Four mouse induced pluripotent stem cell (miPSC) lines were used (2 clones per line): (i) wild-type (WT), (ii) p73KO, (iii) p53 knockout (p53KO) and (iv) p73KO/p53KO (DKO)-miPSCs, whose generation and culture have been described previously (Martin-Lopez et al., 2017). Briefly, miPSCs were cultured on mouse embryonic fibroblast feeder cells (5 × 10⁴ cells/cm²) in Dulbecco’s Modified Eagle Medium (DMEM, Cat# D5671, Sigma-Aldrich, MO, United States) supplemented with 15% fetal bovine serum (FBS, Cat# 17479633, Thermo Fisher Scientific, MA, United States), 2 mM L-glutamine (Cat# 7513, Sigma-Aldrich, MO, United States), 1 mM sodium pyruvate (Cat# 11360039, Thermo Fisher Scientific, MA, United States), 1 mM nonessential amino acids (Cat# 11140035, Thermo Fisher Scientific, MA, United States), 0.1 mM β-mercaptoethanol (Cat# M3148, Sigma-Aldrich, MO, United States) and 1,000 U/mL Leukemia Inhibitory Factor (LIF, Cat# ESG1107, Millipore, MA, United States).

Also, the following mouse embryonic stem cell (mESC) lines were used: (i) parental E14TG2α (Hooper et al., 1987) were kindly provided by Dr. Jim McWhir (former Researcher at the Roslin Institute, Edinburgh, Scotland, United Kingdom); (ii) E14-TAp73KO (3 clones) and (iii) E14-DNp73KO cells (3 clones), which were generated in our lab (Lopez-Ferreras et al., 2021). These cells were cultured on 0.1% gelatin-coated plates (10⁵ cells/cm²) in Glasgow Minimum Essential Medium (GMEM, Cat# G-5154, Sigma-Aldrich, MO, United States) supplemented with 10% FBS, 2 mM L-glutamine, 1 mM sodium pyruvate, 1 mM nonessential amino acids, 0.1 mM β-mercaptoethanol and 500 U/mL LIF.

For mBOs generation, mPSCs were dissociated to single cells using 0.25% trypsin-EGTA and 8 × 10³ cells were seeded and quickly aggregated in each well of 96-well ultra-low attachment plates (Cat# 7007, Costar, ME, United States), in differentiation medium (100 μL/well) containing GMEM supplemented with 10% KnockOut Serum Replacement (KSR, Cat# 10828010, Gibco, MA, United States), 2 mM GlutaMAX (Cat# 35050038, Gibco, MA, United States), 1 mM sodium pyruvate, 0.1 mM nonessential amino acids and 0.1 mM β-mercaptoethanol. A partial medium change (half of the volume) was performed on day 4, and 3 days later, cell aggregates were transferred to a 10-cm bacterial-grade dish in N2 medium: DMEM/F12 (Cat# D8437, Sigma-Aldrich, MO, United States) supplemented with 1:100 N2 (Cat# 17502–048, Gibco, MA, United States) and GlutaMAX. The medium was changed every 3 days until day 14. The day on which mPSCs were seeded in ultra-low attachment plates to differentiate is referred to as differentiation day 0.

The initial optimization of the protocol was performed using the E14TG2⍺ mESC line. We conducted several rounds of experiments, introducing stepwise modifications. These changes were systematically compared to the efficiency of the reference protocol developed by Eiraku et al. (2008), allowing us to iteratively refine and enhance our method. After optimization, we successfully generated mBOs across seven independent batches of mESCs. To further validate the robustness of our protocol, we extended its application to miPSCs, where we conducted experiments with four additional batches.

mBOs were fixed in 4% paraformaldehyde (PFA) containing 6% sucrose at 4°C for 30 min. After fixation, mBOs were washed with PBS, cryoprotected overnight with 30% sucrose and stored at 4°C in PBS. Then, mBOs were embedded in PolyFreeze Tissue Freezing Medium (Cat# SHH0026, Sigma-Aldrich, MO, United States) and frozen in Peel-A-Way^™ embedding molds (Cat# 18986-1, Polysciences, PA, United States) using dry ice. mBOs were finally cryosectioned at 8 μm using a Leica cryostat (Cat# CM 1950, Leica, Germany) and the obtained sections were collected in SuperFrost Plus™ Adhesion slides (Cat# J1800AMNZ, Epredia, NH, United States) and stored at −80°C.

mBO cryosections were permeabilized in PBS containing 0.25% Triton X-100 and blocked in 10% of horse serum and 2% bovine serum albumin (BSA) for 2 h at room temperature (RT). mBO slices were incubated with primary antibodies overnight at 4°C in blocking solution, followed by the appropriate fluorophore-conjugated secondary antibodies incubation for 2 h. Nuclei were counterstained with DAPI.

Primary antibodies used were: anti-SOX2 (1:200, goat polyclonal, Cat# AF 2018, R&D Systems, MN, United States); anti-bIII tubulin (1:1,000, mouse monoclonal, Cat# MMS-435P, Covance, NJ, United States); anti-N-Cadherin (1:200, rabbit polyclonal, Cat# 13116, Cell Signaling, MA, United States); anti-NeuN (1:200, mouse monoclonal, Cat# MAB377, Merck Millipore, MA, United States); anti-phospho Histone H3 (pHH3) (1:500, mouse monoclonal, Cat# 9706, Cell Signaling); anti-Cleaved Caspase-3 (CC3) (1:200, rabbit polyclonal, Cat# 9664, Cell Signaling). The secondary antibodies were: Alexa568 anti-goat (Cat# A11057, Invitrogen, MA, United States); Alexa647 anti-mouse (Cat# A31571, Invitrogen); Alexa488 anti-rabbit (Cat# A21206, Invitrogen); Alexa647 anti-mouse IgG2a (Cat# A21241, Invitrogen); Alexa594 anti-mouse IgG1 (Cat# A21125, Invitrogen); all diluted 1:1,000.

Confocal microscopy images (8 bits) were obtained in a Zeiss LSM800 Confocal Laser Scanning Microscope (Carl Zeiss Microscopy GmbH, Germany) using 25 × Plan-Apo/0.8 numerical aperture or 63 × Plan-Apo/1.4 numerical aperture oil objectives at RT. In some cases, a 0.5 × digital zoom was used. Confocal Z-stack images were acquired, and stacks were z-projected to maximum intensity. Images were processed with the ZEN blue software (Carl Zeiss Microscopy GmbH). All images were further analyzed using ImageJ/Fiji software (MD, United States). Three batches of mESCs-derived organoids and another three batches generated from miPSCs were used for quantifications. For each genotype, five different mBO cryosections per clone were analyzed across the three independent batches resulting in a total number of, at least, 15 mBOs per condition.

For the quantification of the number of ventricles, 25 × 0.5 Z-stack confocal images of mBO slices were employed to manually count the number of ventricles in each mBO cryosection based on the presence of SOX2+ neural rosettes with a defined N-CAD+ lumen.

For the quantification of NPCs, 63 × Z-stack confocal images were used and the number of SOX2+ and the total number of cells (DAPI-stained nuclei) were manually counted. Also, NeuN+ cells were manually quantified to calculate the percentage of NeuN+ cells.

The percentage of proliferating or apoptotic cells was calculated by analyzing 63 × Z-stack pictures to manually count the number of pHH3+ or CC3+ cells and the total number of cells (DAPI-stained nuclei). To quantify the percentage of proliferating progenitors and apoptotic progenitors or neurons, colocalization studies were performed using 63 × Z-stack confocal images. Images of the pHH3 or the CC3 channel were binarized using an automatic threshold. A selection was created (ROI) and applied to the SOX2 or TUJ1 channel. Then, the “Clear Outside” plug-in was applied to eliminate the signal without double staining. Finally, cells co-expressing pHH3/SOX2, CC3/SOX2 or CC3/TUJ1 were counted using ImageJ/Fiji.

Bulk RNA from WT-, TA- and DNp73KO-mBOs was isolated at the indicated time points (see figure legends), using RNeasy mini Kit (Cat# 74106, QIAGEN, Hilden, Germany). RNA concentration and quality were determined using a Nanodrop ND-100 (Thermo Fisher scientific, Waltham, MA, United States). For each genotype and timepoint, RNA samples from three biological replicates were sent to Novogene Company Limited (Cambridge, England, United Kingdom) for RNA-sequencing (RNA-seq). The library preparations were sequenced on an Illumina platform (Illumina Inc., CA, United States) generating paired-end reads with 40 million reads per direction. The quality of sequencing was validated with FASTQC software (Simon, 2010). The alignment and quantification of the reads on the reference sequences were performed using Salmon software (Patro et al., 2017). The differential expression analysis was conducted using the R statistical analysis program, with the DESeq2 package (Love et al., 2014). Mapping of the reads to the reference genome, normalization to obtain the FPKMs (fragments per kilobase of transcript per million mapped reads) and DESeq2 analysis were performed at the core service Bioinformática USAL (Universidad de Salamanca, Spain). The FPKM raw data is available at OPEN SCAYLE: https://open.scayle.es/dataset/alonso-olivares-et-al-2024. Functional annotation analysis on differentially expressed genes (DEGs) was carried out using The Database for Annotation, Visualization and Integrated Discovery (DAVID) (Huang da et al., 2009; Sherman et al., 2022). Principal component analysis (PCA) was performed using Clustvis web tool (Metsalu and Vilo, 2015). Venn diagrams were created using BioVenn (Hulsen et al., 2008).

cDNA synthesis from RNA samples was carried out using the High Capacity RNA-to-cDNA kit (Cat# 4387406, Applied Biosystems, Waltham, MA, United States). Gene expression was analyzed by qRT-PCR using FastStart Universal SYBR Green Master (Cat# 4913850001, Roche, Basel, CH) in a StepOnePlusTM Real-Time PCR System (Applied Biosystems). Primer sequences for target genes used are included in Supplementary Table S1. The comparative threshold cycle method was used to quantify relative mRNA expression.

Statistical analyses were performed using GraphPad Prism software version 10.2.3 for Windows (GraphPad Software, Boston, MA, United States). Values were expressed as mean ± standard error of the mean (SEM). Data distribution was analyzed in all cases. When comparing two groups, an unpaired Student’s t-test was used. For comparisons between multiple groups, statistical differences were determined using either parametric one-way ANOVA followed by Tukey’s correction or nonparametric Kruskal–Wallis tests with Dunn’s correction. Differences were considered significant when p < 0.05. Comparisons with WT mBOs are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001), whereas hash (#) denote comparisons with p73KO organoids and dagger (†) is used to show comparisons with p53KO. ns: non-significant.

Several protocols to generate mBOs have been described (Eiraku et al., 2008; Nasu et al., 2012; Watanabe et al., 2005), but they are not highly reproducible and require some adaptation to each cell line. We sought to establish an improved protocol to consistently generate high-quality, reproducible mBOs. We first used a previous method (Eiraku et al., 2008) to generate mBOs from WT-mPSCs, both mESC and miPSC lines. With this protocol, some mESCs differentiated into the neural lineage and gave rise to TUJ1+ neurons, but neural progenitors (SOX2+) failed to self-organize and form N-CAD+ ventricle-like structures (Supplementary Figure S1). We then adjusted some parameters to optimize the initial size of the organoids and improve cell viability: we increased the number of seeded cells from 3,000 to 8,000 cells/well and reduced the frequency of culture medium changing during the process (Figure 1A). We also modified the composition of the initial differentiation medium, eliminating Dickkopf Wnt Signaling Pathway Inhibitor 1 (DKK-1) and left-right determination factor 1 (LEFTY-1). DKK-1 inactivates Wnt/β-catenin signaling, which is necessary for NSC proliferation (Chenn and Walsh, 2002). On the other hand, LEFTY-1 inhibits Nodal signaling through pSMAD2/3 (Kim et al., 2014), so the presence of LEFTY-1 in the medium may prevent differentiation. Based on this reasoning, and in agreement with (Eiraku and Sasai, 2011), we eliminated both factors from the initial differentiation medium.

To determine whether the modified protocol could generate bona fide mBOs, we analyzed the expression and localization of key markers: SOX2 (a neural stem cell marker), TUJ1 (an early neuronal marker), and neural-cadherin N-CAD (a cell adhesion molecule). These markers are essential for assessing the presence and organization of VZ-like neural rosettes, a crucial indicator of proper brain organoid formation and development (Eichmuller and Knoblich, 2022). Indeed, rosette-like structures represent the in vitro counterpart of the neural tube, the structure of the early nervous system in vivo (Roll et al., 2022). We observed that WT mESC-derived mBOs contained SOX2+ neural progenitors at day 14, which were clearly organized in rosette-like structures surrounded by TUJ1+ neurons (Figure 1B). Rosettes were structurally similar to the VZ of the embryonic forebrain, with an internal lumen nicely decorated with N-CAD corresponding to the apical domain of neuroepithelial cells. This protocol was also efficient at generating mBOs from WT miPSCs (Figure 1B’), again with rosettes structurally similar to telencephalic VZ, thus confirming that the implemented modifications resulted in a robust protocol to generate mBOs from mPSCs.

To substantiate the quality of the mBOs generated with the improved protocol, we performed bulk transcriptomic analyses on mBOs derived from WT E14TG2α mESCs, analyzing them at different time points during development. Analysis of differentially expressed genes (DEGs) identified 3,083 genes that were significantly upregulated from day 3 to day 14 (padj <0.01; log2FC > 1), while 914 genes were downregulated (padj <0.01; log2FC < −1). Functional annotation analysis of the upregulated DEGs confirmed that our culture conditions induced transcriptional programs related to neurogenesis, highlighting Gene Ontology (GO) terms related to nervous system development, axon guidance, neuron fate specification, and forebrain neuron differentiation (Figure 1C). The transcriptomic data also showed progressive upregulation of pan neuronal markers such as Map2, NeuN, Ncam1 and Dcx, as well as genes related to axon guidance (Dpysl4 and Dpysl5, Robo2) and neurite outgrowth (Ncan) during the process of mBO formation (Figure 1D; Supplementary Table S2). Upregulation of some of these DEGs associated with neuronal identity was further validated by qRT-PCR (Supplementary Figure S2). This upregulation was particularly striking in genes related to the organization of the cortical plate, such as Foxg1 and Lhx2, whereas markers of the caudal brain (Gbx2 and Lmx1a) or cerebellum (Pcp2) remained low (Figure 1E). Concomitantly, pluripotency genes such as Pou5f1, Zscan10 or Lin28A were downregulated from day 3 to day 14 (Figures 1D, E; Supplementary Figure S2; Supplementary Table S2), together with genes associated with cell cycle progression and DNA synthesis, indicating a decrease in stem and proliferative cells. Some of the most prominent signaling pathways involved in neural induction and development include Notch, Sonic Hedgehog and Wnt (Manzari-Tavakoli et al., 2022). As expected, these signaling pathways were activated during the differentiation process and generation of mBOs, as shown by the upregulation of genes directly involved in signal transduction (Shh and Notch1), as well as some of their target genes (Emx2, Ascl1 or Neurogenin), involved in neural development (Figures 1C–E).

Altogether, this confirmed that our optimized protocol promoted neural development and differentiation, leading to the generation and organization of mature cortical neurons from mPSCs. Therefore, we have fine-tuned and improved a protocol to generate mBOs by implementing changes in the initial cell number and the cocktail of factors used to induce neural commitment. This resulted in a rapid, affordable and highly reproducible protocol for the generation of mBOs from both induced and embryonic mPSCs.

p53 and p73 are known cell cycle and cell death regulators during brain development (Agostini et al., 2018). Thus, we decided to corroborate these facts in the mBO model and to exploit our improved protocol to delve into the role of p53 and p73 in several aspects of mouse cortical brain development. We used previously generated p73KO-miPSCs, together with p53KO- and DKO-miPSCs (Martin-Lopez et al., 2017). Additionally, we made use of isoform specific mESCs (TAp73KO- and DNp73KO-mESCs) generated by CRISPR/Cas9 genome editing (Lopez-Ferreras et al., 2021). First, we assessed the capacity of these mBOs to recapitulate some of the effects of p53 and p73 deficiency on neural cell proliferation and/or cell death described in vivo (Gonzalez-Cano et al., 2010; Meletis et al., 2006). Similarly to what was observed in neurospheres assays (Meletis et al., 2006; Armesilla-Diaz et al., 2009), histological analysis of p53KO-mBOs after 14 days in differentiation conditions revealed a significant increase in the percentage of proliferative cells (positive for the mitosis marker pHH3) compared to controls (WT-mBOs) (Figure 2A). On the contrary, the absence of p73 led to a significant reduction in the percentage of proliferating cells. Interestingly, elimination of p73 in the context of p53-deficiency (DKO-mBOs), restored the proliferation index to levels close to WT-mBOs, suggesting that p73 is essential to sustain the high proliferation index unleashed by the absence of p53 within the mBO (Figure 2A).

To identify if the reduced proliferation observed in p73KO-mBOs affected their overall cellular populations, or if it was restricted to a specific cell type, we quantified the percentage of proliferating SOX2+ NPCs (pHH3+/SOX2+, Figure 2B). In WT-mBOs, the percentage of proliferating NPCs was around 2%, while this percentage drastically decreased to 0.45% in p73KO-mBOs (Figure 2B). Again, p53KO-mBOs doubled the percentage of SOX2+ proliferating cells, while DKO-mBOs had NPC proliferation rates comparable to their WT counterparts (Figure 2B). It is important to keep in mind that the decrease of proliferating cells in the p73KO could be related to a compensatory p53-activation leading the NPCs to exit the cell cycle, to senesce or to die.

Next, we analyzed apoptosis by quantification of CC3+ cells. In WT-mBOs, nearly 10% of cells were CC3+ (Figure 2C). Interestingly, while the absence of p53 in p53KO-mBOs significantly reduced, but not obliterated, this basal level of apoptosis, lack of p73 increased significantly the percentage of apoptotic cells in p73KO-mBOs compared to the WT-mBOs (Figure 2C). This effect was reduced to WT levels upon p53 loss in DKO-mBOs (Figure 2C), confirming a role of p73 in preventing p53-dependent cell death. However, it is noteworthy that a significant percentage of apoptotic cell death remains in double mutants, compared to mBOs lacking only p53, suggesting that p73 deficiency elicits a certain level of p53-independent apoptosis.

To determine whether p73 function is necessary to prevent overall apoptosis within mBOs or if its role is specific to a particular cell type, we performed double staining for SOX2/CC3 and TUJ1/CC3. As shown in Figure 2D, the lack of p73 did not significantly affect the frequency of apoptotic NPCs (SOX2+/CC3+ cells) but led to a sharp increase in the number of apoptotic neurons (TUJ1+/CC3+ cells).

Next, we sought to discriminate which p73 isoform was responsible for maintaining the proliferative pool of NPCs by analyzing cell proliferation and apoptosis in the p73 isoform-specific knockout-mBOs. In WT-mBOs, over 1% of the cells were pHH3+, a percentage severely reduced in mBOs lacking either TA- or DNp73 (Figure 2E), indicating that both isoforms are required to maintain the mitotic rate within the mBOs. Intriguingly, TAp73 deficiency led to a significant decrease in the number of SOX2+ NPCs, while DNp73 deficiency did not have a significant effect (Figure 2F). Together, these results indicate that TAp73 is the isoform necessary to sustain proliferation and maintain the pool of NPCs, whereas DNp73 is dispensable for the latter function, but essential for sustaining the proliferative capacity of other cell types in the mBO. The analysis of CC3+ cells revealed that while the loss of TAp73 had no effect on the overall apoptosis rate compared to WT-mBOs, DNp73-deficiency significantly increased the number of apoptotic cells (Figure 2G). Moreover, the lack of DNp73 primarily affected early immature neurons, but not SOX2+ progenitors (Figure 2H, right and left panels, respectively).

Collectively, our mBO model confirmed that p73 plays a key role in neuronal survival in vitro, consistent with previous observations in vivo (Tissir et al., 2009), where DNp73 plays a pro-survival role in discrete neuron types in a p53-dependent manner (Tissir et al., 2009; Osterburg and Dotsch, 2022; Pozniak et al., 2002; Pozniak et al., 2000). Moreover, our results evidenced that both TA- and DNp73 are necessary to maintain a functional pool of NPCs within the mBO, but they operate differently depending on the cell type: TAp73 maintains the proliferating NPC pool, while DNp73 sustains general proliferation and prevents neuronal apoptosis.

We next examined the effect of total loss of p73 and/or p53, as well as the elimination of the specific p73 isoforms, in the cellular and molecular events of mouse brain morphogenesis, beyond the already described impact in cell proliferation and apoptosis. mBOs at day 14 displayed obvious morphological differences between the four genotypes (Figure 3A). WT-mBOs exhibited multiple buds with translucent lumens resembling neuroepithelia (Figure 3A, yellow arrows) whereas the lack of p73, p53 or both (DKO) resulted in mBOs with less visible neuroepithelial structures, especially in the case of p73KO-mBOs which had almost smooth borders (Figure 3A, dashed arrows). This suggested that deficiencies in these p53 family members interfere with the correct development and morphogenesis of mBOs. A more detailed analysis of these epithelia-like buds revealed that WT-mBOs exhibited rosette-like structures formed by SOX2+ NPCs with an N-CAD+ ventricular surface (Figure 3B, arrows), surrounded by TUJ1+ neurons on the organoid surface, resembling the cortical plate in developing embryos (Roll et al., 2022). In contrast, the existing ventricles in mBOs lacking p73, p53, or both, were ill-defined, with an overall disrupted pattern (Figure 3B, dashed arrows). This was particularly striking in p73KO-mBOs, where neither N-CAD expression, nor defined ventricles, were observed (dashed arrows). Accordingly, while WT-mBOs presented an average of three defined neural ventricles per organoid cryosection, the mutant mBOs, and the p73KO-mBOs in particular, displayed a reduced number of ventricles (Figure 3B’). These observations suggest defects in cell-cell adhesion and cellular self-assembly, highlighting the essential requirement of Trp73 for proper NPC organization and self-assembly into defined VZ-like structures.

Having identified that Trp73 is fundamental for mBO organization and the formation of VZ-like structures, we aimed to disclose the specific roles of the different p73 isoforms in this process. While WT- and DNp73KO-mBOs presented translucent neuroepithelial buds (Figure 3C, yellow arrows), mBOs lacking TAp73 were opaque, indicating defective formation of optically translucent and radially organized neuroectoderm. Besides, the neuroepithelial structures were less numerous and conspicuous in TAp73KO-mBOs (Figure 3C, dashed arrows).

Immunofluorescence analysis of mBO cryosections showed that mBOs generated from WT-mESCs exhibited NPCs organized into spherical rosettes with defined N-CAD+ ventricles, similar to the ones generated from WT miPSCs (Figure 3D, arrows). In contrast, NPCs in TAp73KO-mBOs were rather scattered and loose and did not form neural rosettes nor N-CAD+ ventricles (Figure 3D, dashed arrows), suggesting defects in NPC cell-cell adhesion, polarization and self-organization. In contrast, DNp73KO-mBOs showed well defined ventricles (Figure 3D, arrows), which were even more numerous than in WT-mBOs (Figure 3D’). This increased abundance of rosettes might be due to elevated TAp73 levels detected in these cells (Lopez-Ferreras et al., 2021), which could enhance their ability to establish correct cell polarity and form cell-to-cell interactions. In this differentiation context, this may favor the formation of VZ-like structures.

Considering all the above, the defective organization observed in p73KO-mBOs appears to be primarily due to the absence of the TAp73 isoform, seemingly essential for the organization of NPCs into rosette-like structures, and hence for the proper formation of ventricles within the mBO.

In the course of brain morphogenesis, NSCs differentiate to progressively acquire a specific neuronal identity and to eventually establish the complex structures of the brain. The processes of fate commitment and spatial organization are tightly regulated and, in fact, closely coupled. Thus, we explored whether the observed defects in neural structure organization within the mBOs were related to irregularities in the process of cellular decision-making and cell fate determination. In a typical differentiation process, SOX2+ progenitors give rise to TUJ1+ neurons which further differentiate into mature, post-mitotic neurons expressing the nuclear marker NeuN. Hence, we analyzed the expression of SOX2 and NeuN in mBOs at day 14 by immunofluorescence (Figure 4). In WT-mBOs, nearly 40% of cells were SOX2+ neural progenitors and 14% were NeuN+ neurons (Figures 4A, A’). The number of SOX2+ progenitors in mBOs lacking p53 was not significantly affected, while they exhibited a minor, but significant, increase (up to 23%) in the percentage of NeuN+ neurons (Figure 4A’), indicating that lack of p53 slightly affected the differentiation programs involved in neuronal maturation rather than the generation of neurons from SOX2+ progenitors. This is in agreement with previous studies highlighting the role of p53 in neuronal differentiation and maturation (Armesilla-Diaz et al., 2009; Forsberg et al., 2013; Marin Navarro et al., 2020). In contrast, mBOs lacking p73 showed significantly fewer SOX2+ neural progenitors and a noticeable significant increase in mature NeuN + cells (from 14% to 65%) (Figure 4A’). This indicates that the absence of p73 is associated with deregulation of neural cell fate, leading to a premature neuronal differentiation in these organoids, supporting previous data obtained using neurosphere assays (Fujitani et al., 2010; Gonzalez-Cano et al., 2010). Interestingly, the elimination of p53 in this context (DKO), decreased significantly, but not completely, this effect, suggesting a functional interaction between p73 and p53 in regulating cell fate and neural differentiation.

When analyzing the contribution of each p73 isoform, we observed that TAp73 deficiency did not significantly affect the number of SOX2+ progenitors nor the percentage of NeuN + neurons compared to WT-mBO (Figures 4B, B’). In contrast, the absence of DNp73 led to a notable decrease in the number of SOX2+ progenitors, accompanied by a highly significant increase in the number of mature neurons (from 8% in WT-mBOs to 50% in DNp73-mBOs). In DNp73-mBOs, these neurons were detected throughout the organoid rather than mainly in the cortical border, as was the case in WT- and TAp73KO-mBOs (Figure 4B, white arrows). This is in line with the phenotype observed in p73KO-mBOs and confirms that accelerated neuronal differentiation in these organoids is caused by the absence of DNp73. These results demonstrate that DNp73 has a remarkable effect in orchestrating neuronal differentiation from neural progenitors and may specifically act as a negative regulator of neuronal cell fate.

To further pinpoint the transcriptional changes occurring in the p73 isoform-deficient mBOs, we conducted global transcriptomics analyses using bulk RNA sequencing at days 7 and 14 (Supplementary Table S3). PCA analysis showed that clustering of biological replicates from WT, TA- and DN-p73KO organoids improved over time, with a clear separation between days 7 and 14 (Supplementary Figure S3A). Moreover, the major source of variation was linked to the mBOs genotype (with principal component PC1 accounting for almost 40% of the variance). The specific loss of either TA- or DN-p73 resulted in altered transcriptional profiles (Supplementary Figure S3B), with the effect being more pronounced at later stages. Moreover, DNp73 deficiency had a greater impact on the number of DEGs, particularly at day 14: 2,835 genes were upregulated (padj <0.01; log2FC > 1) and 2,350 genes became downregulated (padj <0.01; log2FC < −1) compared to WT-mBOs.

TAp73 is a bona fide transcriptional activator, while DNp73 acts mostly, but not always (Toh et al., 2008), as a repressor. Thus, we focused on genes that were downregulated in the absence of TAp73 and upregulated upon DNp73 deficiency. Functional annotation analyses of the DEG lists revealed that downregulated genes in TAp73KO-mBOs were associated with cell adhesion, extracellular matrix organization, animal organ morphogenesis (e.g., heart tube development, mammary gland development, etc.) and axis specification (Figure 5A). This reinforces the role of TAp73 in regulating a transcriptional module related to tissue morphogenesis (Maeso-Alonso et al., 2021) and may explain the observed inability to form archetypal structures, such as proper VZ-like structures, in the absence of TAp73. Notably, genes associated with functions previously linked to TAp73, such as angiogenesis, stem cell differentiation, and cytoskeleton regulation (Gonzalez-Cano et al., 2010; Fernandez-Alonso et al., 2015; Fuertes-Alvarez et al., 2018), were also enriched. On the other hand, upregulated genes in DNp73KO-mBOs were highly significantly associated to biological terms such as nervous system development, synaptic transmission, axonogenesis or axon guidance (Figure 5B), correlating with the phenotype of premature neuronal differentiation observed in DNp73KO-mBOs. Additionally, GO terms linked to DNA replication, cell cycle progression, cell division and stem cell population maintenance were enriched among downregulated DEGs in DNp73KO-mBOs (Supplementary Figure S3C), indicating that DNp73 sustains the capacity of NPCs to proliferate.

To evaluate how the elimination of TA- or DN-p73 affected neural cell fate determination, we intersected the up- and downregulated DEGs at different stages (compared to their WT counterparts), with gene sets that have been previously linked to NSCs, NPCs and mature neurons (Bifari et al., 2020; Cahoy et al., 2008; Ciarpella et al., 2021). As shown in Figure 5C, none of the genes related to the NPCs- or the mature neurons-profiles were downregulated in DNp73KO-mBOs, and only a few were repressed in TAp73KO. It called our attention that three NSC-associated genes were commonly downregulated in DNp73KO and TAp73KO-mBOs both at day 7 and 14: NADPH-Dependent Carbonyl Reductase 3 (Cbr3), Thiosulfate Sulfurtransferase (Tst) and ETS Variant Transcription Factor 4 (Etv4). The downregulation of Etv4 is of particular interest, since it works as a mechanical transducer that drives spatiotemporal lineage specification and its inactivation in human embryonic stem cell epithelia derepresses the neuroectoderm fate (Yang et al., 2024). On the other hand, DNp73KO-mBOs remarkably upregulated genes related to the neural fate when compared to the WT counterparts, whereas the lack of TAp73 had a milder outcome. Indeed, elimination of TAp73 had almost no impact at early stages (day 7) (Figure 5C). On the contrary, the effect in DNp73KO-mBOs was striking at early stages, with DNp73 deficiency resulting in the upregulation of 60% of the genes included in the NPC profile and a 37% overlap with genes related to mature neurons at day 7. This suggests that DNp73 absence accelerates neuronal lineage induction.

The importance of DNp73 repression, and TAp73 to a minor extent, in establishing neuronal identity is further illustrated by the significant upregulation of individual markers related to various neuron types (Figure 5D). For instance, isoform-deficient mBOs upregulate genes that have been shown to be important for the function of interneurons, such as Dlx1, Scn1a or Erbb4 (Ciarpella et al., 2021). In addition, DNp73KO-mBOs robustly overexpress genes related to excitatory and inhibitory synapses. The enrichment in glutamatergic synapses was evidenced, for example, by the upregulation of the glutamate transporter Slc17a6 (VGLUT2), as well as the overexpression of NMDA and AMPA receptors (Grin2b, and Gria1a/Gria2, respectively). On the other hand, the presence of GABAergic synapses was indirectly assessed by the increased expression of GABA transporter (Slc32a1) and gephyrin (Gphn). Conversely, markers indicative of other neuronal lineages (e.g., serotonergic) were not consistently expressed. Collectively, our results indicate the presence of a diversely enriched population of neurons in p73 deficient organoids. Finally, we examined a previously described gene set of mature neuron markers (Ciarpella et al., 2021). The expression levels of genes related to: i) metabolism and axon growth, ii) synaptic functionality and vesicle trafficking and iii) signal transduction and neurotransmission regulation were increased in both genotypes, most notably in DNp73KO-mBOs, indicating an enhanced neuronal maturation (Figure 5E). However, genes related to beta-oxidation pathways at peroxisomal (Acox1, Ech1) or mitochondrial level (Acadsb, Echdc2, and Slc25a20) (Watanabe et al., 2007) were not significantly upregulated, suggesting that at the latest analyzed time point the metabolic switch to oxidative metabolism did not accompany the neuronal maturation observed in those organoids.

Brain cortex development is a highly orchestrated sequence of events that requires precise signaling inputs. Shh signaling is vital for the formation and maturation of cortical layers, as it regulates the timing of progenitor cell differentiation and promotes their proliferation (Cai et al., 2023; Saade et al., 2017). These processes affect the thickness and cell composition of each cortical layer (Belmonte-Mateos and Pujades, 2021; Nowakowski et al., 2017). The Notch signaling pathway is pivotal in maintaining NPC populations and regulating their differentiation into specialized neurons and glia (Gaiano and Fishell, 2002), thereby influencing the overall size and complexity of the cortical region within the organoid. Additionally, the timing of Wnt activation is particularly crucial for establishing neural identity and suppressing unwanted cell fates (Rosso and Inestrosa, 2013), ensuring that cortical brain organoids develop with the appropriate cellular composition and organization. Supporting the role of DNp73 in regulating neural cell fate, we observed significant overexpression of some key genes involved in these pathways in DNp73KO-mBOs at day 7 (Figure 6A). This was evidenced by the upregulation of key genes involved in the activation and transduction of these pathways (Notch1, Shh, and Wnt5a), as well as their target genes (Emx2, Ascl-1, and Neurog1) (Figures 6A, B). Additionally, the induction of the neural lineage requires dual SMAD inhibition (Chambers et al., 2009). In accordance, DNp73KO-mBOs also exhibited downregulation of Bmp-4 and Nodal, which promote the activation of Smad1/5/8 and Smad2/3 signaling, respectively, suggesting that the absence of DNp73KO leads to dual-SMAD inhibition.

Gene expression dynamics in DNp73KO-mBOs from day 3 to day 14, as indicted by our transcriptomic data and validated by qRT-PCR analysis, reflected the activation of the Shh, Wnt, and Notch signaling pathways (Figures 6B, C). Regarding the Activin/Nodal signaling, it is known that this cascade regulates Nodal and Lefty1 gene expression and, that while Lefty1 blocks Nodal signaling (Chen and Shen, 2004), Lefty1 expression pattern follows essentially that of Nodal in early stages of development (Besser, 2004; Zhang et al., 2019). Thus, in agreement with a repression of this signaling cascade, we detected a significant reduction of Lefty1 expression in DNp73KO-mBOs from days 7–14 (Figures 6B, C). This repression of Nodal signaling could account for the switch towards a predominant neuroectodermal fate observed in the absence of DNp73 at early stages (Figure 5C), as previously reported in ESCs (Vallier et al., 2004; Wang et al., 2017), pointing to DNp73 as a node regulator of neuronal fate. Altogether, these characteristic molecular signatures highlight the importance of p73, particularly DNp73, for a precise and timely regulation of the onset of neural cell fate commitment and shed new light on the specific role of each p73 isoform, and how they tightly coordinate the maintenance of a pool of proliferating neural progenitors with the timely generation of neurons, and their maturation.