iPSC-derived mammary-like organoids harbor characteristics of the human mammary gland
Pluripotent stem cell-derived organoids represent 3D, self-organizing cellular assemblies, which recapitulate organ-like features and thus, find application in disease modeling and precision oncology [25]. For the generation of a co-culture platform consisting of benign tumor-adjacent mammary gland tissue and breast cancer spheres in an autologous and allogenic manner, we first aimed at differentiating iPSC into MLOs expressing markers of the mammary (myo-) epithelium. Our differentiation protocol comprises two steps and is based on previously described protocols [23, 24]. In a first step, solid iPSC spheres were directed into the non-neural ectodermal lineage for a period of 10 days by the incubation in serum-free culture medium optimized for culturing human breast tissue, as schematically illustrated in Fig. 1A. The non-neural ectodermal iPSC spheres are referred to as “mammary embryoid bodies” (mEBs). As a control for the up- or downregulation of differentiation- and pluripotency markers, we concomitantly kept iPSC spheres in regular stem cell maintenance medium resulting in normal embryoid bodies (EBs). We examined the mEBs and EBs for their expression of CK18 and AP2γ, crucial markers associated with the mammary epithelium and mammary gland differentiation, and of the pluripotency transcription factor Sox2. In line with a successful differentiation, our data revealed a significant upregulation of AP2γ and CK18 in mEBs compared to EBs. Contrary, Sox2 expression was maintained in EBs but downregulated in mEBs (Fig. 1B). The second stage of the protocol comprises the generation of a 3D floating system, consisting of day-10 mEBs that are cultured in the center of floating Matrigel®-collagen droplets. The matrix approximately recapitulated the stiffness of a normal mammary gland of about 170 Pa [23, 26]. Since our overall aim comprises the co-culture of MLOs and PDMs or cancer cell line-derived spheroids, we intended to combine our breast cancer spheres with mEBs at the beginning of stage 2 of the differentiation protocol and to monitor the co-culture for 10 days. Therefore, we sought to ensure that day-20 MLOs express markers associated with the mammary (myo-) epithelium. Whole mount IF staining revealed that day-20 MLOs were positive for the luminal epithelial markers CK8, CK18 and EpCAM and for the basal myoepithelial markers CK14 and p63 (Fig. 1C). As a proof for functionality, day-35 MLOs were incubated for another 5 days in prolactogenic medium and stained positive for the expression of the milk protein β-casein in gland-like structures (Fig. 1C, lower panel). Furthermore, day-20 MLOs formed acinar structures with hollow lumen (yellow arrow) as shown by parallel staining of F-actin and Nuclei (Fig. 1D). To confirm that our organoids are viable prior to embedding in Matrigel®-collagen matrices and during the downstream differentiation, we performed a viability test of day-10 mEBs and day-35 MLOs via parallel staining with Calcein-AM™ (viable cells), SYTOX™ Orange (dead cells) and Hoechst nucleic acid stain. Both the mEBs and the MLOs exhibited a mean viability of at least 88% ± 6.9 (Fig. 1E). Representative 2D projections of confocal 3D images illustrated the high overall viability of day-35 MLOs (Fig. 1E, lower panel). Taken together, we demonstrated the successful iPSC-derived differentiation into MLOs, which express markers of the mammary epithelium and myoepithelium and harbor phenotypical and functional properties of the human mammary gland such as the formation of acinar structures or β-casein expression.
Generation of viable, heterogenous patient-derived microtumors from histologically different types of primary breast cancers
Breast cancer is a heterogenous malignant lesion of viable origin and can be classified according to molecular and morphological features. To better understand breast cancer’s invasive properties and interaction with the TME, it is crucial to use cancer models that reflect this diverse nature of breast carcinomas. We previously introduced the isolation of ovarian and brain microtumors from primary tumor tissue (PTT) and described the heterogenous nature of PDMs and their functionality in compound testing [17–19]. Here, we transferred this knowledge to breast cancer lesions and isolated breast microtumors from primary breast tumor tissue. To this end, residual tumor specimen from n = 3 treatment-naïve patients diagnosed with histologically different breast cancers were obtained after completion of histopathological examination. Anonymized clinicopathological data including histopathological and clinical staging of respective tumors are summarized in Table 1. The PTT was mechanically dissected into 1–2 mm fragments and enzymatically digested overnight (Fig. 1A). The residual cell aggregates represented the PDMs, which typically showed an average diameter of 120 µm. The total number of isolated PDMs from individual tumor specimens ranged from 200 to 4500, depending on the size and biological composition of the PTT. To compare cellular composition and viability of PDMs with widely used breast cancer model systems we generated cancer spheroids from MCF-7 and MDA-MB-231 cell line. Image-based analysis revealed a mean PDM viability from 82% (range: 68–95%), which resembled the mean viability of cell line-derived spheroids of 84% (range: 73% − 95%) (Fig. 2B). Representative images of the life-dead cell staining indicated the high overall viability of PDMs and cell line spheroids (Fig. 2C). H&E staining reflected the heterogenous cellular morphology and cell density of the different breast cancer spheres (Fig. 2E, upper panel). To determine the proliferation activity of PDMs and cell line-derived spheroids in suspension culture, we stained the cancer models for Ki67, a commonly used proliferation marker. For the investigated PDM models, low-level expression of Ki67 expression was observed (range: 0% − 3%), while the cancer cell line-derived spheroids showed higher mean proliferative activity of 11% in suspension culture (Fig. 2D). Furthermore, we assessed the expression of primary tumor tissue resident ECM components such as collagen in isolated PDM models as compared to spheroids derived from cell lines. Our data revealed a significant difference of the collagen amount in BCP 1 and BCP 2 as compared to BCP 3 confirming the heterogenous inter- and intratumoral heterogeneity of individual breast tumors. (Fig. 2D). In contrast to cell line-derived spheroids, PDMs contained substantial amounts of extracellular matrix (ECM) components (Fig. 2E, black arrows). CK18 acts as a breast cancer prognostic marker due to its frequent downregulation during epithelial-mesenchymal-transition (EMT) [27]. We observed that CK18 was significantly higher expressed (> 50%) in PDMs from BCP 3 as compared to BCP1 and BCP2. As expected, spheroids derived from highly metastatic MDA-MB-231 cells expressed significantly higher levels of CK18 as compared to spheroids derived from non-metastatic MCF-7 cells (Fig. 2D). Representative images of immunohistochemistry (IHC) stainings reflected the distribution of the investigated markers (Fig. 2E, lower panel).
Table 1
Overview of the anonymized patient cohort.
Patient | Age | Histology | ER | PR | HER2 | T | N | M | L | V | Pn | Ki67 |
BCP 1 | 48 | IDC-NST | + | + | - | 3 | 3a | 0 | 1 | 0 | 0 | 21% |
BCP 2 | 86 | ILC | + | + | - | 3 | 1a | 0 | 0 | 0 | 1 | 10% |
BCP 3 | 71 | Mucinous | + | + | - | 2 | 0 | 0 | 0 | 0 | 0 | 5–10% |
BCP 1 was diagnosed with an invasive ductal carcinoma of the no special type (IDC-NST), BCP 2 developed an invasive lobular carcinoma (ILC) and BCP 3 was diagnosed with a mucinous ductal carcinoma. All tumors were primary tumors and derived from treatment-naïve patients. TNM-classification identifies the size of the tumor (T), regional lymph node involvement (N) and the presence of distant metastases (M). T2: 2–5 cm; T3: > 5cm. N0: no regional lymph node spread. N1: invasion to 1–3 lymph nodes. N3: invasion to 7 + lymph nodes. L, V, and Pn indicate the invasion to the lymphatic pathways (L), venous (V) and perineural (Pn) invasion, respectively. Ki67 values indicate the respective cancer proliferation rate.
Invasive behavior of breast cancer PDMs in the presence of extracellular matrix
In vivo, solid tumors are under the dynamic influence of connective tissue and the ECM, which play a dual role in tissue homeostasis and pathogenesis. After the isolation of PDMs from primary tumor tissue, we aimed at investigating the influence of the ECM on PDMs and cancer cell line spheroids with respect to morphology and invasive behavior. First, the breast cancer spheres were cultured in suspension in absence of ECM for at least 10 days. Representative brightfield images displayed compact, circular, and thus, non-invasive phenotypes among all breast cancer models (Fig. 3A). Next, singlets of PDMs and cancer cell line spheroids were individually embedded in the center of liquid Matrigel®-collagen matrix droplets recapitulating the normal mammary gland stiffness. We monitored the phenotypes of the PDMs and spheroids in the floating 3D system over a period of 10 days. Our data revealed a > 10-fold increase in the invasiveness of BCP 1 over this experimental period (f 1.3 ± 0.12 at day 0, to 14.7 ± 12.29 at day 10). As expected, spheroids generated in parallel from the highly metastatic MDA-MB-231 cell line also showed an increase in invasive behavior, which, however, was less pronounced as compared to BCP 1 PDMs. In contrast, PDMs from BCP 2 and BCP 3 as well as spheroids derived from non-metastatic MCF7 cells maintained a circular, non-invasive phenotype in ECM (Fig. 3B). Based on their morphology, we classified the observed phenotypes into four groups: acinar structures (yellow arrows), spindle-like protrusions (purple arrows), intermediate structures showing both acinar and spindle-like properties as well as circular morphologies (blue arrows) (Fig. 3C). 65% of the highly invasive BCP 1 PDMs showed the spindle-like phenotype, while 23% formed acinar protrusions. 8% of BCP 1 PDMs were in an intermediate state at day 10. In contrast, MDA-MB-231 spheroids lacked the formation of prominent acinar phenotypes, however, 9% showed the formation of intermediate structures, while the majority of 91% developed a spindle-like morphology (Fig. 3D). To exclude the possibility, that the maintenance of circular phenotypes in the non-invasive breast cancer models is caused by different extent of cell death among tested models, we assessed their viability after 10 days in the 3D floating ECM system. Results confirmed a mean cell viability of at least 50% and no significant difference between tested models (Fig. 3E). Taken together, our results show that the presence of ECM can lead to a pronounced invasive phenotype in patient-derived IDC-NST microtumors.
Co-culture of invasive PDMs and early MLOs potentiates the development of a metastatic tumor phenotype
Breast tumor-adjacent cell populations, such as mammary epithelial cells, might exert a tumor-promoting function during the complex process of cancer progression. We attempted to extend our current knowledge about the potential tumor-promoting function of healthy mammary gland organoids on breast cancer growth and invasion by generating a 3D co-culture system which closely resembles the characteristics and heterogeneity of (cancerous) human breast tissue. To this end, we combined MLOs with breast cancer spheres and monitored growth and invasiveness of the respective cancer models over a period of 10 days. One single mEB and one single breast cancer sphere were placed in proximity to each other at the center of a liquid Matrigel®-collagen droplet resulting in a 3D floating co-culture system. Representative brightfield images of the co-cultures at different time points are shown (Fig. 4C). Within the first 3 days, cancer spheres (blue arrows) and MLOs (red arrows) moved towards each other and formed a compact two-component structure. For BCP 1 PDMs, which showed the highest degree of invasiveness in previous experiments with ECM (see above), we observed a directed invasion of the PDMs towards the MLO at day 3 (Fig. 4C, d3–20x). We investigated the effect on the invasiveness of breast cancer spheres under the influence of the co-culture with MLOs. Our data revealed that the presence of MLOs potentiates this process depending on the invasiveness of the individual breast tumor (Fig. 4A). In support of this observation, a similar but less pronounced trend could be observed for MDA-MB-231 spheroids in co-culture with MLOs. For BCP 2, BCP 3 and MCF-7, which did not show invasive behavior in previous experiments, our results show no significant difference in invasiveness between the breast cancer spheres in co-culture with or in absence of MLOs. Next, we investigated whether the presence of MLOs influences the growth of breast cancer spheres in addition to its impact on morphology and invasion. To this end, we examined the area fold change of PDMs and cell line spheroids in co-culture or cultured alone. We detected a significant increase in size for highly invasive BCP 1 PDMs as well as MDA-MB-231 spheroids co-cultured with MLOs but not for non-invasive BCP 2 and BCP 3 PDMs and MCF7 spheroids (Fig. 4B). In a next step, we analyzed differences in the secretion of previously described metastasis-related markers (FN and MMP2) among the different cultures. We found that soluble FN and MMP2 were significantly elevated in all co-cultures of BCP 1, 2, 3 and MDA-MB-231 but not MCF7 at day 10 compared to mono-cultured MLOs and breast cancer spheres (Fig. 4D, E). Within this cohort, the co-culture of BCP 1 PDMs + MLOs expressed the highest levels of metastasis-markers of which FN (mean: 47.6 ± 12.9 pg/mL) differed significantly compared to BCP 2, 3, MCF-7 and MDA-MB-231 (Fig. 4D). Taken together, our data show that tumor-adjacent, benign breast tissue supports tumor growth and invasiveness of invasive in ductal breast carcinoma microtumors and might play a crucial role in activating the metastasis-related markers MMP2 and FN in breast cancer.