Cell-Type-Specific Impact of Glucocorticoid Receptor Activation on the Developing Brain: A Cerebral Organoid Study
This article has been corrected.
VIEW CORRECTIONPublication: American Journal of Psychiatry
CME has expired
Abstract
Objective:
A fine-tuned balance of glucocorticoid receptor (GR) activation is essential for organ formation, with disturbances influencing many health outcomes. In utero, glucocorticoids have been linked to brain-related negative outcomes, with unclear underlying mechanisms, especially regarding cell-type-specific effects. An in vitro model of fetal human brain development, induced human pluripotent stem cell (hiPSC)–derived cerebral organoids, was used to test whether cerebral organoids are suitable for studying the impact of prenatal glucocorticoid exposure on the developing brain.
Methods:
The GR was activated with the synthetic glucocorticoid dexamethasone, and the effects were mapped using single-cell transcriptomics across development.
Results:
The GR was expressed in all cell types, with increasing expression levels through development. Not only did its activation elicit translocation to the nucleus and the expected effects on known GR-regulated pathways, but also neurons and progenitor cells showed targeted regulation of differentiation- and maturation-related transcripts. Uniquely in neurons, differentially expressed transcripts were significantly enriched for genes associated with behavior-related phenotypes and disorders. This human neuronal glucocorticoid response profile was validated across organoids from three independent hiPSC lines reprogrammed from different source tissues from both male and female donors.
Conclusions:
These findings suggest that excessive glucocorticoid exposure could interfere with neuronal maturation in utero, leading to increased disease susceptibility through neurodevelopmental processes at the interface of genetic susceptibility and environmental exposure. Cerebral organoids are a valuable translational resource for exploring the effects of glucocorticoids on early human brain development.
Human developmental trajectories are influenced by both genetics and the environment (1). An important environmental factor is activation of the glucocorticoid (GC) system during critical brain development periods, which is associated with cognitive, behavioral, and psychiatric health outcomes (1, 2). The glucocorticoid receptor (GR; NR3C1 gene) mediates key processes in fetal organ development, and its activation is essential for proper formation and maturation of many organs (3). Upon activation, GR translocates to the nucleus, where it functions as a transcription factor by binding to GC response elements (4) in many genes, as well as through protein-protein interactions with other transcription factors (5).
GCs are critical for development, and thus alterations profoundly affect fetal outcome. Synthetic GCs readily cross the placenta (3, 6) and are prescribed throughout pregnancy (7), for example, for maternal autoimmune conditions or fetal congenital adrenal hyperplasia (8). Higher doses are given at certain times, such as during postconception weeks 0–12 to prevent miscarriages (9). Most commonly, synthetic GCs are given to promote organ maturation as early as postconception week 22 in pregnancies at risk of premature delivery (10). Synthetic GC treatment has clear benefits for pregnancy and birth outcomes, but negative long-term outcomes are increasingly being reported (11). Accumulating evidence from human cohorts indicates that fetal synthetic GC exposure in mid to late pregnancy may result in adverse postnatal outcomes such as dysregulated hypothalamic-pituitary-adrenal axis activity and altered brain structure and development (12). While earlier studies (N=∼2,000) did not find major long-term effects of antenatal synthetic GCs on neurobehavioral outcomes (13), more recent findings from larger population-based cohorts reveal a significant increase in neurocognitive and behavioral risk later in life. For example, a Finnish study of children (N>600,000) followed up to age 8 (14) found increased prevalence of mental and behavioral disorders in exposed children. A Canadian study (N>500,000) found increases in neurocognitive disorders at age 5 in those exposed to antenatal synthetic GCs (15). Extended synthetic GC treatment for congenital adrenal hyperplasia has been linked to altered cognitive performance, including inattention, and increased fearfulness in childhood (8). Furthermore, elevated endogenous maternal GCs can affect brain development (2, 16–18), for example, with altered neonatal amygdala connectivity, differences in sensory processing and integration, and subsequent internalizing symptoms in girls (19). Together with the increasing prevalence of prenatal synthetic GC administration, this has a large societal impact and underlines the necessity of better understanding how prenatal GR activation outside the normal balance in the developing brain affects health outcomes. For this, model systems are necessary.
GR activation can be modeled robustly in vitro and in vivo and is thus amenable to system-wide investigation. Rodent studies have provided important insights into the role of GCs during development, including that synthetic GC exposure during gestation can affect neurodevelopment (20), and additionally can lead to depression-like phenotypes and other negative outcomes in adult offspring (21). Although important, these models have limitations. During brain development, important differences exist between species in the abundance, lineage complexity, and proliferative potential of certain neural progenitor cell types such as basal radial glia. Additionally, GR response elements and enhancers are not always conserved. Human induced pluripotent stem cells (hiPSCs) represent a promising tool for filling this gap. In addition to having human genetic background, hiPSC-derived cerebral organoids recapitulate the three-dimensional complexity and cell-to-cell interactions (22, 23) of fetal brain development during early to mid gestation (24–26), a period critical for neuron production, migration, connection, and differentiation (1).
Environmental impacts during these sensitive periods have rarely been studied. We aimed to test whether cerebral organoids are suitable for studying the impact of prenatal GC exposure on the developing brain. Here, GR activation with dexamethasone (Dex) in organoids led to cell-type-specific transcriptional responses mapping to genes that moderate risk for negative neurodevelopmental trajectories and psychiatric and behavioral traits, especially in neuronal cells.
Methods
Three hiPSC lines were used to generate cerebral organoids (27), starting with 9,000 cells. Line1 cells were reprogrammed from NuFF3-RQ male human newborn foreskin feeder fibroblasts (GSC-3404, GlobalStem) (28). Line2 cells were generated using a plasmid-based protocol for integration-free hiPSCs from peripheral blood mononuclear cells (29) from a female donor through the BeCOME study (30). Line3 hiPSCs were reprogrammed skin fibroblasts (HPS0076:409b2, RIKEN BRC cell bank, female) (31, 32). All donors gave informed consent.
Downstream analyses used the Scanpy package (33) unless stated otherwise, and MAST (34) was used for all single-cell differential expression analyses. To test for enrichment of selected gene sets in the differentially expressed genes (DEGs; false discovery rate–corrected p≤0.05), we used hypergeometric tests or permutations based on equal-sized gene sets with either the same mean expression distribution as the significant DEGs from each cell class or random sampling.
The study methods are described in more detail in the Supplementary Methods section of the online supplement.
Results
Cerebral Organoids Model the Early Developing Brain In Vitro and Express the Molecular Machinery for Glucocorticoid Response
Organoid transcriptomes were explored using bulk RNA sequencing (RNAseq) at seven successive developmental times from day 17 to day 158 (see Table S1 in the online supplement). We delineated a robust developmental trajectory at the whole transcriptome level, with age-defining genes clearly clustering in a progressive temporal manner (Figure 1A). Established cell type markers displayed predictable developmental patterns at the RNA (see Figure S1A,B and Table S2 in the online supplement) and protein level (see Figure S1C,D in the online supplement). We compared this data set with other RNAseq data sets including fetal and adult postmortem brain (35, 36) and hiPSC-derived in vitro differentiation models. We found that organoids older than 70 days clustered with early fetal brain samples (Figure 1B).
FIGURE 1. Glucocorticoid receptor (GR) activation paradigm in organoidsa
a Panel A is heatmap representing all genes peaking at each of seven tested time points (days 17 to 158; D17–D158 in the figure) of organoid culture (N=3), organized by temporal trajectory. Gene lists are provided in Table S2 in the online supplement. Organoids show a robust and linear progression of gene expression across developmental time. Panel B presents results of singular value decomposition analysis of published bulk transcriptome data sets including fetal and adult postmortem brain from BRAINSPAN and CORTECON, as well as two-dimensional human induced pluripotent stem cell–derived in vitro differentiation models, and finally blood gene expression as a non-brain control. Our organoid data from mid to older age organoids cluster most closely with fetal brain samples and advanced two-dimensional in vitro neuronal cultures. Organoid transcriptome data sets (N=21), represented by black diamonds, follow the linear trajectory of progressively maturing cell types or tissues according to organoid age. Panel C is an immunofluorescence image of GR protein staining in day-30 organoids, showing protein-level expression of the receptor in a fraction of cells throughout the three-dimensional structure. Magnification is 25×, obtained with a confocal microscope; blue=nuclei (DAPI [4',6-diamidino-2-phenylindole]); green=neurons (DCX [Doublecortin]); white=GR (glucocorticoid receptor). Panel D shows RNA expression of the primary GR protein complex genes across organoid development. NR3C1 expression significantly increased only between days 23 and 40 (analysis of variance and Holm-Šídák post hoc pairwise test, p<0.0001; fold change=1.29; N=3 per time point). NR3C1=glucocorticoid receptor; HSPA1B=heat shock protein 70; DNAJA1=heat shock protein 40; HSP90B1=heat shock protein 90; STIP1=stress-induced phosphoprotein 1; PTGES3=prostaglandin E synthase 3; HDAC6=histone deacetylase 6; FKBP5=FKBP prolyl isomerase 5. Panels E and F show results of quantitative PCR analysis of key GR-regulated genes across two different acute time paradigms (4 hours and 12 hours, respectively) and three concentrations of dexamethasone (Dex) (10 nM, 100 nM, and 1000 nM) in stimulation paradigms. Organoids were stimulated at day 45 in culture (N=3). Gene expression was normalized to the geometric mean of endogenous genes GAPDH, POLR2A, and YWHAZ. (Statistical analysis and results are reported in Table S3 in the online supplement.) Fold changes are reported in reference to vehicle (dimethyl sulfoxide). Expression is graphically represented as normalized to the vehicle expression for each gene (indicated by the dotted horizontal red line). Statistical significance (two-sided) is denoted with an asterisk, representing a p value ≤0.05. Bars indicate mean, and error bars indicate standard error of the mean.
Acute Glucocorticoid Exposure in Cerebral Organoids Elicits Differential Expression of Established Glucocorticoid-Responsive Genes
We found the GR (NR3C1 gene) to be present at the protein level in a fraction of cells throughout the ventricle structure of organoids, from basal to apical, as well as outside the ventricle in stromal cells. Thus, GR is neither exclusive to any cell type nor ubiquitously expressed (Figure 1C). To determine the availability of the molecular machinery for GR activation in organoids, we tested expression of chaperones and co-chaperones of the GR protein complex (37) at the RNA level (Figure 1D; see also Table S1 in the online supplement). NR3C1 expression increased with organoid age, significantly between day 23 and day 40 (p<0.0001; fold change [FC]=1.29), plateauing after day 40 (Figure 1D). However, since cell type composition is also dynamic with age, we cannot conclude that there is an age effect on gene expression levels, as opposed to a change in the abundance of gene-expressing cells. Genes encoding the GR protein complex were abundantly and stably expressed across organoid development (Figure 1D).
To test GR activation dynamics, we used the synthetic agonist Dex. Organoids were grown until day 45, that is, after GR expression plateau, and tested with two acute paradigms (4 hours and 12 hours), with three different medically relevant Dex concentrations (10 nM, 100 nM, and 1000 nM) and vehicle control (dimethyl sulfoxide). Neither progenitor (SOX2, PAX6) nor neuronal cell markers (TUBB3, MAP2), nor the GR itself (NR3C1) were significantly Dex reactive at the RNA level (Figure 1E,F; see also Table S3 in the online supplement). This was contrary to transcripts known to be up-regulated by GCs: FKBP5 (38), SGK1 (39), TSC22D3 (37), and ZBTB16 (40). Most of the concentration-time combinations tested resulted in significant up-regulation of these genes (Figure 1E,F; see also Table S3). The strongest response was elicited by 12-hour stimulation (Figure 1F), with a 39-fold change for TSC22D3 and a 20-fold change for FKBP5 (Table S3). This combination (100 nM Dex/12 hours) was chosen for subsequent experiments.
Single-Cell RNA and Protein Characterization Shows Non-Cell-Type-Specific Expression and GR Activation in Organoids
We next aimed to understand cell-type-specific GC responses. We performed single-cell transcriptome sequencing in hiPSC Line1 at three time points with relevant GR expression: at day 30, day 60, or day 90 (N=4 each) in organoids exposed to Dex (100 nM/12 hours). After quality control procedures, we profiled 14,002 cells across time points and treatment conditions (day 30, N=5,035 cells; day 60, N=4,315; day 90, N=4,652). We applied graph-based cluster analysis using Scanpy on all Line1 organoid cells and identified 17 fine cell types (Figure 2A) and three coarse cell classes (Figure 2B) based on established marker genes (see the Supplementary Results section, Table S4, and Figures S2 and S3A,B in the online supplement) (22, 23, 41). When separating cells by organoid age, the same cell types or cell classes were identified and represented by the same markers, with the main difference being relative cell abundance indicative of advancing neuronal maturation (see Supplementary Results and Figure S3C,D in the online supplement).
FIGURE 2. Cell-type-specific differential gene expression and phenotypic characterization in organoidsa
aPanel A is a dimensionality reduction uniform manifold approximation and projection (UMAP) plot depicting 14,002 single cells that passed quality control, from all time points and treatment conditions (N=12 experiments) from Line1 organoids. Colors depict each of 17 clusters representing individual cell types (details are provided in Figure S2 in the online supplement). Panel B is a UMAP plot representing classification of cells into three cell classes (nonneural progenitors, neural progenitors, and neurons). Panel C is a UMAP plot depicting expression of the GR gene NR3C1 across all 14,002 cells in the analysis, showing even distribution of NR3C1-positive cells across cell types. Panel D is a UMAP plot depicting expression of the mineralocorticoid receptor gene NR3C2 across all 14,002 cells in the analysis. Panel E shows average NR3C1 expression by cell cluster. Cell types are listed along the y-axis, and the x-axis is the log2-normalized average gene expression across each cluster (the color scheme corresponds to that in panel A). Panel F depicts the quantification of nuclear translocation in all cell types from day-60 organoids. Panel G is an immunofluorescence image of an organoid in wild-type conditions showing GR protein localization in the cytoplasm of a neuronal cell. Magnification=40×; scale bar=50 μm; blue=nuclei (DAPI [4',6-diamidino-2-phenylindole]); green=young neurons (DCX [Doublecortin]); white=GR (glucocorticoid receptor); red=neurons (NeuN [RNA Binding Fox-1 Homolog 3]). Panel H is an immunofluorescence image of an organoid following 100 nM dexamethasone (Dex) treatment for 12 hours, showing nuclear translocation of GR in a neuronal cell. Colors and scale are as in panel G.
Treated and untreated organoids yielded comparable cell numbers and distribution across organoids and cell types (see Figure S3E in the online supplement). NR3C1 (Figure 2C) had detectable expression in 15% of all cells. High Dex concentrations have been shown also to induce activation of the mineralocorticoid receptor (NR3C2 gene). Given that NR3C2 (Figure 2D) is detected in ≤2% of cells as opposed to 15%−34% of cells expressing NR3C1, consistent with other organoid (22, 42, 43) or fetal brain (42, 44) data, it is more likely that the majority of Dex-induced transcriptional responses are conferred by GR activation. This difference between the two glucocorticoid receptors was consistent when separating cells by organoid age (see Figure S4D–I in the online supplement). NR3C1 was distributed across cell types, although on average more abundantly in nonneural progenitors. Specifically, in the ES, S, and M clusters, >50% of cells expressed NR3C1 (Figure 2E). NR3C1 expression at specific organoid ages was not significantly different, either in all cells (see Figure S4J in the online supplement) or in individual clusters (i.e., mesenchymal stromal cells or radial glia; see Figure S4H,I in the online supplement).
For protein-level evidence of GR activation (37) in single cells, we applied 100 nM Dex/12-hour exposure in day-60 organoids and queried cellular localization using immunofluorescence and confocal microscopy. In the vehicle condition, GR was preferentially localized in the cytoplasm but also sometimes detectable in the nucleus (Figure 2F,G). Following Dex there was a significant (χ2=100.9, p<0.0001) shift in detectable GR toward the nucleus (Figure 2F,H), suggesting Dex-induced nuclear translocation, and thus GR activation. This was true regardless of cell type, with significant shift to nuclear GR following Dex happening in nonneural progenitors, neural progenitors, and neurons (see Figure S5 in the online supplement).
Cell-Type-Specific Responses to Acute Glucocorticoids Point to a Disruption of Neuronal Processes With Lasting Effects
Given the comparable cell types observed at day 30, day 60, and day 90 (see Figure S4 and Tables S5–S7 in the online supplement), we performed the main differential expression analyses in all Line1 cells combined (referred to as “Alldays”) to maximize statistical power. In the 17-cluster annotation, we found significant DEGs in nine clusters, ranging from one to 55 DEGs (q≤0.05; see Figure S6 and Table S8 in the online supplement). More detail on specific genes and dysregulation patterns is presented in the Supplementary Results section in the online supplement.
When analyzing differential expression within the three broader cell classes, we found 899 DEGs in nonneural progenitors, 407 DEGs in neural progenitors, and 322 DEGs in neurons (q≤0.05) (Figure 3A; see also Table S8 in the online supplement). The number of DEGs increased with increasing NR3C1 expression, and fold changes were higher in NR3C1-positive cells (Figure 3B; see also the Supplementary Results section in the online supplement). Forty-eight DEGs were significant in all three cell classes (Figure 3C), of which 33% were down-regulated, including neuronal genes FOXG1 and NEUROD6, and 67% were up-regulated, including Dex-responsive genes FKBP5 and TSC22D3 and the progenitor gene MGARP. Gene ontology analyses of these 48 shared genes revealed “forebrain development” and “cerebral cortex development” among the most significantly enriched (see Table S9 in the online supplement). When separating cells by organoid age, the fold change direction for all significant DEGs was always consistent with the direction in Alldays in at least two ages (q≤0.05; see Table S10 in the online supplement), and switched direction extremely rarely (≤2%) only in one age, supporting the decision to combine cells across organoid age. However, few genes were significant across all organoid ages (q≤0.05) reflective of power limitations due to cell abundance.
FIGURE 3. Cell-type-specific differential gene expression in organoidsa
aPanel A is a bubble plot of number of differentially expressed genes (DEGs) by average NR3C1 expression in all three cell classes. Panel B shows differential gene expression by cell class in glucocorticoid receptor (GR)–positive versus GR-negative cells; absolute fold changes of genes previously associated with glucocorticoid response are plotted in nonneural progenitors, neural progenitors, and neurons. HPC=hippocampal progenitor cells. Panel C shows the overlap of differentially expressed transcripts (q cutoff, 0.05) across three cell classes. Panels D–G show lasting effects of acute (12 hours, 100 nM) dexamethasone (Dex) stimulation up to 8 days after treatment had stopped. Gene expression is quantified by real-time quantitative reverse transcription PCR in RNA from full organoids, in four genes found differentially expressed across all three cell classes: FKBP5, TSC22D3, MGARP, and PAX6. Expression is graphically represented normalized to the vehicle expression for each gene. Statistical significance (two-sided) is denoted with asterisks, representing a p value ≤0.05. Bars indicate mean, and error bars indicate standard error of the mean.
To test whether this acute stimulation has a sustained effect beyond the immediate GC response, we exposed organoids to 100 nM Dex for 12 hours and collected them up to 8 days after Dex removal. At the bulk RNA level, the effect of this acute GR activation lasted for >6 days (Figure 3D–G; see also Figure S7 and Table S3 in the online supplement). When exploring this effect on cell-type composition at the protein level, after an 8-day washout period, we found an increase in new CTIP2-positive neurons (p=0.0007; FC=2.5; deep layer V marker) and an increase in TBR1 that fell short of significance (p=0.07; FC=2.4; layer VI marker) in the ventricular zone. There was no significant difference in upper layer marker SATB2, as these neurons have a different birth and migration window. These findings are in line with lasting effects of the Dex-related up-regulation of progenitor markers such as PAX6 (see Figure S7) and evidence from mouse studies (45).
Dex-Regulated Transcripts in Neurons Are Enriched for Genes Implicated by GWAS for Behavioral Traits
To better gauge a potential impact of altered GR activation during neurodevelopment on phenotypes later in life, we mapped DEGs to gene lists implicated by over 700 genome-wide association studies (GWASs) in different quantitative traits and diseases recorded in the GWAS Catalog (46) (see Table S11 in the online supplement). We tested enrichment of DEGs from the three cell classes (q≤0.05) within genes significantly associated with GWAS traits using FUMA (47). To increase specificity, we reduced the test set to only those traits with at least five shared genes with our DEG lists. No significant enrichment emerged for nonneural progenitors. Neural progenitor DEGs were enriched for genes genome-wide associated with 38 traits, and neuron DEGs with 22 traits (enrichment q≤0.05) (Figure 4A,B; see also Table S11). GWAS phenotypes not related to brain function were labeled as “other.” Brain function–related traits were labeled “brain/behavior.” The latter were overrepresented in traits enriched for neuron DEGs, making up 59% of the traits with significant enrichment for this cell class (see Table S11). This distribution was significantly different from the full list of tested traits (N=59) (χ2=4.23, p=0.04) (Figure 4C). Traits classified as “brain/behavior” included “depression,” “neuroticism,” “chronotype,” and “adventurousness,” suggesting that GR activation in the developing brain can affect genes relevant for behavioral phenotypes and psychiatric disorders later in life. This was not significant for neural progenitors, where 89% of traits fell in the “other” category (Figure 4D; see also Table S11).
FIGURE 4. Enrichment analyses in gene lists associated with disease and disease-related traitsa
aPanels A and B show significant results from enrichment analysis of differentially expressed genes from neurons and neural progenitors, respectively, versus all currently published traits with genome-wide association study (GWAS)–significant results (q≤0.05; neural progenitors: N=38 significant traits; neurons: N=22 significant traits). MTAG=multi-trait analysis of GWAS. Panels C and D show comparisons of significantly enriched traits classified as either “brain/behavior” or “other” (percent of total) in neurons or neural progenitors, respectively, compared with the full trait lists with at least five genes’ overlap with neural progenitors or neurons. ns=nonsignificant. Panels E and F show permutations (N=1,000) of cross-disorder enrichment tests in mean-matched nonsignificant (q>0.05) genes from differential expression analyses of neurons and neural progenitors, respectively. Panel G–I show results of enrichment analyses of neurodevelopmental disorders. Significant results are shown from enrichment analysis of differentially expressed genes from nonneural progenitors, neural progenitors, and neurons, respectively, versus gene lists from the DisGeNET database for autism spectrum disorder (ASD) and neurodevelopmental disorders (ND), as well as height as a control. The vertical dotted line represents the false discovery rate–corrected significance threshold (q≤0.05).
Since prenatal GC exposure has been associated with cognitive and behavioral outcomes (14, 15), we tested DEG enrichment among genes carrying common variants identified by the Cross-Disorder Group of the Psychiatric Genomics Consortium GWAS of eight mental illnesses (48). We found a significant enrichment only for neuron DEGs (q=0.00475 and odds ratio=3.91; see Table S12 in the online supplement). To test whether the enrichment results were reflective of cell-type-specific gene expression abundance, rather than strictly GR activation, we used a permutation-based approach. Gene sets comparable in size and mean expression distribution but not significantly differentially expressed (q≤0.05) were tested. A significant enrichment of these gene sets was observed only in 2/1,000 iterations (corrected empirical p=0.009) (Figure 4E; see also Table S12). This indicates that the enrichment of DEGs among genes in the Cross-Disorder GWAS is specific to the GR responsiveness of transcripts and not their expression level in neurons. This test was not significant in neural progenitors (Figure 4F; see also Table S12). An approach where permuted gene sets were randomly distributed yielded comparable results (see Table S12), further reinforcing the specificity of the DEG enrichment finding in neurons.
GR Activation in Neurons Uniquely Regulates Neurodevelopmental Disease-Related Genes
For neurodevelopmental phenotypes, we used gene lists from the DisGeNET database (49) associated with autism spectrum disorder (ASD), neurodevelopmental disorders, and height (as a non-brain-related control). We found a significant enrichment for ASD genes in DEGs from all three cell classes (Figure 4G–I; see also Table S13 in the online supplement). The most pronounced result emerged for neuron DEGs, where we found a significant enrichment with both ASD and neurodevelopmental disorders but not height (ASD: p=8.89 × 10−5, odds ratio=3.04; neurodevelopmental disorders: p=3.41 × 10−6, odds ratio=5.59; height: p=0.83, odds ratio=0.67) (Figure 4I; see also Table S13). Size- and expression-level-matched as well as random permuted gene set enrichments for ASD and neurodevelopmental disorders never reached a p value less than or equal to the nominal p value in the neurons (see Figure S8A,B in the online supplement), while this was not the case for height, making this finding specific to the Dex-regulated genes in this cell class (empirical q=0.009; see Table S13). We further validated these findings with genes carrying loss-of-function mutations associated with intellectual disability from the Developmental Brain Disorders Database (50) (see the Supplemental Results section, Table S14, and Figure S8C,D in the online supplement).
Validation in Day-90 Organoids of Two Additional Genetic Backgrounds Supports Disease Relevance of Neuronal Glucocorticoid Response
Since the most interesting findings emerged in neurons, and this cell class was most abundant at day 90, we generated validation data sets in same-age organoids derived from two additional hiPSC lines. When cells were classified as before (nonneural progenitors, neural progenitors, and neurons) (Figure 5A–C; see also Figure S9 in the online supplement), the proportions of neurons were different across Lines1–3 (Figure 5D), supporting previous reports of heterogeneity of organoids from different donor hiPSCs (24, 43). In neurons (Line1: N=1,831; Line2: N=1,772; Line3 N=2,612) (Figure 5D), we observed 641 DEGs in Line2 and 446 DEGs in Line3 (q≤0.05; see Table S15 in the online supplement), compared with 337 DEGs from the comparable analysis in Line1 at day 90. Of the significant DEGs, 32%, 26%, or 39%, respectively, for Line1, Line2, and Line3 were shared with at least one other line. Thirty-one genes were shared across all three lines, including progenitor staples such as PAX6, FABP7, and CRABP1, and neuronal genes such as C1orf61, GAD2, and NEUROD1. When focusing on the direction of differential expression, we found that 120 of the DEGs that were significant in Line1 (N=337) shared fold-change direction across Lines1–3, three times more than would be expected by chance (binomial distribution p<0.0001).
FIGURE 5. Validation of neuron differentially expressed gene (DEG) enrichment profile in day-90 organoids derived from three independent cell linesa
aPanels A–C are uniform manifold approximation and projection (UMAP) plots of three cell class distributions in three distinct cell lines: Line1, Line2, and Line3. Panel D shows the distribution of cells along three cell classes in three cell lines. Panels E–G show the top significant results from enrichment analysis of differentially expressed genes from neurons versus all currently published traits with genome-wide association study (GWAS)–significant results (q≤0.05), presenting comparisons of significantly enriched traits classified as either “brain/behavior” or “other” (% of total) in neurons in Lines1–3 against the full GWAS trait lists. Panel H presents results of enrichment analysis of neurodevelopmental disorder risk. Significant results from enrichment analysis of differentially expressed genes in neurons from organoids derived from three cell lines versus gene lists from the Developmental Brain Disorders Database of genes associated with intellectual disability. The vertical dotted line represents the false discovery rate–corrected significance threshold (q≤0.05).
Despite differences in cell numbers and proportions (Figure 5D), a similar pattern emerged in the pathways and gene sets enriched. When testing for enrichment of DEGs from neurons (q≤0.05) within genes significantly associated with GWAS traits using FUMA, traits classified as “brain/behavior” were overrepresented in significantly enriched traits, representing 82%, 88%, and 74%, respectively, in Lines1–3 (Figure 5E–G; see also Table S16 in the online supplement). This distribution was significantly different from the full list of traits (Line1: χ2=30.97, p<0.0001 [Figure 5E]; Line2: χ2=78.42, p<0.0001 [Figure 5F]; Line3: χ2=35.41, p<0.0001 [Figure 5G]). Many of the shared enriched GWAS traits across cell lines were related to neuropsychiatry, such as “adventurousness,” “irritable mood,” and “feeling worry.” There were 12 shared enriched GWAS terms across at least two of Lines1–3 (see Table S17A in the online supplement), of which 11 (92%) belonged to the “brain/behavior” category, while “irritable mood” was shared across Lines1–3 (see Table S17B). The 12 shared enrichments were driven by 107 DEGs (Table S17C). TCF4, a transcription factor associated with multiple psychiatric disorders, was the strongest contributor, with 25 instances (trait-by-Line) where it drove the enriched GWAS trait (see Table S17C).
Line2 and Line3 also validated the finding of an enrichment in Cross-Disorder GWAS associations (see Table S18 in the online supplement). Finally, we also tested the relationship between DEGs and intellectual disability genes with loss-of-function mutations in the Developmental Brain Disorders Database (49), where all lines showed a similarly strong enrichment among neuron DEGs (Figure 5H; see also Table S18). Random permutations of nonsignificant genes did not reach a p value less than or equal to the nominal p value (see Table S18). Furthermore, these gene set enrichment findings were specific to neurons in all three lines (see Table S18), validating the finding of a unique profile of GC response in this mature cell class.
Discussion
In epidemiological and clinical studies, GR activation via synthetic GC administration during critical periods of human brain development is associated with negative health outcomes, such as endocrine dysfunction and brain structural and functional alterations, and emotional, behavioral, cognitive, or mental health problems (12, 51, 52). We explored whether hiPSC-derived cerebral organoids can improve our mechanistic understanding of in utero GC exposure. We found GR expression and activity across all cell types in this system. Single-cell RNA sequencing showed that transcriptional GC response was cell-type-specific, pointing to interference with neuronal differentiation and maturation processes. Transcriptional response in neurons was significantly enriched for genes associated with behavioral and neurodevelopmental traits, suggesting a possible convergence of environmental and genetic factors on the same cell-type-specific neurodevelopmental pathways. Thus, cerebral organoids open the possibility of interrogating human-specific genetic vulnerability to prenatal exposures.
The most profound GC response was in neurons, with decreased expression of neuronal-specific genes NEUROD6, FOXG1, TBR1, MYT1L, and NFIA but an increase of progenitor genes PAX6 and MGARP across several cell types and time. This confirms cell and animal study findings of GC involvement in neurogenesis by increasing proliferation while decreasing differentiation (39, 53, 54). In addition to acute effects after 12 hours, we showed Dex effects that lasted several days, both at the RNA level, on GR-responsive genes FKBP5 and TSC22D3 and progenitor genes PAX6 and MGARP, and at the tissue level, with an increase in new CTIP2-positive deep layer neurons. This is in line with mouse models treated with a single acute GC dose (45). Thus, this study further supports the evidence of dysregulated GR activation leading to cell-type-specific effects on neuronal maturation and differentiation.
Since genes associated with psychiatric risk are also important players during fetal development (55), we wanted to understand how these genes respond to GC exposure during neurodevelopment. In our data, GC-responsive transcripts were enriched among gene loci carrying genetic variants associated with neurodevelopmental, behavioral, and psychiatric phenotype associations. Uniquely for GR-activated transcripts in neurons, we found an enrichment of genes linked to behavioral phenotype GWAS findings and to neurodevelopmental disorders. Importantly, validation showed that despite cell-type and individual gene-level heterogeneity between organoids derived from three donor hiPSCs, the pathways of GC-responsive transcripts in neurons are reproducibly consistent. Many of the genes driving these convergent enrichments were transcription factors essential for neural cell specification (TCF4, AUTS2, PAX6, RELN, OTX2) and are implicated in neurodevelopment as well as in brain-related disorders (see Table S17B in the online supplement). A limitation of these transcriptome-wide analyses is that the mechanism by which activated GR induces these expression changes (DNA binding or protein-protein interactions with other transcription factors) was not investigated, and the timing of GC exposure allows for both direct and indirect transcriptional responses. Mechanistic investigation of direct versus indirect GR effects would be highly interesting.
Overall, our results highlight the interplay between environmental exposure and inherited genetic factors on neurodevelopment. Organoids allow exploration of the interaction between human-specific genetic variation and environmental exposures in a three-dimensional system. While we showed that organoids constitute a suitable model for in utero GC exposure, they have limits in modeling complex exposures, such as prenatal stress. Nonetheless, this study helps fill knowledge gaps on how prenatal environmental disturbances affect neurodevelopment, leading to negative outcomes. In the future, this system could help screen compounds to counteract GC effects on brain development.
Supplementary Material
File (ajp_21010095_correction01.pdf)
- View/Download
- 19.54 KB
File (appi.ajp.2021.21010095.ds001.pdf)
- View/Download
- 1.69 MB
File (appi.ajp.2021.21010095.ds002.xlsx)
- Download
- 14.53 MB
References
1.
Monk C, Lugo-Candelas C, Trumpff C: Prenatal developmental origins of future psychopathology: mechanisms and pathways. Annu Rev Clin Psychol 2019; 15:317–344
2.
Krontira AC, Cruceanu C, Binder EB: Glucocorticoids as mediators of adverse outcomes of prenatal stress. Trends Neurosci 2020; 43:394–405
3.
Bonanno C, Wapner RJ: Antenatal corticosteroids in the management of preterm birth: are we back where we started? Obstet Gynecol Clin North Am 2012; 39:47–63
4.
Love MI, Huska MR, Jurk M, et al: Role of the chromatin landscape and sequence in determining cell type-specific genomic glucocorticoid receptor binding and gene regulation. Nucleic Acids Res 2017; 45:1805–1819
5.
Martens C, Bilodeau S, Maira M, et al: Protein-protein interactions and transcriptional antagonism between the subfamily of NGFI-B/Nur77 orphan nuclear receptors and glucocorticoid receptor. Mol Endocrinol 2005; 19:885–897
6.
Leviton LC, Goldenberg RL, Baker CS, et al: Methods to encourage the use of antenatal corticosteroid therapy for fetal maturation: a randomized controlled trial. JAMA 1999; 281:46–52
7.
Miracle X, Di Renzo GC, Stark A, et al: Guideline for the use of antenatal corticosteroids for fetal maturation. J Perinat Med 2008; 36:191–196
8.
Cartier J, Zeng Y, Drake AJ: Glucocorticoids and the prenatal programming of neurodevelopmental disorders. Curr Opin Behav Sci 2016; 7:1–7
9.
Alijotas-Reig J, Garrido-Gimenez C: Current concepts and new trends in the diagnosis and management of recurrent miscarriage. Obstet Gynecol Surv 2013; 68:445–466
10.
Kajantie E, Raivio T, Jänne OA, et al: Circulating glucocorticoid bioactivity in the preterm newborn after antenatal betamethasone treatment. J Clin Endocrinol Metab 2004; 89:3999–4003
11.
Carson R, Monaghan-Nichols AP, DeFranco DB, et al: Effects of antenatal glucocorticoids on the developing brain. Steroids 2016; 114:25–32
12.
Buss C, Entringer S, Moog NK, et al: Intergenerational transmission of maternal childhood maltreatment exposure: implications for fetal brain development. J Am Acad Child Adolesc Psychiatry 2017; 56:373–382
13.
Asztalos EV, Murphy KE, Willan AR, et al: Multiple courses of antenatal corticosteroids for preterm birth study: outcomes in children at 5 years of age (MACS-5). JAMA Pediatr 2013; 167:1102–1110
14.
Räikkönen K, Gissler M, Kajantie E: Associations between maternal antenatal corticosteroid treatment and mental and behavioral disorders in children. JAMA 2020; 323:1924–1933
15.
Melamed N, Asztalos E, Murphy K, et al: Neurodevelopmental disorders among term infants exposed to antenatal corticosteroids during pregnancy: a population-based study. BMJ Open 2019; 9(9):e031197
16.
Green CG, Babineau V, Jolicoeur-Martineau A, et al: Prenatal maternal depression and child serotonin transporter linked polymorphic region (5-HTTLPR) and dopamine receptor D4 (DRD4) genotype predict negative emotionality from 3 to 36 months. Dev Psychopathol 2017; 29:901–917
17.
McEwen BS, Nasca C, Gray JD: Stress effects on neuronal structure: hippocampus, amygdala, and prefrontal cortex. Neuropsychopharmacology 2016; 41:3–23
18.
Zou R, Tiemeier H, van der Ende J, et al: Exposure to maternal depressive symptoms in fetal life or childhood and offspring brain development: a population-based imaging study. Am J Psychiatry 2019; 176:702–710
19.
Graham AM, Rasmussen JM, Entringer S, et al: Maternal cortisol concentrations during pregnancy and sex-specific associations with neonatal amygdala connectivity and emerging internalizing behaviors. Biol Psychiatry 2019; 85:172–181
20.
Shoener JA, Baig R, Page KC: Prenatal exposure to dexamethasone alters hippocampal drive on hypothalamic-pituitary-adrenal axis activity in adult male rats. Am J Physiol Regul Integr Comp Physiol 2006; 290:R1366–R1373
21.
Hiroi R, Carbone DL, Zuloaga DG, et al: Sex-dependent programming effects of prenatal glucocorticoid treatment on the developing serotonin system and stress-related behaviors in adulthood. Neuroscience 2016; 320:43–56
22.
Kanton S, Boyle MJ, He Z, et al: Organoid single-cell genomic atlas uncovers human-specific features of brain development. Nature 2019; 574:418–422
23.
Pollen AA, Bhaduri A, Andrews MG, et al: Establishing cerebral organoids as models of human-specific brain evolution. Cell 2019; 176:743–756.e17
24.
Amiri A, Coppola G, Scuderi S, et al: Transcriptome and epigenome landscape of human cortical development modeled in organoids. Science 2018; 362:eaat6720
25.
Kelava I, Lancaster MA: Dishing out mini-brains: current progress and future prospects in brain organoid research. Dev Biol 2016; 420:199–209
26.
Luo C, Lancaster MA, Castanon R, et al: Cerebral organoids recapitulate epigenomic signatures of the human fetal brain. Cell Rep 2016; 17:3369–3384
27.
Lancaster MA, Knoblich JA: Generation of cerebral organoids from human pluripotent stem cells. Nat Protoc 2014; 9:2329–2340
28.
Cardenas A, Villalba A, de Juan Romero C, et al: Evolution of cortical neurogenesis in amniotes controlled by Robo signaling levels. Cell 2018; 174:590–606.e21
29.
Dowey SN, Huang X, Chou BK, et al: Generation of integration-free human induced pluripotent stem cells from postnatal blood mononuclear cells by plasmid vector expression. Nat Protoc 2012; 7:2013–2021
30.
Brückl TM, Spoormaker VI, Sämann PG, et al: The biological classification of mental disorders (BeCOME) study: a protocol for an observational deep-phenotyping study for the identification of biological subtypes. BMC Psychiatry 2020; 20:213
31.
Koyanagi-Aoi M, Ohnuki M, Takahashi K, et al: Differentiation-defective phenotypes revealed by large-scale analyses of human pluripotent stem cells. Proc Natl Acad Sci USA 2013; 110:20569–20574
32.
Okita K, Matsumura Y, Sato Y, et al: A more efficient method to generate integration-free human iPS cells. Nat Methods 2011; 8:409–412
33.
Wolf FA, Angerer P, Theis FJ: SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 2018; 19:15
34.
Finak G, McDavid A, Yajima M, et al: MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol 2015; 16:278
35.
Fromer M, Roussos P, Sieberts SK, et al: Gene expression elucidates functional impact of polygenic risk for schizophrenia. Nat Neurosci 2016; 19:1442–1453
36.
Miller JA, Ding SL, Sunkin SM, et al: Transcriptional landscape of the prenatal human brain. Nature 2014; 508:199–206
37.
Petta I, Dejager L, Ballegeer M, et al: The interactome of the glucocorticoid receptor and its influence on the actions of glucocorticoids in combatting inflammatory and infectious diseases. Microbiol Mol Biol Rev 2016; 80:495–522
38.
Gassen NC, Fries GR, Zannas AS, et al: Chaperoning epigenetics: FKBP51 decreases the activity of DNMT1 and mediates epigenetic effects of the antidepressant paroxetine. Sci Signal 2015; 8:ra119
39.
Anacker C, Cattaneo A, Musaelyan K, et al: Role for the kinase SGK1 in stress, depression, and glucocorticoid effects on hippocampal neurogenesis. Proc Natl Acad Sci USA 2013; 110:8708–8713
40.
Menke A, Arloth J, Pütz B, et al: Dexamethasone stimulated gene expression in peripheral blood is a sensitive marker for glucocorticoid receptor resistance in depressed patients. Neuropsychopharmacology 2012; 37:1455–1464
41.
Velasco S, Kedaigle AJ, Simmons SK, et al: Individual brain organoids reproducibly form cell diversity of the human cerebral cortex. Nature 2019; 570:523–527
42.
Polioudakis D, de la Torre-Ubieta L, Langerman J, et al: A single-cell transcriptomic atlas of human neocortical development during mid-gestation. Neuron 2019; 103:785–801.e8
43.
Bhaduri A, Andrews MG, Mancia Leon W, et al: Cell stress in cortical organoids impairs molecular subtype specification. Nature 2020; 578:142–148
44.
Nowakowski TJ, Bhaduri A, Pollen AA, et al: Spatiotemporal gene expression trajectories reveal developmental hierarchies of the human cortex. Science 2017; 358:1318–1323
45.
Tsiarli MA, Rudine A, Kendall N, et al: Antenatal dexamethasone exposure differentially affects distinct cortical neural progenitor cells and triggers long-term changes in murine cerebral architecture and behavior. Transl Psychiatry 2017; 7:e1153
46.
MacArthur J, Bowler E, Cerezo M, et al: The new NHGRI-EBI catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res 2017; 45(D1):D896–D901
47.
Watanabe K, Taskesen E, van Bochoven A, et al: Functional mapping and annotation of genetic associations with FUMA. Nat Commun 2017; 8:1826
48.
Cross-Disorder Group of the Psychiatric Genomics Consortium, Lee PH, Anttila V, et al: Genome wide meta-analysis identifies genomic relationships, novel loci, and pleiotropic mechanisms across eight psychiatric disorders. bioRxiv, January 26, 2019 (https://www.biorxiv.org/content/10.1101/528117v1)
49.
Piñero J, Bravo À, Queralt-Rosinach N, et al: DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants. Nucleic Acids Res 2017; 45(D1):D833–D839
50.
Mirzaa GM, Millen KJ, Barkovich AJ, et al: The Developmental Brain Disorders Database (DBDB): a curated neurogenetics knowledge base with clinical and research applications. Am J Med Genet A 2014; 164A:1503–1511
51.
Khalife N, Glover V, Taanila A, et al: Prenatal glucocorticoid treatment and later mental health in children and adolescents. PLoS One 2013; 8:e81394
52.
Wolford E, Lahti-Pulkkinen M, Girchenko P, et al: Associations of antenatal glucocorticoid exposure with mental health in children. Psychol Med 2020; 50:247–257
53.
Odaka H, Adachi N, Numakawa T: Impact of glucocorticoid on neurogenesis. Neural Regen Res 2017; 12:1028–1035
54.
Provençal N, Arloth J, Cattaneo A, et al: Glucocorticoid exposure during hippocampal neurogenesis primes future stress response by inducing changes in DNA methylation. Proc Natl Acad Sci USA 2020; 117:23280–23285
55.
Birnbaum R, Jaffe AE, Hyde TM, et al: Prenatal expression patterns of genes associated with neuropsychiatric disorders. Am J Psychiatry 2014; 171:758–767
Information & Authors
Information
Published In
History
Received: 26 January 2021
Revision received: 21 June 2021
Accepted: 16 August 2021
Published online: 26 October 2021
Published in print: May 2022
Keywords
Authors
Competing Interests
Dr. Theis has served as a consultant for Immunai and has ownership interest in Dermagnostix and Cellarity. The other authors report no financial relationships with commercial interests.
Metrics & Citations
Metrics
Citations
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
For more information or tips please see 'Downloading to a citation manager' in the Help menu.
View Options
View options
PDF/EPUB
View PDF/EPUBLogin options
Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.
Personal login Institutional Login Open Athens loginNot a subscriber?
PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.
Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).