Scientific Understanding of Consciousness
Consciousness as an Emergent Property of Thalamocortical Activity

Brain Model via Cerebral Organoids


Nature 501,373–379(19 September 2013)

Cerebral organoids model human brain development and microcephaly

Institute of Molecular Biotechnology of the Austrian Academy of Science (IMBA), Vienna 1030, Austria

Madeline A. Lancaster, Magdalena Renner, Daniel Wenzel, Josef M. Penninger & Juergen A. Knoblich

MRC Human Genetics Unit, Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh EH4 2XU, UK

Carol-Anne Martin, Louise S. Bicknell & Andrew P. Jackson

Wellcome Trust Sanger Institute, Cambridge CB10 1SA, UK

Matthew E. Hurles

Department of Clinical Genetics, St. George’s University, London SW17 0RE, UK

Tessa Homfray


The complexity of the human brain has made it difficult to study many brain disorders in model organisms, highlighting the need for an in vitro model of human brain development. Here we have developed a human pluripotent stem cell-derived three-dimensional organoid culture system, termed cerebral organoids, that develop various discrete, although interdependent, brain regions. These include a cerebral cortex containing progenitor populations that organize and produce mature cortical neuron subtypes. Furthermore, cerebral organoids are shown to recapitulate features of human cortical development, namely characteristic progenitor zone organization with abundant outer radial glial stem cells. Finally, we use RNA interference and patient-specific induced pluripotent stem cells to model microcephaly, a disorder that has been difficult to recapitulate in mice. We demonstrate premature neuronal differentiation in patient organoids, a defect that could help to explain the disease phenotype. Together, these data show that three-dimensional organoids can recapitulate development and disease even in this most complex human tissue.

Mammalian brain development begins with the expansion of the neuroepithelium to generate radial glial stem cells (RGs). These RGs divide at the apical surface within the ventricular zone (VZ) to generate neurons and intermediate progenitors. Intermediate progenitors populate the adjacent subventricular zone (SVZ), whereas neurons migrate through the intermediate zone to populate specific layers within the cortical plate. In humans, the organization of progenitor zones is markedly more elaborate; the SVZ is split by an inner fibre layer (IFL) into an inner SVZ and an outer SVZ (OSVZ). The OSVZ represents an entirely separate progenitor layer populated by intermediate progenitors and a unique stem cell subset termed outer radial glia (oRG), which are only present to a limited degree in rodents. Both the IFL and OSVZ are completely absent in mice. These key differences allow for the striking expansion in neuronal output and brain size seen in humans.

Primary microcephaly is a neurodevelopmental disorder in which brain size is markedly reduced. Autosomal-recessive mutations have been identified in several genes, all of which encode proteins localizing to the mitotic spindle apparatus. Heretofore, primary microcephaly pathogenesis has primarily been examined in mouse models. However, mouse mutants for several of the known genes have failed to recapitulate the severely reduced brain size seen in human patients.

Given the dramatic differences between mice and humans, methods that recapitulate paradigms of human brain development in vitro have enormous potential. Although considerable progress has been made for in vitro models of whole-organ development for other systems—such as intestine, pituitary and retina — a three-dimensional culture model of the developing brain as a whole has not been established. Previous studies have modelled certain isolated neural tissues in vitro, including dorsal cerebral cortical tissue, which could recapitulate many aspects of early development. However, later events such as the formation of discrete cortical layers with stereotypical inside-out organization as well as human characteristics, such as the presence of oRG cells and the unique organization of progenitor zones, were not described.

Here we describe a three-dimensional culture system for deriving brain tissue in vitro. These so-called cerebral organoids develop a variety of regional identities organized as discrete domains capable of influencing one another. Furthermore, cerebral cortical regions display an organization similar to the developing human brain at early stages, as well as the presence of a considerable oRG population. Moreover, cerebral cortical neurons mature to form various pyramidal identities with modest spatial separation. Finally, we use patient-derived induced pluripotent stem (iPS) cells and short hairpin RNA (shRNA) in these organoids to model CDK5RAP2-dependent pathogenesis of microcephaly, a disorder that has been difficult to model in mice.

Generation of cerebral organoids

Recent progress with in vitro models of various organ systems has demonstrated the enormous self-organizing capacity for pluripotent stem cells to form whole tissues. We built upon this concept and developed a protocol without the use of patterning growth factors, focusing instead on improving growth conditions and providing the environment necessary for intrinsic cues to influence development. We began with a modified approach to generate neuroectoderm from embryoid bodies. Neuroectodermal tissues were then maintained in three-dimensional culture and embedded in droplets of Matrigel to provide a scaffold for more complex tissue growth. These Matrigel droplets were then transferred to a spinning bioreactor to enhance nutrient absorption. This method led to rapid development of brain tissues, which we termed cerebral organoids, requiring only 8–10days for the appearance of neural identity and 20–30days for defined brain regions to form.

Cerebral organoids at early stages (15–20days) formed large, continuous neuroepithelia surrounding a fluid-filled cavity reminiscent of a ventricle with characteristic apical localization of the neural specific N-cadherin.

Although tissues reached maximal size by 2months, cerebral organoids formed large (up to 4mm in diameter), complex heterogeneous tissues, which could survive indefinitely (currently up to 10months) when maintained in a spinning bioreactor. Histological and gross morphological analysis revealed regions reminiscent of cerebral cortex, choroid plexus, retina and meninges. Notably, tissues typically reached a size limit, probably because of the lack of a circulatory system and limitations in oxygen and nutrient exchange. Consistent with this, extensive cell death was visible in the core of these tissues, whereas the various brain regions developed along the exterior. Furthermore, cerebral organoids could be reproducibly generated with similar overall morphology and complexity from both human embryonic stem (ES) cells and iPS cells.

Cerebral organoids display discrete brain regions

Brain development in vivo exhibits a striking degree of heterogeneity and regionalization as well as interdependency of various brain regions. Histological analysis suggested that human cerebral organoids might similarly display heterogeneous brain regions. To examine this further, we first tested the efficiency of initial neural induction in these tissues by performing reverse transcriptase PCR (RT–PCR) for several markers of pluripotency and neural identity. As expected, pluripotency markers OCT4 (also known as POU5F1) and NANOG diminished during the course of organoid differentiation, whereas neural identity markers SOX1 and PAX6 were upregulated, indicating successful neural induction.

To test for early brain regionalization in whole organoids, we performed RT–PCR for forebrain (FOXG1 and SIX3) and hindbrain (KROX20 (also known as EGR2) and ISL1) markers, revealing the presence of both populations within the tissue. However, as tissue development proceeded, forebrain markers remained highly expressed whereas hindbrain markers decreased, reminiscent of the developmental expansion of forebrain tissue during human brain development.

In order to test whether cells with these brain region identities developed as discrete regions within the organoids, as gross morphology would suggest, or were randomly interspersed within the tissue, we performed immunohistochemical staining for markers of forebrain, midbrain and hindbrain identities during early development of these tissues (16days); PAX6 expression revealed several regions of forebrain identity, and OTX1 and OTX2 expression marked forebrain/midbrain identity. These regions were located adjacent to regions that lacked these markers but that were positive for hindbrain markers GBX2, KROX20 and PAX2, which was reminiscent of the early mid–hindbrain boundary, suggesting similar regional communication and probably mutual repression.

In vivo brain development involves increasing refinement of regional specification. Therefore, we examined further-developed cerebral organoid tissues for regional subspecification. We performed staining for the forebrain marker FOXG1, which labelled regions displaying typical cerebral cortical morphology. Many of these regions were also positive for EMX1, indicating dorsal cortical identity. We also tested for further subregionalization by staining for cortical lobe markers, namely AUTS2, a marker of prefrontal cortex; TSHZ2, a marker of the occipital lobe; and LMO4, a marker of frontal and occipital lobes but absent in parietal lobes. These markers could be seen in neurons labelling distinct regions of dorsal cortex, suggesting subspecification of cortical lobes.

Furthermore, staining for markers of the hippocampus and ventral forebrain revealed specification of these regions, although they did not organize to form the overall structure seen in vivo. Notably, interneurons produced in ventral forebrain regions exhibited a morphology and location consistent with migration from ventral to dorsal tissues. Within the dorsal cortex, these neurons displayed neurites parallel to the apical surface, reminiscent of the migratory extensions seen in tangential migration in vivo. Notably, calretinin+ interneurons were absent from the dorsal cortex of organoids lacking a ventral region (4 out of 4 NKX2-1-negative organoids), suggesting that interneurons originate in the ventral forebrain to migrate to the dorsal cortex. This suggests that distant regions can influence one another in developing cerebral organoids.

Finally, other brain structures could be observed, namely choroid plexus and even immature retina. Overall, all tissues examined displayed regions with dorsal cortical morphology (35 out of 35; 100%), most displayed choroid plexus (25 out of 35; 71%) and several displayed ventral forebrain identity as determined by NKX2-1 immunoreactivity (12 out of 35; 34%), whereas only a few displayed retinal tissue (determined by presence of retinal pigmented epithelium; 4 out of 35; 11%). These results indicate that cerebral organoids developed a variety of brain region identities organized into discrete, although interdependent, domains.

Recapitulation of dorsal cortical organization

The most dramatic changes in brain evolution from rodent to human affect the dorsal cortex. Therefore, we analysed the organization of dorsal cortical regions within cerebral organoids. Staining for markers of RGs and newborn neurons revealed typical organization into a layer reminiscent of the VZ with neurons located at the basal surface. Staining for TBR1 revealed proper development of neural identity and radial migration to the developing preplate (precursor to the cortical plate). Furthermore, staining for neural progenitor and neural-specific BAF (mammalian SWI/SNF chromatin-remodelling complex) components revealed the characteristic switch in chromatin-remodelling complexes during neural fate specification. Finally, staining for the intermediate progenitor marker TBR2 (also known as EOMES) revealed the presence of intermediate progenitors adjacent to the VZ. Thus, dorsal cortical tissues displayed typical progenitor zone organization.

In both mice and humans, cortical progenitors undergo a stereotypical nuclear movement called interkinetic nuclear migration (IKNM). Mitotic divisions occur at the apical surface of the VZ whereas the nuclei of cells in S phase are located on the basal side of the VZ. We stained for the mitotic marker phospho-histone H3 and observed most of the cells dividing at the apical surface. Similar observations were evident when we stained for phospho-vimentin (p-vimentin), a marker of mitotic RGs. In addition, as this marker labels the entire cell, we could observe basal cellular processes typical of RGs, which extended to the outer surface of these tissues. Thus, RGs of cerebral organoids exhibited typical behaviour and morphological features.

To examine this in more detail, we used a method to label individual RGs for morphology and live imaging experiments. In the mouse brain, individual cells can be labelled by in utero electroporation of fluorescent-protein-expressing plasmids. Similarly, we injected green fluorescent protein (GFP) plasmid into fluid-filled cavities of cerebral organoids and electroporated RGs adjacent to these ventricle-like cavities. This approach led to reproducible expression of GFP in RGs, revealing typical morphology at various stages of development: early pseudostratified neuroepithelium followed by later bipolar morphology with extended apical and basal processes.

To test for IKNM, we performed live imaging of GFP-electroporated RGs in cerebral organoids and observed many examples of RGs that displayed movement of the cell body along the apical and basal processes. Furthermore, we performed pulse-chase experiments with the S-phase marker BrdU and could observe a shift in RG nuclei from outer VZ localization towards the apical surface with time.

RGs in the VZ of rodents exhibit biased spindle orientation, predominantly horizontal, parallel to the ventricular surface. To examine whether RGs in human cerebral organoids exhibited a similar orientation bias, we used p-vimentin staining to examine the plane of division in mitotic RGs. We observed mainly horizontal orientations (41%), somewhat similar to the orientation bias observed in other mammals. However, we also observed abundant oblique (37%) and vertical (22%) orientations, which were more abundant in these human tissues than has been described for rodent neocortex. Interestingly, these measurements reflected the same trend recently described in the human brain, suggesting that the cerebral organoids could recapitulate aspects of human cortical development.

We examined further the fate potential of these divisions to test whether organoid RGs could divide symmetrically and/or asymmetrically. We performed electroporation of GFP followed by a short 1-h BrdU pulse and a 16-h chase to lineage trace divisions of a small minority of cells. We examined double-labelled daughter-cell pairs and could observe both symmetric proliferative fate outcomes, as well as asymmetric outcomes. This suggests that the RGs in these human tissues can undergo both symmetric and asymmetric divisions.

Formation of functional cerebral cortical neurons

The formation of the radially organized cortical plate begins with the formation of its precursor, the preplate. To test for this initial organization, we stained 30-day organoids for TBR1, a marker of the preplate, as well as MAP2, a neuronal marker. This revealed the presence of a basal neural layer reminiscent of the preplate, and an apically adjacent region reminiscent of the intermediate zone. Furthermore, we could observe reelin+ neurons along the basal surface, suggesting the presence of Cajal–Retzius cells, a cell population important in the generation of cortical plate architecture.

The stereotypical layered structure of the mammalian cortical plate is generated inside-out with later-born neurons migrating through existing layers to populate more superficial layers. Although previous methods of deriving cortical neurons have been able to generate distinct layer identities, they have been unable to recapitulate this spatial separation. To test whether this organization could be recapitulated in cerebral organoids, we stained for cortical layer markers. In less-developed tissues (30 day), early-born CTIP2 (also known as BCL11B)+ neurons were located adjacent and internal to the TBR1+ preplate, suggesting initiation of cortical plate layer formation. Furthermore, neurons exhibited rudimentary separation into an early-born deep layer (CTIP2+) and a late-born superficial layer (SATB2+ and BRN2 (also known as POU3F2)+), which became more distinct as tissues developed (75days). Finally, a cell-poor region reminiscent of the marginal zone was also evident. Notably, although this modest spatial separation was an improvement upon other in vitro methods, organoids could not recapitulate the same degree of mature layer organization as in vivo, suggesting that further cues are needed to generate the complex stereotypical layer II–VI organization.

In vivo, dorsal cortical neurons mature and extend long-range axons. To test for these characteristics, we performed GFP electroporation and examined neuronal morphology. GFP-labelled axon projections displayed complex branching and growth cone behaviour and projected long-range axons in a manner reminiscent of axon bundling.

Finally, we tested whether neurons within cerebral organoids could exhibit neural activity by performing calcium dye imaging to detect Ca2+ oscillations, which revealed spontaneous calcium surges in individual cells. Furthermore, we applied exogenous glutamate and observed more frequent calcium spikes, indicating glutamatergic receptor activity. Finally, we performed action potential blockade by application of tetrodotoxin and observed dampened calcium surges, indicating that calcium spikes were dependent upon neuronal activity.

[end of paraphrase]



Return to — Modeling Brain Functionality of Consciousness