Mapping the brain’s connections — the conundrum of modern Neuroscience
The networks in our brain are long enough to circle the world four times, but we’re just starting to map them fully
Networks permeate every aspect of our lives, from social media to tracking traffic patterns to finding the best-recommended restaurant in a new city. However, the most mysterious of them all arguably can be found inside our skulls. The complexity and density of networks formed by brain cells, or neurons, makes even the most complicated human-made network pale in comparison. However, when it comes to making sense of this complexity of neuronal networks, opinions within the Neuroscience community differ on whether mapping all the brain’s connections is a fruitful approach to explain the brain’s function. After all, modern Neuroscience aims to ultimately gain an understanding of the mind, consciousness, and humanity, but strategies on how to achieve this goal range from investigations at the molecular-to the behavior scale.
Imagine you round our earth at the equator four times, you will have completed a distance of around 160,000km
Can you think of any network that could possibly encompass such a vast distance when measuring all the connections? It may come as a shock that the cumulative length of myelinated nerve fibers in the human brain ranges from 150,000–180,000 km. Impossible, you say? Now the bustling field of Neuroscience known as “Connectomics” aims to convince you with hard facts by mapping every single connection between neurons in the brain.
Much like an electrical circuit on a computer chip, the high density of connections in the brain have to be organized with sophisticated precision, with little margin for error
Molecular programs orchestrate neuronal circuit assembly during brain development. These programs are endowed with a great responsibility to ultimately construct the proper wiring diagram of the networks in our brain, which directly related to how we function and interact with the outside world.
Highway connections that run from our sensory organs or extremities to our brain determine how we process what we see, smell, taste, or where we decide to go. What’s more, a wealth of research shows that differences in wiring diagrams underlie a wide spectrum of neurological diseases, from neurodevelopmental diseases, such as autism spectrum disorders and schizophrenia to neurodegenerative diseases, such as Alzheimer’s or Parkinson’s Disease.
Thus, connectomes have to properly form, adjust to allow flexibility, but are also susceptible to degeneration, making them the common denominator in many neurological diseases
Hence, it comes as no surprise that Neuroscientists have spent significant time, money, and “brain” power on trying to reconstruct these wiring diagrams in model organisms ranging from worms to flies, to mice, and to humans in order to gain insight into healthy and diseased wiring diagrams of the brain.
This year, 2020, was an exciting year for this thriving field of Connectomics. Scientists at the Janelia Research Campus, together with Google, published the fruit fly “hemibrain connectome” termed FlyEM. Therein, 25,000 neurons and their connections in a part of the fruit fly brain were reconstructed at excruciating detail using a technique called FIB-SEM (Focused Ion-Beam Scanning Electron Microscopy), which allows viewing brain tissue structures in three dimensions at a resolution of less than 10nm. For comparison, that is 1/1,000th the size of a human hair, but now imagine that embedded in an extremely dense jungle of fibers of all the different cell types of the brain.
Imagine trying to untangle those headphones from your pocket, but only millions of times more complex, and smaller.
With structure and function thought to be intricately linked, connectomics bears the promise of being able to determine the relationship between behavior and the wiring of the brain
However, during its advent, connectomics has also sparked an active debate that embodies and exemplifies a somewhat soul-searching moment in modern Neuroscience. With such big data projects being the norm rather than the exception in Neuroscience, the question is how much understanding can be gained from describing all the connections in an organism’s brain or in extension, the activity of all neurons in its brain. The realization being that a complete snapshot of connections alone may still be insufficient to fully explain behaviors.
This year, 2020, provided an elegant response to this conundrum when researchers at the Cold Spring Harbor Laboratory published an approach called BRICseq. BRICseq combines multilevel maps including gene expression data, mesoscale connectomes (i.e. the medium-scale anatomical connections between different brain regions) with a functional map of neuronal activity. The researchers establish a new approach that uses genetic barcodes introduced into neurons to identify how brain regions are connected and how such a wiring diagram correlates with and even predicts functional connectivity maps or behavior.
This debate tackles a central tension in the quest for scientific discovery; which of either hypothesis-driven or hypothesis-generating research is superior to the other
Certainly, there are examples when both approaches cross-pollinate, for example when microscale connectomics identified a new circuit in the fruit fly brain, the function of which could then be tested using optogenetics.
Optogenetics is yet another revolutionary tool that allows Neuroscientists to use light to turn neurons on and off. Then, we could argue, have we not shown a causal relationship between behavior and structure-function relationships in the brain?
The question is, what is the most insightful level that will lead to an understanding of the brain?
The nano-, micro-, meso- or macroscale? Or will understanding only be gained from deriving the fundamental principles of the brain in a theoretical framework, akin to theoretical physics in the 20th century? After all, Einstein’s general theory of relativity published in 1915 made bold predictions that could only experimentally be verified several years later.
Ultimately, all these reflections converge on the underlying question that many Neuroscientists find themselves wondering about: will we be able to unify modern Neurosciences from the nano- to the macroscale, from experiments to theory?
Despite this philosophical debate, there are many other aspects to connectomics that illustrate important advances in Neuroscience
The field is notoriously hard to commercialize, owing to our still limited understanding of the basic principles of how the brain works.
“Bridging a gap between two different fields — such as biology and computer science — that addresses a big need is where innovation and business opportunities arise”, says Dr Yusuke Hirabayashi, an Associate Professor at the University of Tokyo, who collaborates with a startup company on deep learning algorithms that can be applied to the reconstruction of connectomes from electron microscopy data.
Precisely, Dr. Hirabayashi investigates the “intracellular connectome”, studying in exquisite detail how different organelles (the “organs” of a cell) form sophisticated three-dimensional interactions and networks in neurons — yet an even smaller scale of connectomics. Indeed, the application of algorithms specializing in feature recognition, extraction, and prediction have been existential to the success of connectomics, as well as smart data storage solutions that can handle the petabytes of data that result from dense EM reconstructions.
Interestingly, this is not the end of the many ways that connectomics impacts innovation and democratization in scientific discovery. Another remarkable advance that came out of connectomics is the emergence of amazing citizen science projects that provide a crowd-sourced solution to big data problems. Therein, big Neuroscience problems, such as the connectome of our Retina, protein folding, or even clinical datasets are gamified in order to attract citizen scientists to help solve outstanding questions and also serves scientific dissemination.
Another remarkable advance that came out of connectomics is the emergence of amazing citizen science projects that provide a crowd-sourced solution to big data problems.
Many outstanding promises remain that connectomics may provide an unprecedented solution to
In fact, several diseases are co-coined “connectopathies” referring to changes in connectivity as the common denominator of multi-faceted clinical diseases from Neurodevelopment to Neurodegeneration.
The above-referenced BRICseq approach shows the use of multi-level connectomics to detect changes in connectivity in a mouse model of autism spectrum disorders that, amongst others, lack the nerve tracts that connect the left and right hemispheres of the brain (corpus callosum).
Another recent study in humans shows the usefulness of connectomic analyses in predicting how well patients will respond to antidepressive treatment. Thus, hopes are rising to use analyses of connectomes more broadly as clinically relevant biomarkers. It is important to point out that in a translational context, in which connectomic research has to remain non-invasive for safe use in humans, rely mostly on magnetic resonance imaging (MRI) technology, with much lower structural resolution than can be used in animal models.
Pioneering new multilevel correlation maps in humans is the Allen Brain Institute in Seattle, as part of the Human Connectome Project, which aims to integrate gene expression data, MRI-level imaging with gene expression data, and cellular level analysis from ex vivo human brain samples. A lot of potential lies in combining such big data and connectivity map analyses, as it might provide breakthrough opportunities to understand disease pathology comprehensively, offering a departure from the one-gene-one-target approach.
Lastly, I want to leave the readers with a reflection on the number of resources that went into reconstructing just a tiny part of the brain of one single fruit fly at the microscale
This begs the question as to whether and how these approaches may ever be applied on a large scale to reflect individual differences in brain wiring.
Individuality is an important aspect not only in humans but even in fruit flies, as a recent study exemplifies. While much of the debate has focused on the dichotomy between genes or experience to determine behavior (the age old nature vs. nurture debate), this study beautifully shows how random events (“noise”) during brain development lead to measurable differences in behaviors in adult fruit flies.
Maybe, one day, we will all be able to reconstruct our individual connectomes across scales and also understand what it means? Will we be able to track how and why our brain behaves a certain way and where our memories are stored?
That certainly is a big question. One we’ll leave for another day.
Heike Blockus obtained her PhD in Neuroscience from the Sorbonne University in Paris, France, and is currently an Associate Research Scientist at Columbia University in the Department of Neuroscience at the Zuckerman Mind Brain Behavior Institute in New York City.