As BMP4 treatment has been shown to generate the three germ layers in 2D models, we developed a method to explore whether BMP4 triggers changes in shape of a 3D hiPSCs model. hiPSCs single cells were cultured under defined medium conditions that provided consistent shape variation in spheroid morphology (Figure 1A). E8 medium promotes self-renewing conditions, while KSR BMP4 triggers differentiation (Chen et al., 2011; Tewary et al., 2019). We then monitored spheroid formation of hiPSCs aggregates (Figure 1B; Supplementary Image S1).

To quantify dynamic changes in the morphology of hiPSCs spheroids in suspension, we first set up a robust high-content imaging-based platform to assess morphology using frame-to-frame variation analysis to obtain efficient segmentation of simple phase-contrast images and precise morphology quantification (i.e., size and shape) (Alsehli et al., 2021). Consistent with our previous results, we observed distinct phenotypic variations in each medium condition (Figures 1B–E). Cells cultured in E8 medium formed round spheres (Figures 1B, D, E; Supplementary Video S1), whereas in KSR BMP4, spheroids broke symmetry and exhibited axial elongation (Figures 1B, D, E; Supplementary Video S2). The control conditions produced intermediate phenotypes: in E8 BMP4, spheres were smaller in size, and tended to form dynamic small protrusions (Figures 1B, D–E; Supplementary Video S3) while culturing in KSR alone produced a budded morphology (Figures 1B, D, E; Supplementary Video S4). These results demonstrate that adding BMP4 to KSR medium enabled consistent morphological changes, including axial elongation.

To confirm differentiation concomitant with axial elongation, 3D hiPSCs were first stained for pluripotency marker OCT4 after 96 h of culture. This revealed that spheroids cultured in E8 medium expressed high levels of OCT4, while in the other three conditions, OCT4 was detected at low levels (Figure 2A).

We next interrogated markers for trilineage differentiation of germ layers as OCT4 expression is reduced. After 96 h, we stained for endodermal marker SOX17, mesodermal marker BRA, and ectodermal marker SOX2 to verify that 3D hiPSCs were able to form all lineages when cultured in KSR BMP4 media. Our results showed consistent emergence of the three lineage markers’ expression upon culture in KSR BMP4 compared to KSR medium (Figures 2A–H), which is known to only induce spontaneous differentiation (Taha et al., 2016; Ungrin et al., 2008).

Immunofluorescence staining showed that SOX17 was absent in spheroids grown in E8; however, strong expression of SOX17 in KSR BMP4 conditions was detected consistently in the distal domain of elongated spheroids in a mutually exclusive expression pattern with OCT4 (Figure 2A). Notably, in the KSR BMP4 medium condition, despite its ability to drive the most substantial shape change, we observed residual OCT4 expression, as well as another cell population co-expressing SOX17⁺/OCT4⁺ neighbouring SOX17⁺/OCT4⁻ cells in the elongated tip (Figure 2A). Consistent with this, when spheroids were cultured in E8 BMP4, low SOX17 expression was observed randomly. Morphological changes as previously described in (Figure 1B), demonstrated that a few spheroids in the E8 BMP4 medium condition formed small protrusions in which SOX17 was detected at low levels with marked polarisation in the protrusion area (Figure 2A). When spheroids were cultured in KSR medium, we observed lower expression of SOX17 (Figure 2A) that distributed across the entire spheroid compared to the KSR BMP4 condition (Figure 2A). SOX2, an ectoderm marker, and BRA, a marker for mesoderm, were expressed in KSR and KSR BMP4 conditions, but their expression patterns did not appear to correlate strongly with elongation (Figures 2B, C). Altogether, these observations indicate an associated lineage specification suggesting that morphogenesis and SOX17 polarisation toward the elongated area in KSR BMP4 could be coupled.

Next, we quantified SOX17 expression in 3D using Imaris software to estimate the percentage of cells expressing markers (Figures 2D–H). Unsurprisingly, hiPSCs spheroids cultured in E8 medium maintained high expression of OCT4 (Figure 2D). Conversely, a significant reduction of OCT4 expression was detected in all other medium conditions (E8 BMP4, KSR, and KSR BMP4), indicating that cells ceased to be pluripotent. Spheroids in KSR BMP4 medium had high levels of SOX17 positive cells compared to other conditions (Figure 2E). Quantification of ectoderm and mesoderm markers showed that spheroids cultured in E8 medium express low levels of the ectodermal marker SOX2 with no sign of BRA expression (Figures 2G, H). In the differentiating condition (KSR BMP4), SOX2 and BRA expression were observed (Figures 2G, H). When spheroids were grown in E8 BMP4 medium, neither SOX2 nor BRA was detected. Altogether, these observations demonstrate that in KSR BMP4 medium, spheroids give rise to cells that express markers for all three germ layers. In contrast, in KSR alone, spheroids undergo spontaneous differentiation primarily toward the ectodermal lineage.

Next, we explored whether biochemical cues, including BMP4 signalling, influenced the proliferation of free-floating hiPSCs spheroids as the shape changes. This was investigated using an EdU proliferation assay on 3D spheroids in suspension. In E8 medium, cells were highly proliferative and exhibited homogenous EdU expression (Figures 3A, D). When spheroids were cultured in KSR BMP4 medium, cell proliferation was reduced and restricted to the tips of elongated spheroids (Figures 3A, D). In the intermediate conditions, the proliferation rate in E8 BMP4 was significantly decreased compared to E8, whereas spheroids cultured in KSR notably polarised toward the budding area (Figures 3A, D). These results indicate that biochemical cues in the medium regulate cell proliferation, suggesting that low cell proliferation is associated with BMP4 treatment. In contrast, KSR medium prompts proliferative cell polarisation, which induces morphogenesis and differentiation.

Having observed variation in the number of proliferative cells post-BMP4 treatment and under KSR, we next explored the F-actin network, which has been reported as an indicator of cellular tension (Clarke and Martin, 2021). Here, we postulate that cellular tension regulates spheroids’ morphology and can be inferred based on F-actin intensity and its orientation within spheroids (Figures 3B, C, E). Phalloidin staining revealed that spheroids cultured in E8 medium exhibited organised and tightly packed F-actin across the entire spheroid (Figure 3B). In KSR BMP4 conditions, we observed a disorganised and stretched F-actin network localised preferentially around the edges (Figure 3B). However, in spheroids cultured in E8 BMP4, F-actin accumulated in the core of the spheroids (Figure 3B). In KSR medium, we also observed a disorganised F-actin network stretching around the edges of spheroids (Figure 3B). Overall, these observations show that changes in morphology correlate with changes in cellular tension (F-actin organisation) under distinct medium conditions and may suggest a causal relationship between the two readouts. Following these indications, we quantified F-actin intensity distributions (Figure 3C). F-actin orientation changed upon treatment with different media. In KSR BMP4, elongated spheroids showed more F-actin expression in the periphery (Figure 3E). In E8 BMP4 and E8 conditions, the F-actin is distributed in the inner core (centrally). Altogether, these observations indicate that in differentiating conditions, KSR and KSR BMP4, cellular tension is associated with changes in morphogenesis.

We next aimed to investigate further whether there was a relationship between changes in shape and the patterning of germ layers in 3D hiPSCs spheroids. When spheroids were exposed to BMP4, we observed the emergence of SOX17-expressing cells, which coincided with axial elongation (Figures 1B, 2A). Therefore, we hypothesised that SOX17 expression was dependent on this shape change and that impeding elongation would impact the expression of SOX17. To test this, spheroids were physically confined using fully elastic PEG-based hydrogels. We harnessed our previously reported synthetic platform that could support the culture of hiPSC-derived intestinal organoids (Jowett et al., 2021; Norman et al., 2021). These PEG-based hydrogels are formed at low polymer concentrations using two sequential reactions. First, four-arm PEG nitrophenyl carbonate (PEG-4NPC) is conjugated with hetero-bifunctional peptides to form PEG-peptide conjugates. These conjugates are then cross-linked with four-arm PEG vinyl sulfone (PEG-4VS) to form the non-degradable hydrogel network (non-degPEG) (Figure 4A). We observed that confinement within non-degPEG prevented morphological changes in spheroids under differentiating conditions after BMP4 treatment (Figure 4B). Confined spheroids in E8 medium exhibited high levels of pluripotency marker OCT4, and no SOX17 expression was detected, similar to suspension conditions (Figures 4B, 2A). Interestingly, encapsulated spheroids in KSR BMP4 medium showed significantly reduced OCT4 and SOX17 expression (Figures 4B, 4F, G). Similarly, the E8 BMP4 condition showed a significant reduction in expression of OCT4 and very few SOX17 positive cells (Figures 4B, F, G). This was in contrast to the KSR medium condition in which spheroids maintained high expression of OCT4, but no SOX17 expression was detectable, which is comparable to the undifferentiated condition E8 (Figures 4B, 4F, G). These results indicate that when cultured within confining hydrogels, cells within hiPSC spheroids do not express SOX17 as they do in free-floating (suspension) differentiating conditions (Figures 2A, E). Overall, this suggests that embedding hiPSCs in a confining environment blocks morphogenesis, and despite the addition of BMP4, this was insufficient to prompt SOX17 expression.

The emergence of SOX17 lineage was dramatically reduced compared to free-floating culture when hiPSCs spheroids were cultured in confining hydrogels. We thus asked whether encapsulating spheroids in a hydrogel that was not confining would permit differentiation. We, therefore, modulated our design by swapping PEG-4VS with tetra-arm PEG molecules functionalised with acrylate groups (PEG-4Acr) to form a degradable hydrogel network (deg-PEG) (Figure 4C). PEG hydrogels cross-linked with acrylate have been reported to undergo controlled softening over time (Gjorevski et al., 2016). Using small amplitude oscillatory rheology, we confirmed that while standard PEG hydrogels cross-linked with PEG-4VS did not exhibit significant changes in stiffness over 5 days under standard culture conditions (G′ ∼ 488.3 Pa to 329.9 Pa) (Figure 4E), hydrogels formed by swapping 75% of the PEG-4VS with PEG-4Acr softened over 5 days and were significantly softer than day 0 controls (∼ 270 Pa to 78.7 Pa) (Figure 4E).

To determine whether encapsulation within deg-PEG hydrogels influences lineage differentiation, we stained for OCT4 and SOX17. Our findings show that in self-renewing conditions (E8 medium), encapsulated spheroids within deg-PEG maintained high expression of OCT4 and did not express SOX17 (Figure 4D). As expected, in the KSR BMP4 medium condition, we observed a significant reduction of OCT4 expression. We also identified an increase in SOX17 expression compared to non-degPEG (Figures 4B–D, F, G). Furthermore, when spheroids encapsulated in deg-PEG were cultured in E8 BMP4, OCT4 was expressed in very few cells; however, we did observe a significant increase in SOX17 expression compared to non-degPEG cultures (Figures 4B–D, F, G). Spheroids grown in the KSR medium showed reduced expression of OCT4 in deg-PEG compared to non-degPEG conditions (Figures 4D, F), and SOX17 expression remained low (Figures 4D, G). Altogether, these findings demonstrate that deg-PEG conditions partially restored SOX17 expression in both the KSR BMP4 and E8 BMP4 conditions.

hiPSCs line (Hoik_1) were selected from HipSci biobank (www.hipsci.org). As previously described (Kilpinen et al., 2017), hiPSCs were derived from skin fibroblast and reprogrammed using Sendai virus vectors (CytoTune) expressing the four factors OCT4, SOX2, MYC and KLF4. Briefly, iPSCs were cultured on a previously coated 6-well plate (Thermo Fisher Scientific) with 10 μ g/mL vitronectin (Stem cell Technologies). Cells were cultivated in feeder-free medium using Essential 8 (E8) medium with 2% E8 supplement (50x) (Thermo Fisher Scientific), and 1% Penicillin Streptomycin 5,000 U/mL (Gibco) incubated at 37°C, 5% CO₂. Cells were used at passages 28–31 and were not cultured for more than 6 passages for experiments.

When colonies reached 70% confluency, typically after 3 days, cells were washed with Hank’s Balanced Salt solution (HBSS) (Gibco). Cells were incubated in Accutase (Bio Legend) for 3 min at 37° C, 5% CO₂, and resuspended in E8 medium (Thermo Fisher Scientific) and 10 μM Y-27632 Rho-kinase inhibitor (ROCKi) (ENZO Life Sciences). Prior to cell seeding, low attachment 96-well V-bottom plates (Thermo Fisher Scientific) were coated with 5% (w/v) Pluronic solution (Sigma) as previously described (Alsehli et al., 2021). Plates were centrifuged at 500 g for 5 min and incubated at room temperature (RT) for 30 min, followed by washing with Phosphate Buffer Saline (PBS) (Gibco). The chosen cell seeding density was optimised for size and morphology consistency. Single cells were seeded at a density of 750 cells/well and cultured in E8 medium (Thermo Fisher Scientific) as described previously and 10 μM ROCKi (ENZO Life Sciences) to prevent cell apoptosis and enhance cell viability. hiPSCs were incubated for 48 h with a sufficient medium at 37° C, 5% CO₂ to form spheroids (Figure 1A).

The spheroids formation protocol was modified and adapted from previous studies (Ungrin et al., 2008; van den Brink et al., 2014). After 2 days in culture, the spheroids were formed, and medium was replaced either to self-renewing (1), differentiation condition (2) or their controls (3,4) at 37° C, 5% CO₂ for 96 h, as follows; (1) E8 medium supplemented with 2% E8 supplement (50x) (Thermo Fisher Scientific), 1% Penicillin Streptomycin 5,000 U/mL (Gibco), and 10 μM ROCKi (ENZO Life Sciences); (2) Knockout serum medium (KSR) consisting of Advance DMEM/F-12 medium, supplemented with 20% Knockout serum replacement, 1% L-Glutamine, 1% Penicillin Streptomycin 5000 U/mL (all Gibco), 0.1 mM β-mercaptoethanol (Sigma), 10 ng/mL basic fibroblast growth factor (bFGF), 50 ng/mL BMP4 (Invitrogen), and 10 μM ROCKi (ENZO Life Sciences). (3) E8 medium composition as in (1) supplemented with 50 ng/mL BMP4; and (4) KSR medium as described in (2) with no addition of 50 ng/mL BMP4 (see Figure 1A).

Hydrogels were synthesised as previously described (Jowett et al., 2021; Lust et al., 2021) using 4-arm PEG-peptide conjugates and 4-arm PEG-VS 20 kDa (non-degradable) or a mixture of 25% 4-arm PEG-VS and 75% 4-arm PEG-4Acr 20 kDa (degradable) (see Figures 4A, C). Briefly, peptide conjugates were synthesised with Ac-KDW-ERC-NH2, >98% purity (custom synthesis Peptide Protein Research, Ltd. (United Kingdom)) and 4-arm 10 kDa PEG-NPC (JenKem Technology, United States). The N-terminal primary amine (lysine) of Ac-KDW-ERC-NH₂ was reacted with PEG- NPC to form conjugates. Purified conjugates were cross-linked with either PEG-4VS or PEG-4Acr to form the hydrogel network through a Michael addition. Here, the reaction between peptide conjugate and cross-linkers (PEG-4VS and PEG-4Acr) was performed in a stoichiometric ratio 1:1 in 30 mM HEPES buffer at pH 8 (Sigma) diluted in 1x HBSS (Gibco). Hydrogels were allowed to form for 30–45 min. The solid content of the polymer was 2.5% (w/v) for both hydrogel conditions.

Hydrogel storage modulus G′ was measured on a strain-controlled ARES from TA Instruments using a parallel plate geometry (8 mm plate). 50 μL hydrogels were prepared in 8 mm glass rings and placed onto the rheometer plate, and measurements were carried out at 37 °C while sealed with paraffin oil to prevent evaporation. A frequency sweep was recorded, measuring G′ as a function of shear frequency in the range 100–0.1 rad s⁻¹ at a fixed strain of 1%. (Orchestrator software, version 7.2.0.2).

hiPSCs spheroids were obtained 48 h post seeding in 96-well V-bottom plate as previously described. The spheroid in each well was washed once with 30 mM HEPES buffer (pH 8.0), then resuspended in 30 mM HEPES buffer (pH 8.0) (Sigma). Spheroids were encapsulated within 10 μL of hydrogel, formed as described above in μ-slides angiogenesis glass bottom (ibidi) by placing the mixture in each well and incubating for 30–45 min at 37°C, 5% CO₂. Medium containing 10 μM ROCKi (ENZO Life Sciences) was added, and cells were cultured for 96 h at 37°C, 5% CO₂ replenish medium daily without BMP4.

Time-lapse imaging was performed at 37°C, 5% CO₂ using JuLI™ Stage Real-Time Cell History Recorder (NanoEnTek). hiPSCs-derived spheroids in 96-well V-bottom plate in suspension and encapsulated in PEG-peptide hydrogels were imaged every hour for 4 days (96 h). Brightfield time-lapse images were acquired by ×10 objective, using the following set-up: exposure time = 65 ms, brightness = 18, and focus at ∼8,851 to capture spheroids morphological changes over time. The setup could vary slightly between experiments, which require adjustment for focus and brightness.

Spheroids in suspension were fixed using 4% paraformaldehyde (PFA) (Sigma) incubated at room temperature (RT) for 45 min followed by three washes with PBS 5 min/wash. Cells were permeabilised and blocked with 0.3% Triton X100 (Sigma) in PBS, and 3% bovine serum albumin (BSA) (Sigma) for 1 h at RT on a shaker. Primary antibodies used were Rabbit anti-OCT4, Goat anti-SOX17, Goat anti-BRA, Goat anti-SOX2 (R&D); Mouse anti-YAP1 (Santa Cruz) were antibodies diluted in the blocking buffer, and spheroids were incubated in primary antibodies overnight in the dark at 4°C. Following three washes with 0.1% TritonX100 in PBS 5 min/wash, cells were incubated in secondary antibodies anti-rabbit Alexa Fluor 488, anti-goat Alexa Fluor 633, and Alexa Fluor 555 Phalloidin (Thermo Fisher Scientific) and DAPI, for 2 h RT on a shaker. Finally, spheroids were washed three times with 0.1% TritonX100 in PBS for 10 min/wash.

Encapsulated spheroids in PEG-4VS and 75% PEG-4Acr were fixed with 4% PFA incubated at RT for 45 min. Then, cells were washed three times with PBS for 10 min/wash, then permeabilised using 0.3% Triton X100 in PBS for 1 h. Cells were incubated in a blocking buffer consisting of 0.1% Triton X100 and 3% BSA overnight at 4°C. Primary antibodies: Rabbit anti-OCT4 (abcam); Goat anti-SOX17, and Goat anti-BRA (R&D); were diluted at 1:50 in the blocking buffer; cells were stained and incubated for 36 h in the dark at 4°C. Following five extensive washes using 0.1% Triton X100 in PBS for 10 min/wash, cells were stained with secondary antibodies at 1:50 anti-rabbit Alexa Fluor 488 and anti-goat Alexa Fluor 633 (Thermo Fisher Scientific) and DAPI overnight in the dark at 4°C. Then, cells were washed five times with 0.1% TritonX100 in PBS for 10 min/wash.

Spheroids in suspension at day 4 were labelled with 4 μM EdU and incubated for 6 h using a Click-iT EdU Alexa fluor 488 imaging kit (Thermo Fisher Scientific). Spheroids were fixed with 4% PFA, incubated at RT for 45 min and washed three times with PBS. Cells were then blocked and permeabilised using 3% BSA and 0.3% TritonX-100 in PBS for 1 h at RT, followed by washing three times with 3% BSA. A click-iT reaction buffer was prepared (following the manufacturer’s instructions) at the desired volume. Cells were incubated in the reaction buffer for 30 min at RT, followed by washing once with 3% BSA. Nuclear staining, Hoechst 33342 (1:1000, Invitrogen) was added to the cells and incubated for 30 min at RT protected from light. Then, cells were washed twice with PBS for 5 min/wash.

Spheroids in suspension culture were transferred into a 96-well plate with a flat bottom, μclear black for imaging (Greiner). A Z-stack images of the selected markers and cell viability were acquired using Leica TCS SP8 confocal microscope with a ×40 oil objective. Encapsulated spheroids in μ-slides angiogenesis glass bottom (ibidi) were imaged for markers expression and live/dead assay. Embedded spheroids were imaged with a ×63 oil objective. Image processing was performed and visualised using ImageJ version 2.0.0 (Fiji) and OMERO—insight Version 5.5.9.

Image analysis was performed using Imaris software version 9.9. A surface mask was created on the DAPI channel to measure sphericity after changes in morphology. This provides readouts for changes in shape via sphericity and deformation values (values range closer to 1 corresponds to spherical structure, while values below 0.7 indicate shape deformation (Szmańda and Witkowski, 2021).

Nuclei were quantified using spot-counting methods, pluripotency, and cell proliferation markers to determine the germ layer expression intensity. The percentage of positive expression was determined from the total number of cells in the DAPI channel for the respective spheroid. Colocalisation of OCT4 and SOX17 was calculated based on expression thresholds for co-expressed markers, and then spot detection was used to quantify the number of cells.

To analyse the F-actin distribution around the spheroids, a pipeline was created in Icy software, an open community platform for bioimage informatics (de Chaumont et al., 2012) (Figure 3C). Pre-processing of the images included conversion into maximum intensity projections (MIP) and a Gaussian blur to improve the segmentation quality of the whole spheroid area based on the DAPI channel and the F-actin network based on the F-actin channel. First, the two channels (DAPI and F-actin) were extracted separately. The DAPI channel was used for segmentation to detect the whole spheroid area using HK-means thresholding, creating a mask of the entire spheroid. Here, in the HKmeans method, gaussian pre-filter (value = 3) was used to remove noise, and the minimum object size was selected at 10,000 pixels to generate a region of interest (ROIs) that represents the whole spheroid area. As DAPI provides consistent and reliable staining, using this channel for mask creation improves the quality and reliability of segmentation and removes any bias due to changes in intensity in the F-actin channel. Next, A “fill-holes in ROI” step was included to create a single mask that evenly covered the spheroid’s whole area. Therefore, we determine our ROIs by dividing the entire area mask into an inner ring area (to measure F-actin within the centre of the spheroid) and an outer ring area (measuring F-actin within the periphery). To create these ROIs, for each image, the whole area mask was reduced by a scale percentage of 25 along both the X and Y-axis to reduce the ROI to the central region in the inner core. Next, the inner ROI was subtracted from the whole area mask to obtain the outer ring region. Reduction of each mask by a set scale percentage of 25 was chosen to remove bias and maintain as even as possible volume within each ROI. After identifying the ROI, the F-actin channel was used as input to extract the average intensity within each ROI area. The pipeline was run in an automated manner to remove bias (see details pipeline in Supplementary Image S3).

Data from the live images pipeline and Imaris software were compiled into Microsoft Office 365 Excel 16.62. Data were exported to GraphPad Prism version 9.3.1., Results are represented as means with standard deviation (SD). Statistical analysis was performed via ordinary one-way ANOVA test for multiple comparisons (Šídák test) to analyse markers expression (OCT4, SOX17, SOX2, and BRA) between medium conditions. An unpaired t-test was used to analyse F-actin staining intensity and hydrogel shear modulus measurements.