At a look: 

  • Wei-Chung Allen Lee @HarvardMed is learning the neural circuits in the mind that underlie conduct
  • Lee is working in a brand new subject of neuroscience known as connectomics that goals to comprehensively map connections between neurons
  • Lee’s lab is in understanding the normal rules by which neural circuits are organized into useful networks
  • “The brain is structured so that each neuron is connected to thousands of other neurons, and so to understand what a single neuron is doing, ideally you study it within the context of the rest of the neural network,” Lee defined.

Newswise — Many of us have seen microscopic photographs of neurons in the mind — every neuron showing as a glowing cell in an unlimited sea of blackness. This picture is deceptive: Neurons don’t exist in isolation. In the human mind, some 86 billion neurons kind 100 trillion connections to one another — numbers that, sarcastically, are far too giant for the human mind to fathom.

Wei-Chung Allen Lee, Harvard Medical Faculty affiliate professor of neurology at Boston Kids’s Hospital, is working in a brand new subject of neuroscience known as connectomics, which goals to comprehensively map connections between neurons in the mind.

“The brain is structured so that each neuron is connected to thousands of other neurons, and so to understand what a single neuron is doing, ideally you study it within the context of the rest of the neural network,” Lee defined.

Lee lately spoke to Harvard Medication Information about the promise of connectomics. He additionally described his personal analysis, which mixes connectomics with info on neural exercise to discover neural circuits that underlie conduct.

Harvard Medication Information: To start out with a primary query, what’s connectomics?

Lee: We outline connectomics as understanding how particular person neurons are related to each other to kind useful networks. The objective is to create connectomes, or detailed structural maps of connectivity the place we are able to see each neuron and each connection. What’s distinctive is the comprehensiveness of connectivity: In an ideal connectome, we’d know the way each neuron was related to each different neuron.

We consider that the connectivity of neurons is key to how they perform, since they need to obtain info from one another in order to use this info. Having complete information about connectivity permits us to have a look at higher-order interactions between populations of neurons which are vital for mind perform and conduct. It’s difficult to examine higher-order interactions with out connectomics.

Some have argued that you’re your connectome. If you go to sleep at night time, your mind exercise dramatically modifications, interrupting your ideas and emotions — however while you get up, you resume your ideas and emotions with none break in your sense of self. That is possible as a result of your mind connectivity has remained largely intact by the night time. In essence, the construction of how our neurons are wired is our “self,” and connectomics is the key to understanding this construction.

HMNews: What are you learning inside the context of connectomics?

Lee: My lab is in understanding how computations come up in the mind, or the normal rules by which neural circuits arrange themselves into useful networks. To do that, we goal to comprehensively map how particular person neurons are related to each other in advanced networks. At the identical time, we wish to perceive how these neurons are lively inside the functioning circuit. We do that in the context of conduct, starting from making choices to executing actions.

We are attempting to couple connectomics with recordings of neural exercise to do what we name useful connectomics. Primarily, we take the map of the place each neuron is and the way it’s related to each different neuron, and we layer on details about the exercise of these neurons in a dwelling animal. We additionally use genetic engineering approaches to label particular cell varieties, which is extra info that we are able to layer on high of connectivity.

HMNews: What instruments do scientists use to map connectomes?

Lee: We’re growing and making use of high-throughput microscopy, computational approaches, and machine studying to generate connectomes and translate these detailed maps of neural connectivity into organic and computational insights. One key element of our strategy is serial transmission electron microscopy, or EM, which has unsurpassed spatial decision, signal-to-noise ratio, and pace relative to different serial EM strategies. This method permits us to establish excitatory and inhibitory neurons, in addition to the synapses, or small gaps the place neurons join to one another. We will additionally study connectivity patterns of neurons, and examine the group of synaptic connections.

Traditionally, high-resolution EM has been sluggish and tedious, however we’ve engineered a high-speed EM platform that permits us to seize the complete nervous system of an grownup fruit fly in a number of months, producing 5 to 10 terabytes of information a day. We have now additionally developed computational infrastructure and instruments that allow us to deal with and visualize the giant quantities of information that we’re producing. For instance, we use synthetic deep neural networks to extract details about cells and their connectivity from these huge datasets.

HMNews: What fashions do you utilize in your analysis?

Lee: We have now primarily labored with mice and fruit flies, that are highly effective and well-studied mannequin methods. The sphere has refined genetic instruments that permit us to label totally different populations of neurons throughout the central nervous methods of these species. In fruit flies, we are able to use the applied sciences we’ve been growing for connectomics to seize the complete mind and nervous system at synapse decision. In the mouse, we are able to goal related neural circuits or subcircuits. We’re utilizing these fashions to examine the primary rules of how neural circuits are constructed and function — principally how neural networks are related to one another to carry out totally different computations that underlie conduct.

We additionally work in nontraditional mannequin methods equivalent to the mosquito. Mosquito brains are about the identical dimension as fruit fly brains, however the genetics is more difficult. Scientists have used genetics to entry the first-order neurons that begin carrying info into the mosquito mind, however the relaxation of the mind is a black field in many respects. We don’t know a lot about its elementary neurobiology, together with how the mosquito mind integrates totally different sensory modalities to drive conduct.

For instance, grownup feminine mosquitoes which are making an attempt to reproduce combine info on human odors, warmth, and carbon dioxide. We all know that these totally different sensory cues enter the mind, however we don’t know the way they’re built-in and converge onto neural circuits that drive mosquitoes’ host-seeking conduct.

We hope that mapping the complete mosquito mind will present a brand new basis for understanding how sensory integration and motion choice works for innate conduct. Moreover, the particular mosquito species we examine is a vector for ailments equivalent to malaria, West Nile, Zika, yellow fever, and dengue fever, so there’s a medical and public well being facet of this that makes it a extremely vital mannequin system.

HMNews: You lately printed a paper in Nature on mind connectivity and sample affiliation in mice. What was the premise of the examine?

Lee: This was a collaboration with Wade Regehr, professor of neurobiology at HMS. The paper focuses on info processing in the cerebellum, which is a mind area that, amongst different issues, is vital for clean, coordinated motion. One of the issues the cerebellum is assumed to do is make fine-scale error corrections in motion by evaluating patterns from supposed and executed actions. For instance, in the event you strive to contact your nostril and also you miss, there may be info coming out of your motor system that tells your cerebellum what the supposed motion was, and there may be sensory info coming out of your finger about what truly occurred, together with the location of your finger in area. The cerebellum is assumed to compute the distinction between the supposed motion and the precise motion, and to assist appropriate the error.

We studied the cerebellar cortex, which is packed full of small neurons known as granule cells that make up greater than half the neurons in the mind. These granule cells every have, on common, 4 dendrites, or branching constructions that obtain info from different neurons. On this case, the dendrites join to neurons known as mossy fibers that carry info into the cerebellum. The granule cells then course of this info and talk it to different neurons known as Purkinje cells, with every Purkinje cell integrating info from 100,000 to 200,000 granule cells, and sending this info to different mind areas. These three cell varieties make up the “feedforward” circuit we wished to higher perceive.

HMNews: What was your key discovering in the Nature paper?

Lee: Beforehand, scientists and computational fashions assumed that the dendrites on granule cells randomly related to totally different mossy fibers, and this randomness contributed to the complexity and encoding capability of the info communicated to Purkinje cells. Nonetheless, utilizing connectomics, we mapped the connections between mossy fibers, granule cells, and Purkinje cells. We discovered that the dendrites on granule cells don’t join to mossy fibers in a random means. As an alternative, they join to mossy fibers selectively, with extra granule cells connecting to the identical mossy fibers than anticipated. This selectivity ought to lower the encoding capability of the info that may be conveyed — but it surely seems that for under a really small lower in capability, you get extra robustness in sample affiliation. We expect it’s because there may be extra redundancy in the connections between granule cells and mossy fibers, and granule cells could also be connecting to more-informative mossy fibers.

This can be a discovering that leverages connectomics to set up extra complete circuit construction by permitting us to have a look at how giant populations of neurons are related to one another in the identical circuit. We want this connectivity info to make detailed and complete fashions of how info flows by the community. This paper demonstrates how connectomics can be utilized to present information to take a look at long-standing theories about info processing and complicated neural networks.

HMNews: What else do you suppose connectomics might help scientists work out?

Lee: One thing that I feel goes to be actually highly effective in the close to future is what individuals are calling “comparative connectomics,” or evaluating totally different connectomes. I’m notably enthusiastic about how behavioral variations throughout people correlate with variations in their connectomes. I’m additionally in evaluating connectomes for various species to see what rules are conserved in totally different sorts of brains. As well as to discovering conserved rules that may be generalized throughout species, I need to discover differentiating rules that make people distinctive. Finally, our widespread humanity could lie in the shared construction of how our brains are wired.

HMNews: Why do you suppose connectomics is such a rising subject?

Lee: Progress has been in half pushed by advances in expertise, together with advances in mechanical engineering that permit us to scale information acquisition, in addition to advances in genetic engineering that permit us to label particular cell varieties. Moreover, the subject has been reworked by machine studying, which can be utilized to analyze these datasets to extract organic perception. The connectomics subject is an fascinating convergence of neurobiology, engineering, computing energy, and synthetic intelligence.

We’ve been growing quite a bit of totally different applied sciences for scaling up information technology and information evaluation that I feel might be helpful in different scientific disciplines. We’re producing some of the greatest picture datasets in the world proper now, and there are extra to come. For instance, the NIH has a objective of mapping an entire mouse mind connectome in the subsequent 10 years, which might be a few zettabyte of information, or a trillion gigabytes. Researchers additionally need to map human and nonhuman primate brains.

We’ve solely scratched the floor of understanding how neurons are related to each other to kind useful networks, however connectomics is arguably remodeling neuroscience. I consider we’re on the cusp of understanding circuit mechanisms underlying how neurons and networks of neurons compute. We’re on the precipice of understanding the primary constructing blocks of neural networks, together with the guidelines by which they join to each other and the guidelines that underlie the computations they perform. To me, that’s actually, actually thrilling.

Authorship, funding, disclosures

Extra authors on the paper embrace Tri Nguyen, Jeffrey Rhoades, Xintong (Cindy) Yuan, Logan Thomas, Ilaria Ricchi, and David Hildebrand of HMS; and Jan Funke and Arlo Sheridan of the Howard Hughes Medical Institute.  

Help for the analysis was supplied by the Nationwide Institutes of Well being (R21NS085320; RF1MH114047; R01MH122570; R35NS097284), the Bertarelli Program in Translational Neuroscience and Neuroengineering, The Stanley H and Theodora L Feldberg Basis, and the Edward R. and Anne G. Lefler Heart.

This interview was edited for size and readability.

Paper DOI: 10.1038/s41586-022-05471-w

Paper URL: https://www.nature.com/articles/s41586-022-05471-w 


What's Your Reaction?

hate hate
0
hate
confused confused
0
confused
fail fail
0
fail
fun fun
0
fun
geeky geeky
0
geeky
love love
0
love
lol lol
0
lol
omg omg
0
omg
win win
0
win
The Obsessed Guy
Hi, I'm The Obsessed Guy and I am passionate about artificial intelligence. I have spent years studying and working in the field, and I am fascinated by the potential of machine learning, deep learning, and natural language processing. I love exploring how these technologies are being used to solve real-world problems and am always eager to learn more. In my spare time, you can find me tinkering with neural networks and reading about the latest AI research.

0 Comments

Your email address will not be published. Required fields are marked *