DISCUSSION
The need for innovative biomanufacturing platforms that sustain cell viability and functionality are critical to broadly advancing cell therapies. In this study, we demonstrated production of insulin from metabolically competent pancreatic cells seeded onto a wicking matrix cellulose scaffold. We tested multiple surface-modification conditions for cell attachment and expansion in a multi-well test platform and small-scale bioreactor. Cells seeded onto the wicking matrix cellulose scaffold in the bioreactor exhibited greater expansion (up to 10-fold) and increased insulin secretion, when compared with traditional 2D cultures. Metabolic profiling indicates that the oxygenation and physiological conditions in bioreactor were favorable. This biomanufacturing platform provides an efficient 3D microenvironment for pancreatic cells to support mass production of these insulin producing cells for potential use in cell therapy to treat Type 1 diabetes patients. The ability to use human induced pluripotent stem cell-derived pancreatic cells also opens up the possibility of patient-specific treatments.
Multiple surface modifications enabled attachment and growth of primary pancreatic cells as evaluated by MTT metabolic assay and scanning electron microscopy. Before seeding cells, a fraction of the culture was validated in parallel for relevant biomarker expression by immunocytochemistry, including PDX-1, NGN3, NKX6-1, NeuroD1 and insulin, which are indicators of pancreatic lineage-committed cells (Takahashi 2016, Kieffer 2014, Benthuysen 2016). Pancreatic cells seeded at a density of 20,000 cells/well on various modified scaffolds in a multi-well platform were able to attach similarly to cellulose with all six surface modifications; however, their expansion varied considerably. Growth on uncoated, amine-modified cellulose (A) and gelatin-coated, NaOH-modified cellulose (NG) scaffolds provided the most favorable surface conditions for expansion of cells with 5 to 6-fold expansion. By high resolution SEM imaging of cells on day 1 and day 3, we observed spreading of cells on these scaffolds with visible lamellipodia formation. As discussed by several authors (Banik 2015, Le 2013, Krishnamurthy 2009), surface topography of various scaffolds contributes to cell attachment and can impact long-term viability and differentiation of cells. The cellulose scaffold supports cell spreading and retention of phenotypic function. Seeded pancreatic cells displayed an affinity towards all modified surfaces of the porous cellulose scaffold that was apparent by formation of structures that resemble lamellipodia and production of a film of extracellular matrix. In addition, cells were able to form larger aggregates over 5-10 days, consistent with survival and proliferation seen in other studies (Kim 2016, Salvatori 2014). Between day 1 and day 10, sustained culture on the amine-modified scaffolds resulted in better expansion, based on a greater distribution of cell clusters throughout the scaffold.
Our analysis revealed that the uncoated, amine-modified (A) and gelatin-coated, NaOH-modified (NG) scaffolds were most favorable, and these were chosen for additional analysis using hiPSC-derived pancreatic cells. Viability, morphology and insulin secretion of hiPSC-derived pancreatic cells on these two scaffold modifications were examined. The human iPSC-derived endocrine cells were generated by multiple stages of differentiation. In the first stage of pancreatic differentiation, expression of definitive endoderm biomarkers, such as Sox 17 and Brachyury increase while pluripotency markers decrease. Moving developmentally toward posterior foregut, pancreatic biomarkers such as PDX-1 appear. Following commitment to the pancreatic lineage, expression of biomarkers such as NKX2-2, NKX6-1, Neurogenin-3 and insulin increase (D’Amour 2006, Rezania 2012, Russ 2015). From committed pancreatic cells, the hormone-releasing beta cells arise; however, these cells are initially immature, and the amount of insulin released in response to changes in glucose level is low. Further differentiation to mature, beta islet cells is facilitated by addition of Triiodothyronine (T3) which is a Thyroid receptor agonist, with observed upregulation of MafA in cultures (Aguayo-Mazzucato 2013, Millman 2016). We tested hiPSC-derived pancreatic progenitors, NKX6-1+/PDX-1+ on our uncoated, amine-modified (A) and gelatin-coated, NaOH-modified (NG) cellulose scaffolds for 10 days as single cells and as lightly dissociated cell aggregates. By MTT assay, we observed a predictable trend of decreased cell proliferation as cells underwent differentiation on amine-modified scaffolds from seeded aggregated cells. We observed that cell aggregates had higher viable cell density on amine-modified (A) scaffolds compared to gelatin-coated, NaOH-modified (NG) scaffolds on all days of testing. Viable cell density from single-seeded cells on amine-modified scaffolds (A) increased linearly throughout the experiment. However, single cells seeded on gelatin-coated, NaOH-modified scaffolds (NG) showed a significant decrease in viable cells on day 5. These findings suggest that the differentiating seeded pancreatic cells were more compatible with the amine-modified scaffolds. Images taken by SEM support this observation. SEM images acquired on day 5 showed production of significant amount of ECM on the amine-modified scaffolds seeded with cell aggregates. The presence of ECM was also observed on day 5 and day 10 on amine-modified scaffolds seeded with single cells. As noted previously (Hammer 2004, Stendhal 2009, Llacua 2018, Smink 2018), ECM production is a necessary step for support of multiple cell activities on a scaffold including survival and migration, and in some cases, ECM components can integrate into the scaffold structure to enhance cell attachment, differentiation and proliferation for tissue engineering.
A sustained ability to produce insulin is critical to biomanufacturing efforts with pancreatic cells. Insulin secretion was measured for cells grown on both the uncoated, amine-modified (A) scaffold and gelatin-coated, NaOH-modified (NG) scaffold, and both demonstrated greater insulin release than was observed from the same cells in 2D culture (data not shown). This analysis demonstrates the enhanced functional potential of chemically modified 3D cellulose scaffolds for supporting hiPSC-derived pancreatic cells and differentiation into mature, functional, insulin-secreting pancreatic cells. The amount of insulin secreted from single cells seeded on both cellulose surface-modified conditions decreased over time, in contrast to cell aggregates seeded, which showed an increase in insulin secretion on day 5. We speculate that the higher insulin secretion and ECM production by cell aggregates on amine-modified scaffolds is due to higher cell density and increased cell-cell contact on the scaffolds. As a follow up, we tested insulin release from single cells seeded on the scaffold in the bioreactor at two seeding densities. The hiPSC-derived endocrine precursors were seeded onto the amine-modified wicking matrix scaffold in the bioreactor, and insulin release and metabolic activity of cells were assessed for 13 days. We observed a significant increase in insulin secretion at both cell seeding densities that remained high until the conclusion of the experiment, demonstrating an improvement of cell functionality in this 3D environment. The porous structure of cellulose fibers along with adequate oxygenation and continuous accessibility of all cells to media in the bioreactor appear to meet critical microenvironment needs enabling ~2-fold increase in insulin release in the bioreactor. Doubling the cell density enhanced insulin production in the bioreactor, reaching the higher insulin secretion by day 2 rather than day 7. Given that cell seeding density affects differentiation and maturation of pancreatic cells in 2D cultures (Gage 2013), we suspect that delayed insulin production at the lower cell seeding density in the bioreactor is a result of an arrested or slowed maturation process.
Metabolically, glucose consumption and lactate production in the bioreactor were reflective of good cell viability and activity. Low glucose consumption levels, which can be a sign of lowered cell viability, were not observed. Higher inoculation densities (1 x 107) exhibited increased glucose consumption and lactate production compared to lower cell densities (i.e., 5 x 106 cells, Supplemental Figure 2). The relatively low levels of lactate produced in the bioreactor indicate sufficient aeration, as would be expected in the wicking matrix bioreactor with the thin film of liquid. In addition, the low lactate concentration demonstrates a low rate of aerobic glycolysis. Elevated aerobic glycolysis is a hallmark of tumor cells as well as dedifferentiation in culture. In summary, the metabolic profile further supports the design and use of the bioreactor for cell biomanufacturing.