Modeling Breast Cancer Using CRISPR-Cas9-Mediated Engineering of Human Breast Organoids

Abstract

Breast cancer is characterized by histological and functional heterogeneity, posing a clinical challenge for patient treatment. Emerging evidence suggests that the distinct subtypes reflect the repertoire of genetic alterations and the target cell. However, the precise initiating events that predispose normal epithelium to neoplasia are poorly understood. Here, we demonstrate that breast epithelial organoids can be generated from human reduction mammoplasties (12 out of 12 donors), thus creating a tool to study the clonal evolution of breast cancer. To recapitulate de novo oncogenesis, we exploited clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 for targeted knockout of four breast cancer-associated tumor suppressor genes (P53, PTEN, RB1, NF1) in mammary progenitor cells from six donors. Mutant organoids gained long-term culturing capacity and formed estrogen-receptor positive luminal tumors on transplantation into mice for one out of six P53/PTEN/RB1-mutated and three out of six P53/PTEN/RB1/NF1-mutated lines. These organoids responded to endocrine therapy or chemotherapy, supporting the potential utility of this model to enhance our understanding of the molecular events that culminate in specific subtypes of breast cancer.

Figure 1.

Generation and transformation of human breast organoids. A) Schematic overview of (i) generation of organoids from basal and luminal progenitor cells derived from normal human breast tissue (n = 10), sorted by flow cytometry; (ii) sequential CRISPR-Cas9–mediated gene editing in organoids of four tumor suppressor genes implicated in breast cancer, P53, PTEN, RB1, and NF1; and (iii) functional readouts used in this study. B) Organoids derived from mature luminal (ML), luminal progenitor (LP), or basal subsets (11) after 2 weeks in culture. Scale bar = 200 μm. C) Whole-mount 3-dimensional confocal image (left) and optical sections (right) of a normal human breast organoid derived from admixed basal and luminal progenitor cells labeled for keratin 5 (K5), E-cadherin, F-actin, and 4',6-diamidino-2-phenylindole (DAPI). Scale bar = 30 μm. D) Representative brightfield images of control organoids or organoids that were CRISPR-Cas9–edited (12) for P53 and PTEN and treated with Nutlin-3a (10 μM). Scale bar = 200 μm. E) Immunoblot analysis of MCF10A cells for P53, PTEN, and RB1 following CRISPR-Cas9 editing for P53 and PTEN or RB1 using two independent single-guide RNAs (sgRNAs) for each gene compared to control cells (control sgRNA). Cells were treated with Nutlin-3a (10 μM) as indicated. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. F) Quantification of the number of DNA reads for mutant P53, PTEN, and RB1 as determined by next-generation sequencing (miSeq) of organoids derived from three independent specimens that were genetically edited for P53, PTEN, and RB1 and treated with Nutlin-3a. Results are shown for guide set 1, with similar results obtained for guide set 2. G) Quantification of the proliferation of organoids that were genetically edited for P53, PTEN, and RB1 and treated with Nutlin-3a compared to control organoids. The multiplication number per week was calculated using cell counts performed during each passage (P) and compared using a two-sided unpaired Student t test. P values are indicated. The cultures were passaged weekly, and the indicated passage count was started after genetic editing, three to five passages after isolation and sorting of cells. Mean (SD) (n = 3 individual donors).

Figure 2.

Mutant organoids form breast tumors in vivo. A) Schematic overview of the in vivo experimental design. Genetically engineered tumor organoids were transplanted together with irradiated fibroblasts (TERT-immortalized human fibroblasts) subcutaneously or into the mammary fad pads (MFPs) of immunodeficient NOD-SCID-IL2Rγ–/– mice (NSG) (n = 4–6 mice per condition) and tumor growth monitored. Tumors were retransplanted into new mice or analyzed for histology, DNA sequence, and drug response. B) Growth curves and Kaplan-Meier plots of tumors edited for P53/PTEN or P53/PTEN/RB1 using two different guide sets and comparison of transplantation into the MFPs or subcutaneous sites (SQs). The triple-mutant line 14PM0932 gave tumors in 10 of 11 and 11 of 12 cases at the SQ and MFP sites, respectively. Mean (SD) is shown for n = 6 mice per condition. C) Histological assessment of tumors generated from P53/PTEN/RB1– or P53/PTEN/RB1/NF1–mutated organoids. Scale bar = 50 μm. D) Representative organoids generated from tumors presented in (C) were treated with increasing concentrations of tamoxifen or docetaxel for 7 days, after which viability was assessed using CellTitre-Glo (Promega, Wisconsin). Data are shown as mean (SD) for two independent tumors (14PM0932) or one tumor (16PM0462). E) Analysis of the clonal landscape of P53/PTEN/RB1–mutated organoids at the time of transplantation (six passages after cell isolation) and tumors generated from these organoids. Results for both guide sets are shown. The bar graphs represent quantification of the number of specific P53-indels detected by next-generation sequencing (miSeq) as a percentage of the total number of miSeq reads analyzed. ER = estrogen receptor; H+E = hematoxylin and eosin stain; PR = progesterone receptor.