Design and fabrication of 3D geometrically-engineered, permeable membrane for organoid culture platform
To generate organoids that are not only uniform but also mature at a scalable level, we first aimed to establish an organoid culture platform. This platform offers both geometrical constraints for precise organoid regulation and an unrestricted supply of soluble factors, ensuring the efficient diffusion and exchange of nutrients, growth factors, and oxygen (Fig. 1a). We named this platform “UniMat” (Uniform and Mature organoid culture platform), and developed it based on three primary design criteria:
(1) UniMat should efficiently partition individual organoids, providing geometrical constraints that ensure organoid uniformity;
(2) The individual microwells in UniMat should promote cell aggregation, which is essential for the initial formation of cohesive 3D aggregates crucial to organoid development;
(3) UniMat should possess a porous structure, creating a highly permeable environment that facilitates unrestricting supply of soluble factors necessary for organoid maturation.
To realize this concept, we engineered a flat nanofiber (NF) membrane into a 3D permeable microwell array, maintaining a highly porous membrane structure to ensure high permeability to both gases and solutes.
To fabricate the UniMat, we employed the electrospinning process, followed by a subsequent matched-mold thermoforming process, based on our previous work22 (Fig. 1b). Notably, this matched-mold thermoforming process can precisely microstructure a thin and flat NF membrane, ranging in thickness from 50 µm to 150 µm, into a 3D shape. This results in superior shape flexibility and replication fidelity without compromising the porous structure of the NF membrane. Leveraging this capability, we successfully engineered an electrospun free-standing NF membrane with a thickness of 50 µm into a 3D microwell array-structured membrane, resulting in a UniMat. Subsequently, the UniMat was incorporated into the bottom opening of a custom-made 24-well insert wall in a free-standing configuration, allowing for compatibility with standard cell culture plates (Fig. 1c). The successful fabrication of UniMat, with its 3D microwell array-structured NF membrane, was confirmed using scanning electron microscopy (SEM) analysis (Fig. 1d).
In this study, we fabricated the UniMat by microstructuring an electrospun NF membrane to form a V-shaped microwell array (Fig. 1e), promoting the collection and aggregation of seeded cells at the bottom regions of microwells. The V-shaped design enhances cell confinement owing to its inclined walls, ensuring a more defined region for cell placement within every microwell. Importantly, by merely modifying the mold geometry used in the matched-mold thermoforming process, the size of the microwells could be tuned to accommodate the growth of organoids of various sizes. We fabricated three different types of UniMats, each with distinct width and depth dimensions: UniMat400 (width: 400 µm, depth: 343 µm), UniMat600 (width: 600 µm, depth: 517 µm), and UniMat800 (width: 800 µm, depth: 691 µm) (Supplementary Fig. 1). The microwell in the UniMat preserved a highly porous architecture similar to the original NF membrane (Fig. 1f), and possessed a thin wall thickness of approximately 30 µm (Fig. 1g). These characteristics enabled the UniMat to display a high permeability for molecules of varying molecular weights, which was more than twice that of a conventional PET porous membrane (Fig. 1h).
To validate the capability of the UniMat to generate uniform and mature organoids, we selected kidney organoids as a representative lineage of organs. Kidney organoids are notably complex and hold significant biomedical interest because of their potential applications in disease modeling and drug testing. For differentiation of kidney organoids, we meticulously optimized the Morizane protocol23 (Supplementary Notes 1 and 2 and Supplementary Figs. 2 and 3) and seamlessly incorporated this refined protocol into the UniMat. In accordance with this protocol, nephron progenitor cells (NPCs) derived from hiPSCs were seeded onto the UniMat400 on day 9 and subjected to the subsequent differentiation processes within the UniMat400, resulting in the formation of kidney organoids by day 21 (Fig. 1i). Upon differentiation, a vast majority of kidney organoids exhibited nephron-like structures, including podocytes (PODXL+), proximal tubules (LTL+), and distal tubules (CDH1+), within the UniMat (Fig. 1j). Remarkably, approximately 87 ± 5% of all pretubular aggregates were successfully developed into nephron-like kidney organoids, achieving around 5 organoids per mm2 within the UniMat400.
UniMat improves uniformity of organoids
Having demonstrated the successful production of kidney organoids in the UniMat400, we utilized it as a model to explore its potential to reduce the variability of organoids. We assessed the morphological consistency in size and structure of kidney organoids in the UniMat platforms (Fig. 2a), which provide geometric guidance from the NPC stage. This was further compared to the uniformity observed in organoids that were exclusively generated on a plate coated with Geltrex hydrogel (hereinafter referred to as “hydrogel layer”), without any physical constraints (Fig. 2a). Notably, kidney organoids cultured in each UniMat demonstrated a significantly more consistent size distribution compared to those on the hydrogel layer (Fig. 2b and Supplementary Table 1). Moreover, the size of organoids appeared to be regulated through the UniMat-based culture, correlating with the size of the V-shaped microwells in the UniMat (Fig. 2b).
To delve deeper into the mechanism by which UniMat controls the organoid size, we examined the initial aggregate size, which is known to influence organogenesis and organoid formation. Given the distinct sizes and arrays of V-shaped microwells in each of the UniMat400, UniMat600, and UniMat800 (Supplementary Fig. 4a), we noted that within 6 h of seeding, aggregates were formed in sizes corresponding to each specific UniMat (Supplementary Fig. 4b). These measured sizes closely matched with the estimated sizes based on the geometrical data of the microwells and the average number of cell sedimentations per microwell in each UniMat (Supplementary Note 3 and Supplementary Fig. 4c-f). Our findings suggest that by adjusting both the microwell size and the cell seeding density in the UniMat, the initial aggregate size can be effectively modulated. Notably, the size variances among the organoids grown in different UniMats were consistent with the trends observed in the initial aggregate sizes, underscoring the pivotal role of UniMat in determining organoid size.
Having established that the UniMat can effectively regulate organoid size, we further investigated the structural variability of kidney organoids. Using confocal imaging rendered with the Imaris 3D surface rendering software, we quantified the presence of glomerular podocytes (PODXL+), proximal tubules (LTL+), and distal tubules (CDH1+) in individual kidney organoids (Fig. 2c). Notably, the kidney organoid cultured in the UniMat exhibited significantly less structural variability compared to those cultured on the hydrogel layer. These findings indicate that the geometrical constraints provided by the UniMat promote the production of organoids with more uniform sizes and structures (Fig. 2d and Supplementary Table 2). Especially in the UniMat800, we observed a remarkable reduction in structural variability, with an 8-fold decrease in podocytes, a 9-fold decrease in proximal tubules, and a 5-fold decrease in distal tubules, as compared to the organoids cultured on the hydrogel layer (Supplementary Table 2). Based on our data, the UniMat800-cultured kidney organoids exhibited the highest degree of uniformity and they had an average size comparable to those cultured on the hydrogel layer. In this regard, UniMat800 was selected as the representative UniMat for further experiments.
To further assess the potential of UniMat in improving the functional uniformity of organoids, we randomly selected 10 organoids from both the hydrogel layer and UniMat. We then performed a quantitative real-time polymerase chain reaction (qRT-PCR) analysis to quantify the expression levels of key genes associated with podocytes (PODXL), proximal tubules (SLC34A1), and distal tubules (CDH1). Intriguingly, in accordance with the reduced structural variability observed in UniMat, the organoids cultured in UniMat exhibited less gene expression heterogeneity compared to those grown on the hydrogel layer (Fig. 2e and Supplementary Table 3). These results underscore the UniMat’s ability to enhance functional uniformity in organoids, demonstrating its potential to yield more precise and consistent experimental results. Moreover, intriguingly, UniMat showed a capacity to improve the uniformity of organoid formation efficiency, which is quantified as the number of organoids formed per mm2. In determining this efficiency, the variability in the UniMat-based culture was significantly lower than in the hydrogel layer-based culture (Fig. 2f and Supplementary Table 4). Impressively, the formation efficiency was relatively consistent across three independent batches for the UniMat-based cultures (Fig. 2f and Supplementary Table 4). These findings highlight the reproducibility of organoid formation through the UniMat, obviously indicating that the UniMat provides a consistent and reliable platform for generating uniform organoids in a scalable manner.
UniMat enhances maturity of organoids
After examining organoid uniformity, we investigated if the utilization of the permeable UniMat could promote the maturation of organoids, thereby addressing the limitations associated with conventional impermeable microwell platforms. For this experiment, we employed the AggreWell™800 Microwell Plate (hereinafter referred to as “AggreWell”), as a representative conventional microwell platform, with a width of 800 µm, similar to the dimensions of the UniMat used in this study. Confocal microscopic analysis revealed the formation of kidney organoids with nephron-like structures, including glomerular podocytes (PODXL+), proximal tubules (LTL+), and distal tubules (CDH1+), in both the AggreWell and UniMat (Fig. 3a and Supplementary Fig. 5a). Notably, the differentiated regions of glomerular podocytes, proximal tubules, and distal tubules were more prevalent in the UniMat (Fig. 3b), indicating a higher presence of nephron cells in kidney organoids cultured in the UniMat compared to the AggreWell. Furthermore, the podocytes in the UniMat displayed mature characteristics with the basally localized expression of ZO-1 at the edge of cell-cell joints (Fig. 3c), which is a marker associated with tight junction observed within podocyte slit diaphragms in the native kidney tissue24 and mature kidney organoids25. qRT-PCR analysis also highlighted the upregulated expressions of NPHS1 and PODXL (Supplementary Fig. 5b), both essential for the formation of the glomerular filtration barrier. This suggests that our UniMat can generate mature kidney organoids with mature glomerular cells, a contrast to those produced by the AggreWell-based culture. Furthermore, the kidney organoids cultured in UniMat exhibited morphogenesis of tubular structures, as evidenced by polarized proximal tubules with apical enrichment of the brush border marker LTL26 (Fig. 3d). Consistent with this polarization, the mRNA expression levels of AQP1, SLC34A1, and ABCB1 (proximal tubule) as well as UMOD (Loop of Henle) and CDH1 (distal tubule) were also upregulated in the UniMat-cultured kidney organoids (Supplementary Fig. 5c). This indicates enhanced functional potential compared to the kidney organoids cultured in the AggreWell. To further examine the physiological relevance of the tubular functions of these kidney organoids, we performed an in vitro dextran uptake assay. After a 24-h exposure to 10 kDa dextran, selective uptake of dextran within LTL+ proximal tubules confirmed the absorptive functionality of the UniMat-cultured kidney organoids (Fig. 3e). Specifically, these kidney organoids demonstrated the ability to uptake and retain dextran within LTL+ proximal tubule epithelial cells (Fig. 3e), indicating their capacity for reabsorption.
Having identified that kidney organoids formed in the permeable UniMat exhibited higher maturity than those in the impermeable microwells, we next addressed the question of how the unconstrained supply of soluble factors through the 3D geometrically-engineered, permeable membrane in the UniMat influences the maturity of organoids. This question also arose in light of the limitation of conventional impermeable microwell platforms. To explore the impact of UniMat’s permeability on organoid maturity, we fabricated two different UniMat variants with distinct permeability characteristics. We achieved these characteristics by strategically blocking the pores of the NF membrane through precise adjustments of the matched-mold thermoforming process conditions (Fig. 3f). One variant, named UniMat-LP, was designed to have a lower permeability than the original UniMat, resembling the permeability of a conventional PET membrane (Fig. 3g). The second variant, UniMat-ZP, exhibited zero permeability (Fig. 3g). These modifications allowed for controlled studies on the influence of different levels of UniMat permeability on organoid maturation. Our spatiotemporal numerical simulations predicted active glucose diffusion around a single kidney organoid situated in a microwell of the original UniMat, supplied from the culture medium through its highly permeable nanofibrous wall within 72 h (Fig. 3h). However, with the reduced permeability, the glucose diffusion level around the kidney organoid decreased, as observed in both UniMat-LP and UniMat-ZP (Fig. 3h). This glucose diffusion simulation highlighted the effectiveness of UniMat’s permeability in facilitating nutrient supply to organoids. Notably, in concordance with the simulation results, mRNA expression levels of NPHS1, PODXL (podocyte), ABCB1, AQP1, SLC34A1 (proximal tubule), UMOD (loop of Henle), and CDH1 (distal tubule) significantly increased in response to the enhanced permeability (Fig. 3i). Therefore, our findings obviously indicated that the unconstrained supply of soluble factors within the UniMat, facilitated by high permeability through the 3D geometrically-engineered, permeable membrane, enables the generation of mature kidney organoids.
Engineering vascularization of organoids stands as one of the primary objectives in ongoing research efforts to advance organoid technology27. In a previous study, researchers transplanted kidney organoids generated in the impermeable microwell platform into mice to promote vascularization15. Our study aimed to determine if our UniMat could promote in vitro vascularization in kidney organoids without relying on an animal host. We observed a significant increase in vascularization in the kidney organoids when cultured in the UniMat, as evidenced by the increased presence of renal vasculature (PECAM1+) (Fig. 3j). To quantify this vascularization, we evaluated confocal z-stack images of kidney organoids using the AngioTool28 (Supplementary Fig. 6a). The vessel percent area of PECAM+ vasculature in the UniMat-cultured organoids exhibited an increase of over two-fold compared to the AggreWell-cultured organoids (Supplementary Fig. 6b). Additionally, we found that the PECAM+ vasculature of UniMat-cultured organoids had more than a two-fold increase in both junctional density (i.e., branch points per unit area) and average vessel length (i.e., interjunctional distance) compared to the AggreWell-cultured organoids (Supplementary Fig. 6b). qRT-PCR analysis further confirmed the upregulated expression of PECAM1 in kidney organoids cultured in the UniMat (Fig. 3k). To ascertain whether the vascularization induced by a permeable environment extended to the glomerular compartments of organoids, we employed confocal imaging to quantify PODXL+ podocyte clusters invaded by PECAM1+ vascular structures in the kidney organoids cultured in both AggreWell and UniMat (Supplementary Fig. 6c and Supplementary Videos 1 and 2). While PECAM1+ vascular invasion into the glomerular structures was rarely observed in the AggreWell-cultured organoids (Supplementary Fig. 6d and Supplementary Video 1), there was a significant increase in PECAM1+ vascular invasion in the UniMat culture (Supplementary Fig. 6d and Supplementary Video 2). Intriguingly, our analysis using confocal 3D rendering revealed that the vascular structures in the kidney organoids cultured in the AggreWell displayed the characteristics of endothelial cell precursors29, expressing both PECAM1+ and PODXL+, compared to the kidney organoid in UniMat (Fig. 3l), suggesting an immature vascularization of organoids. These findings suggest that the vascular structures within the kidney organoids cultured in the UniMat are more advanced in their development compared to those cultured in the AggreWell.
Single-cell transcriptomic profiling of kidney organoids in UniMat
To investigate the comprehensive capability of UniMat, we performed single-cell RNA sequencing (scRNA-seq) u sing the 10X Genomics platform and analyzed the data using the Seurat R package. We compared the kidney organoids cultured in UniMat to those cultured in the AggreWell to assess the functionality of UniMat’s permeability in the maturation of kidney organoids. After implementing quality control measures, we isolated 17,412 cells from day-25 kidney organoids in AggreWell and 13,098 cells from those in UniMat. These 30,510 cells were subsequently integrated and visualized using the Uniform Manifold Approximation and Projection (UMAP). The cells were categorized into 16 distinct clusters, each of which was annotated by comparing differentially expressed genes to known markers specific for cell types (Fig. 4a). A dot plot across different clusters exhibited unique transcript expression patterns for each cluster (Fig. 4b). Notably, 11 of these clusters were identified as differentiated kidney cells: podocytes (POD; PODXL, MAFB, and NPHS2), early proximal tubules (EPT; SPP1 and FXYD2 with low LRP2 and SLC3A1 expression), proximal tubules (PT; UGT3A1, LRP2, and SLC3A1), loop of Henle/distal tubules (LOH/DT; MAL, WFDC2, and EPCAM), endothelial cells (EC; PECAM1, CDH5, and CAV1), juxtaglomerular cells (JG; REN and PDGFRB), and five mesenchymal clusters (Mesen; COL1A1 and COL3A1). The proportion of differentiated kidney cells in both platforms was similar, accounting for 78.70% of the cells in UniMat and 74% in AggreWell among whole cells (Fig. 4c). The remaining cells included nephron progenitor cells (NPC), defined by expression of SIX2 and EYA1, along with proliferative premature tubular cells (Tub.pre; MKI67, SPP1, and FXYD2), proliferative mesenchymal cells (Prolif.Mesen; CENPF, COL1A1, and COL3A1), and muscle cells (Muscle; MYPLF and MYOG) (Fig. 4c).
We further examined the cell-type proportions in each kidney organoid from both AggreWell and UniMat (Supplementary Table 5 and Supplementary Fig. 7). These findings were then juxtaposed with a recently published scRNA-seq dataset for human and mouse nephrons with a focus on core nephron segments such as podocytes, broad proximal tubules (including early proximal tubules), and loop of Henle/distal tubules (Fig. 4d). In the AggreWell samples, podocytes constituted 35.74%, broad proximal tubules 48.58%, and loop of Henle/distal tubules 15.68% (Fig. 4d). This distribution aligns with prior findings using the Morizane protocol30, 31, which showed high percentages of podocytes and proximal tubules. Remarkably, kidney organoids from the UniMat exhibited a 4.5-fold decrease in podocytes (7.97%), but higher proportions of broad proximal tubules (60.93%) and loop of Henle/distal tubules (31.10%). These proportions more closely resemble the cellular composition of natural nephrons in both mice32, 33 and humans34, 35 (Fig. 4d). Significantly, within the broad proximal tubule category, the proportion of early proximal tubules was 18.7% in UniMat, as compared to 62% in AggreWell, highlighting a higher presence of mature proximal tubules in the former. Beyond nephron cells, there was a noticeable difference in the NPC cluster: AggreWell exhibited a fraction 1.7 times higher than that of UniMat (Fig. 4e). This further demonstrated the enhanced maturity of kidney organoids cultivated on the UniMat platform (Fig. 4e and Supplementary Table 5). Given that tissue structure and function hinge on cellular compositions, these results emphasize the superior in vivo relevance of kidney organoids generated using UniMat.
We next examined gene expression profiles of podocytes, proximal tubules, loop of Henle/distal tubules as well as endothelial cells clusters (Fig. 4f and Supplementary Fig. 8). The analysis revealed elevated expression of genes related to differentiation in the podocytes, loop of Henle/distal tubules, and endothelial cells clusters. This suggests a greater degree of maturity in kidney organoids cultivated using the UniMat. Although the gene expression profiles in the proximal tubule cluster appeared similar between UniMat and AggreWell, a low proportion of early proximal tubules in UniMat (Supplementary Fig. 9) indicates a more mature state of these tubules.
To explore the potential pathways and functions that differentiate kidney organoids in UniMat compared to those in AggreWell, we carried out Biological Process Gene Ontology (GO) enrichment analysis on genes that were significantly upregulated in podocytes, proximal tubules, loop of Henle/distal tubules, and endothelial cells clusters within the UniMat (Fig. 4g). In the podocyte cluster, UniMat showed elevated expression of genes associated with glomerulus development (GO:0032835) and glomerular basement membrane development (GO:0032836) (Supplementary Table 6). For both the proximal tubule and the loop of Henle/distal tubule clusters, the most enriched GO terms were associated with the negative regulation of mesenchymal cell apoptosis in metanephros development (GO:1900212), metanephric nephron tubule formation (GO:0072289), and ureter morphogenesis (GO:0072197) (Supplementary Table 7). These observations corroborate the advanced tubular structures observed in the UniMat-cultured kidney organoids. In the loop of Henle/distal tubule cluster, enriched GO terms revealed elevated expression of genes like CLDN3, CLDN19, and CRB3, which are linked to the positive regulation of cell junction assembly (GO:1901890) (Supplementary Table 7). This suggests enhanced epithelial junctions in kidney organoids cultured in UniMat. Additionally, the GO term related to metanephric distal convoluted tubule development (GO:0072221) was enriched in UniMat (Supplementary Table 7), indicating a higher propensity for kidney organoids to differentiate into distal tubules in UniMat as compared to AggreWell, as shown in Fig. 4d. In the endothelial cell cluster, upregulated genes such as SOX17, SOX18, NRP1, and CLIC4, played a role in endothelial cell differentiation (GO:0060956) (Supplementary Table 8). These findings are in line with the advanced development of kidney organoids in UniMat.
Potential of UniMat in standardizing organoid-based assays
While earlier researches have utilized kidney organoids to simulate polycystic kidney disease (PKD) and mimic its pathological features, the inconsistent morphology of these organoids often resulted in considerable variability in both size and cystic regions36, 37. Recognizing that UniMat facilitates the generation of kidney organoids with shape and functional consistency and higher maturity in tubular structures across different nephron segments, we employed these organoids for PKD modeling to evaluate UniMat’s potential as a drug testing platform (Fig. 5a). We exposed the UniMat-cultured kidney organoids to a 30 µM concentration of forskolin for 48 h, replicating the cAMP-induced cystic growth commonly seen in PKD. This enabled us to evaluate the platform's suitability for generating PKD organoids38. In response to the forskolin treatment, we observed both an enlargement and the formation of cysts within organoids. This confirmed the UniMat’s ability to reliably produce PKD organoids on a scalable level (Fig. 5b).
To evaluate if the uniformly mature kidney organoids generated in UniMat enhance the reliability of PKD modeling, we compared the cyst-forming features of organoids cultured in the hydrogel layer, AggreWell, and UniMat. To facilitate closer observation, the kidney organoids cultured in UniMat and AggreWell were transferred to individual ultra-low attachment plates prior to 48-h exposure to forskolin at concentrations of 10 and 30 µM (Fig. 5a). Notably, cysts became evident within 24 h of forskolin exposure, predominantly in the LTL+ and CDH1+ tubular regions (Supplementary Fig. 10). We quantified the dimensions of kidney organoids collectively, both before and after the induction of cyst formation, without monitoring each organoid individually. As expected, a considerable variability in the size and morphology of kidney organoids cultured on the hydrogel layer resulted in inconsistent cyst formation, in terms of both location and dimensions (Fig. 5c). A similar pattern of variability and incomplete cyst formation was observed in organoids cultured in the AggreWell, implying the existence of undifferentiated, non-renal structures unresponsive to forskolin treatment (Fig. 5c). Additionally, cyst formation in kidney organoids from both hydrogel layer and AggreWell was erratic, regardless of forskolin dosage and time of exposure (Fig. 5e). In contrast, organoids in the UniMat exhibited uniform cysts throughout their entire structure, resulting in consistent increases in overall sizes. This can be attributed to the high and consistent differentiation of nephrons (Fig. 5c). When comparing the percentage area of individual forskolin-treated PKD organoids to their average size before treatment, we found that significant dose-dependent responses were observed uniquely in the UniMat (Fig. 5e).
To explore the impact of uniform disease modeling on drug testing results, we exposed PKD organoids, induced by forskolin (30 µM), to CFTR inhibitor-172 at concentrations of 50 or 100 µM (Fig. 5a). This PKD drug, which targets the cytoplasmic side of CFTR, had previously demonstrated its potential to inhibit cyst growth in PKD mice and in vitro models39, 40. The PKD organoid induced from kidney organoids cultured in UniMat displayed consistent reductions in size (Fig. 5d), with the percentage area decreasing significantly in a dose-dependent manner (Fig. 5f). In contrast, the PKD organoids from both hydrogel layer and AggreWell experienced inconsistent changes in cyst dimensions. Our results indicate that, within the UniMat platform, we can validate not only the formation but also the reduction of cysts by simply comparing average size changes across the organoid population, without the need to track individual organoids. In summary, our study clearly demonstrated that UniMat offers significant advantages in disease modeling and drug testing by facilitating the uniform and mature differentiation of organoids.