Generation of fetal organ spheroids (FOSs)
An overview of the steps involved in the generation of chick fetal organ spheroids (FOSs) is shown in Figure S1. Brain, heart, liver, stomach, and epidermis from the top of the head, were collected from embryonic day (E) 12, and E14 embryos were used to harvest lung and intestine, stages when respective organs were clearly defined (Figs. S1a, b). The dissected organs were maintained in phosphate-buffered saline (PBS) on ice until further processing, including mincing and enzymatic dissociation of the tissue (Fig. S1c). The duration of enzymatic dissociation ranged from 5 to 20 min depending on organ structure (Fig. S1d). After analyzing the percentage of viable cells from each organ, single cells were seeded at a density of 10 × 106 cells/flask in agarose-coated flasks in supplemented RPMI medium. Over time, seeded cells clustered together, with visible spheroid-like structures already from day one (Fig. S2).
Morphological characterization and growth of FOSs
FOSs (presented from anterior to posterior) derived from brain, lung, heart, liver, stomach, intestine and epidermis, consistently displayed a 3D spheroid-like morphology with a general increase in size over time from day 4–12 (Fig. 1). Different FOSs exhibited some morphological variation depending on the organ of origin; for example, FOSs generated from brain mostly exhibited a spherical morphology (Fig. 1a), whereas stomach FOSs were more ellipsoidal (Fig. 1e). Smaller FOSs generated from lung, heart, intestine and epidermis were all irregular in shape, whereas the larger spheroids were more spherical (Fig. 1b, c, f, g). Interestingly, from day 3–4 of culture, lung FOSs exhibited the presence of differently sized globular sac-like structures adjacent to and around the spheroids, and these sac-like structures remained intact for several days in culture (Fig. 1b).
Differences in growth kinetics of various FOSs were observed (Table S1). Brain spheroids showed the most rapid increase in perimeter over time, with a 2.9-fold increase from day 4–12, when reaching 1.88 ± 0.06 mm (Fig. 1a′). Long-term culture up to a month of the fetal brain spheroids did, however, not increase the size further (data not shown). The perimeter of heart spheroids increased 2.2-fold from day 4–12, and reached 0.94 ± 0.03 mm at day 12 (Fig. 1c′). Meanwhile, liver, stomach and intestinal spheroids all showed a 1.3-fold increase of the perimeter (Fig. 1d′-f′), whereas lung and epidermal spheroids did not grow significantly during the culture time (Fig. 1b′, g′). Notably, stomach and epidermal spheroids were already 1.21 ± 0.06 mm and 0.94 ± 0.04 mm in perimeter, respectively, at day 4 (Fig. 1e′, g′), which might explain lack of or slower growth between day 4 and 12. Thus, the perimeter and morphological appearance of the spheroids varied in some extent from one organ type to another.
The increase in size of the FOSs implicated ongoing proliferation. To identify the presence of mitotic cells in FOSs, the proliferative marker phosphorylated (p) histone H3 (pHH3), which labels proliferating cells in late G2 and M phases [19], was analyzed by immunocytochemistry. pHH3+ proliferative cells were observed in all FOSs at all time points analyzed, with no clear difference in proliferation rate between spheroids or throughout culture period (Fig. 2a-g and data not shown). Even if lung and epidermal spheroids did not show an apparent increase in size over time between days 4 and 12 in culture, the presence of pHH3+ cells indicates that some degree of cell renewal is also occurring in these spheroids. Guided by the growth and proliferation results, all subsequent studies were performed on FOSs from day 8 of culture to represent a median time point.
Next, we examined whether the pluripotent and progenitor marker SRY-box 2 (Sox2) [20, 21] was expressed in any of the FOSs. By applying immunocytochemistry, scattered Sox2 staining was detected throughout the brain, lung and liver spheroids (Fig. 3a-c), while stomach spheroids showed Sox2 expression primarily in cells lining the surface of the spheroids (Fig. 3d). No expression of Sox2 was observed in heart, intestinal or epidermal spheroids (Figs. S3a-c). In addition, only a few cleaved (c) Caspase3+ cells, indicative of apoptotic cells, were detected in the different spheroids (Fig. S4), ruling out a major apoptotic contribution to morphology and growth patterns of the FOSs. Thus, all FOSs appears healthy, contain mitotic cells, and several of the spheroids also express Sox2 indicative of progenitor cells.
FOSs emulate typical characteristics of in vivo organs
To validate the conservation of tissue-like expression profiles of the generated FOSs, the presence of key genes associated with a particular organ type was analyzed by RT-PCR in the FOSs, compared to their corresponding in vivo embryonic organs. In the brain, GFAP is expressed in astroglial cell types both in the developing and mature chicken brain [22, 23]. In the lung, the appearance of surfactant-producing alveoli is a hallmark of lung maturation during the transition from the saccular to alveolar stage of lung development [24]. Also in the lung, Surfactant protein (SP) genes, including SP-C, are expressed in a cell-type restricted manner by Clara and/or alveolar type II cells [25]. In the heart of chick and other vertebrates, cardiac troponin T (cTnT), encoded by TNNT2, is expressed throughout heart development and at postnatal stages [26]. Albumin, which is exclusively synthesized by the liver, is the most abundant protein in the blood plasma with several physiological roles. In chick, embryonic liver Albumin is first detectable at E6.5 and remains expressed during adult stages [27]. In the stomach, the expression of Homeobox protein BarH-like 1 (Barx1) is restricted to the stomach mesenchyme during gut organogenesis [28], and in the mouse, loss of mesenchymal Barx1 prevents stomach epithelial differentiation [29]. In the intestine’s brush border membrane of mammals and birds, Sucrase-isomaltase (SI), an α-glucosidase enzyme involved in sugar digestion, is expressed [30] . In chicken and other mammals, genes within the epidermal differentiation complex (EDC) are enriched in epidermal cells, one of which is EDMTFH in avian skin [31]. The enriched key genes in specific organs mentioned above were examined in both the FOSs and in vivo organs of interest, as well as in two other selected spheroids/organs. As expected, RT-PCR confirmed that similar gene expression profiles typical of the organs of origin, from which the FOSs were derived, were also present in the generated FOSs, but no or low expression in the other two FOSs/organs studied (Fig. 4; Fig. S5). These results indicate that all seven FOSs maintain a good degree of organ-like characteristics for downstream applications.
It is known that mesenchymal-epithelial crosstalk plays a crucial role in organ development [32]. To evaluate whether mesenchymal or mesenchymal-derived cells are present in the FOSs, the expression of Vimentin [33] was examined in the seven FOSs using RT-PCR. All FOSs, expressed Vimentin, albeit with lower expression in the liver FOSs, which are in line with observed Vimentin expression in the corresponding in vivo embryonic organs (Fig. S6). These results suggest that the majority of the established FOSs include cells of mesenchymal character.
Next, analyzes of typical tissue and/or cell markers by immunocytochemistry were performed on sectioned FOSs. Epithelial cadherin (encoded by CDH1), better known as E-cadherin, is the major adhesion component of epithelial adherens junctions that are essential for tissue barrier formation. E-cadherin plays an important role in epithelial tissues, such as lung, liver, stomach and intestine, for organ homeostasis during development and disease [34, 35]. Therefore, we examined E-cadherin expression in FOSs of lung, liver, stomach and intestine, and observed that all four FOSs expressed E-cadherin (Fig. 5). The liver FOSs showed E-cadherin expression throughout the whole spheroid structure (Fig. 5b), indicating membrane expression of hepatic epithelial cells, as well as clusters of auto-fluorescence (Fig. 5b), most likely caused by the accumulation of lipofuscin known to be present in Kupffer cells [36]. By contrast, lung, stomach and intestinal spheroids were lined by an epithelial-like E-cadherin+ layer, in a barrier-like formation (Fig. 5a, c, d).
To further exemplify how the FOSs can be characterized, we focused our analyzes to three FOSs; brain, lung and stomach. The presence of important differentiated cell types in these three spheroids was evaluated by immunocytochemistry. As expected, GFAP and the post-mitotic neuronal marker HuC/D, (encoded by Elavl3/4) [37], identifying astroglial cell types and post-mitotic neurons, respectively, were both expressed in brain FOSs (Fig. 6a, b). GFAP+ astroglial cell types were located throughout the entire spheroids, whereas the HuC/D+ neurons were located mostly in the periphery (Fig. 6a, b), indicating some extent of zonation in the 3D cellular organization of brain spheroids. In chick, the expression of NK homeobox 2–1 (Nkx2–1; also called thyroid transcription factor-1 [TTF-1]) is detected at the onset of lung bud formation and throughout lung development [38], and Nkx2.1 knock-out mice are born dead due to lack of lung parenchyma [39]. Consistently, both SP-C and Nkx2.1 expression were detected in lung FOSs (Fig. 6c, d), with SP-C+ Clara and/or alveolar type II cells being observed broadly throughout the spheroids, and Nkx2.1 expression being restricted to the periphery (Fig. 6c, d). Gastrin, produced by G cells, is a hormone that regulates gastric acid production, which facilitates the digestion of proteins, as well as the absorption of various vitamins and minerals. To regulate gastric acid in the stomach, the peptide hormone Somatostatin, produced by δ-cells, acts to decrease acid production by preventing the expression and secretion of Gastrin. Somatostatin is also expressed in the gut of chick embryos [40]. Accordingly, Somatostatin+ δ-cells were observed in restricted clusters within stomach FOSs (Fig. 6e). Brain, lung and stomach FOSs exhibited the presence of tissue-specific differentiated cell-types that are consistent with their organ of origin. Collectively, this further validates the use of chick FOSs to mimic in vivo organs at the cellular level.
Electron microscopy of FOSs
To study the morphological surface and cellular structures of different cell types present in the generated FOSs, EM was performed on brain and lung FOSs. Specifically, both scanning electron microscopy (SEM) and transmission electron microscopy (TEM; the Tokuyasu technique) were used to image fixed brain and lung FOSs.
During SEM processing, the large fetal brain spheroids (FBSs) were easily broken, whereas the smaller fetal lung spheroids (FLuSs) remained intact. SEM images of the fractioned FBSs showed nerve fibers between neuronal cells (Fig. 7a), indicating establishment of neuronal connections. Presumed astrocytes with characteristic multiple processes were also observed (Fig. 7a). Moreover, the TEM images of the FBSs indicated the presence of neuro-filament, astrocytes and dendrites (Fig. 7b, c). SEM views of FLuSs revealed cells with protrusions that are characteristic of ciliated lung cells (Fig. 7d). Furthermore, TEM images of the FLuSs indicated alveolar lumen, alveolar septal walls and lamellar bodies that are found in type II alveolar epithelial cells (Fig. 7e, f). As expected, the overall structural organization of spheroids derived from fetal brain and lung cells did not completely mimic in vivo organs, nevertheless, EM successfully identified important characteristic organ-specific structures in both brain and lung FOSs.
Confrontation assay of FOSs with tumor spheres
To assess a functional application of FOSs, we performed a confrontation assay between specific FOSs and tumor spheres to analyze and monitor potential cell-cell contact and tumor invasion patterns. For this purpose, FBSs and FLuSs were confronted separately with tumor spheres derived from either the glioblastoma cell line U251 or the lung cancer cell line A549. To facilitate the monitoring of potential invasion, U251-GFP and A549-GFP cell lines that stably express green fluorescent protein (GFP) [17] were used to form spheres in a similar manner as described for the FOSs. Briefly, similarly sized FOSs and tumor spheres were placed together to establish a cellular confrontation, such that each well in an 8-well chamber contained one tumor sphere and one FOS. In total, four confrontation assays containing up to eight replicates each were seeded. After seeding, potential fusion between the tumor spheres and FOSs, and invasion of GFP-expressing cancer cells into or on the surface of the FOSs were examined by bright field and fluorescence microscopy, respectively, each day up to 10 days. In particular, image analyzes was focused at the interface between the tumor spheres and FOSs to determine the degree of cell-cell contact and tumor invasion.
Already at confrontation day 1, all U251 glioblastoma tumor spheres had fused with adjacent FBSs (Fig. 8a, m). Furthermore, fluorescence imaging provided evidence that from day 3, U251 cancer cells had started to invade the FBSs, a phenomenon observed in 100% of the confrontation pairs (n = 7; Fig. 8b-c′′, n and Movies S1-S2). Next, the combination of U251 tumor spheres and FLuS showed that 57% of the U251 spheres and FLuSs had fused at confrontation days 2–7, reaching 86% at day 10 (n = 7; Fig. 8d-f, m). Four out of seven U251 cancer spheres exhibited invasion patterns in combination with FLuS, starting from day 4 (Fig. 8f′, f′′, n and Movies S3-S4). Analyzing the A549 lung tumor sphere experiments showed that after 2–4 days, most (86%) A549 tumor spheres had fused with the FBSs, and thereafter a 100% of these fused tumor spheres: fetal spheroids was observed (n = 7; Fig. 8g-i, m). At day 7, tumor cells from all A549 spheres had invaded their paired FBSs (Fig. 8i′, i′′, n and Movies S5-S6). By contrast, up to 4 days, no fusion between A549 tumor spheres and FLuSs was observed, however, a minority (12.5–37.5%) of these tumor spheres did slowly fuse with fetal lung spheroids between day 5 and day 10 s (n = 8; Fig. 8j-l, m). Moreover, only minor invasion of A549 tumor cells into the FLuSs was detected during the 10 days of monitoring (Fig. 8l′, l′′, n and Movies S7-S8). Notably, the results indicated that U251 cells started to invade the FBSs within 1–2 days after fusion, whereas no or slower invasion (5–6 days) of U251 cells into the FLuSs was observed after fusion. Also, A549 cells fused to either FBSs or FLuS invaded the FOSs after 3–4 days (Fig. 8m and n). No evident morphological changes of the spheroids were observed during the duration of the confrontation assays.
Additional file 2: Movie S1. Video of confocal images of GFP-expressing U251 cancer cells invading the FBSs at day 10 of the confrontation assay (n = 7/7).
Additional file 3: Movie S2. Video of confocal images of GFP-expressing U251 cancer cells invading the FBSs at day 10 of the confrontation assay (n = 7/7). DAPI indicates cell nuclei.
Additional file 4: Movie S3. Video of confocal images of GFP-expressing U251 cancer cells invading the FLuSs at day 10 of the confrontation assay (n = 4/7).
Additional file 5: Movie S4. Video of confocal images of GFP-expressing U251 cancer cells invading the FLuSs at day 10 of the confrontation assay (n = 4/7). DAPI indicates cell nuclei.
Additional file 6: Movie S5. Video of confocal images of GFP-expressing A549 cancer cells invading the FBSs at day 10 of the confrontation assay (n = 7/7).
Additional file 7: Movie S6. Video of confocal images of GFP-expressing A549 cancer cells invading the FBSs at day 10 of the confrontation assay (n = 7/7). DAPI indicates cell nuclei.
Additional file 8: Movie S7. Video of confocal images of GFP-expressing A549 cancer cells invading the FLuSs at day 10 of the confrontation assay (n = 3/8).
Additional file 9: Movie S8. Video of confocal images of GFP-expressing A549 cancer cells invading the FLuSs at day 10 of the confrontation assay (n = 3/8). DAPI indicates cell nuclei.
Finally, confocal imaging indicated that the tumor cells invaded the organ spheroids, but not the other way around, as we did not detect any GFP-negative spheroid-derived cells in the tumor spheres (Movies S1-S8). Taken together these results indicate a cell-to-cell preference of tumor types and distinct target organs. This show that chick FOSs used in the confrontation assay may serve as a model to validate cell contact, cell fusion and tumor invasion, in which different molecular conditions that might alter fusion and/or invasive properties can be tested.