In a culture flask, 45 lentil-sized globs of neurons swirl in a gentle eddy of liquid medium. These lumpy, 3-D networks of human nerve cells, called brain organoids, have generated more diverse and mature cell types than any other model system of brain tissue to date.

Scientists have increasingly turned to organoids, organ models cultured from induced pluripotent stem cells, to investigate human brain development and disease. However, most brain organoid models to date have been cultured on a scale of weeks to investigate early neural development under various conditions. To accurately model development and pathogenic conditions at later stages, researchers need to generate and map more-mature human brain organoids that can gain additional features of the developed organ.

In a Nature paper published this week, researchers from Harvard University and the Stanley Center for Psychiatric Research at the Broad Institute of MIT and Harvard developed protocols and culture conditions for human brain organoids that allow the tissues to grow and develop over nine months or longer and enable unprecedented levels of cell maturation.

The team saw some striking developments in the organoids, including spontaneously active neural networks and the development of light-sensitive cells. The paper provides the largest-to-date molecular map of the diversity and reproducibility of cell types generated in brain organoids, and suggests that 3-D brain organoids have the potential to model higher-order functions such as cellular interactions and neural circuit dysfunctions related to developmental and psychiatric disorders.

“Researchers have used cocktails of signals to develop pluripotent stem cells into different types of cells from the nervous system — neurons, astrocytes, sometimes even specific classes of neurons,” said senior author Paola Arlotta, co-director of the nervous system disease program at the Harvard Stem Cell Institute, professor of stem cell and regenerative biology at Harvard, and institute member in the Stanley Center. “But the brain contains an incredible diversity of cell types that interact and form connections. We took on the challenge of investigating to what extent such complexity and diversity of cells can be produced in the dish, and how closely the organoid cell types mirror those in the endogenous tissue.”

Giorgia Quadrato, a postdoctoral fellow in Arlotta’s lab and first author on the paper, and colleagues adapted lab protocols to enable the organoids to develop long term from healthy induced pluripotent stem cell lines. To identify the diverse cell types that the organoids generated, the team used single-cell RNA sequencing (a recent invention named “drop-seq”) to analyze gene expression in more than 80,000 cells isolated from 31 brain organoids, and then compared the results to expression documented in human brain tissue samples.

They found that the longer the organoids developed, the more cell types they formed.

“The cellular diversity that the organoids generated stunned all of us,” said paper co-author Steve McCarroll, associate professor of genetics at Harvard Medical School, director of genetics for the Stanley Center, and institute member at the Broad. “The ability of stem cells within organoids to generate so many of the brain’s cell types — using their own genetic and molecular instruction book — evokes how development works inside the body.”

These cells included subtypes of neurons and progenitors of the cerebral cortex and many cell types of the visual system. Analysis showed that the organoids had generated a virtually complete array of cell types documented from the human retina, including photoreceptor-like cells. When the researchers shined light on the organoids, the stimulus affected neuronal activity.

The team proposes that this model someday may make it possible to study the response of neuronal networks to physiological sensory stimuli. “This is something that could allow detection of even subtle functional abnormalities in the neurons of organoids, generated from neuropsychiatric patients, as they respond to physiological stimuli,” said Quadrato.

After about eight months in culture, the neurons in the organoids had also developed dendritic spines (protrusions from the dendrites where synapses are formed) that were visible via electron microscopy imaging, a key feature of mature neurons that has been notoriously difficult to obtain in culture.

“The presence of dendritic spines in these organoids is an important step for studying development and disease,” said McCarroll. “Genetic studies indicate that disorders such as schizophrenia involve dysfunction at synapses and perhaps in the regulation or pruning of dendritic spines. An experimental model that develops spines can open the door to understanding how genes, and perhaps new pharmacological therapies, shape synaptic biology.”
Organoids could serve as important models for neuropsychiatric disorders, for which there are inadequate models to reflect both the genetic diversity that exists and the importance of understanding this diversity’s impact on disease manifestation.

“Human brain organoid models such as these can be produced from human cells with any genetic background,” said Steve Hyman, director of the Stanley Center, core institute member at the Broad, and professor in the Department of Stem Cell and Regenerative Biology at Harvard. “Given the large number and diversity of genetic variants that contribute to normal human cognitive and behavioral traits, and brain disorders such as schizophrenia, brain organoids are an extremely promising platform to study the effects of genes on the molecular, cellular, and synaptic building blocks of our brains.”

The team is continuing to explore avenues for maturing organoids on a faster timescale, guiding anatomical organization during development, and reducing variability in anatomy and cell composition between organoids to explore neuron maturation and networks.

“Eventually, we hope to produce a framework for next-generation disease modeling, to look at aspects of human neurobiology that are not accessible in today’s models,” said Arlotta. “We know very little about neuropsychiatric disease, and this research suggests that we may be able to produce models that help us learn a lot more. This could someday lead to therapies and cures, if we do it right.”