The story so far: Scientists at Johns Hopkins University (JHU) recently outlined a plan for a potentially revolutionary new research area called “organoid intelligence,” which aims to create “biocomputers”: where lab-grown brain cultures are paired with real-world ones sensors and input/output devices. Scientists expect the technology to harness the processing power of the brain and understand the biological basis of human cognition, learning and various neurological disorders.
What is the premise of this technology?
Understanding how the human brain works has been a difficult challenge. Traditionally, researchers have used rat brains to study various human neurological disorders. While rats offer a simpler and more accessible system for studying the brain, there are several differences in structure and function and apparent differences in the cognitive abilities of rodents and humans.
In the lab, in search of systems more relevant to humans, scientists are building 3D cultures of brain tissue, also known as brain organoids. These “mini-brains” (up to 4mm in size) are constructed from human stem cells and capture many of the structural and functional features of a developing human brain. Researchers are now using them to study human brain development and to test drugs to see how they respond.
However, the human brain also requires various sensory inputs (touch, smell, sight, etc.) to develop into the complex organ it is, and lab-developed brain organoids are not sophisticated enough. The organoids also currently lack blood circulation, which limits their growth.
Aren’t there other ways to study the human brain?
Recently, scientists transplanted these human brain organoid cultures into rat brains, where they made connections with the rat brain, which in turn provided circulating blood. Since the organoids had been transplanted into the visual system, human neurons were also activated when the scientists showed the experimental rats a flash of light, indicating that the human brain’s organoids were also functionally active.
Scientists have touted such a system for studying brain disorders in a human context. However, human brain organoids are still embedded in the rat brain microenvironment, including the non-neuronal cells that we know play critical roles in some neurological diseases. The effect of drugs in this model also needs to be interpreted through various behavioral tests on rats, which may not be sufficiently representative. So we need to address the limitations of lab-grown organoids and develop a system more relevant to humans.
What is the new “bio-computer”?
The JHU researchers’ project combines brain organoids with modern calculation methods to form “bio-computers”. They have announced plans to couple the organoids with machine learning by growing the organoids in flexible structures attached with multiple electrodes (similar to those used to take EEG readings from the brain).
These structures will be able to record the firing patterns of the neurons and also deliver electrical stimuli to mimic sensory stimuli. The response pattern of the neurons and their effect on human behavior or biology are then analyzed through machine learning techniques.
Recently, scientists managed to grow human neurons on a microelectrode array that could both record and stimulate these neurons. By providing positive or negative electrical feedback from the sensors, they were able to train the neurons to produce a pattern of electrical activity that would be produced if the neurons played table tennis.
What chances do “bio-computers” have?
For example, while human brains are slower than computers at simple arithmetic, they outperform machines at processing complex information.
Brain organoids can also be developed using stem cells from individuals with neurodegenerative diseases or cognitive disorders. Comparing data on brain structure, connections and signaling between “healthy” and “patient-derived” organoids can reveal the biological basis of human cognition, learning and memory.
They could also help unravel the pathology and drug development for devastating neurological and degenerative diseases such as Parkinson’s and microcephaly.
Are “bio-computers” ready for commercial use?
Currently, brain organoids are less than 1 mm in diameter and have fewer than 100,000 cells (both on average), which is approximately three-millionths the size of an actual human brain. Therefore, enlarging the brain organoid is key to improving its computational capacity — as is incorporating non-neuronal cells involved in biological learning.
Second, researchers must also develop microfluidic systems to transport oxygen and nutrients and remove waste products. These hybrid systems will generate very large amounts of data (ie neural recordings from each neuron and each connection) that researchers will need to store and analyze using a ‘big data’ infrastructure. They must also develop and apply advanced analytical techniques (using machines) to correlate the structural and functional changes in the brain organoids with the various output variables.
“The first, very primitive forms of learning already exist, like the pong-playing brain cultures,” said Thomas Hartung, professor of evidence-based toxicology at JHU, who is leading this work. “The challenge now is to build long-term memory. We hope to achieve this in 1-2 years. Application to brain organoids derived from patient cells, such as autism and Alzheimer’s donors, is already on the way. We could see benefits for drug development in this decade.”
It is also proposed to establish an ethics team to identify, discuss and analyze in parallel ethical issues as they arise in the course of this work.
Surat Parvatam is Senior Research Associate, Center for Predictive Human Model Systems, Atal Incubation Center – Center for Cellular and Molecular Biology (AIC-CCMB), Hyderabad.
