But now neuroscientists are in a whole new era. Three technological breakthroughs over the past 12 years have given them a revolutionary way to study human brain disease: They can create cultures of brain cells derived from individual patients, and even engineer complex, three-dimensional “organoids” that mimic key aspects brain tissue. Scientists are only beginning to harness the potential of these new human cell and tissue models, and Picower Institute labs are helping lead the way. Professors Li-Huei Tsai, Mriganka Sur and Kwanghun Chung are among a global vanguard that is making and analyzing these new patient-derived testbeds and applying them to study conditions such as Down syndrome, Alzheimer’s disease, Rett syndrome, and Zika virus infection.
While the whole field’s progress with these new capabilities has been rapid, so has been the recognition of their limits. That’s why rather than replacing animal models and other research methods, these new models are becoming integrated as powerful tools to complement broader research programs where findings from multiple methods often enhance each other’s value.
Above: In the developing brain and 3D brain organoid cultures, cells self-organize into patterns. How do we interpret these pattems? In this visualization created by members of the Chung Lab, different cell types (color-coded dots) surround the ventricles (gray blobs). Different regions of the ventricle will produce specific patterns (colored on the right) based on the presence of cell types and their migration away from the ventricle.
The new wave of breakthroughs began in 2007 when scientists showed how to take a cell from an individual’s body (often a skin cell) and to “reprogram” it to become an “induced pluripotent stem cell” (iPSC) that can then be biochemically guided to become any other cell, like a neuron or a supporting astrocyte or microglia. This development allowed scientists to make the brain cells that they of course would never directly extract from patients. By 2013, scientists began using iPSCs to grow 3D cultures of multiple brain cell types, or “organoids,” that can reproduce key aspects of brain development and intercellular interactions. That same year, scientists demonstrated that a technique called CRISPR/Cas9, could be used for precise genetic editing. Scientists quickly began using that to manipulate the genes in their iPSC-grown cultures and organoids, creating “isogenic pairs” where two otherwise identical stem cells contain a disease-causing or healthy version of a gene.
Tsai and lab members Jay Penney and William Ralvenius summarized and celebrated the significance of these breakthroughs for Alzheimer’s disease research in an August 2019 paper in Molecular Psychiatry.
“In little more than a decade since the advent of human iPSC technologies we have developed the ability to generate all the main brain cell types from pluripotent cells,” they wrote. “Increasingly complex 3D co-culture systems are also emerging that allow us to reconstitute many of the key interactions between brain cells. These technologies have already contributed greatly to our understanding of human development and human disease.”
Sur agreed, “It is perhaps the only way one can study a direct human model – it’s astonishing to be able to grow brain cells from a patient with a disease, derived from that person’s genetic material.”
Picower labs have embarked on numerous studies using the models. In 2018, Tsai’s lab, led by Yuan-Ta Lin and Jinsoo Seo, published a paper in Neuron in which they used 2D single-cell-type iPSC cultures and 3D mixed-cell-type organoids to study the differences made by two versions of the leading risk gene for Alzheimer’s disease, APOE. People with the APOE4 variant are at much higher risk for the disease than APOE3 carriers but scientists haven’t been sure why. Tsai and Lin’s team used CRISPR/Cas9 to make isogenic pairs of neurons, astrocytes and microglia and spotted several key differences that likely help explain how APOE4 raises disease risk. For instance, APOE4 neurons secreted more potentially harmful amyloid proteins, APOE4 astrocytes showed dysregulated cholesterol metabolism and cleared less amyloid. APOE4 microglia, too, did a poorer job of getting rid of amyloid buildup. Using CRISPR to change APOE4 to APOE3, meanwhile, improved cell activity.
Elsewhere in the the Tsai lab, Joel Blanchard is using iPSCs to model the blood-brain barrier, which stringently filters what comes into and goes out of the brain through the blood stream, so he can see how APOE variants affect that. Researchers are also looking at how myelination – the process of wrapping neurons in a fatty sheath to improve their electrical conductivity – may differ in Alzheimer’s.
As part of the work in the Alana Down Syndrome Center, Tsai’s lab is also using iPSC-based cultures to study the difference that having a third copy of chromosome 21 makes in gene expression in a variety of brain cells and in the development of organoids. At October’s Society for Neuroscience (SfN) annual meeting, the team presented some initial results. Hiruy Meharena observed significant physical changes within chromosomes in various brain cell types with the syndrome’s three copies of chromosome 21 (trisomy) vs. when they have two copies (disomy). These changes, which appear especially prevalent in neural progenitor cells, occur genomewide and result in substantial differences in gene expression that are associated with brain development. Meanwhile, Lin showed how organoids grown from trisomy iPSCs vs. disomy ones have smaller diameters after 30 days of growth and show gene expression differences that may hinder development. Elana Lockshin focused on differences in glial cells, finding, for instance, that trisomy astrocytes don’t migrate during development as readily as in disomy ones.
Sur’s lab has been able to make important findings by using iPSCs as part of its studies of Rett syndrome, an autism-like disorder. In a study published in 2017 in Molecular Psychiatry, a team led by Nikolaos Mellios used isogenic 2D iPSC cultures to find that the disease-causing mutation in the gene MeCP2 led to misregulated forms of RNA that alter a key molecular pathway in early neural and brain development. They also used organoids to show that if they corrected regulation of those RNAs, they could restore healthier development. The study helped to show that Rett syndrome may begin to affect health even earlier than the onset of systems in toddlerhood.
“It is perhaps the only way one can study a direct human model – it’s astonishing to be able to grow brain cells from a patient with a disease, derived from that person’s genetic material.” -Mriganka Sur
At SfN, Sur lab member Vincent Pham showed how he and Chloe Delepine are carrying on this work. Isogenic organoids with vs. without the MeCP2 mutation show distinct differences in development. They don’t grow as large and the open spaces within, called ventricles, are longer and thinner. The team also documented comparative reductions in the number of newly born neurons. Using a sophisticated imaging method developed by labmate Murat Yildirim, they observed that young neurons in Rett model organoids also struggle to migrate from the ventricles where they are born, to outer layers where they are supposed to settle into cortical circuits. Typically newborn neurons would follow a straight line outward along a pathway provided by radial glia cells. In Rett organoids the radial glia are normal, but the team watched young neurons zig zag along those paths, taking longer to arrive than they should. By applying sophisticated molecular analysis techniques to the organoid testbed, they were able to investigate the molecular pathway to account for the misguided migration. Certain adhesion molecules that are supposed to keep the new neurons on track are altered, they found.
Both the Tsai and Sur labs frequently collaborate with Kwanghun Chung, whose research group is dedicated to developing tools and technologies to help fellow scientists better visualize and quantitatively analyze tissues from the scale of whole human brains down to subcellular components like the synaptic connections between neurons. Chung has not only aided the Sur’ lab’s Rett syndrome studies but has also been working with MIT Health Sciences and Technology Professor Lee Gehrke in his lab’s efforts to quantify differences that may help explain what hinders the growth of organoids – and brains – with Zika infection.
“Organoids were pivotal in helping scientists discover how Zika causes microcephaly,” said Chung lab postdoc Alex Albanese. “We are taking a more in depth look at how Zika is actually modifying the structure and cell populations inside the organoids.”
Innovating past limits
A large part of the work in Chung’s lab, which is led by Albanese and Justin Swaney, aims to overcome some key fundamental limits of organoids. Though they are sometimes called “minibrains,” they really aren’t lab replicas of the real thing. A human brain, while extraordinarily complex, has a well-defined geography. Even though organoids are much simpler – with thousands of cells rather than about 100 billion – they are much more variable in how they turn out. While a real brain will develop highly distinct regions and exactly the right number of ventricles in the right places, organoids will recapitulate only an approximation of, say, cortical structure and may have enough ventricles to look more like Swiss cheese than a brain. Albanese calls this the “snowflake” problem, alluding to how organoids can differ so widely.
So how can they still be valuable models? One has to know how to assess them. Chung’s lab has developed advanced tissue processing technologies that can clarify, preserve, label and enlarge tissues, including organoids, so that properties like physiology and cellular function can be highlighted at all scales. Moreover, his lab has developed an imaging pipeline using light-sheet microscopy that is capable of capturing enormous amounts of data quickly (15 minutes per organoid), so that technicians can thoroughly image many organoids in a day.
Beyond that, Swaney and Albanese are developing algorithms and other analytical tools to quantify a wide array of measurements in organoids, such as the physical form, distribution and numbers of cell types. They examine how cells are arranged in the 3D context of each other, which tells them important information about their type and function. All of these advances work in combination to help researchers rigorously assess which differences that they are measuring in their organoids are attributable to their experimental variable – like a genetic disease mutation – or just to the natural variance in how organoids turn out. This improved analysis is part of the Chung lab’s Rett and Zika collaborations, Swaney said in presenting his poster at the Society for Neuroscience meeting (that same week Albanese gave a podium talk on their work at the Biomedical Engineering Society in Philadelphia).
Another limitation of organoids is that they aren’t actually that small. A millimeter or two of diameter may seem tiny, but that’s still big enough to present challenges. Traditional microscopes can’t image far enough into them to resolve what’s going on with deeper cells. Chung’s technologies overcome that problem both by clarifying tissue and labeling cells and proteins, but that requires chemically fixing the organoids. Yildirim’s “three photon” microscope technology doesn’t label cells as richly, but it can image all the way through organoids while they are alive and active. That’s how Sur’s lab was able to witness the erratic migration of new neurons. They’ve developed microfluidic multiple-well housings for organoids and devised methods for holding them perfectly steady for long-term imaging while still allowing nutrients and oxygen to circulate around them.
Indeed a significant problem associated with large size is that with no blood vessels to carry oxygen and nutrients in or to take waste out, the innermost cells can die. Keeping either the nutrients and oxygen, or the organoid, in motion helps keep them healthier than just growing them in dishes, research has shown. The Tsai lab recently began growing organoids by the thousands using a clever bioreactor device that keeps them spinning in the incubator. Postdoc Ping-Chieh Pao said the system saves space and uses less growth media, while yielding higher quality organoids with more mature neurons and less cell death.
Blood vessels are also integral to brain function. For instance, dysfunction in the blood-brain barrier plays a potentially pivotal role in Alzheimer’s. To build a human model of Alzheimer’s that explicitly accounts for vasculature in sickness and in health, Tsai and Blanchard recently secured a grant from the National Institutes of Health to build a “brain on a chip” that will engineer connections to unify their blood-brain barrier model with a co-culture of many brain cell types. The addition of vasculature and cell types like the oligodendrocytes that produce myelination will simulate a richer degree of Alzheimer’s complexity than human iPSC-based models have to date.
Yet another limitation of organoids has been that they don’t mimic multiple brain regions all that clearly, though they can be chemically coaxed to trend toward one or another. For that reason, organoids don’t provide much of a testbed for understanding the functional significance of inter-regional circuits. Delepine said the Sur lab is interested in experimenting with growing multiple organoids of different regional character on a chip and then nurturing connections between them to see if they can replicate such circuits.
Multiple models together
While Picower researchers and colleagues elsewhere work to make the most of human disease models, there are some limits that seem sure to endure, both for technical and ethical reasons. That means that traditional models, including animals, will remain integral to neuroscience research.
After all, organoids don’t think, behave or remember. They can’t gather any sensory input and do not exhibit consciousness. Lacking all these basic capabilities, they naturally can’t provide any data about how disease affects those vital functions of daily existence. In a recent study where Sur collaborated with MIT biologist Rudolf Jaenisch to test new drugs with the potential to treat Rett syndrome, Jaenisch’s lab screened the compounds on stem-cell derived human neurons, but Sur’s lab tested the most promising ones in mouse models, because that was the only way to see if they improved cognition and behavior.
“The question was, were they valid in a much more functional system,” Sur said.
Mellios’s study, too, drew upon and also utilized findings in Rett model mice.
“Our work in these two studies shows how mouse work can complement human work,” he said.
Especially when fighting disease, researchers will always draw upon whatever systems they can to help them find answers.