Chronic exposure to glucocorticoids amplifies inhibitory neuron cell fate during human neurodevelopment in organoids

Abstract

Disruptions in the tightly regulated process of human brain development have been linked to increased risk for brain and mental illnesses. While the genetic contribution to these diseases is well established, important environmental factors have been less studied at molecular and cellular levels. Here, we used single-cell and cell type-specific techniques to investigate the effect of glucocorticoid (GC) exposure, a mediator of antenatal environmental risk, on gene regulation and lineage specification in unguided human neural organoids. We characterized the transcriptional response to chronic GC exposure during neural differentiation and studied the underlying gene regulatory networks by integrating single-cell transcriptomics with chromatin accessibility data. We found lasting cell type-specific changes that included autism risk genes and several transcription factors associated with neurodevelopment. Chronic GC exposure influenced lineage specification primarily by priming the inhibitory neuron lineage through transcription factors like PBX3. We provide evidence for convergence of genetic and environmental risk factors through a common mechanism of altering lineage specification.

Fig. 1.. Overview of experimental setup and design.

We designed six treatment conditions duplicated in organoids derived from two cell lines. We replicated each of these 12 conditions using four samples. We collected two treatment conditions at day 70: Veh (exposed to the treatment vehicle DMSO for 10 days starting from day 60) and Chr (exposed to the GC Dex for 10 days starting from day 60). We derived the four additional treatment conditions collected at day 90 from the day 70 conditions with sustained culturing in regular medium conditions for a further 20 days (washout period). The two conditions derived from the Veh condition were Veh-Veh and Veh-Acu, with an additional 12-hour acute GC exposure applied. Analogously, the two day-90 conditions derived from the Chr condition were Chr-Veh and Chr-Acu, with an additional 12-hour acute GC exposure applied. Parts of the figure were created with BioRender.com.

Fig. 2.. Chronic GC exposure in neural organoids does not induce significant metabolic stress in cells.

(A) Cell types on UMAP embedding for Line 409b2 and Line FOK4. Each cell type was identified in both datasets. (B) Localization of nonviable cells (charcoal) in the UMAP embedding for Line 409b2 and Line FOK4 organoids. The “Unknown” cluster is primarily made up of nonviable cells (74% in Line 409b2 and 41% in Line FOK4), suggesting that these cells and the entire cluster should be removed. (C) Swarm plots, showing no significant difference in mean viability score between control (gray) and treated (blue) samples following nonviable cell removal using an unpaired t test. Each dot represents a sample from the indicated treatment condition. (D) Top: UMAP embedding colored by cell type for Line 409b2 and Line FOK4 following nonviable cell removal. Middle: UMAP embedding colored by treatment conditions. Bottom: NR3C1 (GC receptor) gene expression.

Fig. 3.. Transcriptional response following chronic GC treatment in organoids includes key neurodevelopmental genes.

(A) Overview of the experimental design for 70-day-old organoids. Created with BioRender.com. (B) UMAP embedding for day 70 data of Line 409b2 and Line FOK4 at day 70 colored by cell type. Cells from day 90 samples are shown in gray. All identified cell types are present at this earlier stage. (C) Upset plot showing consensus DE results per cell type and the number of unique and shared consensus DE genes. Selected genes are highlighted, ASD risk genes from the SFARI Gene database (42) are shown in blue, and further TFs are shown in green. (D) Grouped semantic space representation of the GO-BP enrichment results for the three cell types with the most detected DE genes. The size of the circles corresponds to the number of terms in the cluster; their color corresponds to the log₁₀(q value) of the representative term for each cluster. The integers within the circles enumerate the five most significant clusters, and their representative term is written out in the legend below each plot. (E) Upset plot showing DE results aggregated across all cell types for the effect of chronic GC exposure directly following the treatment (Day 70 Chr), the effect of chronic GC exposure after a 20-day washout period (Day 90 Chr), and the effect of acute GC exposure in 90-day-old organoids (Day 90 Acu).

Fig. 4.. TF regulation causes priming of the inhibitory neuron lineage in neural organoids.

(A) Force-directed graph embedding of organoid data from both cell lines (without the RGS5 Neurons cluster) colored by cell type. Lineage endpoints are labeled with black circles. FA, force-directed graph embedding. (B) Computed lineage probabilities per cell for the three lineage endpoints in the two datasets. (C) Force-directed graph layout of validation data (published 70-day-old organoid data derived from six additional cell lines) (43). Colored by cell type (left) and cell line (right). Lineage endpoints are labeled with black circles. CGE, caudal ganglionic eminence; MGE, medial ganglionic eminence; LGE, lateral ganglionic eminence; IN, interneuron; glutamat., glutamatergic; IPs, intermediate progenitors; NPCs, neural progenitor cells. (D) Computed lineage probabilities per cell for the two neuronal lineage endpoints visualized on the force-directed graph layout of the validation data from six cell lines. (E) Fraction of genes where the directionality of consensus DE effect and driver status was aligned out of all genes both significantly DE (consensus DE genes from Line 409b2 and Line FOK4 data) and in the top 500 significant driver genes (recomputed for each of the three datasets). The dashed line indicates the fraction expected by chance (0.5). Statistical significance was computed using a paired t test. (F) Magnitude of driver gene correlation with the inhibitory neuron lineage in the validation data (43) versus log₂FC of consensus DE effect measured in our two cell lines. Genes with the highest lineage correlation are labeled by name, ASD risk genes from the SFARI Gene database (42) are marked in blue, and genes associated with high risk for ASD in the Li et al. publication (29) are marked in blue with a magenta core.

Fig. 5.. GC exposure results in an increased abundance of inhibitory neurons in organoids.

(A) Left: Force-directed graph layout colored by expression of the inhibitory neuron marker GAD1. Right: Force-directed graph layout colored by absorption probability per cell for the inhibitory neuron lineage. (B) Fraction of GAD1-positive cells and mean GAD1 expression across all cells in Veh Chr, Veh-Veh, and Chr-Veh conditions and organoids from both cell lines. (C) Representative images of whole slice ventralized organoids at day 70 in culture, following 10 days of chronic treatment with GCs (100 nM Dex; Chr condition; right) and control (Veh condition; left), show an increased abundance of GAD1⁺ cells in the treated condition. Images were acquired at ×20 magnification, showing 4′,6-diamidino-2-phenylindole (DAPI) (blue) and GAD1 (green). Lower panel: Zoomed-in insets. (D) Cell counting quantification of GAD1⁺ cells across entire organoid tissue slices (n = 5 per condition) and graphically represented as cells/mm². Means per condition are indicated as a dotted black line. Statistical significance was computed using an unpaired t test. IF, immunofluorescence.

Fig. 6.. PBX3 regulation through chronic GC exposure supports inhibitory neuron priming.

(A) Magnitude of driver gene correlation with the inhibitory neuron lineage in the validation data (43) versus log₂FC of GC day 70 DE effect measured in our two cell lines. Genes with aligned direction of log₂FC and lineage correlation are marked in purple. (B) Correlation of PBX3 expression with lineage probability across the excitatory and inhibitory neuronal lineages in all three datasets. The percentile of PBX3 among all significant driver genes ranked by driver strength is shown on the side of every bar. (C) Expression of PBX3 on a force-directed graph embedding of Line 409b2 data (left) with cell type reference (right). (D) Expression patterns of PBX3 across pseudotime for each of the two neuronal lineage endpoints in Line 409b2. (E) Coexpression of GAD1⁺ (green) cells with PBX3⁺ (magenta) cells in 70-day-old ventralized organoids of Line 409b2 in Veh and Chr conditions. The ×63 magnification zoom-in images are shown to the right of the respective 20× whole-slice images. Examples of double-positive cells are marked by white arrows. (F) Cell counting quantification of PBX3⁺ cells across entire organoid tissue slices (n = 5 per condition) and graphically represented as cells/mm². Means per condition are indicated as a dotted black line. Statistical significance was computed using an unpaired t test. (G) Cell counting quantification of PBX3⁺GAD1⁺ double-positive cells across entire organoid tissue slices (n = 5 per condition) and graphically represented as cells/mm². Means per condition are indicated as a dotted black line. Statistical significance was computed using an unpaired t test.

Fig. 7.. Analyses of multimodal GRNs associate PBX3 with the regulation of inhibitory neuron priming in organoids from Line 409b2.

(A) Magnitude of driver gene correlation with the inhibitory neuron lineage versus log₂FC of Line 409b2. First- and second-order PBX3 target genes in the inferred chronic GRN are labeled in pink. Left: Directly following treatment (70 days in culture). Right: After 20 days of washout (90 days in culture). Genes with an absolute lineage correlation greater than 0.45 are labeled by name. (B) UMAPs of integrated scRNA-seq and scATAC-seq data of Line 409b2 at 90 days in culture. ScRNA-seq data are colored by cell type, and scATAC-seq data are shown in gray. Left: Vehicle organoid data. Right: GC-exposed organoid data. (C) GRN centered around PBX3 in vehicle organoids with DE genes (consensus DE genes from any of the three DE comparisons: D70 Chr, D90 Chr, D90 Acu) colored in red and TFs labeled by name. (D) GRN centered around PBX3 in treated organoids with top 500 inhibitory neuron driver genes colored in green and TFs labeled by name. The bar chart shows the fraction of newly gained direct PBX3 downstream targets. (E) Fraction of inhibitory neuron drivers in direct TF downstream targets for control (Veh-Veh) and GC-exposed organoids (Chr-Veh) of Line 409b2 at 90 days in culture.