Morphogenesis is a complex but tightly regulated multicellular process where cells selforganize into tissues with specialized macroscale form and function via dynamic integration of cues from the mechanical microenvironment [1] , chemical morphogens [2] and local cell states [3] .
Forces and mechanical cues play a key role in several morphogenetic processes, especially during embryonic development. [4,5] For example, in preimplantation mouse embryos, differences in cell contractility regulate differentiation to trophectoderm and inner cell mass fates by controlling YAP nuclear localization [6] , and leads to the formation of microlumens which coalesce to form the blastocyst cavity. [7] However, morphogenetic processes during the earliest stages of human embryogenesis are much less understood. Human embryogenesis involves a series of complex morphogenetic events which cannot be studied in-vivo or via longterm in-vitro culture of human embryos due to ethical concerns. [8] Recently, there has been a tremendous effort towards developing human pluripotent stem cell-based models of embryonic development. [9][10][11][12][13] Lumen formation is the first morphogenetic event that pluripotent stem cells undergo in-vivo during post-implantation human embryogenesis at the epiblast stage. [14][15][16][17][18][19] Interestingly, human pluripotent stem cells self-organize to form lumens when cultured in reconstituted basement membrane-based (rBM) matrices. [14][15][16] However, these matrices suffer from loss of hiPSC pluripotency over longer timescales [14] and are poorly defined and heterogeneous with limited tunability of matrix properties. Thus, the role of the matrix, in regulating lumen formation by human pluripotent stem cells remains unknown.
Engineered biomaterials are often used for 3D culture of cells to model morphogenetic processes in-vitro and to elucidate the role of different matrix properties in mediating morphogenesis. [11,[20][21][22] Matrix stiffness, degradability, cell-matrix adhesion ligand type and ligand density have been shown to impact intestinal stem cell organoid formation and budding morphogenesis [23,24] , neural tube formation [25] , liver organoid formation [26] , and MDCK lumenogenesis. [27] More recently, scaffold geometry was engineered using hydrogel coated microchips to obtain intestinal organoids with in-vivo like morphology. [28] On the other hand, the impact of viscoelasticity on morphogenetic processes is much less explored. Viscoelastic materials dissipate mechanical energy, like viscous liquids, while exhibiting some degree of elastic recovery, like elastic solids. Under a constant deformation or strain, viscoelastic materials will exhibit stress relaxation over time, corresponding to a decrease in resistance to deformation, while elastic materials will maintain a constant stress. A faster rate of stress relaxation corresponds to a greater rate of viscous energy dissipation, or "loss", in the material per unit time. Many viscoelastic materials, including biomaterial systems such as collagen or alginate, are also viscoplastic, or sustain permanent deformations in response to an applied stress. [29] Viscoelasticity and viscoplasticity are properties of natural ECM and in-vivo tissues including embryonic tissue [30] which have been shown to regulate fundamental cellular processes, such as cell spreading, proliferation, migration and differentiation. [31][32][33][34][35] Further, organoid and embryoid formation has been most successful in 3D rBM matrices such as Matrigel or Geltrex, which are viscoelastic, and wherein cells self-organize and form lumens (Table 1). However, the role of matrix viscoelasticity in directing morphogenetic events such as apicobasal polarization and lumen formation, has not been studied in any model system.
Here, we examined the role of matrix viscoelasticity, cell-matrix adhesion ligand density and matrix stiffness in lumen formation by human induced pluripotent stem cells (hiPSCs). Fast stress relaxation and high RGD ligand density resulted in larger lumens as well as higher rate of lumen formation. Further, matrix stress relaxation and ligand density impacted hiPSC viability, apoptosis and proliferation. Interestingly, hiPSC clusters in fast relaxing, high RGD gels recapitulated key morphological features and dimensions of human epiblasts. Together, these studies reveal the impact of matrix mechanical properties on lumen formation by hiPSCs and provide insights into how matrix viscoelasticity regulates morphogenetic processes.
Organoid class Organoid Culture material (3D / 2D) Culture material viscoelastic? Organoid contains lumen? Starting cell type Refs Embryoids Human epiblast rBM matrix and PEG IPN (3D)
Yes Yes Human ESCs [14] Human epiblast rBM matrix (3D) Yes Yes Human ESCs/ iPSCs
[15] [16] [36] Human amnion rBM matrix (3D) Yes Yes Human ESCs/ iPSCs [15] [37] Human gastruloid Suspension culture (3D) n/a No Human ESCs [38] Human micropatterned germ layers Micropatterned glass coated with rBM matrix (2D) n/a n/a Human ESCs [39] Mouse epiblast rBM matrix (3D) Yes Yes Mouse ESCs [36] [40] Mouse ETS embryos rBM matrix (3D) Yes Yes Mouse ESCs, TSCs [41] Mouse ETX embryos Suspension culture (3D) n/a Yes Mouse ESCs, TSCs, XENs [42] Mouse gastruloid Suspension culture (3D) n/a No Mouse ESCs [43] [44] Mouse somitogenesis and TLS rBM matrix (3D) Yes No Mouse ESCs [45] [46] Organoids Intestine rBM matrix (3D) Yes Yes Mouse Lgr5 + cells [47] Collagen (3D) Yes Yes Mouse intestinal tissue [48] Alginate (3D) Yes Yes Human ESCs/ iPSCs [49] PEG (3D) No Yes Mouse intestinal crypts/ ISCs [23] Brain rBM matrix (3D) Yes Yes Human ESCs/ iPSCs [50] [51] Lung rBM matrix (3D) Yes Yes Human ESCs/ iPSCs [52] Prostate rBM matrix (3D) Yes Yes Mouse prostate epithelial cells [53] Gastric rBM matrix (3D) Yes Yes Human ESCs/ iPSCs [54] Pancreas rBM matrix (3D) Yes Yes Mouse and human pancreatic tissues/ human ESCs [55] [56]
Table 1. Material systems for embryoid and organoid culture. Materials used for embryoid and organoid culture are largely viscoelastic, and the resulting self-organized multicellular clusters often contain a lumen. rBM, reconstituted basement membrane; PEG, polyethylene glycol; IPN, interpenetrating network; ESC, embryonic stem cell; iPSC, induced pluripotent stem cell; n/a, not applicable; TSC, extraembryonic trophoblast stem cell; ETS, ESC and TSC embryos; XEN, extra-embryonic-endoderm; ETX, ESC, TSC and XEN embryos; TLS, trunklike structures; ISC, intestinal stem cell. "rBM matrix" represents use of commercially available matrices derived from EHS (Engelbreth-Holm-Swarm) tumor such as Matrigel or Geltrex.
We used ionically crosslinked hydrogels, with independently tunable viscoelasticity, initial elastic modulus, and ligand density for 3D culture of hiPSCs. [33] Alginate is a polysaccharide that presents no cell adhesion motifs and cannot be degraded by mammalian enzymes. [57] Here, alginate was crosslinked by calcium ions and the cell adhesion motif Arginine-Glycine-Aspartate (RGD), was covalently coupled to alginate chains. Initial elastic modulus was tuned by changing the concentration of calcium ions, RGD density was tuned by changing the concentration of RGD peptides coupled to alginate chains, and viscoelasticity was tuned by changing the molecular weight of alginate while keeping the overall alginate concentration at 20 mg ml -1 (Figure 1a). Alginate hydrogels with 3 different molecular weights (280 kDa -"slow relaxing", 70 kDa -"medium relaxing" and 28 kDa -"fast relaxing") and 3
different RGD densities (0, 150 µM and 1500 µM) were formed with an initial elastic modulus of either 3 kPa or 20 kPa by adjusting calcium crosslinking density (Figure 1b; Figure S1a).
Stress relaxation tests in compression and shear showed that t1/2, which is defined as the time taken for the stress in hydrogel to drop to half of its initial value, ranged from ~1000 s in slowrelaxing gels, to ~180s in medium relaxing gels and ~70 s in fast relaxing gels (Figure 1c-e, Figure S1b-d). Consistent with the stress relaxation results, slow relaxing gels exhibited the lowest loss tangent, which is a measure of energy dissipation, while fast relaxing gels exhibited the highest loss tangent (Figure 1c).
hiPSCs were encapsulated in the 20 kPa alginate hydrogels with a range of stress relaxation times and RGD ligand densities as single cells and cultured in mTeSR1 media, which promotes hiPSC self-renewal (Figure 1a). Alginate initial elastic modulus was 20 kPa, unless mentioned otherwise. A ROCK inhibitor (Y-27632) was added to culture media for 24hrs post encapsulation to prevent dissociation-induced apoptosis in hiPSCs. [58] By day 7, hiPSCs formed lumen-containing clusters depending on alginate stress relaxation rate and RGD density (Figure 1f). In general, faster stress relaxation at high RGD density promotes this morphogenetic process. Single hiPSCs on day 1 divide to give rise to small clusters by day 3 and form lumens by day 5 which grow in size with time (Figure 1g). of culture in different alginate formulations with an initial elastic modulus of 20 kPa. Scale bar is 100 µm. g. Representative bright-field images of hiPSC clusters in fast relaxing 1500 µM RGD gels on different days of culture. Scale bar is 50 µm.
First, we explored the effect of matrix stress relaxation rate and RGD density on hiPSC viability, apoptosis, proliferation and pluripotency in 20 kPa hydrogels. Live/ dead staining was performed on day 1 to quantify cell viability. Slow relaxing gels promoted extensive cell death while medium and fast relaxing gels supported high cell viability (Figure 2a, Figure S2a). Cell viability improved with increase in RGD density across all rates of stress relaxation (Figure 2a, Figure S2a). Slow relaxing gels with 0 µM RGD had only ~6% cell viability while slow relaxing 1500 µM RGD gels had ~70% cell viability and fast relaxing 1500 µM RGD gels had ~90% cell viability. To quantify levels of apoptosis, the TUNEL assay was performed on day 1. Low viability in slow relaxing gels was attributed to high levels of apoptosis, while medium and fast relaxing gels had significantly fewer apoptotic cells (Figure 2b, c; Figure S2b). RGD density also impacted cell proliferation. After day 5, differences in overall cell number, as characterized by the PicoGreen assay, began to emerge, with hiPSCs in 1500 µM RGD gels proliferating more than those in 150 µM RGD gels for both medium and fast stress relaxation (Figure 2d; Figure S2c). In addition, hiPSC clusters remained in an active proliferative state until day 14 of culture, with >50% EdU+ and Ki-67+ cells per cluster (Figure S3a-d).
Strikingly, hiPSCs maintained pluripotency in all gel formulations across long timescales. hiPSCs in clusters both with and without lumens expressed pluripotency markers OCT4, SOX2 and NANOG, as quantified by immunohistochemical staining of day 7 samples (Figure 2e-h; Figure S4). Further, hiPSCs maintained pluripotency till at least day 14 of culture (Fig. 2f-h; Figure S3e; Figure S4). This is in sharp contrast with 3D Matrigel culture where hiPSCs form lumens but lose pluripotency after 3 days of culture (Figure S5). [14] Overall, these findings show that slow stress relaxation triggers apoptosis and high RGD density improves cell viability and proliferation while promoting hiPSC self-renewal. formulations. Bars indicate means and SEM (**p<0.01, ****p<0.0001, ns: not significant (p>0.05), one-way ANOVA). All gels had an initial elastic modulus of 20 kPa.
Next, we sought to study the impact of matrix stress relaxation rate and RGD density on lumen formation in hiPSC clusters. To quantify the dynamics of lumen formation and growth, cell membranes of hiPSCs were stained with octadecyl rhodamine B chloride (R18) dye and imaged on day 3, 5, 7 and 9 (Figure 3a, Movie S1). Both matrix stress relaxation rate and ligand density independently impacted percentage of clusters forming lumens (Figure 3b) and rate of increase in lumen volume (Figure 3c, d). Firstly, slow stress relaxation at 0 µM and 150 µM RGD did not allow any cluster formation due to extensive hiPSC apoptosis (Figure 3a).
Concentration of RGD ligands regulated the onset of lumen formation or alternatively, % clusters forming lumen by a given day, say day 7 (Figure 3b). In the absence of RGD ligands, lumen formation was not observed, suggesting that cell-matrix adhesion is necessary for lumen formation in hiPSCs (Figure 3b). Higher RGD density promoted earlier lumen formationlumens formed by day 5 in 1500 µM RGD gels but in 150 µM RGD gels, lumens formed only after day 5 (Figure 3d). Matrix stress relaxation rate regulated the lumen volume (Figure 3c) as well as percentage of clusters forming lumens (Figure 3b). Lumen volume was highest in fast relaxing gels (Figure 3c), even though the total cell volume was similar for medium and fast relaxing 1500 µM RGD gels (Figure 3d). Further, matrix stress relaxation regulated percentage of clusters forming lumens independent of RGD density -in 1500 µM RGD gels, only ~55% of hiPSC clusters in slow relaxing gels formed lumens by day 7 while in medium and fast relaxing gels ~98% clusters formed lumens (Figure 3b). Consistent with the observed differences in proliferation between 1500 µM and 150 µM RGD gels, clusters in 1500 µM gels had higher total cell volume as compared to those in 150 µM RGD gels for both medium and fast stress relaxation (Figure 3d). Lumen and total cell volumes increased with time but plateaued by around day 9 (Figure 3d, e). These findings indicate that faster stress relaxation at high RGD density promotes lumen formation and growth.
Interestingly, in addition to affecting the rate of lumen formation, RGD density also impacted the morphology of the lumen-containing hiPSC clusters. By day 7, clusters in 1500 µM RGD gels had a monolayer of hiPSCs whereas clusters in 150 µM RGD gels had a thicker cell layer (Figure 3f). This thicker cell layer in 150 µM RGD gels persisted with lumen growth till at least day 14 of culture (Figure S3a, c, e). Clusters in 1500 µM RGD gels are reminiscent of the human epiblast where a monolayer of OCT4+ SOX2+ cells with low NANOG expression enclose the pro-amniotic cavity. [14,36] For 150 µM, n = 50 clusters and for 1500 µM, n = 50 clusters. For single cell length, n = 500 cells, measured from 100 clusters. All gels had an initial elastic modulus of 20 kPa.
Next, we examined the impact of the initial elastic modulus, or stiffness, of the hydrogel on hiPSC behavior given extensive previous work showing the impact of elastic modulus on hiPSC behavior in 2D. [59][60][61] The initial elastic modulus was varied between 3 kPa and 20 kPa, a range that was previously shown to significantly impact differentiation of mouse mesenchymal stem cells in 3D culture [33,62] and also that falls within the range of elastic moduli of soft tissues. [31] No significant differences in cell viability or cluster size were observed between 3 kPa and 20 kPa gels with similar levels of stress relaxation rate and RGD density (Figure 4a, b; Figure S6a). Similar amounts of DNA were observed after 7 days of culture, indicating similar levels of proliferation (Figure 4c). Further, there was no difference in apoptosis between 3 kPa and 20 kPa gels, as quantified by TUNEL assay (Figure 4d; Figure S6b). hiPSC pluripotency was also maintained at similar levels between 3 kPa and 20 kPa gels (Figure 4e). Further, lumen formation was unaffected by the initial elastic modulus of the hydrogel (Figure 4f). Together, these data show that hydrogel stiffness does not impact hiPSC viability, apoptosis, proliferation, pluripotency and lumen formation. Bars indicate means and SEM (ns: not significant, one-way ANOVA).
Next, we focused on analyzing apico-basal polarity and basement membrane formation of the lumen-containing hiPSC clusters in fast relaxing 1500 µM RGD gels, as this condition promoted the most robust and rapid lumen formation. For characterization of apico-basal polarity, immunohistochemical staining of apical markers PAR6 and EZRIN was performed.
PAR6 is a key component of the PAR (CDC42-PAR3-PAR6-aPKC) polarity complex, a master regulator of cell polarity [63,64] and EZRIN is a membrane-actin crosslinker necessary for proper lumen formation. [65] PAR6 and EZRIN were found to be distributed selectively at the apical surface suggesting that hiPSCs in lumen-containing clusters are polarized along the apical-basal axis (Figure 5a-d). As expected, at the apical surface, EZRIN co-localizes with actin, suggesting that EZRIN crosslinks actin and apical membrane in hiPSC clusters (Figure 5d). The basement membrane is a thin layer of matrix containing laminin and collagen IV, and cell adhesion to basement membrane is a key regulator of morphogenetic events during mouse embryogenesis. [40,66] Immunohistochemical staining of hiPSC clusters for laminin and collagen IV in fast relaxing 1500 µM RGD gels showed no continuous basement membrane layer surrounding hiPSC clusters (Figure 5e-h). Laminin and collagen IV were sparsely present along the cluster boundary (< 25%), suggesting that cell-matrix adhesion via RGD is sufficient for lumen formation. On the other hand, b1 integrin, one of the key integrins that binds to RGD, laminin and collagen IV, was present all along the cluster boundary. Similar apico-basal polarization and minimal presence of basement membrane proteins at the cluster boundary were also observed in medium relaxing 150 µM RGD gels (Figure S7a-c). Together, these data show that lumen-containing hiPSC clusters exhibit apicobasal polarity but do not form a fully continuous basement membrane at the basal surface.
Finally, we examined the molecular events associated with lumen formation of hiPSCs in the hydrogels. We first examined the transcriptional regulator YAP, which mediates the mechanotransductive response to shear stress, cell shape, and matrix elasticity on 2D
substrates [67] , and has been found to be critical for hiPSC self-renewal [68] as well as intestinal stem cell (ISC) expansion and morphogenesis. [23] In hiPSC clusters, YAP is initially localized in the nucleus, which is typically assumed to be indicative of YAP being active. However, as the lumens form over time, YAP is translocated to the cytoplasm, where it is necessarily inactive (Figure 6a-c). This behavior mirrors that of lumen formation in ISC organoids. [23,24] Thus, YAP regulated signaling pathways might play a crucial role in mediating hiPSC lumen formation.
Next, we investigated the mechanisms underlying lumen formation in hiPSCs. Rac1 inhibition resulted in much smaller clusters and failure to form lumens by day 7 (Figure 6d, e).
ROCK inhibition, however, did not significantly impact lumen formation with lumen volume and total cell volume per cluster being similar to that of control samples. Lumen volume on day 7 was significantly reduced when actomyosin contractility was perturbed using blebbistatin or ML-7 and when actin polymerization was inhibited by Latrunculin A (Figure 6d). Similarly, cytoplasmic translocation of nuclear YAP as lumens form was slowed down due to inhibition of myosin activity (Figure 6f; Figure S8). Next, to identify the signaling pathways impacted by matrix viscoelasticity and RGD ligand density in hiPSCs, the expression or activity of 39 kinases, including many that play a key role in cell-matrix adhesion signaling, were screened using a protein array. Both matrix stress relaxation and RGD ligand density impacted the expression or phosphorylation of key signaling kinases (Figure S9). Specifically, matrix stress relaxation altered the expression of b-Catenin and HSP60 and phosphorylation of STAT6 (Figure S9b). RGD ligand density altered the phosphorylation of p53 and STAT1 (Figure S9c).
Put together, these results indicate that matrix viscoelasticity and adhesions impact key signaling pathways in hiPSCs, and that actomyosin contractility and actin polymerization, as well as Rac1 activity, play an important role in mediating lumen formation.
Finally, we compared the lumen-containing hiPSC clusters formed in alginate hydrogels to the morphological features of day 10 human epiblasts containing a pro-amniotic cavity and hiPSC clusters formed in other in-vitro systems. Nuclear morphology metrics like nuclear area, perimeter and circularity were quantified for epiblast cells in human embryos cultured in-vitro in previous studies [14,69] , and were compared to hiPSCs cultured in fast relaxing 1500 µM RGD alginate gels, 3D Matrigel [14,16,36] , suspension culture [60,70] and on 2D tissue culture plastic. [70- 72] Nuclear morphology of hiPSC clusters in fast relaxing 1500 µM RGD gels was similar to the human epiblast cells for all three characteristics (Figure 7a-c). Differences in nuclear morphology between alginate gels of different stress relaxation rates and RGD density were negligible (Figure S10a-c). hiPSCs cultured on 2D tissue culture plastic exhibited similar nuclear areas and perimeters but were significantly more circular as compared to 3D culture systems and human epiblast cells. Cells in 3D Matrigel had significantly larger nuclear area and perimeter as compared to human epiblast cells. hiPSCs in suspension culture had similar nuclear characteristics to the human epiblast cells and hiPSCs in alginate hydrogels. Thus, hiPSC clusters in RGD-conjugated alginate hydrogels are closest to human epiblast cells in terms of nuclear morphology than any other culture system.
After comparing nuclear morphologies, we examined overall morphologies of human epiblasts and lumen-containing hiPSC clusters formed in alginate hydrogels. For this, lumen volume vs total cell volume per cluster were plotted for hiPSC clusters in alginate hydrogels, model epiblasts (hiPSC cultured in an interpenetrating network of Matrigel and PEG) [14] and day 10 human epiblasts [14] (Figure 7d; see Methods). Strikingly, all of the values fall on the same general curve, which therefore represents the characteristic growth pattern of lumencontaining hiPSC structures. Though, lumen-containing hiPSC structures formed in alginate hydrogels grow to be somewhat larger than epiblasts with longer culture times. In general, clusters in fast relaxing 1500 µM RGD gels, had a larger lumen volume than those in medium and slow relaxing gels, for the same total cell volume per cluster. Further, the morphology of hiPSC clusters in fast relaxing 1500 µM RGD gels was close to that of day 10 human epiblasts and model epiblasts (Figure 7d). Thus, to obtain hiPSC clusters which are morphologically similar to the human epiblast, matrix mechanics need to be tuned to provide high ligand density and exhibit fast stress relaxation. [14,16,36] , 2D tissue culture plastic [70][71][72] , suspension culture [60,70] , in comparison to human epiblasts. [14,69] Bars indicate means and SEM (*p<0.05, ****p<0.0001, ns: not significant (p>0.05), Kruskal-Wallis). Numbers on bars indicate number of nuclei analyzed. d. Correlation between lumen volume and total volume of cells per cluster for different alginate formulations, model epiblasts [14] and day 10 human epiblasts. [14] Faster stress relaxation leads to formation of larger lumens for a given total cell volume per cluster.
Taken together, our findings reveal that hydrogel stress relaxation in combination with ligand density impacts hiPSC lumen formation in 3D culture. We studied the morphogenesis of hiPSCs cultured in a range of alginate hydrogels and found that higher RGD density and fast stress relaxation promote hiPSC viability, proliferation and lumen formation while slow stress relaxation at low RGD densities triggers hiPSC apoptosis (Figure 7e; Figure S11). Surprisingly, hydrogel stiffness did not significantly impact hiPSC viability, proliferation, apoptosis, pluripotency or lumen formation, at least over the range of initial elastic moduli from 3 kPa to 20 kPa. However, we note that matrix stiffness has been shown to impact other morphogenetic events such as neural tube formation [25] and intestinal stem cell organoid formation [23] , and stiffness might impact hiPSC morphogenesis over a range not tested here or over a range of ligand density and type not tested here. Lumen containing hiPSC clusters exhibited apico-basal polarization but did not form a continuous basement membrane at the cluster boundary. Given that the basement membrane surrounding pluripotent stem cells during epiblast formation invivo is secreted by extra-embryonic cells and not pluripotent stem cells [17] , it is not surprising that hiPSCs alone in alginate hydrogels did not form a continuous basement membrane. Lumen formation was found to be regulated by actomyosin contractility as well as Rac1 activity and was accompanied by translocation of YAP from the nucleus to the cytoplasm. Finally, hiPSCs in alginate gels better recapitulated key nuclear features of the human epiblast as compared to culture of hiPSCs in 3D Matrigel or 2D tissue culture plastic, and hiPSC clusters in fast relaxing high RGD gels had a similar morphology to human epiblasts. Thus, these results indicate that hiPSC structures formed in alginate hydrogels can be utilized to model the human epiblast.
Morphogenetic processes occur in-vivo in response to both soluble signals and physical signals, under the constraints imposed by the 3D cell microenvironment. [1] Physical signals from cell-matrix adhesion, however, are necessary and sufficient for several lumenogenesis events including neural tube formation [25] , intestinal stem cell organoid formation [23] , MDCK cyst formation [27] and hiPSC lumen formation. [73,74] Accordingly, matrix properties such as stiffness, adhesion ligand density and degradability have been found to be crucial for mediating lumen formation in these systems. However, these studies probe lumenogenesis in elastic hydrogels which do not capture the complex viscoelastic nature of in-vivo extracellular matrix.
Matrix viscoelasticity has been shown to drive mechanotransductive pathways by regulating cell volume and ligand clustering. [33,75] Interestingly, in hiPSCs myosin inhibition perturbs lumen formation but not ROCK inhibition, suggesting volume regulation and associated signaling could be responsible for regulating hiPSC lumen formation. In our culture system, we find that matrix viscoelasticity is a critical regulator of lumen formation and the mechanisms mediating the impact of viscoelasticity on hiPSC lumen formation need to be studied further.
Put together, our results identify matrix viscoelasticity as a key factor regulating hiPSC lumen formation, which could potentially impact other lumenogenesis events as well.
Culture of hiPSCs in alginate hydrogels provides some distinct differences and potential advantages for applications in regenerative medicine or for modeling embryonic development and diseases in-vitro. Ideally, specialized cell types and organoids which mimic in-vivo form and function are needed. This requires precise regulation of hiPSC fate, function, and organization in 3D microenvironments with the help of appropriate biochemical and mechanical cues. [22,76] Several 2D culture platforms, embryoid bodies in suspension culture and 3D hydrogels have been used to identify biochemical and mechanical cues for promoting hiPSC self-renewal and directed differentiation. [76] However, 2D culture systems fail to recapitulate the 3D architectures of native tissues and organs. Embryoid bodies and 3D hydrogels, on the other hand, provide opportunities for generating hiPSC derived organoids with in-vivo like morphology and patterning. This is extremely important because tumorgenicity of transplanted organoids due to incorrect patterning has been one of the major limitations of hiPSC based therapies. [77] Since differentiation and organ formation during human embryogenesis proceeds after pluripotent stem cells form a polarized lumen-containing epiblast, 3D polarized lumencontaining hiPSC clusters might be crucial for activating appropriate developmental programs necessary for generating organoids with in-vivo like features. Basement membrane matrixbased cultures allow hiPSC polarization and lumen formation [14][15][16] but have many limitations:
(i) hiPSCs fail to maintain pluripotency for greater than 3 days under self-renewal conditions [14] ,
(ii) these natural matrices have lot-to-lot compositional and structural variability and lack tunability of mechanical properties, and (iii) tumor-derived matrices such as Matrigel have limited potential for clinical translation. Our culture system using RGD-conjugated alginate hydrogels allows long term culture of lumenal structures, has well defined composition and can be used for clinical applications. Further, in these hydrogels tuning hydrogel properties provides control over lumen size and architecture which may be utilized for directed differentiation of hiPSCs into specialized organoids with in-vivo like features. [78] Put together, hiPSC culture in RGD-conjugated alginate hydrogels provides new avenues for building organoids with better patterning for use in regenerative medicine.
Overall, this study presents a set of design rules that enable tuning of hiPSC cluster morphology, elucidates some of the mechanisms governing hiPSC lumen formation and indicates a broader role for matrix viscoelasticity in regulating morphogenetic processes.
Similar design rules may be used to study the mechanisms underlying other morphogenetic processes or to engineer organoids which closely mimic in-vivo form and function for in-vitro models or use in regenerative medicine and drug discovery.
Sodium alginates rich in guluronic acid blocks were purchased from FMC BioPolymer (LF20/40; 280 kDa MW) and Dupont corporation (ProNova UP VLVG; 28 kDa MW) and prepared as described previously. [33] Mid-MW (70 kDa) alginate was produced by irradiating LF20/40 alginate with a cobalt source for 3 minutes to decrease the average molecular weight to 70 kDa. RGD peptides were coupled to alginate using carbodiimide chemistry. [33] Alginate was dissolved in 0.1 M MES, 0.3 M Sodium Chloride buffer. N-hydroxysulfosuccinimide
hydrochloride (EDC, Sigma-Aldrich) and GGGGRGDSP (Peptide 2.0) peptide were sequentially mixed in the alginate solution for 24 hr. The alginate was then dialyzed in DI water, purified with activated charcoal, sterile filtered, frozen and lyophilized. The final peptide concentration in the alginate gels was either 0, 150 μM or 1500 μM as quantified previously. [79] Lyophilized alginate was then reconstituted at 3% (w/v) in serum-free DMEM (Life Technologies). Reconstituted 3% (w/v) alginate was diluted with solution containing cells and crosslinked using calcium sulfate to make hydrogels with 2% (w/v) final alginate concentration.
The initial modulus of hydrogels was modulated by adjusting the concentration of calcium sulfate used to ionically crosslink the alginate and is shown in Figure S1a.
Compression tests were performed using 5848 MicroTester (Instron) to quantify the initial elastic modulus and characterize the stress relaxation behavior of different alginate formulations. [75] Alginate disks of 2 mm thickness and 6 mm diameter were made and equilibrated in DMEM for 24 hours to allow for hydrogel swelling. Unconfined compression tests were then performed on alginate disks using a 4 mm diameter cylindrical probe. In brief, gels were compressed from 0 to 10% compressive strain at a deformation rate of 1 mm min -1 .
10% compressive strain was then maintained for 10 4 s and the corresponding stress was measured over time (stress relaxation test). To calculate the initial elastic modulus, stress was plotted as function of strain during the initial strain ramp between 5% and 10% compressive strain. A linear model was fitted to experimental data and slope of the linear fit was reported as the initial elastic modulus of the hydrogel. Next, to quantify the stress relaxation behavior at 10% compressive strain, the time at which stress drops to half of its initial value was measured and defined as t1/2. Gels were discarded after each test. Tests were performed on at least 4 biological replicates per alginate formulation.
Rheology measurements were made using a Discovery HR-2 hybrid rheometer (TA Instruments) to quantify the frequency dependent viscoelastic properties including storage modulus, loss modulus, loss tangent and complex viscosity. [34] 900µL of alginate was deposited directly onto the rheometer base plate after mixing with calcium sulfate crosslinker. A 25 mm flat plate was then immediately lowered to create an alginate disk of 1.5 mm thickness and 25 mm diameter. Mineral oil (Sigma) was deposited along the edges of the disk to prevent the gel from drying up. During the gelation period, rheometer geometry was oscillated at an amplitude of 1% shear strain and 1Hz frequency for 1 hr. As the gelation completed, storage and loss moduli reached an equilibrium value. Equilibrium values of storage and loss moduli were used to calculate the loss tangent which is defined as:
Frequency sweep was then performed from 0.1 to 100 rad s -1 at 1% strain and 37°C and storage and loss moduli were measured as a function of oscillation frequency (w). Complex viscosity was then calculated as a function of w using the following equation:
(2)
hiPSC line generated through synthetic mRNA reprogramming of BJ human fibroblast cells [80] was a gift from Dr. Vittorio Sebastiano (Department of Obstetrics and Gynecology, Stanford University). hiPSCs were cultured on TC-treated 6-well plates (Costar, Corning) coated with LDEV-free hESC-qualified Matrigel (Corning, 354277) in mTeSR1 media (STEMCELL Technologies) at 37°C in 5% CO2. [81] hiPSCs cultured in mTeSR1 were passaged at 70% confluency as pluripotent aggregates without manual selection or scraping using ReLeSR (STEMCELL Technologies) following the manufacturer's protocol.
For cell encapsulation in alginate hydrogels, hiPSCs at 70% confluency or less, were passaged as single cells using Accutase (STEMCELL Technologies), washed in mTeSR1 media, and resuspended as a single-cell suspension in mTeSR1 media with 10 µM ROCK inhibitor (Y-27632, STEMCELL Technologies) to prevent dissociation-induced apoptosis. [58] Cell concentration was measured using Vi-CELL cell viability analyzer (Beckman Coulter). To prepare a cell stock for encapsulation, cells were then centrifuged and resuspended in mTeSR1 media with 10 µM ROCK inhibitor to achieve a cell concentration of 10 million cells per mL.
To make hydrogels, 3% (w/v) alginate was drawn in a 3 mL luer lock syringe (Cole-Parmer).
Appropriate volume of cell stock was then added to the syringe with alginate and mixed to achieve a final cell concentration of 1 million cells per mL in the final hydrogel. In a second syringe, appropriate volume of calcium sulfate was added, based on the target hydrogel modulus and diluted with serum-free DMEM (Life Technologies), such that the final alginate concentration after mixing would be 2% (w/v). Alginate cell solution was then homogenously mixed with calcium sulfate using a female-female luer lock coupler (Value Plastics). The mixture of cell, alginate and calcium sulfate solution was deposited on a hydrophobic glass plate and was covered with another glass plate with 1 mm spacing between the plates to form a hydrogel of 1 mm thickness. The cell alginate mixture was allowed to gel for 30 mins.
Hydrogels were punched out using a 6 mm diameter biopsy punch and immersed in mTeSR1 media with 10 µM ROCK inhibitor. 24 hrs post encapsulation, media was changed to mTeSR1 which was replenished daily. Cells were maintained in hydrogels for 14 days.
For cell encapsulation in reconstituted basement membrane (rBM) matrices, single-cell suspension of hiPSCs was mixed on ice with growth factor reduced Matrigel (Corning, 354230) such that the final Matrigel concentration was 8 mg mL -1 and final cell concentration was 1 million cells per mL. Final Matrigel-cell mixture was then incubated at 37°C in 5% CO2 for 20 mins for gelation to occur. After gelation, mTeSR1 media with 10 µM ROCK inhibitor was added. 24 hrs post encapsulation, media was changed to mTeSR1 which was replenished daily.
hiPSCs cultured in alginates hydrogels were fixed with 4% paraformaldehyde (PFA, Alfa Aesar) for 1 hr, and then washed three times with PBS containing Ca 2+ and Mg 2+ (cPBS, GE Healthcare). The alginate gels were placed in 30% sucrose (Fisher Scientific) overnight and then transferred to 50/50 30% sucrose, OCT compound solution (Tissue-Tek) for 5 hr. The alginate gel was then embedded in OCT, frozen, and sectioned to 40µm thickness using a cryostat (Leica CM1950). These sectioned samples were used for immunofluorescence staining.
The sectioned samples were washed three times in DPBS, permeabilized with 0.5% Following three washes with 0.1% Triton X-100 in DPBS, ProLong Gold antifade reagent (Life Technologies) was used to minimize photobleaching. Images were acquired using a Leica HC PL APO 63×/1.4 NA oil immersion objective.
For Live/Dead assay, cells were encapsulated in hydrogels and Live/Dead Viability/Cytotoxicity Kit (cat. #L3224; Thermo Fisher Scientific) was used according to the manufacturer's instructions. In brief, 200µL of mTeSR1 media with 80 µM Calcein AM and 20 µM Ethidium homodimer-1 was added to a 6 mm diameter, 1 mm thick alginate disk and incubated for 20 mins at 37°C in 5% CO2. Gel disk was then washed twice with DPBS and imaged on a laser scanning confocal microscope (SP8, Leica) under standard cell culture conditions (37°C, 5% CO2). Maximum intensity projections of acquired z-stacks were used to quantify number of live cells and dead cells as well as cluster projected areas (Fig. 4b) in ImageJ (NIH) using a custom macro.
For TUNEL assay, cells were encapsulated in hydrogels and fixed 24 hours after encapsulation. The TUNEL (terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling) Assay Kit (Thermo Fisher Scientific) was used according to the manufacturer's instructions. For colocalization analysis, Coloc 2 (https://imagej.net/Coloc_2) plugin was used to calculate the different correlation coefficients. EdU incorporation assay (cat. #C10337; Thermo Fisher Scientific) was performed according to manufacturer's instructions with a 24 h incubation of 10 µM EdU.
Nuclear localization of markers such as OCT4, SOX2, NANOG, Ki-67 and EdU, and nuclear morphology metrics such as area, perimeter and circularity were measured using a custom macro in ImageJ. In brief, number of nuclei were estimated from DAPI channel and number of cells expressing markers such as OCT4, SOX2, NANOG, Ki-67 and EdU were estimated from respective immunofluorescence channels. Images were thresholded, smoothened using median filter (radius of 2 pixels) and processed using Watershed algorithm to separate touching nuclei. Number of nuclei as well as nuclear morphology metrics such as nuclear area, perimeter and circularity were then calculated where:
For quantification of nuclear and cytoplasmic localization of YAP, a custom ImageJ macro was used as described previously. [82] In brief, DAPI channel was used to obtain nuclear traces which were then overlaid on YAP channel as follows:
run("Gaussian Blur...", "sigma=2"); run("Subtract Background...", "rolling=50"); For quantification of % cluster boundary occupied by basement membrane proteins (Laminin-111 or Collagen IV) as well as b1 integrin, a custom ImageJ macro was used. In brief, Laminin-111 or Collagen IV images were auto-thresholded using Yen auto-thresholding algorithm to remove any user bias and the intensity of thresholded images was plotted along the basal surface, by manually drawing a line in the Phalloidin channel and overlaying it on the channel of interest. The resulting plot profile had a series of 1s and 0s. % of 1s in the total profile was reported as the % cluster boundary occupied by the protein.
For quantifying average thickness of cell layer per cluster, 5 lines were manually drawn per cluster from the apical to the basal surface and their corresponding lengths were measured in ImageJ. These 5 lengths were averaged and reported as the average thickness of cell layer per cluster.
To quantify lumen volume and total cell volume of hiPSC clusters, the cell membrane of hiPSCs was stained with octadecyl rhodamine B chloride (R18, Thermo Fisher Scientific; 10 mg ml -1 stock solution). R18 dye was diluted 1:1000 in mTeSR1 media and added to alginate gels for 1hr and incubated at 37°C in 5% CO2. Next, gels were washed three times with DPBS.
A Leica SP8 confocal microscope was then used to take 3D image stacks with a 10×/0.4 NA objective or a 20×/0.75 NA objective. Optical image stacks were obtained with a 5-10 µm zaxis interval. Analysis of volumes was based on single hiPSC clusters which were not in contact with any other clusters.
Using R18 fluorescence, total cell volume per cluster was calculated in Imaris software (Bitplane). In brief, R18 images were thresholded and, luminal and basal surfaces were fitted using Imaris software. Volume enclosed between these fitted surfaces was reported as total cell volume per cluster. Next, to estimate the lumen volume further image processing was done using a custom macro in ImageJ. In brief, R18 z-stacks were thresholded to the same level as in Imaris, smoothened using median filter (radius of 2 pixels) and lumens were filled using the Fill Holes algorithm in ImageJ. These z-stacks were visualized in Imaris, and basal surfaces were fitted. This gave the volume enclosed by the basal surface. Lumen volume was then calculated by subtracting the total cell volume per cluster from the volume enclosed by the basal surface. Custom macro in ImageJ was verified to yield the same basal surface as Imaris by quantifying volumes for clusters without any lumens.
Inhibitors were used to test the role of actomyosin contractility and key regulatory pathways, known to be influenced by matrix stress relaxation [31-33, 35, 75] , on hiPSC lumen formation. Inhibitor concentrations used are as follows: 70 μM NSC23766 to inhibit Rac1 (Tocris); 10 μM Y-27632 to inhibit ROCK (STEMCELL Technologies); 25 μM ML-7 to inhibit myosin light chain kinase activity (Tocris); 10 μM Blebbistatin (Abcam); and 2 μM Latrunculin-A (Tocris). NSC23766, Y-27632, ML-7 and Blebbistatin were all added to the respective gels on day 1 and replenished daily. Latrunculin-A was added to control samples 24 hr prior to imaging. Lumen volume and total cell volume per cluster were quantified, as described in previous section, for each of the inhibitor conditions on day 7.
For phospho-kinase protein array studies, hiPSCs were cultured in fast relaxing 0, 150, 1500 μM RGD and slow relaxing 1500 μM RGD gels for 7 days. On day 7, gels were treated with ice cold 50 mM EDTA in DPBS for 10 min with pipette mixing to extract the clusters from gels. Cells were lysed using lysis buffer (R&D Systems, ARY003C). Total protein amount was estimated using Pierce BCA protein assay (Thermo scientific). The human Phospho-Kinase Array Kit (R&D Systems, ARY003C) was then used according to the manufacturer's recommendation.
For comparing morphologies of hiPSC clusters in alginate hydrogels to human epiblasts as well as hiPSC clusters in Matrigel, lumen volume of hiPSC clusters was plotted against the total cell volume per cluster to give a characterize growth curve for different culture conditions