Read an Excerpt
The Future of the Brain
Essays by the World's Leading Neuroscientists
By Gary Marcus, Jeremy Freeman PRINCETON UNIVERSITY PRESS
Copyright © 2015 Gary Marcus and Jeremy Freeman
All rights reserved.
ISBN: 978-1-4008-5193-5
CHAPTER 1
BUILDING ATLASES OF THE BRAIN
Mike Hawrylycz
With Chinh Dang, Christof Koch, and Hongkui Zeng
A Very Brief History of Brain Atlases
The earliest known significant works on human anatomy were collected by the Greek physician Claudius Galen around 200 BCE. This ancient corpus remained the dominant viewpoint through the Middle Ages until the classic work De humani corporis fabrica (On the Fabric of the Human Body) by Andreas Vesalius of Padua (1514–1564), the first modern anatomist. Even today many of Vesalius's drawings are astonishing to study and are largely accurate. For nearly two centuries scholars have recognized that the brain is compartmentalized into distinct regions, and this organization is preserved throughout mammals in general. However, comprehending the structural organization and function of the nervous system remains one of the primary challenges in neuroscience. To analyze and record their findings neuroanatomists develop atlases or maps of the brain similar to those cartographers produce.
The state of our understanding today of an integrated plan of brain function remains incomplete. Rather than indicating a lack of effort, this observation highlights the profound complexity and interconnectivity of all but the simplest neural structures. Laying the foundation of cellular neuroscience, Santiago Ramón y Cajal (1852–1934) drew and classified many types of neurons and speculated that the brain consists of an interconnected network of distinct neurons, as opposed to a more continuous web. While brain tissue is only semitranslucent, obscuring neuronal level resolution, a certain histological stain Franz Nissl (1860-1919) discovered, and known as the Nissl stain, can be used to stain negatively charged RNA in the cell nucleus in blue or other visible colors. The development of this stain allowed the German neuroanatomist Korbinian Brodmann (1868–1918) to identify forty-three distinct regions of the human cerebral cortex based on cytoarchitectural organization using this Nissl stain. These pioneering works of Brodmann, Constantin von Economo, Marthe Vogt, and others mapped cyto- and myeloarchitectural landscape of the human cortex based on painstaking visual inspection and characterization of a few observable cellular properties such as cell shape, density, packing, and such.
Since Vesalius, most atlases of the brain have been drawn on paper, with the most recent versions in vivid color delineating hundreds of structures. Such atlases have been drawn for most of the important model organisms studied in the laboratory and provide key bench-side experimental references. As with most aspects of modern biology, however, technology has been a driving factor in improved understanding of brain organization. Neuroimaging techniques evolved over the last twenty years have now allowed neuroscientists to revisit the subject of brain mapping, with the modern brain atlas more akin to a digital database that can capture the spatiotemporal distribution of a multitude of physiological and anatomical data. Modern techniques such as magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), diffusion MRI, magnetoencephalography (MEG), electroencephalography (EEG), and positron emission tomography (PET) have provided dramatic improvements in brain imaging for research, clinical diagnosis, and surgery. Digital atlases based on these techniques are advantageous since they can be warped, mathematically or in silico, to fit each individual brain's unique anatomy.
The origin of modern brain mapping for clinical use lies with the seminal work of Jean Talairach, who in 1967 developed a 3D coordinate space to assist deep brain surgical methods. This atlas was generated from two series of sections from a single sixty-year-old female brain, and was later updated by Talairach and P. Tournoux in a printed atlas design for guiding surgery. Today biomedical imaging forms a crucial part of diagnosis and presurgical planning, and much time and resources are invested in the search of imaging biomarkers for diseases. Atlases have been used in image-guided neurosurgery to help plan "stereotaxic," that is, coordinate referenced, neurosurgical procedures. Using this data, surgeons are able to interpret patient-specific image volumes for anatomical, functional, and vascular relevance as well as their relationships.
The field of digital atlasing is extensive and includes high-quality brain atlases of the mouse, rat, rhesus macaque, human, and other model organisms. In addition to atlases based on histology, magnetic resonance imaging, and positron emission tomography, modern digital atlases use gene expression, connectivity, and probabilistic and multimodal techniques, as well as sophisticated visualization software. More recently, with the work of Alan Evans at the Montreal Neurological Institute and colleagues, averaged standards were created such as the Colin27, a multiple scan of a single young man, as well as the highly accessed MNI152 standard. While inherently preserving the 3D geometry of the brain, imaging modalities such as MRI, CT, and PET do not usually allow for detailed analysis of certain structures in the brain because of limitations in spatial resolution. For this reason it is common to use very high-resolution 2D imaging of in vitro tissue sections and employ mathematically sophisticated reconstruction algorithms to place these sections back into the 3D context of the brain.
Today digital brain atlases are used in neuroscience to characterize the spatial organization of neuronal structures, for planning and guidance during neurosurgery, and as a reference for interpreting other data modalities such as gene expression or proteomic data. One ultimate aim of neuroscientific inquiry is to gain an understanding of the brain and how its workings relate to activities from behavior to consciousness. Toward this end, brain atlases form a common coordinate framework for summarizing, accessing, and organizing this knowledge and will undoubtedly remain a critical-path technology in the future.
The Genetic Brain
The development of the techniques of modern molecular biology and eventually whole genome sequencing opened the door for understanding the genetics of the brain, and new perspectives on the study of brain anatomy are emerging with the availability of large-scale spatial gene expression data. The brain consists of at least several hundred distinct cell types whose complete classification is still at present elusive. Each cell type is related to its function with its gene expression pattern, for example, on/off, high/low, as a key determinant. Gene expression data can be collected through a variety of techniques, and exploration of these data promises to deliver new insights into the understanding of relations between genes and brain structure.
Early gene expression studies used methods such as northern blots, which combine electrophoresis separation of RNA molecules followed by hybridizing probes for detection. At one time this method was the gold standard for confirming gene expression, but it ultimately gave way to more quantitative methods. The microarray revolution dramatically increased our ability to profile genes by hybridizing many gene probes on a single gene chip. Today rapid digital sequencing technology can count individual RNA fragments that can subsequently be mapped back to the genome once it is known for an organism.
In 2001, Paul Allen, cofounder of Microsoft, assembled a group of scientists, including James Watson of Cold Spring Harbor Laboratory and Steven Pinker, then at MIT, to discuss the future of neuroscience and what could be done to accelerate neuroscience research. During these meetings the idea emerged that a complete 3D atlas of gene expression in the mouse brain would be of great use to the neuroscience community. The mouse was chosen due to the wealth of existing genetic studies and for practical reasons. Of the potential possible techniques, the project chose a technique for mapping gene expression called in situ hybridization (ISH) (automated by Gregor Eichele of the Max Planck Institute and colleagues), which uses probes that bind to mRNA within sectioned but intact brain tissue and thereby preserves spatial context (see color plate 1).
In 2006, an interdisciplinary scientific team at the Allen Institute for Brain Science, funded by Paul Allen and led by Allan Jones, delivered the first atlas of gene expression in a complete mammalian brain, publically available online at www.brain-map.org. Since then, the Allen Institute has expanded its projects to provide online public resources that integrate extensive gene expression, connectivity data, and neuroanatomical information with powerful search and viewing tools for the adult and developing brain in mouse, human, and nonhuman primate (see figure 3 for an example). In addition to the data there are colorimetric and fluorescent ISH image viewers, graphical displays of ISH, microarray and RNA sequencing data, and an interactive reference atlas viewer ("Brain Explorer") that enables 3D navigation of anatomy and gene expression across these datasets. (Approximately fifty thousand users worldwide access the Allen Brain Atlas resources each month.) Scientists have mined the atlases to search for marker genes in various brain regions associated with diseases, to identify different cell type markers, to delineate brain regions, and to compare gene expression data across species.
Extending this work to humans, the Allen Human Brain Atlas was made public in May 2010 and is the first anatomically comprehensive and genome-wide, three-dimensional map of the human brain. This transcriptional atlas of six adult human brains contains extensive histological analysis and comprehensive microarray profiling of several hundred precise brain subdivisions and has revealed that gene expression varies enormously by anatomical location, with different regions and their constituent cell types displaying robust molecular signatures that are highly conserved between individuals.
In particular, these data show that 84 percent of all genes are expressed somewhere in the human brain and in patterns that while complex are substantially similar from one brain to the next. The analysis of differential gene expression and gene coexpression relationships demonstrates that brain-wide variation strongly reflects the distributions of the major cell types such as neurons, oligodendrocytes, astrocytes, and microglia, all of which are essential to brain function. Interestingly, the neocortex displays a relatively homogeneous transcriptional pattern but with distinct features associated selectively with primary sensorimotor cortices and with enriched frontal lobe expression. Interestingly, the spatial topography of the neocortex is strongly reflected in its molecular topography, that is, the closer two cortical regions are, the more similar their gene expression patterns remain.
Several other significant efforts toward understanding the genetic basis of brain organization are underway, including the Edinburgh Mouse Atlas Project (EMAP) (www.emouseatlas.org), which contains substantial spatial and temporal data for mouse embryonic development, and the Rockefeller University–based GENSAT project of Nathanial Heintz and colleagues that seeks to characterize gene expression patterns using Bacterial Artificial Chromosomes (BAC) in genetically modified mice (www.gensat.org), as well as BGEM (www.stjudebgem .org), GenePaint (www.genepaint.org), EurExpress (www.eurexpress .org), and MGI (http://www.informatics.jax.org), all generally user friendly with useful tutorials.
A Standard Brain?
Does a standard or normal brain exist? This is less likely for humans than genetically bred mice, but mapping neuroscientific and clinical data onto a common frame of reference allows scientists and physicians to compare results between individuals. One main reason for standardization is that multiple and diverse brains can be transformed into a standard framework that maximizes our ability to understand their similar features. Another is that it allows us to identify how unique or unusual features in a particular brain may differ from an average population. With modern advanced image processing capabilities, digital atlases can serve as the framework for building standard atlases and for surveying the information linked to it. In contrast to basic data repositories, which allow for simple access to data through a single interface, sophisticated digital atlases backed by appropriate technology can act as hubs facilitating access to multiple databases, information sources, and related documents and annotations. These may act as a scaffold from which to share, visualize, analyze, and mine data of multiple modalities, scales, and dimensions.
Many of these ideas of standardization grew out of a major initiative of the National Institutes of Health in the 1990s called the "Decade of the Brain," where a number of digital and electronic resources were created to enable the unification and integration of the various subfields of neuroscience. One outcome of this work is the field of neuroinformatics, or the application of computer- and mathematical-based technologies to organize and understand brain data. The ultimate goal of neuroinformatics is to bring together brain architecture, gene expression, and 2D and 3D imaging information into common frames of reference. Major organizations have evolved around mapping brain data, such as the International Consortium for Human Brain Mapping (www.loni.ucla .edu/ICBM/About) and the International Neuroinformatics Coordinating Facility (INCF, www.incf.org). These efforts have led to atlases such as the standard Talairach Atlas and the Montreal Neurological Institute (MNI) standard that have been extensively used in neuroscience.
One consideration in standardizing brain atlases is the type of coordinate system used. As Alan Evans of the Montreal Neurological Institute remarks, "The core concept within the field of brain mapping is the use of a standardized or 'stereotaxic' 3D coordinate framework for data analysis and reporting of findings from neuroimaging experiments. This simple construct allows brain researchers to combine data from many subjects such that group-averaged signals, be they structural or functional, can be detected above the background noise." The concept of a coordinate system is fundamental to digital atlases and requires two basic components: the specification of an origin in the stereotaxic space and a mapping function that transforms each 3D brain from its native coordinates to that of the atlas. A major step in addressing these issues, and a standard tool set that allows different types of neuroscience data to be combined and compared, is now in development for one of the most important subjects in experimental neuroscience: the mouse, Mus musculus. This project is an international collaboration in digital atlasing and is sponsored in part by the International Neuroinformatics Coordinating Facility (INCF).
(Continues...)
Excerpted from The Future of the Brain by Gary Marcus, Jeremy Freeman. Copyright © 2015 Gary Marcus and Jeremy Freeman. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.