Engineering the bone metastatic prostate cancer niche through a microphysiological system to report patient-specific treatment response

Abstract

Bone is the most common site of prostate cancer metastasis, leading to significant morbidity, treatment resistance, and mortality. A major challenge in understanding treatment response is the complex, bone metastatic niche. Here, we report the first patient-specific microphysiological system (MPS) to incorporate six primary human stromal cell types found in the metastatic bone niche (mesenchymal stem cells, adipocytes, osteoblasts, osteoclasts, fibroblasts, and macrophages), alongside an endothelial microvessel, and prostate tumor epithelial spheroids in an optimized media that supports their viability and phenotype. We tested two standard of care drugs, darolutamide and docetaxel, in addition to sacituzumab govitecan (SG), currently in clinical trials for prostate cancer, demonstrating that the MPS accurately replicates androgen response sensitivity and captures stromal microenvironment-mediated resistance. This advanced MPS provides a robust platform for investigating the biological mechanisms of treatment response and for identification and testing of therapeutics to advance patient-specific MPS towards personalized clinical-decision making.

Fig. 1. Stromal cells retain their viability and phenotype in multicellular co-culture media.

A Schematic representation of the MicroDUO platform used for media optimization, wherein, multiple cell types can be co-cultured by bridging media between wells. B Cell viability of epithelial cells co-cultured with iPSC endothelial cells under 6 different media formulations with respect to the monoculture media formulation (n = 3). C Cell viability of epithelial cells and iPSC endothelial cells co-cultured with macrophages under 5 different media formulations with respect to the monoculture media formulation. (n = 3) Media formulations are listed in Supplementary Table 1. D Cell viability of BMMSCs, fibroblasts, osteoblasts, and adipocytes in monoculture under their standard culture media or the multicellular co-culture media. n = 3 experimental replicates. E Representative images and quantification of various phenotype markers were obtained in different cell populations. Cells were differentiated and then cultured for 7 days in either standard differentiation media or multicellular co-culture media (MCM). The staining included DAPI for nuclear staining, CD31 for endothelial cells (n = 3); CD163 for M2 macrophages (n = 4); TROP-2 for primary epithelial prostate cells (n = 4); RANK and phalloidin for osteoclasts (n = 4), Alkaline phosphatase for osteoblasts (n = 3), and LipidSpot for adipocytes (n = 3). Each point indicates a technical replicate. *p ≤ 0.05 **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001(Student’s T-test). Figures show mean ± SD. Scale bar represents 100 µm for endothelial, osteoclast, osteoblast, and adipocyte cell types, and 200 µm for prostate epithelial cells and M2 macrophages. Figure created with BioRender.com.

Fig. 2. Device illustration and representation of the cell differentiation process.

A Visual representation of the cell types used in this model and their relative abundance in the tumor microenvironment along with a B schematic representation of the differentiation process for cell types including in the bone metastasis MPS. C Schematic representation of device assembly. The LumeNEXT devices are fabricated with two layers with a central chamber housing a PDMS rod. The main chamber is filled with a stromal cell/ECM hydrogel solution and allowed to polymerize at RT for 30 min. Removal of the rod enables seeding of iPSC-endothelial cells in the lumen. Finally, media is added to enable co-culture of all seven cell types. D 3D image and E 2D projection of the MPS where stroma cells are stained in magenta, and lumen is stained in green. Scale bars: 300 µm. F Immunofluorescence staining of CD31 (magenta) and DAPI (Blue). Scale bars: 200 µm. Figure created with BioRender.com.

Fig. 3. Stromal cells retain their viability across seven days in the MPS devices.

A Representative images and quantification of stromal cell viability (Calcein A.M; green, PI: red, DAPI:blue) and cell number per field of vision at 10X magnification (Calcein AM; green) across 7 days. n = three experimental replicates *p ≤ 0.05, **p ≤ 0.01 (T-Test + Mann-Whitney test). Scale bars: 200 µm. B UMAP plots showing Harmony integration of single-cell RNA sequencing data at days 4 and 7. Clusters correspond to specific cell types identified by marker genes: endothelial cells (VWF), MSC (POU5F1), unknown, fibroblasts (COL4A1), osteoblasts (COL1A1), M2 macrophages (CD9 and SRGN), and proliferative cells. C. Relative percentage of the different cell types discovered in the MPS at day 4 and 7 by single cell sequencing. D Representative images and quantification of LipidSpot staining (red) for lipid droplets in early adipocyte differentiation, with DAPI (blue) marking nuclei. Scale bars: 200 µm. n = 3 experimental replicates. E Representative images and quantification showing RANK staining (red) for osteoclasts, with DAPI (blue) marking nuclei. Scale bars: 200 µm. n = 3 experimental replicates. F Calcium deposition at day 4 and 7 determined using the colorimetric assay, Stanbio Calcium (CPC) LiquiColor® test. n = 3 experimental replicates. ns = p > 0.05, **p ≤ 0.01 (T-Test + Mann–Withney test). Each point indicates a technical replicate. Figures show mean ± SD.

Fig. 4. Prostate cancer cell characterization.

A Flow cytometric analysis of epithelial and prostate cancer markers, with median fluorescence intensity (MFI) values for each condition shown in parentheses. TROP-2 FMO (Donor 1) served as negative control (gray) for TROP-2 staining. DU-145 staining served as a negative control for PCa biomarker staining, as known to be negative for these markers. Analysis was done in the live cell population. Gating strategy is provided in Supplementary Fig. 4. B Representative image of a MPS with a primary prostate epithelial cell spheroid. Cells were stained with calcein (green), PI (red), and Hoechst (blue), and tumor cells were stained with Cell Tracker (magenta). Scale bar: 400 µm. C Representative images and D quantification of cell viability of the prostate cancer spheroids at day 4 and 7 days of culture in the bone microenvironment MPS. Cells were stained with calcein (green), PI (red), and Hoechst (blue). Scale bars: 200 µm, N = 3 different donors. No statistical significance found through T-Test + Mann–Whitney test. Each point indicates a technical replicate. Figures show mean ± SD.

Fig. 5. Prostate cancer spheroids modify protein secretion in MPS.

A Bone metastasis MPS was assembled on day 0 with 10,000 stromal cells (MSCs, osteoblasts, adipocytes, macrophages, osteoclasts, fibroblasts) seeded in the main chamber. For prostate cancer devices, a spheroid was included in the port of the devices. After 4 days of co-culture, the main chamber was collected for RNA isolation and RT-qPCR analysis of stromal cells. Gene expression fold change of the significantly upregulated and downregulated genes found in the prostate cancer condition compared to the control. n = 3 replicates. Each point represents a technical replicate. *p < 0.05, **p < 0.01, and ***p < 0.001 (Student’s t-test). B Media from the whole device was collected at day 0, 3, and 7, and soluble factors were analyzed with MAGPIX. n = 4. Each data point represents a technical replicate. *p < 0.05, **p < 0.01, and ***p < 0.001 (Student’s t-test).

Fig. 6. Differential responses of androgen receptor-positive and -negative prostate cancer cells to treatment in a bone marrow microenvironment MPS.

A Schematic timeline of killing experiment in the bone metastasis MPS. The experimental timeline consists of five main stages: (1) initial spheroid formation of the MPS with DU-145, LAPC4, and LNCaP spheroids (2) treatment with either 10 µM darolutamide (AR antagonist) or 40 nM docetaxel (chemotherapy agent) over an additional 72-h period, (3) endpoint assessment via microscopy, and (4) data analysis of spheroid viability. B Prostate cancer cell line spheroid cell death across the three different cell lines measured as cell death fold change respective to the control treated with DMSO. n = 5. C, D Prostate cancer cell line spheroid cell death with (bone metastasis chip) and without embedded stromal cells. n = 5 Each point is a technical replicate. *p < 0.05 **p < 0.01, ****p < 0.0001 (A one way ANOVA, B Student’s t-test). Figures show mean ± SD. Figure created with BioRender.com.

Fig. 7. TROP-2 ADC achieves significant killing of prostate cancer cells in 2D and 3D.

A DU-145 cells (TROP-2 medium) were seeded into a 96 well plate and treated with either TROP-2 ADC (SG) or Isotype ADC (5 µg/mL) for 1 h at 37 °C. Wells were subsequently washed three times to remove unbound ADC, and cell death fold change relative to the isotype ADC condition was quantified after 72 h. n = 3 replicates. Each point is a technical replicate. ****p < 0.0001 (Student’s t-test). B LNCaP, DU-145, and LAPC4 prostate cancer spheroids were introduced into LumeNEXT devices without the bone stromal cells and treated with either the TROP-2 ADC or Isotype ADC (5 µg/mL) through the iPSC endothelial cell microvessel for 1 h at 37 °C. Devices were subsequently washed three times to remove unbound ADC, and cell death fold change relative to the isotype ADC condition was quantified after 72 h. n = 3 replicates. Each point represents a spheroid. **p < 0.01, ****p < 0.0001 (One-way ANOVA + Tukey’s post-hoc test). Figures show mean ± SD. Figure created with BioRender.com.

Fig. 8. Sacituzumab-Govitecan killing in the bone metastasis chip is donor dependent.

A Schematic timeline of killing experiment in the bone metastasis chip. B Primary prostate tumor epithelial spheroid and C stromal cell death across three donors with differing TROP-2 expression (medium, and high) at both 5 µg/mL and 10 µg/mL TROP-2 ADC (SG) measured as cell death fold change with respective to the isotype ADC control for each condition. D Primary prostate tumor epithelial spheroid cell death with (bone metastasis MPS) and without embedded stromal cells. n = 3. Each point is a technical replicate. ***p < 0.005, ****p < 0.0001 (Student’s t-test). Figures show mean ± SD. Figure Created with BioRender.com.