Oxygenated Scaffolds for Pancreatic Endocrine Differentiation from Induced Pluripotent Stem Cells

Abstract

A 3D microenvironment is known to endorse pancreatic islet development from human induced pluripotent stem cells (iPSCs). However, oxygen supply becomes a limiting factor in a scaffold culture. In this study, oxygen-releasing biomaterials are fabricated and an oxygenated scaffold culture platform is developed to offer a better oxygen supply during 3D iPSC pancreatic differentiation. It is found that the oxygenation does not alter the scaffold’s mechanical properties. The in situ oxygenation improves oxygen tension within the scaffolds. The unique 3D differentiation system enables the generation of islet organoids with enhanced expression of islet signature genes and proteins. Additionally, it is discovered that the oxygenation at the early stage of differentiation has more profound impacts on islet development from iPSCs. More C-peptide⁺/MAFA⁺ β and glucagon⁺/MAFB⁺ α cells formed in the iPSC-derived islet organoids generated under oxygenated conditions, suggesting enhanced maturation of the organoids. Furthermore, the oxygenated 3D cultures improve islet organoids’ sensitivity to glucose for insulin secretion. It is herein demonstrated that the oxygenated scaffold culture empowers iPSC islet differentiation to generate clinically relevant tissues for diabetes research and treatment.

1. Introduction

Diabetes mellitus has become a global epidemic in recent years and will be the third leading cause of death by 2030.^[1] The diseases induce other conditions, including heart disease and stroke, high blood pressure, kidney disease, and blindness. While islet transplant is a promising alternative to traditional insulin therapy, there is a severe scarcity of donor islet tissues needed for transplant. This shortage has led to considerable efforts in searching for in vitro generation of human endocrine cells.^[2] Potential cell sources for generating pancreatic β cells include stem cells, endocrine progenitors, trans-differentiation of other cells, and β cell itself. Among them, human pluripotent stem cells (hPSCs), including human embryonic stem cells (hESCs) and human induced pluripotent stem cells (iPSCs) have been promising sources for generating β cells.^[2,3]

One approach taken to guide hPSC differentiation into pancreatic β cells is to identify signaling molecules and culture conditions.^[3a,c,d,4] Genetically introducing β cell transcription factors such as PDX1 and NKX6.1 to hPSCs and/or progenitors to enforce formation of mature endocrine β cell is another strategy.^[5] However, this approach requires genetic modification using a viral vector, resulting in the introduction of undesirable genomic edits to cell origin that may have adverse effect on therapeutics. Delivery of synthetic mRNAs encoding PDX1 and NKX6.1 to hESCs have been explored for its efficacy to enhance pancreatic endocrine lineage differentiation recently.^[6] Our recent study discovered that a short peptide motif developed from a pancreatic extracellular matrix (ECM) protein is able to boost the glucose sensitivity for glucose responsive insulin secretion from iPSC-derived islets.^[7] Nevertheless, generating biologically functional islets that consisting of all major types of islet cells, including α, β, δ, and γ cells remains a significant challenge. It has been reported that decellularized pancreatic materials induce the formation of human islets that harbor all major pancreatic endocrine cell types.^[3a,8] A body of evidence suggests that scaffolds offer better physiochemical niches for promoting stem cell lineage-specific commitment, as they can better mimic the microenvironment in the body.^[9] Hence, another approach to generate endocrine cells is to first direct iPSCs differentiation into endocrine progenitors or immature endocrine cells under a 2D environment, and then switch to a 3D culture for endocrine cell maturation.^[3a,d,4b,8] This is due to the fact that iPSCs are sensitive to scaffolding materials and oxygen (O₂) supply becomes a limiting factor in a scaffold culture because of a low O₂ solubility in a culture medium and constrained O₂ diffusion across the scaffold. Our early study revealed that a biomimetic 3D culture could considerably improve the generation of endocrine cells from hESCs as compared to traditional 2D cultures.^[10] These hESC-derived islet-like organoids formed inside the scaffolds consisted of all four subtypes of pancreatic endocrine cells, i.e., α, β, δ, and γ cells. The cytostructural analysis showed an architecture characteristic of human adult islets. Both β cells and non-β cells were mixed to form organoids that secrete insulin in response to glucose levels. Nevertheless, the impaired O₂ supply and O₂ diffusion within the scaffolds hinders the wide use of scaffold systems for generating functional cell lineages. In addition, previous studies indicated that normoxic conditions favor hESC differentiation to definitive endoderm, which is the first and a key step toward the generation of pancreatic endocrine.^[11] Particularly, O₂ is required to promote the development of endocrine progenitors and in turn expedite β cell development from progenitor cells.^[11c,d] Enhanced O₂ supply improved mouse β cell viability.^[12]

On the other hand, fabricating O₂-releasing materials to enhance local O₂ supply has been an alternative approach for tissue repair. These materials have been utilized to improve cardiac cell survival and differentiation, ischemic limb vascularization and skeletal muscle regeneration, diabetic wound healing, cardiac cell engraftment, bone regeneration, mouse β cell culture etc.^[13] In this study, we fabricated O₂-releasing materials and encapsulated iPSCs inside scaffolds for 3D iPSC pancreatic differentiation. We demonstrated that embedding oxygenators does not alter scaffold’s mechanical properties. The in-situ oxygenation increased O₂ supply to the cells, leading to an elevation of islet organoids’ sensitivity to glucose for insulin secretion. Furthermore, the unique 3D differentiation system enabled the generation of islet organoids with enhanced expression of islet signature genes and proteins. More C-peptide/MAFA dual-positive β and glucagon/MAFB dual-positive α cells formed in the iPSC-derived islet organoids under oxygenated 3D cultures. Therefore, this study revealed the critical role of oxygenation played in iPSC pancreatic islet development, filling a gap in iPSC pancreatic differentiation. The knowledge accumulated from the study would help design a better 3D culture system for generating clinically relevant tissues from iPSCs.

2. Results and Discussion

2.1. Construction of an Oxygenated 3D iPSC Differentiation Platform

We first characterized the net O₂-releasing kinetics from the O₂-releasing materials, also referred to as oxygenators (Figure 1A). The O₂ was released when calcium peroxide (CaO₂) in the polydimethylsiloxane (PDMS) was hydrolyzed for more than 20 d (Figure 1A). Next, to determine whether the oxygenators are cytotoxic to cells, we prepared oxygenated collagen/Matrigel scaffolds (Figure 1B) and performed live and dead dual-color staining. We included Matrigel in the collagen scaffold to enhance the mechanical strength of the scaffold to support iPSC proliferation and differentiation, as shown in our previous study.^[10] We observed that there was no significant difference in cell viability in the presence or absence of the oxygenator (Figure 1C–F), suggesting that the oxygenators are not cytotoxic to the cells. Trypan blue assay further confirmed these results. In this assay, we harvested all the cells from the scaffolds and counted live and dead cells. The cell viability in the presence or absence of the oxygenator was similar (Figure 1G).

Figure 1. Oxygenated scaffolds developed for iPSC pancreatic differentiation.

A) Oxygen release kinetics of the oxygenators. Results were from three independent experiments and shown as mean ± SD. B) Schematic diagrams of cell-laden oxygenated scaffolds. C,D) Cell viability in the non-oxygenated (C) and oxygenated (D) scaffolds at 24 h post seeding. E,F) Cell viability in the non-oxygenated (E) and oxygenated (F) scaffolds at 48 h post seeding. G) Cell viability in the non-oxygenated (NOXY) and oxygenated (OXY) scaffolds at 24 and 48 h post cell seeding, quantified by trypan blue assay. Cell viability under oxygenated condition was normalized to that under non-oxygenated condition. Results were from three independent experiments and shown as mean ± SD.

2.2. The Cell-Oxygenator Interaction and its Effect on the Architecture of the Scaffold

To interrogate whether the cell-oxygenator interaction alters the microarchitecture of the scaffolds, we scrutinized the microscopic structure of the acellular scaffolds in the presence or absence of the oxygenator by scanning electron microscopy (SEM) (Figure 2A–O). We examined the areas next to the oxygenator (with direct contact) and the areas away from the oxygenators (no direct contact). We observed grape-like structures in both areas (Figure 2B,C). As the major components of Matrigel are laminin (≈60%), collagen IV (≈30%), entactin (≈8%), etc. and scaffolds were prepared by mixing collagen I with Matrigel at a ratio of 2:1, we speculated that the fibers that we observed under SEM were collagen fibers. In the area where there was a direct interaction between collagen fibers with the oxygenator (Figure 2L–O), larger pore sizes of collagen fibrous network were observed (Figure 2D–G). In the area where there was no direct interaction between collagen fibers and the oxygenators (Figure 2H–K), their microstructures were similar to the scaffolds without the oxygenation (Figure 2D–G). Under a higher magnification (40 000×), we found partially interrupted collagen fibers in the areas where there was a direct interaction between the oxygenator and the collagen fibers (Figure 2O arrows indicated). In addition, due to direct contact between the oxygenator and the collagen, fibrous diameter and pore size became larger (Figure 2P–Q). The average fibrous diameter and pore size in the area having direct contact were 0.48 and 3.04 μm, respectively, whereas the average diameter and pore size in the areas having no contact or in the nonoxygenated scaffold were 0.33–0.36 and 1.88–2.42 μm, respectively.

Figure 2. SEM images of acellular collagen scaffolds in the presence or absence of the oxygenators.

A) Yellow area indicates the space left behind after the oxygenator was removed. The image was taken at 100×. B) Grape-like structures in the presence of the oxygenator at 5000×. C) Grape-like structures in the absence of the oxygenator at 5000×. D–G) Scaffolds in the absence of the oxygenators. H–K) Scaffolds in the presence of the oxygenators but the areas have no direct interaction with the oxygenator material. L–O) Scaffolds in the presence of the oxygenators in the areas next to the oxygenator material. Magnifications: (D, H, and L): 5K; (E, I, and M): 10K; (F, J, and N): 20K; (G, K, and O): 40K. P) Estimation of mean value of fibrous diameter as measured by ImageJ. n = 259 for no oxygenator group; n = 138 for the areas without direct contact to the oxygenator; n = 114 for the areas with direct contact to the oxygenator. *: p < 0.001. Q) Estimation of mean value of pore size as measured by ImageJ. n = 328 for no oxygenator group; n = 281 for the group without direct contact to the oxygenator; n = 182 for the group with direct contact to the oxygenator. *: p < 0.005, **: p < 0.001. ns: not significant.

In the next experiment, we investigated how the cells interact with the collagen fibers and whether such interaction modifies the structure of the cell-laden scaffolds. We mixed iPSCs with collagen/Matrigel and the oxygenator as illustrated in Figure 1B (Figure 3C–F). iPSCs mixed with collagen/Matrigel without the oxygenator were used as a control (Figure 3A,B). As cells got closer to the oxygenator, either a film-like structure formed on the fibers or a much more protein-like structure formed surrounding the cell cluster (Figure 3E,F). Compared to acellular scaffolds (Figures 2 and 3), cell-laden scaffolds exhibited relatively thinner fibers and smaller pore sizes (Figure 3A–F). Statistical analysis revealed that the average of fibrous diameter was 0.146 μm in the non-oxygenated scaffolds (Figure 3G). In contrast, the collagen fibers in the oxygenated scaffolds displayed a similar thickness in the area with or without direct contact to the oxygenator, which was 0.115 and 0.117 μm on average, respectively. The average pore size of cell-laden non-oxygenated scaffolds was 1.255 μm. This number was 1.530 and 1.329 μm, respectively, in the area without or with direct contact to the oxygenator (Figure 3H). Hence, cells grown within the scaffolds slightly increased the pore size of collagen fibers in the area without direct interaction with the oxygenators.

Figure 3. Microstructure of iPSC-oxygenator embedded scaffolds.

A,B) Cell-laden scaffolds in the absence of oxygenators. C,D) Cells next to the oxygenators. E,F) Cells away from the oxygenators. G) Diameters of the collagen fibers measured by ImageJ. n = 72 for the group without oxygenator; n = 103 for the areas without direct contact to the oxygenator (OXY). n = 91 for the areas with direct contact to the oxygenator. ***: p < 0.001. H) Pore sizes of collagen f networks ascertained by ImageJ. n = 248 for no oxygenator group; n = 331 for the areas without direct contact to the oxygenator; n = 354 for the areas with direct contact to the oxygenator. *: p < 0.05, **: p < 0.01. ns: not significant. Scale bars: 2 μm.

2.3. Mechanical Properties of the Oxygenated Scaffolds

Mechanical properties of a scaffold play critical roles in determining cell behaviors.^[14] Accordingly, we ascertained whether the embedded oxygenator alters mechanical properties of the scaffold. As shown in Figure 4A, we detected a continuous increase in the storage modulus (G′) as a function of increasing angular frequency. This suggested a gel-like character with an increase in the oscillatory frequency. The prevalence of the elastic component over the viscous component indicates that the scaffolds follow a typical collagen gel behavior. The loss modulus (G″) tended to fluctuate and remained mostly unchanged. The elastic modulus is linearly related to the storage modulus. The values exhibited the same behavior as shown in Figure 4A. Scaffolds with or without in-situ oxygenation adhered to these behavioral trends with no statistically significant deviation in the moduli values between two groups (Figure 4A). We then evaluated the stiffness of these scaffolds at 12.6 rad s⁻¹ as this appeared to be the end of the linear regime in both types of scaffolds. The scaffolds with and without oxygenation were found to have an average stiffness of 4.03 and 4.74 Pa, respectively, at frequency of 12.6 rad s⁻¹ (p = 0.396) and thus no significant difference was observed (Figure 4B). Furthermore, the oxygenated scaffolds did not appear to affect viscosity (Figure 4C). At very small near zero frequency (0.1 rad s⁻¹), the scaffolds with oxygenation had a viscosity of 27.1 Pa s as compared to 29.2 Pa s in the scaffolds without oxygenation (p = 0.542) (Figure 4D). Julias et al. performed a similar 1% strain oscillatory test on various type I collagen concentrations.^[15] Their 2.0 mg mL⁻¹ collagen exhibited slightly higher storage and loss moduli than ours, albeit the general viscoelastic and gel-like trends were the same. Overall, our results suggested that embedding of the oxygenator inside the scaffolds does not affect mechanical properties of the scaffolds. All these scaffolds possess typical gel-like behaviors.

Figure 4. Mechanical properties of non-oxygenated and oxygenated acellular scaffolds.

A) G′ and G″ moduli at various angular frequencies of the scaffolds with and without in site oxygenation. B) Stiffness of oxygenated and non-oxygenated scaffolds determined at 1% strain and 12.6 rad s⁻¹. C) Viscosity of oxygenated and non-oxygenated scaffolds determined at various angular frequencies. D) Viscosity of oxygenated and non-oxygenated scaffolds measured at 1% strain and 0.1 rad s⁻¹. Results were from three independent experiments and shown as mean ± SD. ns: not significant.

2.4. Oxygen Levels Inside the Scaffolds with and Without In Situ Oxygenation During iPSC Differentiation

To examine the improvement of O₂ supply inside an oxygenated scaffold during iPSC differentiation, we differentiated iPSCs into definitive endoderm, which is the first and a key step toward pancreatic endocrine development using a protocol schemed in Figure 5A. We monitored the spatial distribution of O₂ tension inside the scaffolds, patricianly near the top and bottom during the differentiation. As shown in Figure 5B,D–G, O₂ tension at the top of the scaffolds was around 100% of saturated air regardless the absence (NOXY) or the presence (OXY) of the oxygenator. This suggested that dissolved O₂ from the air in the cell differentiation medium reached to the top of the scaffolds. In contrast, the O₂ tension in the central and peripheral areas at the bottom of the non-oxygenated scaffolds was about 74% and 75% of saturated air, respectively. This number increased to ≈95% of saturated air in the peripheral area and 89% of saturated air in the central area, respectively, in the oxygenated scaffolds (Figure 5B). Representative distributions of O₂ tension across the bottom of non-oxygenated and oxygenated scaffolds were shown in Figure 5D,F. Overall, the O₂ levels increased in the oxygenated scaffolds, improving the O₂ supply to cells at the bottom of the scaffolds (Figure 5D–G). Additionally, visualizing cell spatial distribution indicated that more cells survived at the bottom of the oxygenated scaffolds (Figure 5H–I). More cells were able to grow at the bottom as compared to those in the non-oxygenated group. These results revealed that the oxygenated scaffolds supplied cells with sufficient O₂, improving the iPSC proliferation and differentiation.

Figure 5. Oxygen tension in oxygenated (OXY) and non-oxygenated (NOXY) scaffolds at the end of iPSC definitive endoderm differentiation.

A) A schematic diagram of an iPSC differentiation protocol with key molecules applied for definitive endoderm development in the oxygenated scaffolds. NaB: sodium butyrate. BSA: bovine serum albumin. B) Oxygen tension at the surface (n = 12) and the bottom (n = 12) of the scaffolds measured by a well-established needle-type microsensor for top of the scaffolds and sensor foil for the bottom of the scaffolds. Results were obtained from four independent experiments and shown as mean ± SD. ***p < 0.01; ns: not significant. C) The oxygen sensor foil was placed under the cell-laden scaffold. The portable detector was mounted underneath the plate to measure the distribution of O₂ tension at the bottom of a scaffold. D) Image of a color-coded oxygen map generated at the bottom of the scaffold in the absence of the oxygenator (NOXY). E) Oxygen tension across the center line of the scaffold indicated by the line in (C). F) Distribution of oxygen tension at the bottom in the presence of the oxygenator (OXY). G) Oxygen tension across the center line of the scaffold indicated by the line in (E). (H,I) Spatial distribution of iPSC-derived definitive endoderm inside the scaffolds viewing from top to bottom. Cells were stained with DAPI (blue).

Furthermore, we investigated O₂ tension inside the two types of scaffolds after iPSC differentiation into pancreatic progenitors (Figure 6A). As shown in Figure 6B, the O₂ tension at the bottom and the central area of the non-oxygenated scaffold dropped to 27% of saturated air, while it maintained around 50% of saturated air in the oxygenated scaffolds. In addition, at the bottom edge area, the O₂ tensions were around 42% and 57% of saturated air, respectively, in non-oxygenated and oxygenated scaffolds, suggesting an improved O₂ supply by the oxygenator. We noticed relatively low O₂ tension inside the scaffolds at the end of pancreatic progenitor differentiation as compared to that at the definitive endoderm stage (Figure 6C–F). This is due to a reduced release of O₂ from the oxygenators from day 10 to day 15 as shown in Figure 1A. Nevertheless, there were more cells in oxygenated scaffolds as compared to those in the non-oxygenated scaffolds (Figure 6G–H), indicating the improvement of iPSC survival under oxygenation within the scaffolds during differentiation.

Figure 6. Oxygen tension within the non-oxygenated and oxygenated scaffolds at the end of iPSC pancreatic progenitor differentiation.

A) A schematic diagram of an iPSC differentiation protocol with key molecules applied for pancreatic progenitor development in oxygenated scaffolds. NOXY: without oxygenation. OXY: with oxygenation. RA: retinoic acid. KGF: keratinocyte growth factor. LDN: LDN193189. ILV: (−)-indolactam V. VC: ascorbic acid. T3: 3,3′,5-triiodo-l-thyronine sodium salt. Rep: Repsox. B) Percentage of oxygen tension at the surface (n = 18) and the bottom (n = 16) of the scaffolds. Results were shown as mean ± SD. ***p < 0.001. C) Distribution of oxygen tension at the bottom of a non-oxygenated scaffold. D) Oxygen tension across the center line of the scaffold indicated by the line in (C). E) Distribution of oxygen tension at the bottom of an oxygenated scaffold. F) Oxygen tension across the center line of the scaffold indicated by the line in (E). G,H) Spatial distribution of iPSC-derived pancreatic progenitors inside the scaffolds viewing from top to bottom. Cells were stained with DAPI (blue).

2.5. Oxygenated Scaffolds Enhance Islet Differentiation from iPSCs

To further validate the concept of oxygenated 3D scaffold culture platform, we examined whether the in-situ oxygenation enhances iPSC islet differentiation. We developed a five-stage differentiation protocol to differentiate iPSCs to islet cells in the presence or absence of oxygenators (Figure 7A). 2D cultures served as a control. To determine optimal 3D microenvironment for islet development, we investigated the effects of oxygenation and scaffold’s thicknesses on iPSC pancreatic differentiation. iPSCs were encapsulated inside the scaffolds with 3 and 5 mm thicknesses with or without in-situ oxygenation. As shown in Figure 7B, key marker gene expressions of pancreatic progenitors PDX1 and NKX6.1 in both 3 and 5 mm thick scaffolds under oxygenation substantially elevated. Interestingly, there was no significant difference in the expression of marker genes in cells differentiated in 3 and 5 mm thick scaffolds (Figure 7B). Hence, we established scaffold differentiation platform with thickness of 3 mm for further investigation of the effect of oxygenation on islet endocrine development.

Figure 7. Oxygenated 3D microenvironment promotes the formation of pancreatic progenitors and subsequent pancreatic endocrine development from iPSCs.

A) A 29-day differentiation protocol with key molecules applied in stepwise differentiation procedures for islet organoid development within scaffolds. NOXY: without oxygenation. OXY: with in-situ oxygenation. Nic: nicotinamide. N-Cys: N-acetyl cysteine. Tro: trolox. γ-sec: γ-secretase inhibitor XX. B) Marker gene expressions of pancreatic progenitors under various microenvironments. Results were from five independent experiments and shown as mean ± SD. Gene expression was normalized to IMR90 cells. Human pancreas (hPancreas) served as a positive control. 3D-3-NOXY: 3 mm thickness of the scaffold without oxygenation. 3D-3-OXY: 3 mm thickness of the scaffold with oxygenation. 3D-5-NOXY: 5 mm thickness of the scaffold without oxygenation. 3D-5-OXY: 5 mm thickness of the scaffold with oxygenation. Different letters denote significant differences from one another. C) Gene expression of key markers of islet cells. Results were from four independent experiments and shown as mean ± SD. Groups with different letters denote significant differences from one another. *p < 0.05, **p < 0.01, ***p < 0.001. Significant differences in 3D groups compared to 2D cultures were determined and represented as #: p < 0.05; ## p < 0.01; and ### p < 0.001.

In addition, the expression of signature genes of iPSC-derived islet organoids in oxygenated scaffold increased substantially (Figure 7C). These genes included insulin (INS), glucagon (GCG), somatostatin (SST), NKX6.1, PDX1, and PTF1A. The fold increase in oxygenated scaffolds compared to non-oxygenated scaffolds was 3.36, 1.70, 1.65, 2.60 and 1.75 for genes INS, GCG, SST, PDX1, and NKX6.1, respectively. It seems like although the gene expression of pancreatic polypeptide (PPY) is not affected by oxygenation, the 3D culture permitted significant enhanced generation of PPY-expressing cells as compared to 2D cultures (Figure 7C). In addition, signature markers for islet development NKX6.1, INS, GCG, SST, PPY expressed remarkably higher in iPSC-derived islet organoids formed in the scaffold cultures regardless oxygenation, which is consistent to our early finding comparing 3D to 2D cultures for stem cell differentiation.^[9c,10] These experimental results demonstrated the importance of O₂ supply to iPSC 3D scaffold differentiation.

To determine whether replenishing the oxygenators with new ones in the middle of the differentiation could further enhance islet development from iPSCs, we replaced the oxygenators on day 19 and detected the key pancreatic markers, such as NKX6.1, INS, and GCG, at the end of 5-stage differentiation (Figure 8A). We did not observe any significant enhancement of the expression of these marker genes, suggesting the O₂ supply at the definitive endoderm stage is more indispensable than the later stage of endocrine differentiation. This is consistent with previous study reporting that supplying of 60% of O₂ gas flow at the definitive endoderm stage improves insulin gene expression in hESC differentiation into pancreatic insulin-producing cells.^[11d] By contrast, increase of the O₂ gas flow during other stages of pancreatic differentiation did not increase insulin gene expression.^[11d]

Figure 8. Oxygenation promotes organogenesis and morphogenesis of iPSC-derived islet organoids.

A) A comparison of gene expression of NKX6.1, INS, and GCG with and without the replacement of oxygenators at day 19 during the differentiation. Results were from three independent experiments and shown as mean ± SD. B) A comparison of insulin stimulation index of iPSC-derived islet organoids generated under indicated conditions. Results were from four independent experiments and shown as mean ± SD. 3D-3-NOXY: 3 mm thickness of the scaffold without oxygenation. 3D-3-OXY: 3 mm thickness of the scaffold with oxygenation. Human islets (hIslet) were used as a positive control. *: p < 0.05, and NS: nonsignificant. C–I) Enhancement of organogenesis and morphogenesis of islet organoids generated in oxygenated scaffolds. At the end of differentiation, the organoids were immunofluorescently labeled for (C) C-peptide (CP, red) and glucagon (GCG, green); (E) somatostatin (SST, green) and pancreatic polypeptide (PPY, red); (G) MAFA (green) and CP (red); and (I) MAFB (green) and GCG (red). Cells were counterstained with DAPI (blue). Scale bars, 50 μm. D,F,H,J) The organoid structures generated were semi-quantitatively analyzed using ImageJ to estimate approximate populations of cell subtypes. D) n = 25 images from five independent differentiation experiments for oxygenated scaffolds; n = 12 images from three independent differentiation experiments for non-oxygenated scaffolds. F,H,J) n = 6–10 images per individual condition from three independent experiments. Results were shown as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001. ns: not significant.

Next, we assessed whether islet organoids’ glucose sensitivity is elevated when they are developed under oxygenation. As shown in Figure 8B, the insulin stimulation index of islets developed under oxygenation was 2.09 ± 0.42, similar to human islets’ index that is 2.34 ± 0.42. The stimulation index in the 3D scaffolds without oxygenation was 1.09 ± 0.21, less responsible to glucose. Finally, we investigated the organogenesis and morphogenesis of the islet organoids generated under oxygenation. The images (n = 6–25) of each subtype of islet cells were quantified. We found that islet organoids generated under oxygenation exhibited increased expression of C-peptide (CP+)/GCG− (51.0±15.2%) as compared to those generated without oxygenation (37.9±13.2%) (Figure 8C,D). Remarkably, the percentage of CP⁺/GCG⁺ cells decreased to 5.6 ±0.9% in the oxygenated condition, while as this number was 13.1±8.0% without oxygenation (Figure 8D). In addition, the subsets of PPY⁺/SST⁻, PPY⁻/SST⁺, and PPY⁺/SST⁺ cells increased in the islet organoids generated with oxygenation, having 3.1±1.0% PPY⁺/SST⁻, 11.7±2.0% PPY⁻/SST⁺, and 3.8±1.2% PPY⁺/SST⁺ cells compared to that of 1.2±0.7% PPY⁺/SST⁻, 5.1±2.6% PPY⁻/SST⁺, and 1.4±0.8% PPY⁺/SST⁺ cells generated without oxygenation (Figure 8E,F). Impressively, there were more CP⁺/MAFA⁺ (19.0±5.1% vs 14.0±4.1%) and GCG⁺/MAFB⁺ cell (12.1±4.5% vs 6.4±2.5%) populations in the oxygenated cultures (Figure 8G–J).

In 2D cultures, O₂ is supplied to cells through unidirectional diffusion from the gas–liquid interface. In O₂-permeable cultures, O₂ is suppled to cells from the bottom of a culture plate through an O₂ permeable membrane,^[16] which is more suitable for monolayer cultures. The membrane is less effective for cell cultures within a scaffold, as a diffusion barrier built by the scaffold hinders the O₂ diffusion within the scaffold. In contrast, oxygenators can be placed inside a scaffold. Multiple oxygenators can be included inside a scaffold if needed, ensuring an even distribution of O₂ to a cell-laden scaffold. As we demonstrated in this study, the islet development was considerably enhanced due to the improvement of O₂ supply within the scaffolds. These observations are supported by others work reported previously. For instance, Suvarnapathaki et al. developed an O₂-generating tissue scaffold, comprised of CaO₂/polycaprolactone micro-particles. The implanted O₂-releasing scaffold supported bone remodeling and vascularization at defect site.^[13d] Similar O₂-releasing materials have been proven to be able to advance cardiac cell survival, differentiation and engraftment, vascularization, skeletal muscle regeneration, and wound healing under hypoxic conditions.^{[13b,e–g,17]} McReynolds et al. established a culture model with encapsulation of an O₂-generator made by microspheres with hydrogen peroxide/poly(N-vinylpyrrolidone) as core and poly(lactide-co-glycolic acid) as shell. They found that cell seeding density and the capacity of the oxygenator are two critical parameters in the optimization of the pancreatic β cell cultures.^[18] O₂-relaseing materials have also been employed to examine the impact of O₂ tension on mouse β cell line cultured under hypoxic conditions.^[12] In this study, we demonstrated, for the first time, the oxygenated scaffold culture is beneficial to iPSC islet differentiation.

3. Conclusion

A number of technologies have been developed and tested for improving O₂ supply within a scaffold. Vascularization of a scaffold is one of the prominent approaches to augment O₂ supply to cells within a scaffold. However, the formation of a vascular system inside a scaffold has not been successful, as it involves coculturing multiple cell types for the formation of an efficient vascular network. Recirculating a cell culture medium around a scaffold can maintain a maximum O₂ level, as a saturated O₂ concentration in a medium is near 0.2 × 10⁻³ m under 1 atm air pressure at 37 °C. However, this strategy will significantly increase costs due to the requirement of large amounts of expensive growth factors in a perfusion system. The use of a gas-permeable membrane to enhance the equilibrium between O₂ gas and dissolved O₂ has been explored.^[19] Nevertheless, it is limited by the saturated O₂ concentration in a medium defined by Henry’s Law. Another approach is to increase O₂ molar ratio in the gas. A gas tube containing 60% of O₂ was used to test the effect of O₂ in the gas inlet on the generation of insulin-producing cells from hESCs. It is unclear, however, whether the insulin expressing cells derived from hPSCs are capable of secreting insulin.^[11d] In this study, we developed an O₂ releasing biomaterial that is nontoxic to iPSCs. By embedding it into a scaffold for in-situ O₂ supply to cells during differentiation, we demonstrated that oxygenated scaffolds offer better niches to augment the iPSC pancreatic islet development. The oxygenated 3D iPSC differentiation platform developed in this work is transformative and can be readily applied to produce other human cell types that require O₂ such as liver cells and cardiac cells.

4. Experimental Section

Fabrication of Oxygenators:

Oxygenators were fabricated by encapsulating CaO₂ in the PDMS disk. Briefly, 2.5 g CaO₂ (Sigma) was mixed with 10 g Sylgard 184 silicone elastomer base (Dow) and 1 g Sylgard curing agent (Dow) using a vortex mixer. Then, the mixture was rapidly transferred to a 100-mm petri dish. The dish was vacuum dried at 60 °C for 1h to cure PDMS and remove air bubbles. The materials were UV sterilized before use.

Measurement of Oxygen Release Kinetics:

The net O₂ releasing kinetics of the oxygenators was measured using a needle-type O₂ microsensor NTH-PSt7 (Presens, Germany) after the complete removal of air using an anaerobic glove chamber. Briefly, one small piece, 3 mm in diameter and 2 mm in thickness, of oxygenator was placed into a 20 mL glass bottle containing 5 mL water. The bottle was placed inside an anaerobic glove chamber and purged with pure nitrogen gas to completely remove O₂ inside the bottle. It was sealed tightly with rubber cap. Measurements were recorded by VisiSens Analytical 1 (Presens) at one-hour interval.

Oxygen Tension Measurement:

The O₂ tension was measured on the surface of the scaffolds using a needle type O₂ microsensor NTH-PSt7 (Presens). Three randomly selected positions on a scaffold surface were measured for each scaffold. An O₂ sensor foil SF-RPSu4 (Presens) (Figure 5C) was used to map the O₂ tension at the bottom of the scaffolds following the manufacturer’s instructions. The microsensor was disinfected with 70% ethanol before use. It was calibrated using a two-point calibration method. Briefly, a piece of sensor foil was applied to the bottom of the inner glass bottle for calibration. 100% saturated air (s.a.) was calibrated with the bottle filled with the air atmosphere. 0% s.a. was achieved by filling the container with nitrogen gas. To measure the O₂ tension at the bottom of the scaffolds, the foil sensor was positioned into wells of a 48-well cell culture plate. All the measurements were recorded in a dark environment.

Scanning Electron Microscopy (SEM):

The microstructure of the scaffolds and interaction of cells and collagen were observed and analyzed using a Zeiss Supra 55 VP SEM (Carl Zeiss, Germany). Acellular and cell-laden scaffolds were prepared in 48 well plates with 3 mm in thickness. After 24 h post cell seeding, the scaffolds were processed for SEM. Briefly, the scaffolds were rinsed with PBS and fixed with 2.5% v/v glutaraldehyde (Electron Microscopy Sciences, Hatfield, PA) overnight at 4 °C. After fixation, the scaffolds were washed in PBS three times to remove glutaraldehyde. Subsequently, the scaffolds were snap-frozen in liquid nitrogen. The frozen samples were dried in a lyophilizer for 2 d in a low-temperature and high-vacuum environment. The resultant samples were broken into small pieces with forceps and tweezers and mounted for coating with carbon for imaging. For estimating fibrous diameters and pore sizes of scaffolds, images (n = 3–5 for each condition) were quantified by ImageJ.

Rheological Measurements:

Oscillatory rheological measurements were performed at 37 °C using a Discovery HR 30 Rheometer (TA Instruments) with the following operating parameters: angular frequency 0.1–500 rad s⁻¹, 1% strain, and 5 points per decade. A 20 mm cross-hatched parallel plate geometry was used with a testing gap of 1000 μm. Storage and loss moduli measurements at each individual frequency were obtained by a Fourier transform processing. Scaffolds with 9 mm in diameter and 3 mm in thickness were applied to the measurements. The storage moduli values were converted to elastic modulus using the following equation: $\mathrm{E}\mathrm{l}\mathrm{a}\mathrm{s}\mathrm{t}\mathrm{i}\mathrm{c}\mathrm{m}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{l}\mathrm{u}\mathrm{s}=\frac{G^{\prime }}{\mathrm{c}\mathrm{o}\mathrm{s}\left(\delta \right)}$

Cell Culture:

iPSCs IMR90 (WiCell) were maintained in mTesR1 (Stem-Cell Technologies) culture medium on Matrigel (Corning Life Science)-coated dishes as described in our previous study.^[20] The mTesR1 medium was replenished daily.

Preparation of Cell-Laden Scaffolds:

The scaffolds were prepared as described in our previous study.^[10] Briefly, before harvesting iPSCs with Accutase (Stemcell Technologies), cells were incubated with Rock inhibitor Y-27632 (Selleckchem) for 2 h. The cells were then collected and encapsulated into the scaffold solution with a seeding density around 2.5 million cells per ml as described below. To fabricate cell-laden scaffolds, high concentration type I collagen solution (Corning) was mixed with Matrigel, 10× DMEM (Sigma Aldrich), and cell culture grade water (Fisher Scientific) to achieve final concentration of 2.0 mg mL⁻¹ collagen I and 1.0 mg mL⁻¹ Matrigel, and 1X DMEM.^[10] The gel solution was neutralized to pH 7–7.4 before mixing with cells using 1 N NaOH solution. The oxygenators were placed at the bottom of the wells of a multi-well plate (Figure 1B), followed by transferring cell-laden scaffold solution to the multiwell plate. The plate was then incubated in a 5% CO₂ incubator at 37 °C for 30 min to allow gelation. After gelation, mTeSR1 medium was added to the wells for scaffold culture for 24 h before initiation of differentiation.

Live and Dead Cell Assay:

Live/dead viability/cytotoxicity kit (Thermo Fisher) was used according to the manufacturer’s instruction. Nonfluorescent calcein-AM is converted into fluorescent calcein in live cells showing green color, while ethidium bromide stains dead cells’ DNA with red color. The stained cell-laden scaffolds were then visualized using a Zeiss 880 multiphoton confocal microscope. For Trypan blue staining, cell-laden scaffolds were rinsed with DPBS, followed by digestion with 2 mg mL⁻¹ type 1 collagenase and incubated at 37 °C for 10 min to extract the cells. The cells were collected by centrifugation and resuspended in DPBS for Trypan blue staining and cell counting.

Stepwise Differentiation:

A five-stage differentiation protocol developed in our previous work was used with slight modification.^[3a] Stage I differentiation was initiated 24 h post single cell seeding of iPSCs containing 50 ng mL⁻¹ activin A (PeproTech) and 1 × 10⁻³ m sodium butyrate (SB, Sigma-Aldrich) for one day. The concentration of SB was reduced to 0.5 × 10⁻³ m from day 2 to day 5. The differentiation medium in Stage II consisted of RPMI1640, 1 × 10⁻⁶ m retinoic acid (RA, Sigma-Aldrich), 50 ng mL⁻¹ keratinocyte growth factor (KGF, PeproTech), 100 × 10⁻⁹ m LDN193189 (LDN, Sigma-Aldrich), 300 × 10⁻⁹ m (−)-indolactam V (ILV, AdipoGen), 100 ng mL⁻¹ noggin (Nog, PeproTech), and 250 × 10⁻⁶ m ascorbic acid (VC, Sigma-Aldrich). At Stage III, cells were cultured in DMEM-F12 (Gibco) containing 1 × 10⁻⁶ m RA, 300 × 10⁻⁹ m ILV, 200 × 10⁻⁹ m LDN,1 × 10⁻⁶ m 3,3′,5-triiodo-l-thyronine Sodium Salt (T3, Sigma), 10 × 10⁻⁶ m ALK5 inhibitor II (A5iII, Enzo Life Sciences), and 10 μg mL⁻¹ heparin (HP, Sigma) and glucose (Gibco) was supplemented to 20 × 10⁻³ m. At Stage IV, the differentiation media consisted of RPMI1640 and pyruvate containing 1 × 10⁻⁶ m T3, 10 μg mL⁻¹ HP, 1 × 10⁻³ m N-acetyl cysteine (N-cys, Sigma), 0.5 × 10⁻⁶ m R428 (Selleck Chem), 10 × 10⁻⁶ m Trolox (Tro, Enzo Life Sciences), 100 × 10⁻⁹ m γ-secretase inhibitor XX (γ-sec, Sigma), and 10 × 10⁻³ m nicotinamide (Nic, Sigma), and glucose was supplemented up to 20 × 10⁻³ m. Serum-free B27 (Gibco) was added from Stage I to Stage IV. At Stage V, the differentiation media was prepared with CMRL supplement and pyruvate (Fisher Scientific) containing 5 mg mL⁻¹ bovine serum albumin (BSA, Sigma), 1 × 10⁻⁶ m T3, and 10 × 10⁻⁶ m A5iII. The differentiation media at all stages were exchanged every two days for 2D culture and every day for 3D cultures. For differentiation in 2D culture mode, cells were seeded as single cells at a density of 1.5 million cells per well of 6 well plates after detachment using Accutase.

RNA Extraction and Gene Expression Analysis by Quantitative Real-Time PCR:

To extract RNA from cells, a combination of enzymatic and mechanical disruption was applied. Briefly, the scaffolds were rinsed with DPBS, followed by 2 mg mL⁻¹ of collagenase digestion for 10 min at 37 °C. After rinsing with DPBS, the samples were subject to RNeasy Plus Mini kit (QIAGEN). The samples were homogenized by a 27 G needle and syringe and further steps were followed according to the manufacturer’s instruction. Gene expression was detected using a CFX Connect RT-PCR system (Bio-Rad) as described previously.^[20] Nonreverse transcribed RNA samples served as negative controls.^[3a,20] Primers and probes used were listed in Table S1 in the Supporting Information. Human pancreatic RNA (Clontech) was used as a positive control.

Glucose Stimulated Insulin Secretion Assay:

Scaffolds were rinsed with DPBS twice and incubated in Krebs–Ringer buffer (KRB, Boston BioProducts) containing 1 × 10⁻³ m glucose for 4 h. They were incubated with KRB containing 2 × 10⁻³ m glucose and 20 × 10⁻³ m glucose at 37 °C separately for 60 min. The supernatants were collected from respective glucose challenges and insulin secreted was measured using a human insulin enzyme-linked immunosorbent assay (ELISA) kit (ALPCO). Total DNA content of each constructor was determined using DNeasy Blood and Tissue Kit (Qiagen) to normalize the insulin secretion. The scaffolds were disrupted with a 27 G needle and syringe for DNA extraction. DNA content was measured by a Synergy H1 microplate reader (Bio-Tek). Human islets purchased from Prodo Laboratories served as positive controls. The stimulation index was determined by the ratio between insulin released at high (20 × 10⁻³ m) to low glucose concentration (2 × 10⁻³ m).

Cryosectioning and Immunofluorescence Staining Microscopy:

Samples were pretreated for cryosectioning and imaging as described in our earlier work.^[3a,b] The cryopreserved samples were sectioned into 10 μm thick. The antibodies used in immunostaining were listed in Table S2 in the Supporting Information. The cell nuclei were counterstained with DAPI (Thermo Fisher Scientific) to show the cell distribution within the scaffolds. Images (n = 6–25) of aggregates per cell subtype were quantified to estimate the percentage of each cell subtype in the organoids using ImageJ software (version 2.9.0/1.53t).

Statistical Analysis:

All experimental data are derived from at least three independent experiments and the exact number was indicated in each Figure legend. Data were presented as mean ± SD. For multiple comparisons, one-way ANOVA with Turkey’s test or with Dunnett’s test was carried out to statistically analyze and interpret the data. Unpaired two-tailed student’s t-test was used for statistical analyses between different groups. All statistical tests were evaluated using GraphPad Prism 7.0 software. The means and standard deviations of experimental results were computed and the differences were considered to be significant when the p-value < 0.05.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.