Skip to main content
NCBI home page
Stem Cell Reports logoLink to Stem Cell Reports
. 2024 Aug 22;19(9):1351–1367. doi: 10.1016/j.stemcr.2024.07.008

Cross-site reproducibility of human cortical organoids reveals consistent cell type composition and architecture

Madison R Glass 1,2,8, Elisa A Waxman 3,8, Satoshi Yamashita 4,8, Michael Lafferty 1,2, Alvaro A Beltran 1,2, Tala Farah 1,2, Niyanta K Patel 1,2, Rubal Singla 1,2, Nana Matoba 1,2, Sara Ahmed 1,2, Mary Srivastava 1,2, Emma Drake 1,2, Liam T Davis 1,2, Meghana Yeturi 1,2, Kexin Sun 1,2, Michael I Love 2,5, Kazue Hashimoto-Torii 4,6,9, Deborah L French 3,7,9, Jason L Stein 1,2,9,10,∗
PMCID: PMC11411306  PMID: 39178845

Summary

While guided human cortical organoid (hCO) protocols reproducibly generate cortical cell types at one site, variability in hCO phenotypes across sites using a harmonized protocol has not yet been evaluated. To determine the cross-site reproducibility of hCO differentiation, three independent research groups assayed hCOs in multiple differentiation replicates from one induced pluripotent stem cell (iPSC) line using a harmonized miniaturized spinning bioreactor protocol across 3 months. hCOs were mostly cortical progenitor and neuronal cell types in reproducible proportions that were consistently organized in cortical wall-like buds. Cross-site differences were detected in hCO size and expression of metabolism and cellular stress genes. Variability in hCO phenotypes correlated with stem cell gene expression prior to differentiation and technical factors associated with seeding, suggesting iPSC quality and treatment are important for differentiation outcomes. Cross-site reproducibility of hCO cell type proportions and organization encourages future prospective meta-analytic studies modeling neurodevelopmental disorders in hCOs.

Keywords: reproducibility, cortical organoids, stem cell state, neurodevelopment, scRNA-seq, tissue clearing

Graphical abstract

graphic file with name fx1.jpg

Open in a new tab

Highlights

  • •

    A harmonized cortical organoid protocol was replicated across 3 research sites

  • •

    Organoids produce consistent cell type proportions in neuroepithelial buds

  • •

    Cross-site differences were found in organoid size and metabolism gene expression

  • •

    iPSC handling was associated with differences in long-term differentiation outcomes

Glass and colleagues from 3 research groups demonstrate the cross-site reproducibility of cell type production and organization in a harmonized cortical organoid differentiation protocol. Organoid size, morphology, and gene expression relating to metabolism and biosynthetic processes varied across sites and were correlated to technical differences at the beginning of the differentiation.

Introduction

Human cortical organoids (hCOs) have been widely adopted to model neurodevelopmental disorders, but concerns remain about the influence of non-biological variability in organoid structure, cell type composition, and gene expression that may confound interpretations (Wang et al., 2023). Reproducibility has been assessed by individual research groups, who demonstrated consistent cell type production across multiple cell lines for guided hCO differentiation protocols (Yoon et al., 2019; Velasco et al., 2019; Qian et al., 2018; Uzquiano et al., 2022). To date, hCO reproducibility has not been systematically evaluated across multiple independent labs (also called sites) using harmonized protocols. Cross-site differences in hCO differentiation could be caused by many factors including uncontrolled differences in technical variables like handling or stochastic developmental processes. Identifying aspects of hCO differentiations that are reproducible across sites will enable prospective meta-analysis of neurodevelopmental disorder case-control studies, leading to sample sizes that are difficult to achieve in one group alone.

2D neuronal cultures of an induced pluripotent stem cell (iPSC) line with a presenilin mutation and its isogenic control were previously compared across 5 sites using bulk transcriptomics. Of notable concern, site differences were greater than case-control differences (Volpato et al., 2018). hCO differentiations have been previously compared across labs using pre-existing single-cell RNA sequencing (scRNA-seq) datasets. These studies have compared either consistency in developmental trajectories across hCO protocols (Tanaka et al., 2020; He et al., 2023) or fidelity of hCO cell types to human brain (Bhaduri et al., 2020; Werner and Gillis 2023), but have several limitations: (1) the previous studies could not disambiguate differences caused by cell lines, protocols, or technical artifacts from site-specific implementations of the protocol, (2) hCO phenotypes other than scRNA-seq have not been compared across multiple labs, and (3) previous approaches used site-specific, differentiation methods, rather than a harmonized protocol.

As part of the iPSC working group within the Intellectual and Developmental Disabilities Research Centers (IDDRC) located across the country, we were in the unique position to undertake an hCO cross-site reproducibility reproject. Three different IDDRC sites generated hCOs in multiple batches (differentiation replicates) from PGP1, a publicly available iPSC line (Lee et al., 2009), using a protocol that was pretested across all sites. By using this harmonized protocol, hCO phenotypes were rigorously assessed using multiple techniques including scRNA-seq, phase contrast and 3D fluorescent microscopy, flow cytometry, and qPCR. While there was variation in the visual quality, size, and gene expression patterns of the hCOs, we found that cell type proportions and structure were consistent across sites, allowing for the ability of future meta-analysis of cross-site research.

Results

Study design for multimodal test of hCO reproducibility

To test hCO reproducibility, three IDDRC sites (Children’s Hospital of Philadelphia [CHOP], Children’s National Hospital [CN], and the University of North Carolina at Chapel Hill [UNC]) chose a widely cited guided protocol that demonstrated transcriptomic fidelity to primary fetal cortex and cortical wall-like organization (Qian et al., 2016). The same iPSC line (PGP1), bioreactor parts, and Matrigel lot were used to standardize differentiations, allowing us to detect differences in hCO phenotypes due to handling differences across sites. Each site completed at least 4 differentiation replicates, defined as a set of hCOs seeded at the same time from the same passage of PGP1 (Table S1). We assessed common phenotypes studied in hCOs, including cell type proportions, gene expression, and structure across time using several assays (Figure 1A and supplemental methods). To minimize site-specific batch effects, samples were collected at each site and shipped to one site for specific assays and corresponding analyses.

Figure 1.

Figure 1

Open in a new tab

Cell type characterizations from day 14 and day 84 hCOs

(A) Experimental design showing critical steps for differentiation, collection time points, and assays.

(B) t-SNE plot of detected cell types across all sites and time points (133,980 cells from 28 hCOs). Abbreviations are as follows: NPC, G2: neural progenitor cells, G2 phase; IP, S: intermediate progenitor, S phase; NPC, S: neural progenitor cell, S phase; IP: non-dividing intermediate progenitor; oRG: outer radial glia; RG: radial glia; NSC: neuroepithelial stem cell; Neuron: unspecified neuron; Pan CN: pan-cortical neuron; ULN, a: upper-layer cortical neuron, group a; ULN, b: upper-layer cortical neuron, group b; LLN, a: lower-layer cortical neuron, group a. LLN; b: lower-layer cortical neuron group b; PN/CR: pan-neuron and Cajal-Retzius; and ChP/MZ: choroid plexus and marginal zone.

(C) t-SNE plots colored by differentiation day, inferred cell cycle phase, or gene expression of previously defined markers.

(D) Heatmap of normalized gene expression for selected marker genes that are expressed in at least 5% of the cells in a cluster.

(E) hCO cell type correlations to primary fetal telencephalic tissue (Bhaduri et al., 2020). Some in vivo cell type names (rows) have been simplified for plotting. Pearson’s correlation coefficients above 0.3 and surviving multiple comparison corrections of FDR <0.05 are labeled.

Organoids produce expected cell types

We collected gene expression measurements using a commercially available version of SPLiT-seq (Rosenberg et al., 2018), yielding 133,980 single cells after quality control (Figure S1). hCOs were selected for scRNA-seq based on visible budding at day 14 and smooth edges with expected morphology at day 84 (Figure S2). Cells were clustered based on the similarity of their gene expression and then cell types were annotated using known markers (Figure 1B; supplemental methods). hCOs produced expected cell types found in the developing fetal cortex, including neuroepithelial stem cells (NSCs), radial glia (RG), intermediate progenitors (IPs), outer radial glia (oRG), and cortical upper and lower layer neurons (Bhaduri et al., 2020; Kelley and Pașca 2022; Kanton et al., 2019). Non-cortical or poorly defined cells account for only 14% of all cells, demonstrating that the majority of hCO cell types were consistent with cortical identity.

To assess the fidelity of hCO cell types to the fetal brain, we correlated cell-type-specific hCO gene expression to primary cortical tissue ranging from Carnegie stage 13 to gestation week 22 (Bhaduri et al., 2020) (Figure 1E). hCO neurons and progenitors generally correlated with their in vivo counterparts, whereas cell types absent in the hCOs, like microglia, showed limited correlation. Dividing neural progenitor cells (NPCs) showed the highest correlations (r = 0.72), and non-dividing cortical progenitors modestly correlated to primary progenitors (r = 0.31 to 0.57). The cortical excitatory neuron subtypes identified in hCOs correlated broadly to primary neurons (r = 0.3 to 0.46). These findings are similar to those of Bhaduri et al. (2020) in that in vitro cells represent broad cell classes, but can lack regional or layer-specific identities of the in vivo brain. Overall, scRNA-seq of hCOs demonstrated that the expected cell types of the developing cortical wall were generated with moderate to high fidelity.

Reproducible cell type proportions in hCOs across sites

Next, we calculated cell type proportions at day 14 and day 84 from each differentiation replicate to test for cross-site differences using Propeller (Phipson et al., 2022). Day 14 hCOs were mostly progenitor cell types (>70%), particularly NSCs. No significant differences in cell type proportions were detected at day 14 after correction for multiple comparisons (Table S2 and Figure 2A). However, the pan-neuron/Cajal-Retzius proportions were nominally different (p value = 0.018, false discovery rate [FDR]-corrected p value = 0.27). At day 84, three cell type proportions were nominally different across sites, but did not survive multiple comparison correction: RG (p value = 0.0046, adjusted p value = 0.061), oRG (p value = 0.012, adjusted p value = 0.063), and unspecified neurons (p value = 0.0081, adjusted p value = 0.061) (Table S2 and Figure 2B). Overall, these data suggest high reproducibility of cell type proportions for individual hCOs across differentiation replicates and sites when selecting for visually similar hCOs (Figure S2).

Figure 2.

Figure 2

Open in a new tab

Cell type proportions across sites in hCO and primary tissue samples

Cell type proportions across the 3 sites at (A) day 14 (N = 12 differentiation replicates) and (B) day 84 (N = 13 differentiation replicates). Differentiation replicates are individual hCOs from each site. Site ANOVA FDR corrected p values are indicated in the heatmap.

(C) Cell type proportions from 3 separate day 56 hCOs harvested from the same differentiation replicate at UNC.

(D) Cell type proportions reported by (Polioudakis et al., 2019) from 5 primary fetal tissue samples at similar developmental stages. End, endothelial; ExDp1, excitatory deep layer 1; ExDp2, excitatory deep layer 1; ExM, maturing excitatory; ExM-U, maturing excitatory upper enriched; ExN, migrating excitatory; InCGE, interneuron caudal ganglionic eminence; InMGE, interneuron medial ganglionic eminence; IP, intermediate progenitors; Mic, microglia; OPC, oligodendrocyte progenitor cell; oRG, outer radial glia; Per, pericyte; PgG2M, cycling progenitors (G2/M phase); PgS, cycling progenitors (S phase); vRG, ventricular radial glia.

(E) Cell type proportions reported by Bhaduri et al. (2020). Each sample is a unique primary tissue donor ordered from earlier to later developmental stages. CS, Carnegie stage; GW,gestation week.

In order to quantify cell type variability across hCOs within a single differentiation replicate, three hCOs from the same differentiation replicate at UNC at day 56 were collected (Figure 2C). The coefficient of variation of cell type proportions from multiple hCOs in the same differentiation replicate (ranging from 0.78 for choroid plexus/marginal zone (ChP/MZ) to 0.018 for oRG) was comparable to cell type proportions across differentiation replicates at day 84 (ranging from 0.58 for ChP/MZ to 0.096 for lower layer neuron, a) (Table S2).

Flow cytometry at day 35 yielded similar measurements of populations of expected telencephalic progenitors (SOX2+/PAX6+, FOXG1+, TBR2+), non-cortical progenitors (GSX2+), and newborn cortical neurons (TBR1+) to the cell type proportions found in the scRNA-seq data (Figure S3A). No significant cross-site differences were detected, and the cultures had limited neurogenesis within the first month of differentiation.

Finally, we contextualized the reproducibility of cell type proportions in our dataset compared to other reproducible differentiation protocols (Figures S3B–S3F) (Velasco et al., 2019) and primary tissue collected at similar periods of development. Intraclass correlation coefficients (ICCs) of cell type proportions within and across sites in our study had a similar range of moderate to excellent values as compared to those across cell lines from previous studies (Figures S3B and S3C). ICC values were not significantly different when comparing within site differentiation replicates at day 14 to the previously published study (Velasco et al., 2019). However, increased average ICC values in Velasco suggest higher reproducibility of cell type proportion in their differentiation protocol than this study, though these values could be inflated by using multiple hCOs from the same differentiation replicate (Figure S3E). There was also a larger range of ICC values in Velasco indicating greater variability in cell type proportions from individual differentiations (Figures 2D and S3D). Primary fetal cortex collected at similar time points has uniformly excellent ICC scores (Figure S3E) (Polioudakis et al., 2019).

Differential expressed genes across sites

We performed differential expression on pseudo-bulked scRNA-seq cell types at day 14 and day 84. We detected 786 unique genes with significant differential expression across sites at day 14 (Figure 3A, FDR <0.05). NSCs, the most abundant cell type, showed the most differentially expressed genes (DEGs). DEGs within abundant cell types like NSCs and NPCs were enriched in gene ontology (GO) terms implicating tissue development, suggesting small but detectable differences in early differentiation processes across sites (Figure 3C). For example, UNC had a higher expression of RSPO2, which can bind to LGR4/5 to stabilize WNT signaling (Carmon et al., 2011) (Figure 3E). UNC also had higher expression of CDH18, a type 2 cadherin whose expression peaks in early embryonic development across several brain regions in humans (Li et al., 2020). Slight changes in WNT activation or sensitivity and cadherins expressed in embryonic development across sites may have led to the enrichment of these neurodevelopmental GO terms within DEGs (Harrison-Uy and Pleasure, 2012).

Figure 3.

Figure 3

Open in a new tab

Differential gene expression across sites

Counts of DEGs across sites in pseudo-bulked cell types at (A) day 14 (N = 12 differentiation replicates) and (B) day 84 (N = 13 differentiation replicates). Colors indicate the post hoc pairwise comparison with significant differences if the gene was differentially expressed across all three sites (6 genes total), differentially expressed in one site relative to the other two (“site v both”) or if the gene expression varied between two sites only. Significant GO terms for biological processes associated with DEGs in cell types at (C) day 14 and (D) day 84.

(E) Gene expression differences in RSPO2 in day 14 neural progenitor cells in G2 phase and gene expression of CDH18 in day 14 NSC.

(F) Gene expression differences in HSPA5 and 7SK in day 84 lower layer neuron, group b. Each point is the gene expression from one hCO. Each site collected 3–5 hCOs. Differences in gene expression across site were determined by likelihood ratio tests (LRTs). Abbreviations are as follows: reg, regulation and neg, negative.

By day 84, DEGs across sites were present in all cell types and the total number of DEGs increased to 2,188 unique genes across cell types (Figure 3B). DEGs across sites were enriched in biosynthetic and metabolic processes, and some types of cellular stress, potentially due to differences in growth, necrotic cores, and bursting seen across sites (Figure 3D, see also Figures 4H and 5A). Examples of DEGs include HSPA5 and 7SK, which had lower expression in CHOP upper layer neurons (Figure 3F). HSPA5 mitigates apoptosis in response to oxygen deprivation in cultured neurons (Wang et al., 2015). 7SK is a broadly expressed single-nucleus RNA that inhibits transcription and regulates the response to cellular stressors such as DNA damage and inhibition of RNA synthesis (Studniarek et al., 2021; Bugai et al., 2019), suggesting both of these genes promote survival in response to stressors. CHOP may have lower expression of these cell-stress-related genes due to the loss of the necrotic core after bursting, creating higher media diffusion in smaller hCOs (Figures 4G–4I).

Figure 4.

Figure 4

Open in a new tab

Qualitative assessments of hCO differentiations

(A) Representative images of ranking embedded hCOs at day 14.

(B) Percent hCOs with visible budding at day 14.

(C) Percent hCOs with growth into the Matrigel at day 14.

(D) Representative images of ranking hCOs in the bioreactor at day 35 and day 56.

(E) Percent hCOs with expected morphology at day 35.

(F) Percent hCOs with expected morphology at day 56. Significance of cross-site differences was evaluated with an ANOVA implemented in a linear mixed effect logistic regression model controlling for the non-independence of multiple hCOs within the same differentiation batch using a random effect, and controlling for the ranker who evaluated the images with a fixed effect. Replicates contain between 56 and 120 hCOs.

(G) Example images of burst hCOs from CHOP at day 84.

(H) Percent of wells with burst hCOs across sites for all differentiation replicates.

(I) Count of wells with burst hCOs by motor failure at CHOP across all differentiation replicates.

(J) hCO day 56 cross-sectional area and motor failure in CHOP burst hCOs. Each point is the average area of hCOs in the same well (3–9 hCOs). Significance of motor failure and cross-sectional day 56 hCO area was evaluated in a linear mixed effect logistic regression model controlling for the other factor.

Figure 5.

Figure 5

Open in a new tab

hCO structure across sites

(A) Average cross-sectional area and growth of hCOs across each site. Cross-sectional area (top) or growth rate (bottom) shows significant differences across sites. Each point represents the average area of all hCOs from the differentiation replicate (39–120 hCO). Each site performed 3–6 differentiations.

(B) Example images of tissue cleared and immunolabeled with PAX6 and NCAD day 14 hCOs from each site.

(C) Total volume of day 14 hCOs across sites. Quantification of neuroepithelial bud structure at day 14 as (D) percent PAX6+ volume per hCO and (E) total NCAD+ area normalized to PAX6+ volume and TOPRO+ volume. Each point represents a single hCO. Each site imaged 9–10 hCOs. Examples images of (F) largest hCOs and (G) smallest hCOs from each site at day 84. All scale bars are 500 μm.

(H) Total volume of largest and smallest hCOs across sites.

(I) Representative 3D image of PAX6 and CTIP2 with annotated ventricular zone-like volumes (VZ-like) and annotated cortical plate-like volumes (CP).

(J) Quantification of volume overlapped between VZ-like and CP-like annotations. Each point represents a single hCO. Each site acquired 3–6 hCOs. All scale bars are 200 μm.

Several previous studies have reported cell stress in hCOs with differing interpretations on how impactful the consequences are for modeling cortical development (Bhaduri et al., 2020; Uzquiano et al., 2022). We used GRUFFI to identify stressed cells independent of cluster annotations based on expression of endoplasmic reticulum (ER) stress and glycolysis-associated genes, which are different GO terms than those enriched in the DEG with only 10.9% overlapping genes across the stress-related GO terms (Vértesy et al., 2022). Stressed cells were identified in RG (11.65%) and unspecified neurons (7.06%), and no significant cross-site differences were detected (Figure S4). This finding is consistent with previous observations of increased ER stress and glycolysis (Uzquiano et al., 2022), which may be common across protocols and is not driven by site-specific handling.

Overall, we identified fewer DEGs at day 14, which could be due to greater power to detect gene expression differences at day 84 when more cells were collected for scRNA-seq or an amplification of culture-induced technical variation through time. All cell types at day 84 had DEGs, with GO terms implicating cross-site differences in metabolism, biosynthetic processes, and cell stress distinct from glycolysis and ER stress.

Assaying stem cell state in iPSCs used for hCO differentiation

Rigorous quality controls are recommended for any iPSC differentiation, including validation of undifferentiated cell state (Engle et al. 2018; ISSCR 2023). To assess stem cell state, gene and protein expression were measured using qPCR and flow cytometry in the same iPSC cultures that went into downstream differentiations (Figure S5). Across sites, all iPSC cultures were in an undifferentiated state marked by the expression of OCT4, NANOG, and SSEA3+/4+. CHOP had higher expression than either other site for NANOG, but this did not survive correction for multiple comparisons. Though UNC showed increased variability in SSEA3+/SSEA4+ populations, leading to nominally significant cross-site differences, all sites had greater than 85% double-positive expression of this marker (Figure S5).

We also evaluated markers of naive or primed pluripotent stem cells (Messmer et al., 2019) (Figure S5). The iPSCs exist along a continuum of transcriptomic states that resemble naive or primed stem cells (Messmer et al., 2019). Naive stem cells are hypothesized to be more labile and capable of differentiating to any cell type, whereas primed stem cells are thought to already be directed toward a differentiation fate (Weinberger et al., 2016). We did not observe cross-site differences in naive or primed markers, other than a nominal difference in OTX2 (ANOVA adjusted p value = 0.069). Overall, we detected limited cross-site differences in stem cell gene expression by flow cytometry or qPCR.

Qualitative ranking of hCOs varies by site

hCOs were selected for scRNA-seq using explicit qualitative criteria (Figure S2), so are not representative of all hCOs within the differentiation replicate. Qualitative assessment of hCO morphology using phase contrast microscopy is common in hCO protocols as a way of quickly and cheaply determining the success of hCO differentiation (Qian et al., 2018; Lancaster et al., 2017). Neuroepithelial (NE) buds, phase-bright circular structures in the image, model the developing cortical wall and indicate successful telencephalic differentiation (Figure 4A). Growth into the Matrigel can indicate off-target differentiation or neurite outgrowth and is not expected (Lancaster et al., 2013; Qian et al., 2018). To determine the reproducibility of visual hCO quality, all hCOs per differentiation replicate (80–100 hCOs) at day 14 were imaged and classified (Figures 4B and 4C and Table S3). We did not observe significant differences across sites for visible budding. However, we did observe cross-site differences for growth into Matrigel (p value = 0.014).

At later differentiation time points when hCOs are in the bioreactor, we expect smooth edges with visible budding, but some hCOs also have unexpected transparent cysts or rough edges (Qian et al., 2018). We visually classified all hCOs per differentiation replicate (60–75 hCOs) as having expected (smooth edges) or unexpected (cysts) morphology at day 35 and 56 (Figures 4D–4F and Table S3). Cross-site differences in hCO morphology were detected at day 35 (p value = 0.0047). However, by day 56, significant differences were no longer detected, suggesting unexpected morphology at day 35 could be transient. The majority of hCOs with unexpected morphology at day 56 had regions visually similar to choroid plexus (Pellegrini et al., 2020), which could be due to increased medialization caused by WNT stimulation in the protocol. We identified cross-site differences in morphology, as well as individual replicates having inconsistent morphology across hCOs. Though qualitative assessment using phase contrast microscopy is widely used for monitoring differentiations, these inconsistencies suggest a need for additional quantitative measurements of hCO quality throughout the differentiation, such as flow cytometry or immunolabeling.

Unexpectedly, after day 56, some hCOs burst and expelled their necrotic cores, leading to dead cells in the well, flattened sheets of NE buds, and hollowed cup-like hCOs that can continue to grow (Figure 4G). CHOP had more burst hCOs than the other sites (Figure 4H). Cross-sectional area for CHOP hCOs at day 56 was associated with an increased incidence of bursting when statistically controlling for motor failure (where the motor slowed and stopped spinning for an estimated 14–18 h) and vice versa (Figures 4I and 4J). Bursting first occurred within a few days of motor failure. We speculate that CHOP’s hCOs had an increased necrotic core size earlier in the differentiation, due to less diffusion of media through the larger hCOs. Loss of agitation from the motor failure could also exacerbate impaired media diffusion and increase necrotic core size.

NE bud structure in hCOs

We measured the hCO cross-sectional average area and growth rates using phase contrast microscopy and found that these measurements of structure varied by site at every time point (Figures 5A and S2). Additionally, we evaluated intact hCO structure and volume using tissue clearing (Renier et al., 2016) and fluorescent microscopy in 3 hCOs per differentiation replicate per time point. All sites generated hCOs with NE buds, labeled with NCAD+ lumens and PAX6+ ventricular zone-like (VZ-like) regions at day 14, modeling the developing cortical wall (Figure 5B and Video S1). Size differences across sites were also identified in hCO volume at day 14 (FDR-corrected p value = 0.0032, Figure 5C). When normalizing measures of structure for overall hCO volume, we did not detect cross-site differences in measures of NE bud size (Figure 5D). Lumen expansion did not significantly differ across sites when correcting for multiple comparisons, either when the NCAD+ area was normalized relative to PAX6+ volumes, representing cortical VZ-like regions only, or when the NCAD+ area was normalized to full hCO volumes (Figure 5E), representing buds with either PAX6+ RG or PAX6− regions that are likely NSCs.

Video S1. Example annotation of immunolabeling whole organoids in IMARIS at day 14

ToPRO3+ nuclei (blue), PAX6+ radial glia (magenta), and NCAD+ tight junctions with neuroepithelial buds (green) segmented in IMARIS and classified by hCO (purple, orange, yellow).

Download video file (1.9MB, mp4)

By day 84, NE buds had developed into cortical wall-like structures with dense PAX6+ VZ-like regions and dense CTIP2+ cortical plate-like (CP-like) regions (Figures 5F, 5G, S6, and Video S2). In addition to some burst hCOs, all sites had a range of hCO sizes. The largest and smallest hCO, which are potentially burst hCOs, were imaged for each differentiation replicate to determine if hCO architecture varied by hCO size. There were significant differences in size between the large and small hCOs for every site, as expected (Student’s t test, p value <0.004 for each site). No cross-site differences in total volumes were detected in either large or small hCOs, potentially because the largest hCOs from other sites had burst prior to collection (Figures 5H and 4G–4J). Dense VZ-like regions and CP-like regions were semi-automatically annotated to measure their overlapping volumes, where we expect VZ-like regions to be well separated from CP-like regions in high-quality differentiations (Figure 5I). All hCOs had less than 40% overlapping volumes, and no significant cross-site differences were detected in small and large hCOs (Figure 5J). Using 2D manual annotation of CP-like and VZ-like regions in optical sections from the same hCOs, we did not detect differences across sites in hCO structure. Subventricular zone-like regions were trending larger in UNC, though this did not survive multiple testing corrections (one-way ANOVA FDR-corrected p value = 0.052) (Figure S6). Our data suggest that reproducible measures of organoid structure could be obtained after controlling for overall organoid size.

Video S2. Example annotation of immunolabeling whole organoids in IMARIS at day 84

ToPRO3+ nuclei (green), CTIP2+ deep layer neurons (blue), PAX6+ radial glia (magenta) segmented in IMARIS.

Download video file (12MB, mp4)

Primed markers prior to differentiation predict cell type proportions

In order to find factors that led to cross-site differences or that are predictive of future hCO quality, we tested whether technical variables or assays at earlier time points correlated to cell type proportions measured in scRNA-seq. In day 84 hCOs, passage at seed (r = 0.82, adjusted p value = 0.021) and the percent of hCOs with visible buds at day 14 (r = −0.95, adjusted p value = 0.00012) both correlated to pan-cortical neuron proportions at day 84 (Figure 6A). The effect was driven by two outlier UNC differentiation replicates (Figures S7A–S7D), so it is not possible to disambiguate if handling differences across site or these specific factors resulted in the differences in cell type proportions. The proportion of NPCs in S phase correlated with SSEA3+/SSEA4+ cells and cross-sectional area and growth rate at day 35 (Figures 6A and S7E–S7G). Increased numbers of proliferating progenitors correlating to increased size and growth rate later is predicted by the radial unit hypothesis (Rakic 2009).

Figure 6.

Figure 6

Open in a new tab

Correlations to cell type proportions

Pearson’s correlations of cell type proportions in scRNA-seq data to assay results or technical variables at (A) day 14 or (B) day 84. Nominally significant test are labeled with ∗ and test surviving FDR correction <0.05 are labeled # (N = 7–12 differentiation replicates).

(C) Correlation between primed marker ZIC2 and lower layer neuron, b proportions at day 14 (N = 11 differentiation replicates).

(D) Correlation between primed marker DUSP6 and NPC, G2 phase in day 14 hCOs (N = 11 differentiation replicates).

Similarly, at day 14, the proportion of IPs in S phase was also correlated with the TBR1+ population measured by flow cytometry at day 35. Though the proportion of both cell populations is small (1.57% IPs in S phase, 4.0% TBR1+ populations), it is expected that proliferating IPs will lead to an increase in the total number of neurons (Rakic 2009). Identification of an expected relationship between progenitor and neuronal cell types across assays increases the confidence in our analysis. Two primed stem cell state markers, which are highly correlated to each other (r = 0.75), both predicted cell type proportions at day 14. ZIC2 expression is positively correlated with lower layer neuron, b (r = 0.91, adjusted p value = 0.00012). DUSP6 expression correlated to increased proportions of NPCs in G2 phase (r = 0.89, adjusted p value = 0.00024; Figure 6B). Importantly, no individual site is an outlier driving these results (Figures 6C and 6D). This suggests that the stem cell state can influence downstream differentiation potential, at least at day 14.

Early differentiation technical variables correlate to DEGs

To identify factors that influence differential gene expression between the sites, we correlated DEGs across sites for each cluster and time point to technical factors and other hCO assays. We identified 17,884 correlations between 2,977 genes and 59 technical or biological variables in our dataset that were significant following correction for multiple comparisons (FDR <0.05; Table S4). Technical factors involving seeding the iPSCs prior to differentiation into hCOs were associated with a larger proportion of DEGs across sites in day 84 cell types (Figure 7). These results also suggest that iPSC quality and treatment can influence downstream hCO differentiations. We show several representative examples of technical variables prior to differentiation correlating to DEGs involved in cell stress and early developmental processes later in differentiation (Figures 7C–7F, 3E, and 3F). CDH18 expression in NSCs at day 14 correlates with the proportion of GSX2+ cells from flow cytometry at day 35, suggesting that early cadherin expression may relate to future cell type fate decisions (Figure 7F), a conclusion also supported by normalized NCAD area in 3D images showing nominal correlations with cell type proportions (Figures 6A and 6B) and DEGs (Figure 7B). Cross-sectional area at day 14 and growth from day 7–14 and day 35–56 also correlated with DEGs at day 84, supporting a potential link between hCO size and day 84 gene expression (Figure 7A). Overall, technical factors in iPSC handling at the beginning of the differentiation and hCO size measurements explain some cross-site variability in gene expression and may be important factors to standardize in future hCO studies.

Figure 7.

Figure 7

Open in a new tab

Correlations of technical and biological variables to differentially expressed genes at day 14 and day 84

Proportion of DEGs across site for each cell type which correlate to technical variables or other assays at (A) day 84 and (B) day 14 (FDR <0.05).

(C) Correlation between 7SK expression in day 84 lower layer neuron, b and iPSC viability at day −1 (N = 12 differentiation replicates).

(D) Correlation between HSPA5 expression in day 84 lower layer neuron, b and the time iPSC were in accutase during their aggregation into 3D cultures at day −1 (N = 12 differentiation replicates).

(E) Correlation between RSPO2 expression in day 14 neural progenitors in G2 phase and the iPSC passage number at day −1 (N = 11 differentiation replicates).

(F) Correlation between CDH18 expression in day 14 NSCs and the percent GSX2+ cells detected by fluorescence-activated cell sorting (FACS) at day 35 (N = 6 differentiation replicates).

Discussion

Our study aimed to identify hCO phenotypes that are reproducible across sites using a harmonized differentiation protocol. Like the original protocol, hCOs generated from PGP1 produced similar proportions of mostly cortical cell types (Figures 1 and 2). All sites generated hCOs with NE buds mimicking the cortical wall (Figure 5). We detected DEG across sites with cell stress and metabolic genes differing as hCOs grew in size (Figures 3 and 7). Non-reproducible measures across sites included hCO size, hCO morphology, and bursting (Figures 4 and 5). Additionally, variations in primed stem cell markers and handling prior to differentiation correlated with cortical cell type proportions in day 14 hCOs and DEGs, suggesting stem cell state can influence hCO outcomes (Figures 6A, 7A, and 7B). Overall, our data support the reproducibility of cell type proportions and structure between the sites and across replicates, suggesting prospective cross-site meta-analysis could use these measures. For individual labs, we identified technical factors correlated to variation in cell type proportions and gene expression that could be mitigated or statistically controlled to increase reproducibility.

Reproducible cell type proportions and different transcriptomes

Irreproducibility of cell type composition was cited as a major barrier to using hCOs in high-throughput drug screening and modeling neurodevelopmental disorders (Huch et al., 2017). Unguided hCO protocols have produced an inconsistent variety of non-cortical cell types and cell type composition across hCOs (Lancaster et al., 2013). In contrast, guided protocols show more consistent cell type composition within a given lab and across multiple iPSC lines with minimal non-cortical cells (Velasco et al., 2019; Yoon et al., 2019). We extend these findings to show that despite cross-site differences in handling, cell type composition in a guided differentiation protocol is consistent and mainly cortical. We also recapitulated known limitations in the fidelity of hCO models. All sites had reproducible non-cortical differentiation, glycolytic and ER stressed cells, and broad neuronal cell types that do not closely resemble a particular cortical layer or region (Bhaduri et al., 2020; Kelley and Pașca 2022). These known limitations could be controlled for in future experimental designs by bioinformatically filtering out stressed cells or cells that do not model the brain region of interest (Vértesy et al., 2022).

DEGs implicated metabolic and cellular stress differences across sites in several cell types. Given that we observed size differences in hCOs across sites, and increased cross-sectional area was associated with bursting hCOs at one site, differential expression of stress, biosynthetic, and metabolism genes could be due to the lack of media diffusion through the hCO and large interior necrotic cores. Using a guided hCO protocol that minimizes necrotic cores or protocols that incorporate vascularization into hCO generation could reduce the possibility of these metabolic-related cross-site differences.

Cortical wall-like organization across time and size

We observed consistent NE budding at day 14 leading to cortical wall-like organization at day 84 hCOs, regardless of size. This result demonstrates that a key feature of hCO culture is reproducible across sites. Cortical wall-like organization increases cell-to-cell contact in hCOs, promoting oRG and IP production (Scuderi et al., 2021), and enables the study of defects in neuronal migration. Though not significantly different across sites, some hCO structural phenotypes were still variable within sites possibly due to technical factors like variability in antibody penetration or agarose embedding distorting the light path in addition to biological influences. Tissue clearing and analysis are not yet standardized in hCO studies, though progress is being made in developing analysis tools for hCO structures in 3D that could increase reproducibility in the future (Benito-Kwiecinski et al., 2021; Albanese et al., 2020).

hCO size varies across site and within each differentiation. Size was not previously addressed in single-site studies of hCO reproducibility (Velasco et al., 2019; Yoon et al., 2019). However, differences in hCO size in case-control comparisons are consistently documented when large effects are expected, such as when studying Zika virus (Watanabe et al., 2017; Qian et al., 2016), monogenic causes of macrocephaly (Zhang et al., 2020) and microcephaly (Lancaster et al., 2013). Though large effects may overcome the noise in hCO size measurements at a single site, our study suggests that hCO size is not suitable for cross-site meta-analysis. Instead, hCO size and morphology are important technical factors to report and statistically control.

iPSC cell state influences differentiation

A previous study found that an intermediate pluripotency state between naive and primed led to higher quality hCOs with fewer non-cortical cell types and more robust cortical wall-like organization (Watanabe et al., 2022). We found that even with culturing the same iPSC line under the same conditions, increased expression of primed marker ZIC2 prior to differentiation led to a greater proportion of lower layer neuron production, and increased DUSP6 expression led to increased proliferating progenitors early in the differentiation. ZIC2 is known to regulate gastrulation and dorsal forebrain specification (Houtmeyers et al., 2013). DUSP6 regulates the MAPK pathway (Ahmad et al., 2018). While not directly evaluating the same phenotypes, these results suggest that stem cell state during pluripotency may impact hCO phenotype after differentiation (Watanabe et al., 2022). We recommend that future studies monitor variation in iPSC cell state before hCO differentiations (ISSCR 2023). Though primed and naive markers are currently not standard for quantitative iPSC quality controls, like the qPCR score card (Fergus et al. 2016), gaining an understanding of how stem cell state influences multiple hCO protocols could lead to better standardization of iPSC quality and more reliable hCO phenotypes.

Limitations

Our study evaluated cross-site hCO reproducibility across many phenotypes using one cell line (PGP1) and one differentiation protocol. As such, we are not able to assess variability driven by differences in cell lines or protocol. Line-to-line variability may also influence differentiation outcomes, caused by either genetic differences or technical variation in iPSC generation, as has been previously demonstrated (Kilpinen et al., 2017). The protocol used in this study may not be representative of all hCO protocols. For example, other protocols do not embed hCOs in Matrigel (Yoon et al., 2019; Velasco et al., 2019). The time needed to embed hCOs within a Matrigel cookie was associated with gene expression differences (Figures 7A and 7B) and nominally associated with cross-site differences in RG and oRG at day 14, as well as NPC, S phase, and lower layer neuron, b cell type proportions at day 84 (Figure 6B), suggesting that this step may contribute to variability in differentiation outcomes. Further, protocols without a necrotic core or WNT activation may not see differences in genes related to cell stress and WNT, respectively. We are unable to validate if iPSC gene expression would correlate to hCO outcomes in other protocols or cell lines. Finally, though we measured multiple phenotypes from multiple differentiation replicates, cross-site differences may exist that we were underpowered to detect for all measurements used here. To demonstrate hCO reproducibility in future studies, we recommend using the same iPSC line cultured across sites with the same protocol as a “rosetta” line (Volpato and Webber 2020).

Conclusions

hCO studies of neuropsychiatric disorders are in their infancy and may expect reproducibility problems unless proactively addressed. Other fields increased reproducibility by (1) demonstrating reproducibility of measurements, (2) standardizing data acquisition and sharing, and (3) combining data from multiple sites to increase statistical power (Medland et al., 2022; Sullivan et al., 2018). This study demonstrates that hCOs can generate reproducible cell types and organization when produced across multiple sites, suggesting that comparisons of changes in cell type proportions or hCO cortical wall-like organizations can be properly controlled, enabling future prospective meta-analytic iPSC-derived hCO studies in neuropsychiatric disorders.

Experimental procedures

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Jason Stein (jason_stein@med.unc.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

Scripts used in analysis can be found at https://github.com/mroseglass/hCO-CrossSiteReproducibility.git. The Seurat object can be found at UCSC Cell Browser: https://hco-cross-site-repro.cells.ucsc.edu.

Protocol

iPSC culture is described in the supplemental methods section.

The hCO differentiation protocol was adapted from Qian et al. (2018) to feeder-free conditions. Briefly, the major changes included culturing the iPSCs on hESC-qualified Matrigel in mTeSR+, generating embryoid bodies from a single-cell suspension, and doubling the concentrations of Forebrain First media small molecules.

qPCR for iPSC cell state

300,000 iPSCs were harvested on day −1 from extra iPSCs not used for seeding the hCOs and sent to one site for analysis.

Phase contrast imaging and analysis

At days 0, 7, 14, 35, 56, 70, and 84, phase contrast images were acquired of the hCOs on an EVOS XL Core microscope using the 4× objective at the UNC and CHOP sites and a LabCam lens at CN.

Tissue clearing and imaging

hCOs for imaging were fixed in 4% paraformaldehyde (PFA) for 30 min at CN and UNC, and for 1 h at CHOP. Day 14 and day 84 hCOs were tissue cleared using an adapted iDISCO+ protocol (Renier et al., 2014). See supplemental methods section for details.

Flow cytometry

Flow cytometry samples were harvested by pre-treatment in collagenase (1 mg/mL) + DNase (0.01 mg/mL) at 37°C (days 14 and 35 only), followed by trypsin for 5 min, then fixation with 1.6% PFA for 30 min (days 7, 14, and 35). Fixed singularized cells were processed for flow cytometry at CHOP (see supplemental methods).

scRNA-seq library preparation

A single hCO was selected for each scRNA-seq for each differentiation replicate. At day 14, hCOs were selected for scRNA-seq that had phase-bright budding structures and no residual Matrigel. At day 84, a phase contrast image was taken of each hCO harvest for scRNA-seq, which was selected to have smooth edges and no visible cysts. Frozen aliquots of fixed single cells were stored at −80°C and shipped on dry ice overnight to UNC from CHOP and CN. See supplemental methods section for details.

Acknowledgments

We thank the Intellectual and Developmental Disabilities Research Center organoid working group for support and insightful discussions. This work was supported by the NIH (3-P50-HD103573-03S1, R01 MH130441, R01MH121433, R01 MH12012, and R01MH118349), the Foundation of Hope, UNC TraCS (550KR262111), the UNC School of Medicine to J.L.S., R01AA025215 and R01AA026272 to K.H.-T., F31MH124427-01A1 to M.R.G., and Intellectual and Developmental Disabilities Research Center NIH/NICHD P50 HD105354 (E.A.W. and D.L.F.). The University of North Carolina High Throughput Sequencing Facility is supported by the University Cancer Research Fund, Comprehensive Cancer Center Core Support grant (P30-CA016086), and UNC Center for Mental Health and Susceptibility grant (P30-ES010126). We thank Pablo Ariel at the Microscopy Service Laboratory and Michelle Itano at the Neuroscience Microscopy Core for assistance with imaging. The Microscopy Services Laboratory is supported in part by P30 CA016086 Cancer Center Core Support Grant to the UNC Lineberger Comprehensive Cancer Center and the North Carolina Biotech Center Institutional Support Grant 2016-IDG-1016. The University of North Carolina’s Neuroscience Microscopy Core (RRID: SCR_019060) is supported, in part, by funding from the NIH-NINDS Neuroscience Center Support Grant P30 NS045892 and the NIH-NICHD Intellectual and Developmental Disabilities Research Center Support Grant P50 HD103573.

Author contributions

J.L.S., D.L.F., K.H.-T., S.Y., E.A.W., and M.R.G. conceived the experiments. M.R.G., E.A.W., and S.Y. generated all replicate organoid differentiations in this study. A.A.B. assisted in feeding and imaging the organoids. N.K.P. generated day 56 organoids. M.R.G. performed scRNA-seq library preparation. M.R.G. and M.L. analyzed scRNA-seq data with N.M.’s assistance. E.A.W. performed flow cytometry experiments and analysis. S.Y. performed qPCR experiments and analysis. R.S. performed immunofluorescence and imaging of sections. M.R.G. performed staining and imaging and analysis of 3D images. S.A., M.S., L.T.D., M.Y., and K.S. visually classified the organoids. T.F. supervised organoid ranking and area measurements, and annotated 3D organoid images. M.I.L. provided advice on statistical analysis. K.H.-T., D.L.F., and J.L.S. supervised the work. M.R.G., E.A.W., S.Y., K.H.-T., D.L.F., and J.L.S. wrote the manuscript. All authors read and approved the final manuscript.

Declaration of interests

The authors declare no competing interests.

Published: August 22, 2024

Footnotes

Supplemental information can be found online at https://doi.org/10.1016/j.stemcr.2024.07.008.

Supplemental information

Document S1. Figures S1–S7, Tables S1, S3, and S5
mmc1.pdf (9.2MB, pdf)
Table S2. Cell type proportions across site and coefficients of variation at day 14, 56, and 84. Baseline proportion is the geometric mean across all samples
mmc2.xlsx (14.3KB, xlsx)
Table S4. All differentially expressed gene correlates
mmc3.xlsx (780KB, xlsx)
Document S2. Article plus supplemental information
mmc6.pdf (16.9MB, pdf)

References

  1. Ahmad M.K., Abdollah N.A., Shafie N.H., Yusof N.M., Razak S.R.A. Dual-Specificity Phosphatase 6 (DUSP6): A Review of Its Molecular Characteristics and Clinical Relevance in Cancer. Cancer Biol. Med. 2018;15:14–28. doi: 10.20892/j.issn.2095-3941.2017.0107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Albanese A., Swaney J.M., Yun D.H., Evans N.B., Antonucci J.M., Velasco S., Sohn C.H., Arlotta P., Gehrke L., Chung K. Multiscale 3D Phenotyping of Human Cerebral Organoids. Sci. Rep. 2020;10 doi: 10.1038/s41598-020-78130-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Benito-Kwiecinski S., Giandomenico S.L., Sutcliffe M., Riis E.S., Freire-Pritchett P., Kelava I., Wunderlich S., Martin U., Wray G.A., McDole K., Lancaster M.A. An Early Cell Shape Transition Drives Evolutionary Expansion of the Human Forebrain. Cell. 2021;184:2084–2102.e19. doi: 10.1016/j.cell.2021.02.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bhaduri A., Andrews M.G., Mancia Leon W., Jung D., Shin D., Allen D., Jung D., Schmunk G., Haeussler M., Salma J., et al. Cell Stress in Cortical Organoids Impairs Molecular Subtype Specification. Nature. 2020;578:142–148. doi: 10.1038/s41586-020-1962-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bugai A., Quaresma A.J.C., Friedel C.C., Lenasi T., Düster R., Sibley C.R., Fujinaga K., Kukanja P., Hennig T., Blasius M., et al. P-TEFb Activation by RBM7 Shapes a Pro-Survival Transcriptional Response to Genotoxic Stress. Mol. Cell. 2019;74:254–267.e10. doi: 10.1016/j.molcel.2019.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Carmon K.S., Gong X., Lin Q., Thomas A., Liu Q. R-Spondins Function as Ligands of the Orphan Receptors LGR4 and LGR5 to Regulate Wnt/β-Catenin Signaling. Proc. Natl. Acad. Sci. USA. 2011;108:11452–11457. doi: 10.1073/pnas.1106083108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Engle S.J., Blaha L., Kleiman R.J. Best Practices for Translational Disease Modeling Using Human iPSC-Derived Neurons. Neuron. 2018;100:783–797. doi: 10.1016/j.neuron.2018.10.033. [DOI] [PubMed] [Google Scholar]
  8. Fergus J., Quintanilla R., Lakshmipathy U. Characterizing Pluripotent Stem Cells Using the TaqMan® hPSC Scorecard(TM) Panel. Methods Mol. Biol. 2016;1307:25–37. doi: 10.1007/7651_2014_109. [DOI] [PubMed] [Google Scholar]
  9. He C., Kalafut N.C., Sandoval S.O., Risgaard R., Sirois C.L., Yang C., Khullar S., Suzuki M., Huang X., Chang Q., et al. BOMA, a Machine-Learning Framework for Comparative Gene Expression Analysis across Brains and Organoids. Cell Rep. Methods. 2023;3 doi: 10.1016/j.crmeth.2023.100409. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Houtmeyers R., Souopgui J., Tejpar S., Arkell R. The ZIC Gene Family Encodes Multi-Functional Proteins Essential for Patterning and Morphogenesis. Cell. Mol. Life Sci. 2013;70:3791–3811. doi: 10.1007/s00018-013-1285-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Huch M., Knoblich J.A., Lutolf M.P., Martinez-Arias A. The Hope and the Hype of Organoid Research. Development. 2017;144:938–941. doi: 10.1242/dev.150201. [DOI] [PubMed] [Google Scholar]
  12. ISSCR Standards for Human Stem Cell Use in Research. Int. Soc. Stem Cell Res. 2023;1:16–21. https://static1.squarespace.com/static/611faaa8fee682525ee16489/t/647de42a1a18dd7bfb91e68e/1685972011644/ISSCR_Standards_09_FINAL.pdf [Google Scholar]
  13. Kanton S., Boyle M.J., He Z., Santel M., Weigert A., Sanchís-Calleja F., Guijarro P., Sidow L., Fleck J.S., Han D., et al. Organoid Single-Cell Genomic Atlas Uncovers Human-Specific Features of Brain Development. Nature. 2019;574:418–422. doi: 10.1038/s41586-019-1654-9. [DOI] [PubMed] [Google Scholar]
  14. Kelley K.W., Pașca S.P. Human Brain Organogenesis: Toward a Cellular Understanding of Development and Disease. Cell. 2022;185:42–61. doi: 10.1016/j.cell.2021.10.003. [DOI] [PubMed] [Google Scholar]
  15. Lancaster M.A., Corsini N.S., Wolfinger S., Gustafson E.H., Phillips A.W., Burkard T.R., Otani T., Livesey F.J., Knoblich J.A. Guided Self-Organization and Cortical Plate Formation in Human Brain Organoids. Nat. Biotechnol. 2017;35:659–666. doi: 10.1038/nbt.3906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lancaster M.A., Renner M., Martin C.-A., Wenzel D., Bicknell L.S., Hurles M.E., Homfray T., Penninger J.M., Jackson A.P., Knoblich J.A. Cerebral Organoids Model Human Brain Development and Microcephaly. Nature. 2013;501:373–379. doi: 10.1038/nature12517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Sullivan P.F., Agrawal A., Bulik C.M., Andreassen O.A., Borglum A.D., Breen G., Cichon S., Edenberg H.J., Faraone S.V., Gelernter J., et al. Psychiatric Genomics: An Update and an Agenda. Am J. Psychiatry. 2018;175:15–27. doi: 10.1176/appi.ajp.2017.17030283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Tanaka Y., Cakir B., Xiang Y., Sullivan G.J., Park I.-H. Synthetic Analyses of Single-Cell Transcriptomes from Multiple Brain Organoids and Fetal Brain. Cell Rep. 2020;30:1682–1689.e3. doi: 10.1016/j.celrep.2020.01.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Li J., Lin X., Wang M., Hu Y., Xue K., Gu S., Lv L., Huang S., Xie W. Potential Role of Genomic Imprinted Genes and Brain Developmental Related Genes in Autism. BMC Med. Genom. 2020;13:54. doi: 10.1186/s12920-020-0693-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Messmer T., von Meyenn F., Savino A., Santos F., Mohammed H., Lun A.T.L., Marioni J.C., Reik W. Transcriptional Heterogeneity in Naive and Primed Human Pluripotent Stem Cells at Single-Cell Resolution. Cell Rep. 2019;26:815–824.e4. doi: 10.1016/j.celrep.2018.12.099. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Pellegrini L., Bonfio C., Chadwick J., Begum F., Skehel M., Lancaster M.A. Human CNS Barrier-Forming Organoids with Cerebrospinal Fluid Production. Science. 2020;369 doi: 10.1126/science.aaz5626. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Phipson B., Sim C.B., Porrello E.R., Hewitt A.W., Powell J., Oshlack A. Propeller: Testing for Differences in Cell Type Proportions in Single Cell Data. Bioinformatics. 2022;38:4720–4726. doi: 10.1093/bioinformatics/btac582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Polioudakis D., de la Torre-Ubieta L., Langerman J., Elkins A.G., Shi X., Stein J.L., Vuong C.K., Nichterwitz S., Gevorgian M., Opland C.K., et al. A Single-Cell Transcriptomic Atlas of Human Neocortical Development during Mid-Gestation. Neuron. 2019;103:785–801.e8. doi: 10.1016/j.neuron.2019.06.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Qian X., Jacob F., Song M.M., Nguyen H.N., Song H., Ming G.-L. Generation of Human Brain Region–specific Organoids Using a Miniaturized Spinning Bioreactor. Nat. Protoc. 2018;13:565–580. doi: 10.1038/nprot.2017.152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Qian X., Nguyen H.N., Song M.M., Hadiono C., Ogden S.C., Hammack C., Yao B., Hamersky G.R., Jacob F., Zhong C., et al. Brain-Region-Specific Organoids Using Mini-Bioreactors for Modeling ZIKV Exposure. Cell. 2016;165:1238–1254. doi: 10.1016/j.cell.2016.04.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Rakic P. Evolution of the Neocortex: Perspective From Developmental Biology. Nat. Rev. Neurosci. 2009;10:724–735. doi: 10.1038/nrn2719. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Renier N., Adams E.L., Kirst C., Wu Z., Azevedo R., Kohl J., Autry A.E., Kadiri L., Umadevi Venkataraju K., Zhou Y., et al. Mapping of Brain Activity by Automated Volume Analysis of Immediate Early Genes. Cell. 2016;165:1789–1802. doi: 10.1016/j.cell.2016.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Rosenberg A.B., Roco C.M., Muscat R.A., Kuchina A., Sample P., Yao Z., Graybuck L.T., Peeler D.J., Mukherjee S., Chen W., et al. Single-Cell Profiling of the Developing Mouse Brain and Spinal Cord with Split-Pool Barcoding. Science. 2018;360:176–182. doi: 10.1126/science.aam8999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Scuderi S., Altobelli G.G., Cimini V., Coppola G., Vaccarino F.M. Cell-to-Cell Adhesion and Neurogenesis in Human Cortical Development: A Study Comparing 2D Monolayers with 3D Organoid Cultures. Stem Cell Rep. 2021;16:264–280. doi: 10.1016/j.stemcr.2020.12.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Studniarek C., Tellier M., Martin P.G.P., Murphy S., Kiss T., Egloff S. The 7SK/P-TEFb snRNP Controls Ultraviolet Radiation-Induced Transcriptional Reprogramming. Cell Rep. 2021;35 doi: 10.1016/j.celrep.2021.108965. [DOI] [PubMed] [Google Scholar]
  31. Uzquiano A., Kedaigle A.J., Pigoni M., Paulsen B., Adiconis X., Kim K., Faits T., Nagaraja S., Antón-Bolaños N., Gerhardinger C., et al. Proper Acquisition of Cell Class Identity in Organoids Allows Definition of Fate Specification Programs of the Human Cerebral Cortex. Cell. 2022;185:3770–3788.e27. doi: 10.1016/j.cell.2022.09.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Velasco S., Kedaigle A.J., Simmons S.K., Nash A., Rocha M., Quadrato G., Paulsen B., Nguyen L., Adiconis X., Regev A., et al. Individual Brain Organoids Reproducibly Form Cell Diversity of the Human Cerebral Cortex. Nature. 2019;570:523–527. doi: 10.1038/s41586-019-1289-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Vértesy Á., Eichmüller O.L., Naas J., Novatchkova M., Esk C., Balmaña M., Ladstaetter S., Bock C., von Haeseler A., Knoblich J.A. Gruffi: An Algorithm for Computational Removal of Stressed Cells from Brain Organoid Transcriptomic Datasets. EMBO J. 2022;41 doi: 10.15252/embj.2022111118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Volpato V., Smith J., Sandor C., Ried J.S., Baud A., Handel A., Newey S.E., Wessely F., Attar M., Whiteley E., et al. Reproducibility of Molecular Phenotypes after Long-Term Differentiation to Human iPSC-Derived Neurons: A Multi-Site Omics Study. Stem Cell Rep. 2018;11:897–911. doi: 10.1016/j.stemcr.2018.08.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Volpato V., Webber C. Addressing Variability in iPSC-Derived Models of Human Disease: Guidelines to Promote Reproducibility. Dis. Model. Mech. 2020;13 doi: 10.1242/dmm.042317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Wang L., Owusu-Hammond C., Sievert D., Gleeson J.G. Stem Cell-Based Organoid Models of Neurodevelopmental Disorders. Biol. Psychiatr. 2023;93:622–631. doi: 10.1016/j.biopsych.2023.01.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Wang P., Zhang N., Liang J., Li J., Han S., Li J. Micro-RNA-30a Regulates Ischemia-Induced Cell Death by Targeting Heat Shock Protein HSPA5 in Primary Cultured Cortical Neurons and Mouse Brain after Stroke. J. Neurosci. Res. 2015;93:1756–1768. doi: 10.1002/jnr.23637. [DOI] [PubMed] [Google Scholar]
  38. Watanabe M., Buth J.E., Haney J.R., Vishlaghi N., Turcios F., Elahi L.S., Gu W., Pearson C.A., Kurdian A., Baliaouri N.V., et al. TGFβ Superfamily Signaling Regulates the State of Human Stem Cell Pluripotency and Capacity to Create Well-Structured Telencephalic Organoids. Stem Cell Rep. 2022;17:2220–2238. doi: 10.1016/j.stemcr.2022.08.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Weinberger L., Ayyash M., Novershtern N., Hanna J.H. Dynamic Stem Cell States: Naive to Primed Pluripotency in Rodents and Humans. Nat. Rev. Mol. Cell Biol. 2016;17:155–169. doi: 10.1038/nrm.2015.28. [DOI] [PubMed] [Google Scholar]
  40. Werner J.M., Gillis J. Preservation of Co-Expression Defines the Primary Tissue Fidelity of Human Neural Organoids. bioRxiv. 2023 doi: 10.1101/2023.03.31.535112. Preprint at. [DOI] [Google Scholar]
  41. Yoon S.-J., Elahi L.S., Pașca A.M., Marton R.M., Gordon A., Revah O., Miura Y., Walczak E.M., Holdgate G.M., Fan H.C., et al. Reliability of Human Cortical Organoid Generation. Nat. Methods. 2019;16:75–78. doi: 10.1038/s41592-018-0255-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Zhang W., Ma L., Yang M., Shao Q., Xu J., Lu Z., Zhao Z., Chen R., Chai Y., Chen J.-F. Cerebral Organoid and Mouse Models Reveal a RAB39b-PI3K-mTOR Pathway-Dependent Dysregulation of Cortical Development Leading to Macrocephaly/autism Phenotypes. Genes Dev. 2020;34:580–597. doi: 10.1101/gad.332494.119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Lee J.-H., Park I.-H., Gao Y., Li J.B., Li Z., Daley G.Q., Zhang K., Church G.M. A Robust Approach to Identifying Tissue-Specific Gene Expression Regulatory Variants Using Personalized Human Induced Pluripotent Stem Cells. PLoS Genetics. 2009;5 doi: 10.1371/journal.pgen.1000718. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Harrison-Uy S.J., Pleasure S.J. Wnt signaling and forebrain development. Cold Spring Harb. Perspect. Biol. 2012;4:a008094. doi: 10.1101/cshperspect.a008094. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Watanabe M., Buth J.E., Vishlaghi N., de la Torre-Ubieta L., Taxidis J., Khakh B.S., Coppola G., Pearson C.A., Yamauchi K., Gong D., et al. Self-Organized Cerebral Organoids with Human-Specific Features Predict Effective Drugs to Combat Zika Virus Infection. Cell Rep. 2017;21:517–532. doi: 10.1016/j.celrep.2017.09.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Kilpinen H., Goncalves A., Leha A., Afzal V., Alasoo K., Ashford S., Bala S., Bensaddek D., Casale F.P., Culley O.J., et al. Common genetic variation drives molecular heterogeneity in human iPSCs. Nature. 2017;546:370–375. doi: 10.1038/nature22403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Medland S.E., Grasby K.L., Jahanshad N., Painter J.N., Colodro-Conde L., Bralten J., Hibar D.P., Lind P.A., Pizzagalli F., Thomopoulos S.I., et al. Ten years of enhancing neuro-imaging genetics through meta-analysis: An overview from the ENIGMA Genetics Working Group. Hum. Brain Mapp. 2022;43:292–299. doi: 10.1002/hbm.25311. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Renier N., Wu Z., Simon D.J., Yang J., Ariel P., Tessier-Lavigne M. iDISCO: A Simple, Rapid Method to Immunolabel Large Tissue Samples for Volume Imaging. Cell. 2014;159:896–910. doi: 10.1016/j.cell.2014.10.010. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Video S1. Example annotation of immunolabeling whole organoids in IMARIS at day 14

ToPRO3+ nuclei (blue), PAX6+ radial glia (magenta), and NCAD+ tight junctions with neuroepithelial buds (green) segmented in IMARIS and classified by hCO (purple, orange, yellow).

Download video file (1.9MB, mp4)
Video S2. Example annotation of immunolabeling whole organoids in IMARIS at day 84

ToPRO3+ nuclei (green), CTIP2+ deep layer neurons (blue), PAX6+ radial glia (magenta) segmented in IMARIS.

Download video file (12MB, mp4)
Document S1. Figures S1–S7, Tables S1, S3, and S5
mmc1.pdf (9.2MB, pdf)
Table S2. Cell type proportions across site and coefficients of variation at day 14, 56, and 84. Baseline proportion is the geometric mean across all samples
mmc2.xlsx (14.3KB, xlsx)
Table S4. All differentially expressed gene correlates
mmc3.xlsx (780KB, xlsx)
Document S2. Article plus supplemental information
mmc6.pdf (16.9MB, pdf)

Data Availability Statement

Scripts used in analysis can be found at https://github.com/mroseglass/hCO-CrossSiteReproducibility.git. The Seurat object can be found at UCSC Cell Browser: https://hco-cross-site-repro.cells.ucsc.edu.

Articles from Stem Cell Reports are provided here courtesy of Elsevier

RESOURCES