ki67 (magenta). e Quantification of percentage of Ki67 positive nuclei in ducts treated with each morphogen gradient (n = 4, 3, 3 ducts examined across three independent experiments; one-way ANOVA with Bonferroni post test Vehicle vs FGF2 **p = 0.0036, FGF2 vs TGFβ1 **p = 0.002, mean ± s.e.m). All images are representative of at least three independent experiments. Source Data are provided as a Source Data file.

Fig. 2. Vascular morphogenesis driven by mammary paracrine signaling.

a Cartoon of co-culture paracrine experimental setup. MCF10A cells expressing IRES-GFP or VEGF-IRES-GFP were seeded into the duct channels adjacent to an endothelialized vessel. Phase-contrast images of vessels with control (left) and VEGF expressing ducts (right) after two days in co-culture. b Quantification of endothelial vessel diameter after three days of culture with either GFP or VEGF expressing MCF10A cells (n = 6, 9 vessels examined across three independent experiments; two-tailed, unpaired Student’s t test **p = 0.0019). c Quantification of sprout number per endothelial vessel after three days of culture with acellular basal assay medium (BM), GFP expressing ducts (GFP), or VEGF expressing ducts in which DMSO (VEGF + Veh) or 10 µM Semaxanib (VEGF + Sem) was delivered to the vasculature. (n = 9, 6, 9, 8 vessels examined across three independent experiments; one-way ANOVA with Bonferroni post test ***P = 0.000014, **P = 0.0011). d Maximum intensity projection micrographs of tissue base to medial section of endothelial vessels co-cultured with GFP or VEGF ducts and treated with DMSO or 10 µM Semaxanib; phalloidin (green) and DAPI (blue). Arrows indicate nascent angiogenic sprouts. For all plots, values mean ± s.e.m and all images are representative of at least three independent experiments. Source Data are provided as a Source Data file.

Fig. 3. Ductal and vascular consequences resulting from specific breast cancer mutations.

a Left: Cartoon of co-culture paracrine experimental setup with mosaic mutant/wild-type ducts. Ducts were generated from MCF10A cells stably expressing empty vector (EV), wild-type hemagglutinin (HA)-tagged ErbB2 (ErbB2^amp), or constitutively active HA-tagged PI3Kα H1047R (PI3Kα^H1047R) along with each associated vascular compartment. Right: Western blot of lysates from EV, ErbB2^amp, or PI3Kα^H1047R MCF10A. b Composite stitched maximum intensity projection micrographs of endothelial vessels co-cultured for five days along with their corresponding modified ducts. Endothelial vessels: phalloidin (green), DAPI (blue), epithelial ducts: phalloidin (cyan), DAPI (blue). c Diffusive permeability (P_D) of 70 kDa dextran across the endothelial barrier (n = 8, 8, 6 vessels examined across three independent experiments; one-way ANOVA with Bonferroni post test, **P = 0.0049, *P = 0.012) and (d) endothelial vessel diameter in each co-culture setting (n = 8, 8, 6 vessels examined across three independent experiments; one-way ANOVA with Bonferroni post test). e Left: Quantification of cortical actin measured from phalloidin labeled micrographs (n = 12, 13, 13 cells taken from three independent vessel experiments one-way ANOVA with Bonferroni post test, ****P = 7.6845e−06, ***P = 0.0003). Right: Averaged, cross-sectional actin distribution in endothelial vessels co-cultured with specific mutant ducts (n = 11 cell actin profiles taken from three independent vessel experiments). For all plots, values mean ± s.e.m. and all images are representative of at least three independent experiments. Source Data are provided as a Source Data file.

Fig. 4. PI3Kα^H1047R increases IL-6 secretion to drive vascular dysfunction and remodeling.

a Representative western blot and associated quantification of conditioned media from EV, ErbB2^amp, and PI3Kα^H1047R ducts immunoblotted for human IL-6 and IL-6Rα (n = 3 western blots from three independent experiments). b Top right: Experimental schematic of endothelial vessel-conditioned media setup. Maximum intensity projection micrographs of endothelial vessels treated overnight with EV, EV+ recombinant human IL-6 (rhIL-6) (200 ng/ml), or PI3Kα^H1047R conditioned media; phalloidin (green), DAPI (blue). c Diffusive permeability P_D (n = 4, 5, 5, 5 vessels examined across three independent experiments; one-way ANOVA with Bonferroni post test, BM + rhIL-6 vs EV + rhIL-6 *P = 0.019, BM + rhIL-6 vs PI3Kα^H1047R *P = 0.035, EV vs EV + rhIL-6 *P = 0.0104, EV vs PI3Kα^H1047R *P = 0.0205) and (d) relative cortical actin measured from vessels treated overnight with basal assay medium supplemented with 200 ng/ml rhIL-6 (BM + rhIL-6) or EV, EV + rhIL-6, or PI3Kα^H1047R conditioned media (n = 12, 12, 12, 12 cells examined from three vessel experiments; one-way ANOVA with Bonferroni post test, BM + rhIL-6 vs EV + rhIL-6 ***P = 1.32402e−06, BM+rhIL-6 vs PI3Kα^H1047R ****P = 1.08309e−08, EV vs EV + rhIL-6 ****P = 1.0819e−06, EV vs PI3Kα^H1047R ****P = 8.84689e−09). For all plots, values mean ± s.e.m. and all images are representative of at least three independent experiments. Source Data are provided as a Source Data file.