Biofunctionalized gelatin hydrogels support development and maturation of iPSC-derived cortical organoids

Abstract

Human neural organoid models have become an important tool for studying neurobiology. However, improving the representativeness of neural cell populations in such organoids remains a major effort. In this work, we compared Matrigel, a commercially available matrix, to a neural cadherin (N-cadherin) peptide-functionalized gelatin methacryloyl hydrogel (termed GelMA-Cad) for culturing cortical neural organoids. We determined that peptide presentation can tune cell fate and diversity in gelatin-based matrices during differentiation. Of particular note, cortical organoids cultured in GelMA-Cad hydrogels mapped more closely to human fetal populations and produced neurons with more spontaneous excitatory postsynaptic currents relative to Matrigel. These results provide compelling evidence that matrix-tethered signaling peptides can influence neural organoid differentiation, opening an avenue to control stem cell fate. Moreover, outcomes from this work showcase the technical utility of GelMA-Cad as a simple and defined hydrogel alternative to Matrigel for neural organoid culture.

Keywords: CP: Stem cell research; GelMA; Matrigel; N-cadherin; neural organoids; peptide-functionalized matrices.

Figure 1.. Functionalized gelatin matrix parameters for cortical organoid culture

(A) Schematic of GelMA-Cad, major tunable parameters, and associated organoid culture timeline. GelMA-Cad utilizes a gelatin-based backbone, with a conjugated methacryloyl group (blue), allowing photo-initiated crosslinking. The methacryloyl can be further modified by the addition of an N-cadherin (Cad) peptide mimetic (orange). Adjusted parameters include the LAP crosslinker concentration and N-cadherin peptide presence. (B) Representative ¹H-NMR spectra of gelatin, GelMA, and GelMA-Cad displaying the characteristic peaks associated with methacryloyl. (C) Atomic force microscope characterization of Young’s modulus. N = 5 measurements, each the average of 512 technical replicates. Bars represent data mean. Statistical significance was calculated with a two-way ANOVA, modeled on crosslinker concentration and peptide presence. (D) Average storage modulus traces from GelMA and GelMA-Cad hydrogels. N = 5 measurements, error bars represent the standard deviation. (E) Rheological characterization of average storage modulus across tested frequencies. N = 5 measurements, bar represents data mean. Statistical significance was calculated with a two-way ANOVA, modeled on crosslinker concentration and peptide presence. (F) Rheological characterization of average loss modulus across tested frequencies. N = 5 measurements, bar represents data mean. Statistical significance was calculated with a two-way ANOVA, modeled on crosslinker concentration and peptide presence. (G) Representative scanning electron micrographs of lyophilized, crosslinked hydrogels. Scale bar: 100 μm. (H) Quantification of pore size from scanning electron micrographs. Each point is the mean of measurements made from a single image, N = 6–7 images per condition, each image from a separate preparation. Bars represent data mean. (I) Apparent permeability coefficient of 3 kDa FITC-dextran in each hydrogel. N = 6 measurements, bar represents data mean. Statistical significance was calculated with a two-way ANOVA, modeled on crosslinker concentration and peptide presence.

Figure 2.. Matrix properties instruct early cortical organoid morphology

(A) Representative bright-field images of day 10 organoids generated in Matrigel or GelMA^HC. Scale bar: 500 μm. (B) Measurements of cortical organoid area at day 30 from bright-field images. Each data point represents a single organoid with sample sizes shown in the plot; N = 8–16 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (C) Measurements of cortical organoid circularity at day 30 from bright-field images. Each data point represents a single organoid with sample sizes shown in the plot; N = 8–16 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (D) Representative maximum intensity Z-projections of calcein-stained organoids at day 30. Scale bar: 250 μm. (E) Quantification of neurite length in maximum intensity Z-projections at day 30. Each data point represents a single organoid; N = 3–5 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (F) Representative images of ZO-1 and DAPI generated by whole-mount imaging at day 10. Scale bar: 50 μm. (G) Quantification of ZO-1 basal region diameter from whole-mount confocal images. Each data point represents a single organoid; N = 3–6 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (H) Quantification of number of ZO-1 basal regions per area from whole-mount confocal images. Each data point represents a single organoid; N = 3–6 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (I) Representative single-slice confocal planes of βIIIT (100 μm from organoid base) generated by whole-mount imaging at day 30. Scale bar: 250 μm. (J and K) Quantification of βIIIT width and βIIIT+ area in single-slice confocal planes. Each data point represents a single organoid; N = 4–20 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (L) Cells from each hydrogel condition simultaneously embedded in a uniform manifold approximation and projection (UMAP) plot. Cell annotated by canonical markers: radial glia (SOX2+), cycling cells (TOP2A+), intermediate progenitor cells (IPCs) /neurons (EOMES+, MAP2+), mesenchymal cells (COL1A1+, TWIST1+), and neural placode and neural crest (SIX1+, ISL1+, FOXD3+). (M) Cell proportions in scRNA-seq data from (L) as a function of hydrogel condition. (N) Comparisons between radial glia from organoids cultured in Matrigel versus GelMA-Cad (low and high crosslinker). DEG, differentially expressed gene as calculated from a Wilcoxon test with adjusted p value using a Bonferroni correction; GSEA, gene set enrichment analysis score as calculated using the ClusterProfiler package in R.

Figure 3.. Neuronal fate outcomes and organization in cortical organoids at day 60

(A) Schematic of the embedding procedure used to create free-floating organoid cultures. Organoids were drawn up into glass capillaries, UV exposed, rotated 180°, UV exposed again, and then expelled back into cell culture media. (B) Cells from each hydrogel condition simultaneously embedded in a UMAP plot display. Cell annotated by canonical markers: radial glia (SOX2+, HOPX+), intermediate progenitors/cycling cells (EOMES+, TOP2A+), RELN+ inhibitory neurons, DLX5+ inhibitory neurons, non-specified neurons (MAPT+, MAP2+), deep-layer neurons (BCL11B+, FEFZF2+), upper-layer neurons (SATB2+, CUX2+), and choroid plexus epithelial cells (TTR+, CLIC6+). (C) Cell proportions in scRNA-seq data from (B) as a function of hydrogel condition. (D) Representative single-slice confocal planes (100 μm from organoid base) generated by whole-mount imaging of organoids stained for CTIP2 (BCL11B) and SATB2. Scale bar: 50 μm. (E) Tracing of SATB2 layers (from images in D) and quantification of thickness. Each data point represents a single organoid, and the black bar represents mean; N = 7–12 organoids. Statistical significance was calculated using a one-way ANOVA with Tukey’s post hoc test. (F) Cells scored in order of pseudotime using Monocle as displayed on a UMAP dimension reduction plot. (G) Distributions of pseudotime scores among deep-layer excitatory neurons per hydrogel condition as represented by violin plot. A boxplot showing median and quartile range with outlier points is superimposed. Statistical significance was calculated using a one-way ANOVA with Dunnett’s post hoc test. (H) Comparisons between excitatory neurons from organoids cultured in Matrigel versus GelMA-Cad (low and high crosslinker). DEG, differentially expressed gene as calculated from a Wilcoxon test with adjusted p value using a Bonferroni correction; GSEA, gene set enrichment analysis score as calculated using the ClusterProfiler package in R.

Figure 4.. Organoid cellular composition divergence at later time points

(A) Representative bright-field images of organoids at days 90 and 120 in different hydrogels. In HC conditions, the remaining gel is outlined to increase contrast. Scale bar: 500 μm. (B) t-Stochastic network embedding (tSNE) plots of integrated day 90 and day 120 organoid scRNA-seq samples annotated by major cell types: glioblast/astrocytes (GFAP+), radial glia (SOX2+, HOPX+, CLU+), intermediate progenitors/cycling cells (EOMES+, TOP2A+), RELN+ inhibitory neurons, DLX5+ inhibitory neurons, non-specified neurons (MAPT+, MAP2+), deep-layer neurons (BCL11B+, FEFZF2+), upper-layer neurons (SATB2+, CUX2+), choroid plexus epithelial cells (TTR+, CLIC6+), and mesenchymal cells (COL1A1+, COL4A1+). Data were generated from 8–10 pooled organoids per condition, with an average of 13,000 cells passing quality control encapsulated per condition. (C) Cell proportions in scRNA-seq data from (B) as a function of differentiation time and hydrogel conditions. (D) Cell proportion ratios in scRNA-seq data from (B), dependent on crosslinking concentration (x axis) or peptide presence (y axis). A value of 2 indicates that the estimated cell proportion in condition A (e.g., LC) is double that of condition B (e.g., HC). Comparisons are split between experimental day and indicated with different symbols. Ellipses are superimposed to highlight populations enriched in GelMA^HC (bottom left) or GelMA-Cad^LC (top right). (E) Representative whole-mount image of NeuN and TTR in organoids cultured in GelMA-Cad^HC and GelMA-Cad^LC for 90 days. Scale bar: 100 μm. (F) Uniform manifold approximation and projection (UMAP) plot of day 90 KOLF2.1J-derived organoids samples annotated by major cell types defined in (B) and derived from GelMA-Cad^LC and GelMA-Cad^HC conditions, with cell proportions shown on the bottom as bar plots. (G) Representative images of layer-specific markers (CTIP2 and SATB2), Reelin, and GFAP in cryosectioned organoids that were cultured in GelMA-Cad^HC for 90 days. Scale bar: 100 μm. (H) UMAP embedding of scRNA-seq data aggregated from days 21, 60, 90, and 120, co-embedded with primary fetal tissue from previous work. Primary fetal tissue is colored by cell type, with the colors corresponding to those in (I). Excitatory neurons (gold) and outlier cells (purple) are outlined in the day 120 projections. (I) Proportion of cell type phenotypes over time in both GelMA-Cad^HC- and Matrigel-embedded organoids as transferred from the fetal database.

Figure 5.. Electrophysiological characterization of organoids on day 90

(A) Representative traces of sEPSCs from organoids differentiated in Matrigel or GelMA-Cad^HC for 90 days. (B) Percentage of recorded cells that displayed sEPSCs. (C and D) Summary of sEPSC frequency (C) and amplitude (D) from organoids embedded in Matrigel and GelMA-Cad^HC (n = 10 and n = 13 cells, respectively; sampled from N = 3 organoids per condition). Statistical significance was calculated using a Student’s t test (p value shown). Bars represent mean ± standard error of the mean. (E) Representative traces of action potentials generated with increasing levels of injected current. (F) Summary of action potentials (APs) generated (n = 11 cells, GelMA-Cad; n = 15, Matrigel). Statistical significance was calculated using a two-way ANOVA with Sidak’s post hoc test (**p < 0.01). Points represent mean, and bars represent standard error of the mean.

Figure 6.. Characterization of radial glia identities and distributions at later time points

(A) tSNE embedding of radial glia clusters from scRNA-seq data aggregated at days 90 and 120. Cell populations were annotated based on Gene Ontology of marker genes after unbiased subclustering. (B) Proportion of radial glia phenotypes within organoids at days 90 and 120 with respect to each hydrogel condition. Bar colors indicate corresponding cell populations from (A) (from top to bottom: neurogenic, ciliated, ECM responding 1, ECM responding 2, ribosome biogenesis, apoptotic, and hypoxic). Hashed bars: day 90, solid bars: day 120. (C) Correlation of radial glial phenotypes between primary fetal tissue and organoids as visualized by the number of intersecting differentially expressed marker genes. GW, gestational week. Neurogenic, ciliated, and ECM responding 1 clusters correspond most closely to GWs 18–22. The ribosome biogenesis cluster corresponds most closely to GW 6. Apoptotic, hypoxic, and ECM responding 2 clusters do not show strong correlation to any fetal radial glia populations. (D) Representative image of HOPX, FABP5, and GFAP in cryosectioned organoids that were cultured in GelMA-Cad^HC for 90 days. Scale bar: 20 μm. (E) Trajectory graph of cell populations across days 90 and 120; nodes are colored by cell population, and radial glia populations are additionally annotated as in (A). Edge weight represents average connectedness metric. (F) Correlation between proportions of radial glia in the ECM responding 2 cluster and mesenchymal cells. RG, radial glia; MC, mesenchymal cells. Point shape indicates the experimental time point, while color refers to the embedding matrix. R² indicates the correlation coefficient from linear regression, and the shaded region is the confidence interval.