National Academies Press: OpenBook

Physics of Life (2022)

Chapter: 3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?

« Previous: 2 How Do Living Systems Represent and Process Information?
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

3

How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?

One of the great triumphs of science in the 20th century was the enumeration and characterization of the molecular components of life. But much of what strikes us as most interesting about living systems emerges from interactions among many of these molecular components. Much of our behavior as humans happens on the scale of centimeters or even meters. For a single cell, this behavioral scale is on the order of microns, something visible only through a microscope but still a thousand times larger than the nanometer scale of individual molecules. A major thrust of biological physics is to understand how to bridge these scales, from microscopic to macroscopic.

The emergence of macroscopic behaviors from microscopic interactions is not a question uniquely about the physics of life. In an ice cube, for example, individual water molecules interact only with their near neighbors, over very short distances. But if one pushes on an ice cube, touching only the molecules on the surface, molecules on the other side of the cube also move, even though they are separated from our finger by hundreds of millions of other molecules. If we raise the temperature by just a few degrees, the ice melts, leaving liquid water; now our finger passes through the liquid, leaving distant molecules unperturbed. The same molecules behave very differently with just a slight change in conditions, and this difference emerges clearly only with a very large number of molecules.

An important part of physics is devoted to understanding emergent phenomena in matter. There are solids and liquids, but also different kinds of magnets; a chunk of metal can conduct electricity or act as an insulator, and when made very cold it can become a superconductor; complex molecules can form the liquid

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

crystals that were a mainstay of computer displays and electronic watches a generation ago and are still widely used in digital cameras. All of these phenomena, which happen in matter at thermal equilibrium, are described in a single unifying language, statistical mechanics. A profound result of statistical mechanics is that macroscopic phenomena often can be described, quantitatively, by models that are much simpler than the underlying microscopic mechanisms. This emergent simplification empowers us to explore much more complex systems, and it has long been hoped that statistical mechanics would provide a language for describing emergent phenomena in living systems. The past decade has seen extraordinary progress toward realizing this dream. Crucially, this has involved much more than applying what is known from physics to the living world. Rather in each case, the biological physics community has uncovered exciting new questions, focusing attention on how the emergent phenomena of life are different from their counterparts in the inanimate world. A sampling of these efforts is given in Table 3.1.

PROTEIN STRUCTURE, FOLDING, AND FUNCTION

Even single biological molecules are so large that one can think of their structure and function as emerging from the interactions among their many component parts. Proteins, for example, are polymers of amino acids. There are 20 possible amino acids at each site along the polymer chain, and proteins vary from a few dozen to several thousands of amino acids in length. The typical polymer does not have a well-defined structure, behaving instead like a randomly crumpled string. A large class of proteins have the remarkable property of folding into compact structures, and these structures are linked intimately to the functions that these molecules carry out in the living cell.

It is useful to identify at least three different questions about emergence of protein structure from the interactions among amino acids. First, what are the structures of these complex molecules? Second, what is the nature of the mapping between amino acid sequences and protein structures? Finally, and perhaps most deeply, what is it about these molecules that makes it possible for them to fold into well-defined structures, so unlike typical polymers? There has been enormous progress on all these problems, driven both by theory and experiment.

Protein Structures

The determination of protein structures is one of the great success stories in the interaction between physics and biology. These developments trace back to the early years of the 20th century, when it was discovered that X rays scatter from crystals to form a pattern of diffraction spots, and that this pattern can be

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

TABLE 3.1 Emergence of Macroscopic Functions from Microscopic Interactions

Discovery Area Page Number Broad Description of Area Frontier of New Physics in the “Physics of Life” Potential Application Areas
Protein structure, folding, and function 120 Proteins represent a different organizational state of matter than found in the non-living world, evolved for particular functions but also for the more general task of folding efficiently into compact structures. Revolutionary tools for structure determination; energy/entropy tradeoffs; avoiding frustration and glassiness; statistical mechanics of sequences. Protein design.
Chromosome architecture and dynamics 129 Understanding how 10 feet of DNA is packaged into nucleus just one-thousandth of an inch across; dynamics in the packed state. Combining equilibrium polymer dynamics with non-equilibrium drive; chromosomes as a state of matter. New tools for exploring chromosomal rearrangements in cell function and disease.
Phases and phase separation 134 Understanding when and how structural organization inside cells and membranes can emerge spontaneously, as with the formation of oil droplets in water. New phases and droplet configurations in multicomponent systems; active droplets; departures from generic behaviors. Self-assembly.
Cellular mechanics and active matter 139 Understanding how molecular motors and filaments organize to generate coherent movements on the scale of cells and tissues. Hydrodynamic theories for “active matter;” reconstitution of more and more complex examples of self-organization for building a model cell; cell mechanosensing and feedback. Cell and environmental mechanics in tumors and cancer treatment.
Networks of neurons 144 Perceiving the world, moving in response, and remembering what we did all involve the coordinated electrical activity of thousands of neurons in the brain. Neural networks as a source of statistical physics problems; new experimental tools to monitor large numbers of neurons; data analysis to connect theory and experiment; the connectome. Brain imaging; artificial intelligence.
Collective behavior 152 Understanding the beautiful, coordinated movements of birds in a flock, insects in a swarm; emergence of “construction projects” in social insects. New universality classes; direct inference of statistical physics models from data; non-classical modes of ordering. Active matter; coordination of autonomous vehicles.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

analyzed to reconstruct the positions of individual atoms in the crystal. Not long after this discovery, it was shown that protein molecules could be crystallized, and in some cases they could still carry out their functions as enzymes while in the crystalline state. It was even possible to crystallize viruses and show that they retained their virulence once the crystal was dissolved. These results raised the possibility that one could use X-ray diffraction, or crystallography, to determine the positions of atoms in proteins in the same way that it was used to determine the positions of atoms in a salt crystal. But in salt crystals, the repeating units of the crystal have only a handful of atoms, while a single protein molecule has thousands of atoms, and established methods for the analysis of X-ray diffraction patterns did not generalize to this scale. Using X rays to help understand life required not just the application, but a dramatic expansion, of the methods of physics.

It would take decades until the first protein structures were revealed through the analysis of X-ray diffraction patterns, in the late 1950s and 1960s. These structures were not so precise as to reveal the positions of individual atoms, but showed clearly that the polymer of amino acids folded into local or “secondary” structural elements, helices and sheets, as had been predicted theoretically; these elements then pack into the overall globular structure of the protein. Structures of different proteins accumulated slowly, and the resolution of these structures improved (see Figure 3.1). The community of physicists, chemists, and biologists interested in protein structure realized that more open exchange of these results would accelerate progress, and in 1971 founded the Protein Data Bank (PDB) with just seven structures. Fifty years later, the PDB holds more than 170,000 structures, and has provided a model for open science. This exponential expansion of structural data was enabled in large part by the advent of synchrotron light sources.

The analysis of protein structures by X-ray crystallography had a revolutionary impact on our understanding of many processes in living cells, providing a literal scaffolding on which to build explanations of mechanism, but the constraint of crystallizing the proteins remained significant. Not long after the discovery of nuclear magnetic resonance (NMR) it was realized that resonances are sensitive to the structural and chemical environment and thus NMR spectra have a fingerprint of structure, but this is hard to extract. But when the magnetic moment of one nucleus is excited by radio waves, it “relaxes” by transferring energy to nearby nuclei, and this transfer is very sensitive to the distances between atoms. Thirty-five years after the first theoretical and experimental explorations of these relaxation dynamics, understanding had developed to the point where they were used to determine the structure of a small protein free in solution. An important aspect of this analysis is that, from the start, it provided not a single structure but an ensemble of structures, focusing attention on the flexibility of proteins.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.1 The analysis of X-ray diffraction patterns began in the mid-20th century, and has been developing ever since to allow us a clearer view of protein structures. (A) Early X-ray diffraction pattern from a hemoglobin crystal. The pattern fades out near the rim of the picture, which corresponds to a spacing of 1.8 Å. (B) Reconstruction of the protein structure from data in (A), with heme groups shown as gray discs with oxygen bound. MF Perutz, X-ray analysis of hemoglobin. (C) Modern high-resolution structure of one hemoglobin subunit, focusing on the heme and nearby amino acids. Carbon monoxide (CO) was bound to the heme, and the bond was broken with a flash of light, but at low temperatures in the crystal the CO remains trapped inside the protein. SOURCES: (A–B) From M.F. Perutz, 1963, X-ray analysis of hemoglobin: The results suggest that a marked structural change accompanies the reaction of hemoglobin with oxygen, Science 140:863, reprinted with permission from AAAS. (C) S. Adachi, S.-Y. Park, J.R.H. Tame, Y. Shiro, and N. Shibayama, 2003, Direct observation of photolysis-induced tertiary structural changes in hemoglobin, Proceedings of the National Academy of Sciences U.S.A. 100:7039, Creative Commons License CC BY-NC-ND 4.0.

Most recently, electron microscopy has taken its place alongside X-ray diffraction and NMR as a method for protein structure determination. Electron microscopes have their roots in the early days of quantum mechanics, with the realization that electrons have wavelike properties. There were steady improvements in resolution throughout the 20th century, and many important discoveries about the internal structures of cells. In the early days, viewing biological samples under an electron microscope involved using heavy metal stains to improve the contrast of the images, but the 1970s brought a combination of better microscopes and samples that were hydrated—that is, surrounded by water as in the living cell—and frozen. This began with proteins that naturally form regular arrays, such as the neurotransmitter receptors at a synapse, where the quality of the image was enhanced by averaging over the many repeating units, much as in X-ray diffraction. In the 2010s, there was a “resolution revolution,” driven by better electron sources, more sensitive detectors, and improved analysis methods. Today it is possible to take electron microscope images of hundreds of thousands of individual protein molecules at extremely low (cryogenic) temperatures, combining these to resolve the underlying structure with a precision sufficient to trace the protein chain, as in Figure 3.2.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.2 Combining electron microscope images of many single protein molecules provides enough information to define the molecular structure completely, resolving the positions of each individual atom. Shown here is a simplified Cryo-EM data processing workflow. From left to right, motion-corrected and summed image of mouse apoferritin in vitreous ice, two-dimensional class averages of apoferritin particles, three-dimensional sorting and subsequent “GoldStandard” refinement of the “best” particles, and zoomed in view of the apoferritin EM density (gray mesh, Electron Microscopy Data Bank ID1638) with the atomic model shown as sticks (Protein Data Bank ID 7A4M); carbons colored white, nitrogens colored blue, and oxygens colored red. SOURCE: Courtesy of Mark Herzik, University of California, San Diego.

Being able to see the structures of proteins at atomic resolution has completely changed how the scientific community thinks about the molecular mechanisms of life; many of these implications are explored in Part II of this report. Although rooted in physics, X-ray diffraction, nuclear magnetic resonance, and cryogenic electron microscopy have combined into a field known as structural biology and absorbed into the biological sciences more broadly. Some measure of the impact of these developments is in the stream of Nobel Prizes recognizing many of the breakthroughs: the first structures of proteins and other complex biological molecules (1962, 1964); the use of electron microscopy to solve structures with repeating units, as with the many proteins in certain viruses (1982); the development of NMR methods for structure determination (1991, 2002); and the development of cryogenic electron microscopy (2017). Alongside the recognition of methods, there is recognition of what has been learned from these methods about particular mechanisms, including photosynthesis (1988), the function of ion channels in the cell membrane (2003), the transcription of DNA into messenger RNA (2006), the translation of messenger RNA into proteins at the ribosome (2009), and amplification in cellular signaling (2012).

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

Folding

The extensive knowledge of protein structures provides a solid foundation from which to ask questions about how and why these structures emerge. The 20 types of amino acids come in two broad classes, the “hydrophilic” ones that interact favorably with water and the “hydrophobic” ones, which (like oil) do not. Protein structures pack the hydrophobic amino acids into the core, leaving a shell of hydrophilic amino acids to interact with the surrounding water. Although the interactions are complex, this suggests that the driving force for protein folding into compact structures is the hydrophobicity of the core amino acids.

While hydrophobic amino acids end up packed in the interior of proteins, much as a drop of oil separates from surrounding water, these amino acids are not all close to one another along the polymer chain. If the polymer is crumpled at random, there are many ways for hydrophobic and hydrophilic amino acids to end up as neighbors. More subtly, if the wrong pairs of hydrophobic amino acids come into contact, this favorable interaction could prevent the rest of the polymer from folding into a compact structure. This suggests that if hydrophobic and hydrophilic amino acids appeared in random order along the polymer chain, it would be very hard for the protein to fold into a well-defined structure because of competition among the many different possible contacts.

The intuition that random sequences cannot fold was made precise using theoretical approaches to the statistical mechanics of disordered systems (Chapter 5). These methods allow us to predict the behavior of the typical “random heteropolymer.” In such random systems, the competition among many different possible contacts becomes a genuine frustration in which not all amino acids can find favorable neighbors. Imagining the energy of the system as a function of the protein structure, the picture is like that of a rough landscape, with many valleys inside of valleys. As molecules try to move on this landscape, they get stuck in local valleys, unable to find their way over the mountain pass to some ultimately more favorable structure. In a sense that can be made mathematically precise, almost all randomly chosen sequences of amino acids behave in this way. Experimentally, one can synthesize a random sequence of amino acids, and typically it does not fold. But real proteins do.

Evolution must have selected amino acid sequences that avoid the rough landscapes that are typical of random sequences. Certainly, the sequences found in today’s organisms are a tiny fraction of the possible sequences, but it might have been that this is just a consequence of evolution not having had enough time to try out more possibilities. Instead, the physics of the folding problem teaches us that only a tiny fraction of sequences are allowed if proteins are to be functional. This shift from a historical view to a functional view is profound, and is encapsulated in the image of a funnel-like landscape for protein folding, as in Figure 3.3.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.3 If proteins are to be functional, only a tiny fraction of amino acid sequences are allowed. Schematic energy landscape for protein folding, showing how the ensemble of unfolded structures is “funneled” to the unique native structure. Importantly, the molecule never visits states in which large numbers of amino acids form incorrectly, as opposed to “native” contacts. This behavior is atypical of polymers, and correspondingly only a tiny fraction of all possible amino acid sequences can become folded, functional proteins. SOURCE: Reprinted by permission from Springer: C.M. Dobson, 2003, Protein folding and misfolding, Nature 426:884–890, copyright 2003.

One can idealize the problem of avoiding rough landscapes by saying that evolution selects for sequences that minimize frustration, sculpting the energy landscape into a funnel that guides proteins to their final folded configuration. Energy landscape theory has created a formalism that quantifies the funnel picture and provides a candidate principle for the dynamics of folding and, importantly, makes detailed predictions, in quantitative agreement with experiments. While X-ray crystallography, NMR, and cryo-EM are the methods needed to determine the final folded structures of proteins, probing the dynamics of folding requires a wider variety of methods involving single molecule experiments and microfluidic devices, atomic force microscopy, several optical spectroscopies including fluorescence resonance energy transfer (FRET) and circular dichroism, and a variety of imaging approaches.

This discussion has emphasized the conceptual problems of protein folding, how and why well-defined structures emerge from interactions among amino acids. But there also are practical versions of these problems. If a new protein is discovered and its amino acid sequence is determined, is it possible to predict the structure? This is the usual formulation of the “protein folding problem.” Conversely, if there

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

is a function that would be useful to implement, is it possible to design a sequence that will realize the required protein structure? This is the “protein design problem.” These questions are discussed in Chapter 6, which describes the connections among biological physics and molecular and structural biology. Applications of these ideas in the search for proteins with engineered functions are discussed in Chapter 7. As will be emphasized in Part II, these more practical formulations of the folding problem have had recent and dramatic input from artificial intelligence. This seems a good place to note the continuing importance of experiments on protein structure, both to explore uncharted territory in the universe of possible structures and to probe structural fluctuations and their dynamics, especially as we gain more appreciation for the functional importance of partially disordered proteins.

Statistical Mechanics in Sequence Space

Although the typical random sequence does not fold, neither does every single amino acid along the polymer chain need to be chosen correctly in order for the protein to fold into a particular structure. Looking across the tree of life, and sometimes even within a single organism, there are many proteins with related but not identical sequences, and the different proteins in these families fold into very similar structures. If these structures are stabilized by contacts among particular amino acids, and by the need to avoid competing interactions, then when mutations occur, changing the identity of one amino acid along the chain, it is reasonable to expect that evolution will select for compensating mutations in other amino acids. This leads to statistical patterns of amino acid covariation that encode physical and functional constraints.

During protein folding, the structure of the molecule changes, while the sequence of amino acids stays fixed. During the evolution of proteins that belong to a well-defined family, the amino acid sequence changes while the structure stays approximately fixed. While for many proteins folding really does correspond to the physical process of coming to equilibrium, and hence will be described by statistical mechanics, the dynamics of evolution in the presence of functional constraints is more complex. Nonetheless, many physicists have explored the idea that there can be a “statistical mechanics in sequence space.” There is a path from the measured covariation in amino acid identities to construct the simplest models that capture these correlations, and these models are equivalent to statistical mechanics problems known as Potts glasses.

The analysis of the Potts glass models, which are determined entirely by the observed sequence variations in protein families, has produced two startling results. First, having constructed a statistical mechanics in sequence space allows the simulation of new sequences in much the same way that one can simulate the positions of molecules in a liquid. But with the modern tools of genetic engineering, these

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

new sequences can be synthesized, and experiments show that, with high probability, they fold and function as do other members of the protein family. Second, the models can be interpreted as describing interactions between the amino acids, and these interactions turn out to be much more spatially restricted than the directly measurable correlations. In many cases, this spatial restriction is enough to identify the amino acids that are in contact, and hence infer the three-dimensional structure of the protein from the patterns of sequence variation alone.

A more general lesson from statistical mechanics in sequence space is that the mapping from sequences to structures is a many-to-one mapping. There is not one sequence that allows a given structure and function, but a whole ensemble of sequences. This idea that functions emerge from microscopic details in a many-to-one mapping is a theme in biological physics, and appears more explicitly in thinking about how organisms navigate the large space of possible parameters that is available to them (Chapter 4). The example of protein folding allows us to see very clearly that functional, living systems do not emerge by setting parameters at random, nor does function require every parameter to be set precisely. Variation in this view is not biological messiness, but rather an exploration of what is allowed by physics. Indeed, looking more closely at the sequence to structure mapping in simplified models, one can show that this level of allowed variation itself is extremely inhomogeneous. There are structures that could be reached by only one sequence, but there are structures even of short proteins that can be reached by thousands of sequences. In this sense, some structures are easier to “design,” and it is plausible that evolution selects for maximally designable structures.

Perspective

Proteins represent a different organizational state of matter than is found in the non-living world, selected by evolution for particular functions but also for the more general task of folding efficiently into compact structures. The great expansion of experimental methods for determining protein structures, now largely exported from biological physics into the structural biology community, encourages us to think more globally about the mapping between sequence and structure. It now is almost possible to state how physical constraints of folding and function shape the ensemble of allowed protein sequences, and it is reasonable to expect that this problem will be solved fully in the coming few years. Beyond compact structures, there are families of intrinsically disordered proteins, whose role in many cellular functions is being appreciated more deeply. All of these developments support a view of life’s functional mechanisms as belonging to ensembles, as in statistical physics. Success in understanding the physics of life requires us to construct these functional ensembles, and progress on protein structure and folding provides a model for the export of this view to function on larger scales.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

CHROMOSOME ARCHITECTURE AND DYNAMICS

The human genome is divided among 23 pairs of chromosomes. If the DNA in these chromosomes, from just one cell, were stretched to its full length, it would be nearly 10 feet long. Along this length, there are more than 20,000 genes and millions of shorter sequences that have been identified as regulatory elements, helping to control which genes are “expressed” as proteins. Thanks to generations of experiments, one can point to the physical location of these many functional pieces along the genome. But despite this detailed information, it still is not known how these regulatory elements, which are often separated from their target genes by thousands to millions of base pairs along the DNA, manage to find and activate their specific target genes. Part of what is missing is the three-dimensional arrangement of these elements in space, as opposed to their one-dimensional arrangement along the genome. Ten feet of DNA is packaged into a nucleus just a thousandth of an inch across, smaller than the thickness of a human hair, and this packing is not random. Genomic regions that are separated by large distances along the DNA may end up in spatial proximity and thus become interacting neighbors. Their interactions, in turn, may activate the transcription of a gene that changes the fate of the cell, enable genomic recombination that leads to the production of an antibody, or result in a recombination event that leads to cancer.

While the structure of proteins is more dynamic than one might expect from simple pictures, this is even more true for the genome, because different segments of the genome can move relative to one another over relatively long distances. Nonetheless, some organizational rules are obeyed. The past decade has brought increasing appreciation for the role of chromosomal structure and dynamics in cellular function. In parallel we have seen glimpses that chromosomes organize themselves into a state quite unlike that of inanimate, polymeric materials, and that this new physics is important for function.

Architecture

How does 10 feet of DNA fit into a box less than one thousandth of an inch across? As shown schematically in Figure 3.4, the double helix of DNA is spooled around protein assemblies known as histone octamers to form nucleosomes. The nucleosome encompasses roughly 150 DNA base pairs, and is small enough that its structure now is known from X-ray crystallography, to atomic resolution. These “beads on a string” condense into fibers, and this can be seen with purified materials in a test tube but it has been challenging to connect these observations to what happens in the living cell. New methods combining chemical labels with electron microscopy, as in Figure 3.4, hold promise, but there still is no full picture connecting the nanometer scale of the nucleosome to micron scale of the nucleus.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.4 The double helix of DNA is packed into the cell nucleus through a series of higher order structures, first looping around proteins, then organizing into fibers and folding. Here, chromatin fibers in the three-dimensional (3D) space of the mammalian cell nucleus are visualized with methods that combine chemical labeling and electron microscopy. Large 3D sampling volumes (rear block) reveal that the vast majority of chromatin in the nucleus is a disordered polymer of 5 to 24 nm in diameter, shown schematically. The polymer is packed at different densities in interphase nuclei and mitotic chromosomes (front block): high density (red); medium density (yellow); and low density (blue). SOURCE: H.D. Ou, S. Phan, T.J. Deernick, A. Thor, M.H. Ellisman, and C.C. O’Shea, 2017, ChromEMT: Visualizing 3D chromatin structure and compaction in interphase and mitotic cells, Science 357:370, reprinted with permission from AAAS.

A very different experimental approach uses chemical methods to crosslink elements of DNA, which are close in space, and then cuts the whole chromosome into small pieces. By sequencing these pieces and comparing the results with the full DNA sequence, one can identify segments of the chromosome that are separated by long distances along the chain but folded to be close in space. Progress in sequencing methods makes it possible to do this in a “genome-wide” survey mode, sampling essentially all of the DNA in an ensemble of cells. An alternative approach attaches fluorescent labels to particular locations along the DNA, and then locates these labels with nanometer precision using super-resolution optical microscopy, building

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

up a map of the chromosome as shown in Figure 3.5. These maps, here for human chromosome 21, illustrate the domains or long segments of the chromosome that are in close proximity, with these domains having relatively sharp boundaries. In addition, the maps are different in different cells. The chromosome is a dynamic structure, and the experiments on single cells capture snapshots of these dynamics.

Some sense for the frontier of this exploration comes from the fact that both the optical methods of Figure 3.5 and the chemical/genomic methods currently are limited to locating segments of the chromosome that are tens of thousands of base pairs in length. This scale is roughly 100× larger than the nucleosome, leaving a substantial gap in our understanding. Importantly, the missing scales overlap the scale of DNA regulatory elements in higher organisms, so what is missing is very relevant for how information flows through genetic networks (Chapter 2) and how cells navigate their parameter space (Chapter 4).

Image
FIGURE 3.5 Super-resolution tracing optical microscopy makes it possible to map the three-dimensional distances between small segments of DNA along a chromosome, one cell at a time. (Top) Many rounds of hybridization label 30 kilobase segments of human chromosome 21, and the labels are then located with super-resolution fluorescence microscopy. (Left) The pseudocolor images of the positions of individual chromatin segments in single cells and the corresponding matrices of intersegment distances reveal domain-like structures with a globular conformation. (Right) The population-average matrix reveals domains at the ensemble level. SOURCE: From B. Bintu, L.J. Mateo, J.-H. Su, N.A. Sinnott-Armstrong, M. Parker, S. Kinrot, K. Yamaya, A.N. Boettiger, and X. Zhuang, 2018, Superresolution chromatin tracing reveals domains and cooperative interactions in single cells Science 362:419, reprinted with permission from AAAS.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

Dynamics

Chromosomes are not static occupants of the nucleus; they constantly wiggle around, and therefore their spatial organization changes over time. Recent developments in microscopy make it possible to follow these movements of individual chromatin regions on sub-second time scales, in live cells and in real time. As chromosomal organization becomes susceptible to these sorts of quantitative experiments, it has become conventional to analyze and interpret these experiments with reference to sophisticated models from physics. More profoundly, these physics approaches to chromosomal organization in specific systems are beginning to reveal general theoretical principles across different systems. Chromosomes have been found to exhibit a highly nontrivial dynamical behavior, whose properties are similar across different species (see Figure 3.6). Ideas from statistical physics have been used to identify the origin of these dynamics and to demonstrate that the robust scaling laws observed for chromosomal motion in vivo can arise from physical principles rather than system-specific biological mechanisms. In particular, trajectories followed by a point along the genome are consistent with anomalous diffusion, governed by the viscoelastic nature of the cellular environment and strongly influenced by the spatial confinement imposed by the hierarchical folding of chromosomal DNA. Evidence is accumulating that these dynamics control crucial processes such as the editing of the genome in the generation of antibody diversity.

While geneticists have known for decades that the genome forms loops, and that the loops bring regulatory elements into close proximity with genes that they control, it was unclear how these loops formed. A scenario of loop extrusion executed by molecular motors, mathematically described by physicists nearly two decades ago, recently began to gain experimental evidence as one of the governing mechanisms of chromosomal compaction.

A long polymer such as the chromosome, which can be crosslinked by other molecules, has multiple possible phases. Ideas from polymer physics have been used to show that recent experiments on chromosomal dynamics are consistent with the chromosome being poised at a special point in its phase diagram, near a sol-gel transition. There are also hybrid theoretical descriptions that treat the chromosome as a flexible polymer but add constraints consistent with measured contacts (as in Figure 3.5). These theories also point to the proximity of chromosomal DNA to phase transitions and hint at their functional role.

Perspective

Understanding chromosomal structure and dynamics would feed into understanding a wide range of life’s essential functions. The advent of powerful new experimental methods is providing unprecedented looks at snapshots and even

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.6 The dynamics of chromosomes. (A) Snapshot of chromosome ends (telomeres) in a mammalian cell nucleus, shown in a two-dimensional projection. (B) Three-dimensional motion of a single telomere. (C) Mean-square displacement versus time, averaged over an ensemble of trajectories, with fixed cells as a control. There is clean scaling over a 100-fold range of times, from seconds to minutes, with an anomalous exponent. SOURCES: (A–B) Reprinted with permission from I. Bronstein, Y. Israel, E. Kepten, S. Mai, Y. Shav-Tal, E. Barkai, and Y. Garini, 2009, Transient anomalous diffusion of telomeres in the nucleus of mammalian cells, Physical Review Letters 103:018102, copyright 2009 by the American Physical Society. (C) Reprinted with permission from S.C. Weber, A.J. Spakowitz, and J.A. Theriot, 2010, Bacterial chromosomal loci move subdiffusively through a viscoelastic cytoplasm, Physical Review Letters 104:238102, copyright 2010 by the American Physical Society.

dynamics of chromosome structure on multiple spatial scales. Gaps in these experiments likely will be filled over the next decade. Existing theories capture different aspects of the existing data, in some cases fitting the data and in other cases trying to derive at least global features of the data from more general principles. A more fully integrated, and more principled, theoretical understanding seems within reach, perhaps even within the next decade. Importantly, there are hints that the dynamic organization of chromosomes is not a simple or generic example drawn from the world of inanimate materials. Rather it seems likely that these crucial building blocks of the cell instantiate new forms of order, exploiting new physics to achieve their functions.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

PHASES AND PHASE SEPARATION

Cells are the smallest basic structural, functional, and biological unit of all known organisms and are capable of independent self-replication. Cells consist of cytoplasm enclosed within a lipid bilayer membrane; the cytoplasm contains water soluble biomolecules such as proteins and nucleic acids that carry out the basic functions of energy production, nutrient uptake and processing, self-replication, and shape control and movement, as well as specialized functions unique to cells in specific tissues such as information processing and transmission, secretion of extracellular signals and structures, or detection and response to threats. To carry out specific activities within this diverse array of functions with spatial and temporal precision, cells organize their contents into specialized subunits known as organelles. Specific organelles were initially discovered by light or electron microscopy, techniques which reveal them as entities distinct from the surrounding cytoplasm. The revolution in chemical preservation (“fixation”) of cellular ultrastructure in the 1960s led to the identification of most cellular organelles known today. The functional autonomy of organelles was validated by the fact that they could be isolated by biochemical fractionation and maintain their activities.

Although the classical textbook picture evolved, the 21st century brought revolutionary changes, grounded in new discoveries about the physics of these systems. It had long been known that purified versions of biological materials had interesting phases and transitions, but except in special cases—such as the behavior of proteins in the lens of the eye—it was never clear that this physics was relevant to the business of life. Over the course of a decade, this has changed dramatically, with novel phases, phase transitions, and phase separation becoming central to discussions of myriad processes in living cells.

Membranes

Physicists have long been interested in the cell membrane as an example of self-assembly. The lipid molecules in these membranes have “oily” tails and charged heads, so they are driven to organize themselves in ways that hide the tail from the surrounding water, exposing only the charged head groups; the bilayer is the simplest structure that does this. This can be reproduced in a test tube, even with just one species of lipid in water. But bilayers typically have at least two distinct phases, one liquid and one more rigid. Lipids in water have a much richer phase diagram, organizing into many different three-dimensional structures, including beautiful labyrinths. These phases, and the transitions between, provide fascinating examples of soft matter physics (Chapter 5), but it has been challenging to connect these phenomena to the life of the cell.

Real biological membranes have several different lipid components in carefully regulated proportions. These multicomponent systems have new axes to their phase

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

diagram, and new phases where droplets or domains of different composition can condense in the membrane. These condensed domains were first observed in model systems made from mixtures of three naturally occurring lipids, then in vesicles formed from membrane extruded by live cells, and finally in fully natural cellular systems such as the yeast vacuole (see Figure 3.7). The transition into the phase with condensed domains is a liquid-liquid phase transition, and there is a critical point at a particular lipid composition. Near criticality, there are fluctuating domains on long length scales, and the spatial and temporal statistics of these fluctuations are predicted theoretically, by general statistical physics principles, with no free parameters; these predictions have been confirmed in detailed experiments on these membrane systems. The surprise is that real biological membranes have lipid compositions close to the critical point.

Image
FIGURE 3.7 The membranes surrounding living cells are composed of many kinds of lipids, allowing for multiple phases and phase separation of droplets or domains. (A) Phase diagram of vesicles formed from mixtures of three lipids, DOPC, DPPC, and cholesterol, at 30°C. Gray region is where condensed domains are found, and images show the appearance of the vesicles near the transition into this region. Images are from a fluorescent probe molecule that partitions differentially between the coexisting phases. Vacuoles in living yeast cells, expressing a fluorescent version of the vacuole membrane protein Vph1. Images are taken with standard wide-field epifluorescence illumination (B) or wide-field illumination with z-sectioning followed by iterative deconvolution (C). SOURCES: (A) Reprinted from S.L. Veatch and S.L. Keller, 2003, Separation of liquid phases in giant vesicles of ternary mixtures of phospholipids and cholesterol, Biophysical Journal 85:3074, copyright 2003, with permission from Elsevier. (B–C) Reprinted from S.P. Rayermann, G.E. Rayermann, C.E. Cornell, A.J. Merz, and S.L. Keller, 2017, Hallmarks of reversible separation of living, unperturbed cell membranes into two liquid phases, Biophysical Journal 1113:2425, copyright 2017, with permission from Elsevier.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

It has long been known that cell membranes have in-plane organization on long length scales, into domains and “rafts” that play a functional role in signaling and other processes; it was assumed that these structures are imposed on the membrane by other mechanisms. Discoveries about phase separation and criticality show that such large-scale organization will happen spontaneously, perhaps needing only to be stabilized by interactions with structure inside the cell. The proximity of the critical point also means that proteins embedded in the membrane will interact with one another over long distances, through the analog of Casimir forces known from other physics problems. The full implications of these results still are being explored, as is the mechanism by which cell membranes become tuned near their critical points. The observation of liquid-liquid phase separation in the two dimensions of a membrane prepares us for the possibility that something similar happens in three dimensions with proteins and nucleic acids in the cytoplasm.

Phase Separation in the Cytoplasm

Understanding the principles that drive the organization of biological molecules into function-specialized machines known as organelles has been largely undertaken by biologists, not physicists. The electron microscopy heyday of the 1960s, with the advent of glutaraldehyde fixation for ultrastructural preservation and staining methods dependent on composition, gave rise to the notion that there were two general classes of organelles: membrane-bounded and non-membrane bounded. Membrane-bounded organelles physically isolate and concentrate specific components in their interiors relative to the bulk cytoplasm, thereby forming reaction vessels containing all the necessary ingredients to perform their task. Phospholipids that make up biological membranes form bilayers in the aqueous cytoplasm by hydrophobic driving forces, and can enclose contents within vesicles, as described above. Some examples of membrane-bounded organelles include the nucleus, which contains the genome wherein genes are transcribed; the endoplasmic reticulum, in which much of RNA translation into protein and protein folding takes place; mitochondria, where oxidative metabolism occurs to generate ATP; and lysosomes, where proteins are degraded and processed into nutrients for cell growth.

Non-membrane-bounded organelles were identified by their lack of specific membrane staining. These include the highly ordered filamentous scaffolds of the microtubule, actin, and intermediate filament cytoskeletons; large semi-ordered macromolecular assemblies such as ribosomes and centrioles; and local concentrations of proteins that appear disordered, variously described by biologists as “bodies,” “aggregates,” “granules,” “clusters,” “plaques,” or “osmophilic clouds.” These include, for example, nucleoli, stress granules, the centrosome, Balbiani bodies, and focal adhesions. Identification of the protein and/or nucleic acid components of the more ordered non-membrane-bounded organelles such as the cytoskeleton and

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

ribosomes, together with structural analysis and in vitro reconstitution, has led to a reasonably high level of understanding of how the assembly of these structures is driven by the same physical principles driving any protein-protein or protein-RNA interaction. However, although the components of the disordered non-membrane-bounded organelles were identified, the disordered nature of their structure and difficulty in achieving in vitro reconstitution made it difficult for biologists to decipher the physical principles driving their highly ordered formation.

It came as a surprise that structurally disordered, non-membrane-bounded organelles form by the process of liquid-liquid phase separation. Two key discoveries led to this idea. First was the direct observation of fluid behavior in one particular organelle, the P granule, through high-resolution microscope movies of living zygotes (see Figure 3.8). The second was the discovery that purified multi-valent proteins or repetitive RNAs that are made up of repeated, low affinity interaction motifs undergo liquid-liquid phase separation into organelle-sized droplets in aqueous solution in vitro (see Figure 3.9). These observations were supported by

Image
FIGURE 3.8 The direct observation of fluid behavior in the “p-granule” of a living embryo was one of two discoveries that some of the organelles identified in classical cell biology are really phase separated liquid droplets. (A) Images of a fertilized egg from the worm Caenorhabditis elegans. A single protein (PGL-1) has been tagged with the green fluorescent protein, and this protein localizes to “P granules,” which eventually become germ line cells. Over the 10 minutes surrounding the meeting of the two pronuclei (pnm = 0), the P granules migrate from the anterior (left, marked A) to the posterior (right, marked P). The embryo is ∼50 µm long. (B) P granules (outlined in red) dripping from a dissected cell, with the nucleus N (outlined in white). This is one of many liquid-like behaviors of the droplets. SOURCE: From C.P. Brangwynne, C.R. Eckmann, D.S. Courson, A. Rybarska, C. Hoege, J. Gharakhani, F. Jülicher, and A.A. Hyman, 2009, Germline P granules are liquid droplets that localize by controlled dissolution/condensation, Science 324:1729, reprinted with permission from AAAS.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.9 The observation that purified multi-valent proteins made up of repeated, low affinity interaction motifs undergo liquid-liquid phase separation into organelle-sized droplets in aqueous solution in vitro was the other key discovery leading to the finding that structurally disordered, non-membrane-bounded organelles form by the process of liquid-liquid phase separation “for highlighted part. Liquid droplets formed from solutions of weakly interacting proteins with repeating units. Differential interference contrast microscopy (A) and wide-field fluorescence microscopy (B), with a small fraction of the SH3 proteins carrying a fluorescent label. Concentrations are well below the affinity measured between individual SH3 and PRRM4 domains, so this is a collective effect. Scale bars: 20 µm. (C) Time-lapse imaging of merging droplets that were formed as in (A). Scale bar: 10 µm. SOURCE: Reprinted by permission from Springer: P. Li, S. Banjade, H.-C. Cheng, S. Kim, B. Chen, L. Guo, M. Llaguno, J.V. Hollingsworth, D.S. King, S.F. Banani, P.S. Russo, Q.-X. Jiang, B.T. Nixon, and M.K. Rosen, 2012, Phase transitions in the assembly of multivalent signalling proteins, Nature 483:336, copyright 2012.

a quantitative description of the relationship between valency, affinity, concentration, and phase separation, which was similar to transitions from small complexes to large, dynamic supramolecular polymers that had been described in non-living systems. Subsequent demonstrations that phase separation actually affects protein activity led to the notion that phase transitions may be used to spatially organize and biochemically regulate information throughout biology.

Perspective

In many of the examples described in this report, learning something essential about the physics of life required looking in places where nobody had looked be-

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

fore, often with new methods. In the case of intracellular condensates, on the other hand, many people had looked through a microscope at these organelles, over many decades. The paradigm-changing discovery that has launched immense progress was that these objects are not what they appear to be, and are governed by different physics. In little more than a decade, this has become a major focus of research not just in the biological physics community but in biology as well, with implications for medicine. These connections are discussed in Chapter 6. In the meantime, progress in understanding the phase behavior of real cell membranes continues. It took decades to realize that these membranes are not at generic points in their phase diagrams, and that this may have functional consequences. It remains to be seen if similarly non-generic behaviors are seen in cytoplasmic condensates. More generally, it will be exciting to learn if cytoplasmic condensates take advantage of novel phase separation behaviors that have not been uncovered in the exploration of inanimate systems. Finally, there are new physics questions about phase separation in the fundamentally non-equilibrium environment of the living cell.

CELLULAR MECHANICS AND ACTIVE MATTER

Living cells move. They change shape, divide, crawl over surfaces, and squeeze past each other even in dense tissues. The forces that drive these movements are generated by motor proteins (Chapter 1) such as myosin and kinesin, which act on filamentous proteins such as actin and tubulin. Filaments in turn can be bundled and cross-linked. The result is that the whole collection or proteins and filaments inside the cell, called the cytoskeleton, forms an active medium. These mechanical behaviors of the cell do not stop at the cell membrane. Instead, cells are responsive to the mechanical properties of their surroundings, which can affect their motility, shape and even decisions about which genes to express. The biological physics community has been interested in all these problems, and has had strong interactions with the larger community of cell biologists, as described more fully in Chapter 6.

Activity and Organization

A major effort in the cell biology community has been to purify the protein components of the cytoskeleton and reconstitute their behavior outside the cell. This makes possible very controlled and quantitative physics experiments, characterizing the mechanical behavior of the reconstituted material. As an example, one can put micron-sized beads into the medium, apply controlled forces with optical tweezers (see Box 1.3), and monitor the displacements of these beads. There will be random displacements, but also displacements in response to applied forces. In an equilibrium system, the random or Brownian motions are related, quantitatively,

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

to the responses through the fluctuation-dissipation theorem (FDT). In an active system, one should see violations of the FDT.

Figure 3.10 shows tests of the FDT in a simple reconstituted system of actin and myosin. Remarkably, motors do not change the response of the medium to applied forces, but the spontaneous motions are an order of magnitude larger than thermal motions predicted by the FDT. On one hand, this is unambiguous evidence of non-equilibrium behavior. On the other hand, this shows that the active motions are not enormously larger than thermal motions, at least on micron length scales.

If the microscopic behavior of motors and filaments is understood, one can try to build a theory that averages over these details and describes the densities and flows of molecules on a scale of microns and larger. This is the same spirit as the derivation of fluid mechanics from molecular dynamics, with the difference that now the constituent particles are active. Pioneering efforts to derive these sorts of hydrodynamic theories for “active matter” were motivated by flocks and swarms (Chapter 3), but in the same way that fluid mechanics is the same for many different kinds of molecules, the hydrodynamics of active matter should be the same for all constituents that have the same symmetry properties. As discussed in Chapter 5, active matter now is a lively field of physics independent of its origins in the physics of living systems.

Image
FIGURE 3.10 Observing fundamental signatures of irreversible, non-equilibrium behavior in simplified mixtures of motor proteins. Spontaneous movements of small beads are characterized by their power spectrum C(ω), which measures the amplitude of motion at each frequency ω. Displacements in response to applied forces also can be measured as a function of frequency, to give the response function a(ω). The response function has an elastic component a´(ω) and a viscous components a´´(ω); the fluctuation-dissipation theorem (FDT) connects C(ω) to a´´(ω) and the thermal energy kBT if the system is in equilibrium. (A) In the absence of myosin, or in the first few hours after myosin is added, the FDT is obeyed. (B) After a few hours, the activity of myosin molecules leads to a large violation of the FDT at low frequencies. SOURCE: From D. Mizuno, C. Tardin, C.F. Schmidt, and F.C. MacKintosh, 2007, Nonequilibrium mechanics of active cytoskeletal networks, Science 315:370, reprinted with permission from AAAS.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

An important feature of filaments such as actin and tubulin is that they are polar: Particular species of motor molecules move primarily in one direction along the filament, and even the polymerization of the filament itself is directional. Early theoretical work showed that active polar fluids have defects analogous to those in liquid crystals, including asters and spirals or vortices. This is provocative because such organization of microtubules happens in cells, especially during the complex process of segregating newly copied chromosomes during cell division. These organized structures can be seen in simple reconstituted mixtures of tubulin and kinesin, conforming to the predicted phase diagram, as illustrated in Figure 3.11.

Image
FIGURE 3.11 Protein filaments such as microtubules can organize in ways similar to what happens in liquid crystals, including forming defects with characteristic geometries. Here we see such self-organization in a mixture of microtubules with the motor protein kinesin. (A) Disordered array of microtubules. (B) Addition of modest amounts of the motor kinesin generates a spiral organization of the microtubules. (C) Higher concentration of motors generates asters. (D) Phase diagram for the hydrodynamics of an active polar medium. Axes are the elastic anisotrpy δK and the motor activity ζ∆µ. SOURCE: Reprinted with permission from M.C. Marchetti, J.F. Joanny, S. Ramaswamy, T.B. Liverpool, J. Prost, M. Rao, and R.A. Simha, 2013, Hydrodynamics of soft active matter, Reviews of Modern Physics 85:1143, copyright 2013 by the American Physical Society.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

One direction for this work is to try to reconstitute more and more complex examples of self-organization, perhaps to the point of building something that could be called a model cell. The other direction is to take these ideas back into real cells. Recently, quantitative theories and related experiments based on active matter ideas have addressed questions such as the size and shape of mitotic spindles and the cortical flow leading to polarization of worm embryos.

Connecting to the World

The cytoskeletal networks of filaments and motors inside the cell are linked to the environment outside the cell through integrin protein assemblies (see Figure 3.12). These have the unusual property of catch bonds—unlike most bonds, which weaken and detach more quickly when they are pulled apart, catch bonds strengthen and detach more slowly. Such bonds allow the cells to sense and respond to applied stresses and to the mechanical stiffness of their environments. The extracellular matrix that surrounds cells was long considered a simple passive scaffolding that simply houses and supports cells. The discovery of cell mechanosensing and feedback led to the recognition that the extracellular matrix—the cell microenvironment—has a profound role in regulating cell behavior and mediating interactions between cells.

Image
FIGURE 3.12 The many pathways for mechanics of the cellular environment to influence internal states via integrin molecules at the surface. SOURCE: Reprinted from F.B. Kai, A.P. Drain, and V.M. Weaver, 2019, The extracellular matrix modulates the metastatic journey, Developmental Cell 49:332, copyright 2019, with permission from Elsevier.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

In particular, the stiffness of the microenvironment is important to how cells sense and respond to external forces, through a process called mechanotransduction.

As an example, stem cell proliferation and differentiation processes were long considered to be controlled solely by biochemical cues, but it is now recognized that these processes depend critically on mechanical properties of the constantly evolving microenvironment. The stiffness of the microenvironment helps determine the cell type during stem cell differentiation: Cells in stiff environments differentiate into stiff cells such as those in bone, while cells in soft environments differentiate into soft ones like neurons.

The understanding of cancer has been similarly transformed by the recognition that a tumor is intimately linked to its microenvironment and can be considered in itself as an organ—cells within tumors cannot be understood in isolation. Tumor cells can subvert their microenvironments to promote the tumor; conversely, targeting the microenvironment may be an effective way of inhibiting a tumor. Mechanics are critical to the interaction; breast tumors are found by palpation because they are stiffer than normal tissue. The tumors promote remodeling of the extracellular matrix that stiffens the matrix, further promoting tumor growth, which further promotes stiffening of the matrix in a downward cycle of malignant tumor progression. Even worse, stiffening of the extracellular matrix promotes tumor metastasis, perhaps through local force cues and increasing expression of proteins that promote cell migration, among other mechanisms.

These collective phenomena governing the behaviors of tumors and stem cells, as just two examples, arise from the many-body interactions that link cells to their microenvironments and thereby link cells to each other. Many of these phenomena are explicitly non-equilibrium phenomena. An important step in cancer progression is the epithelial-to-mesenchymal transition (EMT). During this process, interactions among cells and between cells and the extracellular matrix are modified, leading to detachment of epithelial cells from each other and from the underlying substrate membrane so that cells can migrate away and invade normal tissue, seeding metastasis. Thus, this process marks a transition from a solid state, in which cells do not change their neighbors, to a fluid state, in which cells migrate and change neighbors. Once in the fluid state, the system becomes a realization of a classic active matter system, namely a collection of self-propelled particles.

Perspective

Active matter provides a perspective on the emergence of structure and function from interactions among motile components. From this perspective, in looking at the cytoskeleton the “particles” are molecules, while in tissues the particles are cells, but the physical principles are the same. Thus, movements of cells in tissues recapitulate some of the phases and transitions seen for movements in-

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

side single cells, connecting to questions about active matter beyond the living world (Chapter 5). As recently as a decade ago, there was a substantial divide between (roughly) physicists interested in the mechanics of cells and biologists interested in the myriad pathways by which these mechanics are regulated. As experimental tools become more powerful, it becomes possible to explore the physics of the fully regulated mechanical system, with quantitative probes of force, displacement, and signaling molecule concentrations, simultaneously. It will be interesting to see whether these richer systems—as with flocks and swarms (Chapter 3)—have ways to generate behaviors outside the universality classes of conventional active matter theories. Concepts from active matter such as self-organization and self-healing are impacting thinking on engineered micromechanical systems, with much more to be explored.

NETWORKS OF NEURONS

Perceiving the world, moving in response to stimuli, and remembering past events all involve the coordinated electrical activity of thousands of neurons in the brain. It is an old dream of the physics community, dating back into the 1940s and receiving a major stimulus in the 1980s, that this coordination could be understood as an emergent phenomenon in the language of statistical mechanics. Some of the first theoretical ideas were prompted by experiments on the all-or-none nature of the action potential in single neurons, and on the coarse-grained behavior of large numbers of neurons seen in the electroencephalogram (EEG). Subsequent decades have seen great progress in both theory and experiment, as well as in our ability to bring theory and experiment together. Important successes often have become part of the mainstream of neuroscience, but the effort to understand collective behavior in networks of neurons continues to occupy a significant part of the biological physics community, as experimentalists develop new instruments for quantitative exploration of network dynamics and theorists use neural networks as a source of new problems in statistical mechanics.

Observing the Human Brain

Humans have a special interest in the dynamics of their own brains. In addition to the EEG, all of the methods for observing electrical activity in the human brain have had major contributions from the biological physics community. Magnetoencephalography (MEG) measures the magnetic fields that result from coordinated current flow among neighboring neurons in the brain, and the high sensitivity needed for these measurements has led to the use of superconducting quantum interference devices (SQUIDs) as field sensors. EEG and MEG both have high time resolution, but limited spatial resolution.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

The electrical activity of neurons requires energy, and so the consequences of this activity are detectable in blood flow and metabolism. Positron emission tomography (PET) follows, for example, the uptake of radio-labeled glucose molecules that provide the (almost) unique energy source for the brain. For cells to extract energy from glucose requires oxygen, and oxygen binding to hemoglobin in the blood changes the nuclear magnetic relaxation behavior of protons in the surrounding water; this is the basis of functional magnetic resonance imaging (fMRI, Chapter 6). PET and fMRI have high spatial resolution, but the metabolic signals to which they are sensitive are slower than the electrical activity itself.

These physics-based experimental methods have provided dramatic glimpses of our brains in action. Experiments show that the pattern of brain activity when humans imagine an image is very similar to that when they see the image, and in some individuals this is true even in the primary visual cortex. When two people have a conversation, their brain activity becomes synchronized; while the listener’s brain largely follows the speaker’s brain with some delay, specific brain regions have activity that is predictive. These and other observations provide boundary conditions for theories of how the emergent behavior of neural networks underlies our experience of the world.

Functional magnetic resonance imaging has joined with PET scanning, EEG, and MEG to form the field of human brain imaging, and this has become part of the mainstream of neuroscience and psychology (Chapter 6), as well as playing a key role in the understanding of brain injuries and disease (Chapter 7). In many ways this parallels the merger of X-ray crystallography, NMR, and cryogenic electron microscopy into structural biology (Chapters 3 and 6). Physicists continue to improve these technologies—creating better scintillation detectors for PET studies and denser arrays of electrodes for EEG, and increasing the resolution of MRI through the use of stronger magnetic fields and more sophisticated pulse sequences.

Monitoring Many Single Neurons Simultaneously

At the opposite extreme from measuring the EEG is the measurement of electrical activity of single neurons in laboratory animals. The first such experiments in the 1910s strained the sensitivity of instruments in the physics laboratory, and there is continuing input from the physics community into these measurement techniques. Although much has been learned about the brain by studying the responses of single neurons, there was a gap between these measurements and ideas about emergent and collective behavior in networks of neurons. Closing this gap requires monitoring the electrical activity of many individual neurons, simultaneously. Some of the first efforts in this direction were aimed at the retina. Arrays of electrodes deposited on glass allow a relatively straightforward interface to a piece of retina dissected from the eye and kept alive in a dish. By the early 2010s,

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

these methods had advanced to the point where one can monitor almost all of the hundreds of signals that the eye sends to the brain from a small patch of the visual world, giving new opportunities to study how visual information is represented (Chapter 2) and to search for collective behavior in this relatively simple network. More broadly applicable methods adapt semiconductor microfabrication techniques to build electrode arrays that can be inserted into three-dimensional brain tissue. Major efforts to make such tools available to the wider research community were mounted in the late 2010s. As with detector arrays in experimental particle physics, there are challenging problems in transforming the data from multiple electrodes into meaningful signals from multiple individual neurons; in parallel with hardware developments these problems have attracted attention from the physics, applied mathematics, and computer science communities. Today it is possible to resolve hundreds or even thousands of single neurons, and importantly one can track these signals continuously and stably over many months.

During an action potential, the electric field across the cell membrane changes by ∼107 V/m. This large field is enough to generate large changes in the optical properties of molecules in the membrane, and there were efforts dating back to 1970 to use voltage-sensitive dye molecules that would dissolve in the membrane and literally make the electrical activity of neurons visible as a change in fluorescence. Although there were dramatic early demonstrations of activity in large populations of cells, the dyes suffered from various limitations that prevented them from becoming a viable alternative to electrodes.

The idea of recording electrical activity by optical imaging methods received a revolutionary push from the discovery of the green fluorescent protein, described in Chapter 6. These proteins have been modified to have their fluorescence depend on a variety of signals in the surrounding solution, including calcium. Because action potentials trigger an influx of calcium into neurons, which is pumped out (or into internal stores) more slowly, calcium concentration is a signal that traces a temporally smoothed version of electrical activity. Animals genetically engineered to express calcium-sensitive fluorescent proteins in their neurons thus make it possible to visualize electrical activity as a flickering movie of fluorescent signals. To make the most of these signals requires development and deployment of specialized optical methods, including scanning two-photon microscopy, microendoscopy, head-mounted miniature microscopes, and microscopes with adaptive optical capabilities for imaging deeper into brain tissue. The result of these developments is that one can monitor hundreds or even thousands of individual neurons simultaneously, with high signal-to-noise ratio, as illustrated by the experiments in Figure 3.13. These methods are undergoing continual development, with steady progress in the number of individual neurons that can be resolved and the quality of the recordings. There are also fluorescent proteins that insert into the membrane and respond directly to voltage. These methods are on the threshold of general use, which will realize a 50-year-old dream of directly visualizing electrical activity in the brain.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.13 Genetically encoded fluorescent proteins allow us to monitor electrical activity in many neurons simultaneously, at high signal-to-noise ratio. (A) Image of neurons in the CA1 region of a mouse hippocampus (left). The cells express a protein whose fluorescence is sensitive to the calcium concentration, which changes in response to electrical activity. Cell bodies appear outlined because the protein is excluded from the nucleus. Fluorescence images are collected by scanning two-photon microscopy. Selected cells are outlined (right). (B) Fluorescence signals from four cells as the mouse runs along a (virtual) linear track, receiving rewards at the end. Note the low level of background noise. SOURCE: Reprinted by permission from Springer: D.A. Dombeck, C.D. Harvey, L. Tian, L.L. Looger, and D.W. Tank, 2010, Functional imaging of hippocampal place cells at cellular resolution during virtual navigation, Nature Neuroscience 13:1433, copyright 2010.

Theory

Recording the electrical activity of thousands of neurons creates the opportunity to search for collective, emergent behaviors in these connected networks. But such high-dimensional data cannot be explored without some guidance from theory. Theories of neural network dynamics date back to the 1940s, with the first efforts to understand what neurons could compute. This early work showed that arbitrary patterns of connections (synapses) between neurons can generate very complex dynamics. To make progress, two extreme simplifications emerged from the biological physics community. As often the case in the physics of interacting many-body systems, neither of these simplifications are literally correct for networks of neurons in real brains, but both have been powerful sources of ideas.

The first simplification is to imagine the neurons are organized into layers, and that synaptic connections carry signals from one layer to the next, with no

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

feedback. This “perceptron” or feed-forward architecture was proposed around 1960, and reappeared in the 21st century as the foundation for the deep network revolution in artificial intelligence, as explained in Chapter 7. The second, alternative simplification is to imagine that all synaptic connections are symmetrical, in which case the dynamics of the network are equivalent to motion on an energy landscape. In both cases, ideas from statistical physics play a key role in the analysis; more deeply, these model neural networks have been the source of new statistical mechanics problems.

An essential step in the theoretical analysis of neural networks is to think not about a particular network, in which there is some specific pattern of connections among all the neurons, but rather about the behaviors that are expected in an ensemble of networks, where the patterns of connections are chosen from some distribution. In the limit that networks are large, there is self-averaging, so that a single network becomes typical of the whole ensemble. This approach connects the theory of neural networks to the statistical physics of disordered systems such as glasses and spin glasses (Chapter 5). Indeed, the symmetric model of neural networks maps exactly to a novel family of spin glasses.

In the symmetric model of neural networks, the dynamics in the absence of noise is just a downhill slide on an energy landscape. The network stops at local minima of the energy, or attractors. In the first proposal, these dynamics were envisioned as a model for the recall of a memory; a cue for recall would initialize the network in the basin of attraction for one memory, and the dynamics would recover that memory. The structure of the landscape, and hence the stored memories, depends on the detailed pattern of connections between neurons; the network can be “programmed” by changing the strengths of these synapses, sculpting the energy landscape. Importantly, in certain limits this programming can turn the current state of the network into an attractor by changing the synaptic connection between two neurons solely in relation to the activity of those two neurons. More generally, many problems that the brain has to solve—and many classical problems in the theory of computational complexity—can be formulated as minimizing a cost function and mapped into the dynamics of a network.

The symmetric model of neural networks thus connected brain dynamics, statistical physics, computational complexity, and the rules for synaptic plasticity. As the 20th century ended, many of these connections were solidified, for example, the use of statistical physics methods to identify phase transitions in large computational problems, and to understand the conditions under which these problems become hard (Chapter 5). For the collective behavior of neurons in real brains, perhaps the most important prediction of these models is that memories are stored in locally stable patterns of activity distributed across the whole network, patterns in which the activation of each neuron is maintained self-consistently by the activity of the other neurons. This gives us a mathematically precise version of

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

classical ideas about reverberating activity, and connects directly to a large number of experiments that probe persistent activity of neurons under conditions in which animals remember and compare distinct sensory inputs. In both symmetric and feed-forward architectures, there is a notion of capacity for the network, and this capacity depends on the distribution of synaptic strengths. Maximizing capacity leads to nontrivial predictions for this distribution, which agree with experiment, including the large number of silent or nearly silent connections.

Beyond the symmetric and feed-forward simplifications, there have been substantial efforts to understand fully dynamical regimes of neural networks. Much has been learned from the study of relatively small networks, where it is possible to make nearly complete, microscopically realistic models and then analyze these models with all the tools of modern dynamical systems theory. These systems have been especially important for thinking about the mapping between microscopic parameters and macroscopic functions, as described in Chapter 4. There is a well-developed mean field theory for large networks with random synaptic connections, and these systems exhibit a transition from a quiescent to a chaotic state. While random connections may seem non-functional, these systems have rich dynamics, especially near the transition, which can be harnessed to generate or analyze temporal sequences on time scales much longer than the transient response times of individual neurons. There are several lines of evidence that real networks of neurons may be poised near critical points, although this remains controversial.

An important challenge in searching for collective behavior in networks of neurons, as in many other living systems, is the absence of the usual macroscopic, thermodynamic probes. Even in models that map to well-defined statistical physics problems, order parameters are complex combinations of activity across the network; available experimental manipulations do not couple naturally to these order parameters (as with applying a magnetic field to a ferromagnet), nor is it clear how to change the analog of the temperature. In the absence of such probes, there are efforts to infer a statistical or quasi-thermodynamic description directly from experiments on the activity of large numbers of neurons. Ideas along these lines include searching for low dimensional structure in the activity patterns; studying the behavior of these patterns under coarse-graining, inspired by the renormalization group; and constructing minimally structured or maximum entropy models for the distribution of activity patterns, matching measured correlations. The maximum entropy approach connects with the analysis of sequence variation in protein families (Chapter 3) and velocity fluctuations in flocks of birds (Chapter 3), and in some cases makes predictions that agree with experiment in quantitative detail. Still, it is not clear that the community has found the compelling theoretical framework to link the rapidly growing body of data on large populations of neurons with the ideas of emergent, collective behavior in these networks.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

Connectomics

All theories of neural dynamics agree that the collective behavior of a network depends on the pattern of connections among neurons. There is a long history of probing these connections between pairs of individual neurons, but as the 21st century began many people started to take seriously the possibility of mapping connections—network architecture—on a much larger scale. Higher throughput methods of electron microscopy are being combined with machine learning to trace neurons and their connections through very densely packed brain tissue. There have been important successes in using these methods to study smaller systems, for example showing that the nearly crystalline, orderly structure of connections in the early stages of sensory processing in insects gives way to more nearly random connectivity deeper in the brain. These results have provided proof of principle, and there are now serious proposals to chart the full “connectome” of a mouse, a primate, or perhaps even a human. Such a large-scale project would provide a scaffolding on which to build a description of collective neural dynamics. The first such effort dates back to the mid-1980s, with the reconstruction of all the synaptic connections among the 302 neurons of the nematode worm Caenorhabditis elegans.

There remains considerable debate within the scientific community as to whether the enormous effort and funding needed to determine one or more truly complete mammalian connectomes by electron microscopy would constitute a prudent allocation of resources. Nonetheless, there is general agreement that the biological physics community has played, and will continue to play, a crucial role in the development of imaging techniques for the acquisition of data and in the development of analysis techniques for image processing and the elucidation of neural circuits. Extensive challenges concern the successful visualization and tracing of trillions of axons and their synaptic connections, key constituents of a complete connectome, as illustrated in Figure 3.14. As discussed in Chapter 9, the National Institutes of Health and the Department of Energy are now actively engaged in discussions about supporting the largest scale versions of such a project, building on the success of intermediate scale projects sponsored by the Allen Institute, the Howard Hughes Medical Institute, the Intelligence Advanced Research Projects Activity, and other agencies. Thus far, the impact of connectomics has been greatest when focused on smaller circuits where we have a substantial body of knowledge about the computational functions being carried out. An impending effort of national scale, however, could bring us to an inflection point at which the impact of large-scale connectomics research rises dramatically.

Perspective

The search for emergent behavior in networks of neurons has been enormously productive, but is far from over. The last decade has seen dramatic advances both

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.14 Electron microscopy to reconstruct the connections between neurons. (A) The anterior quarter of a larval zebrafish was captured at 56.4 × 56.4 × 60 nm3 per voxel resolution from 16,000 sections. (B) The Mauthner cell (M), axon cap (AC), and axon (Ax) illustrate features visible in the 56.4 × 56.4 × 60 nm3 per voxel image volume. (C) Posterior Mauthner axon extension. (D) Targeted re-acquisition of brain tissue at 18.8 18.8 60 nm3 per voxel (dashed) from 12,546 sections was completed after 56.4 × 56.4 × 60 nm3 per voxel full cross-sections (solid). (E, F) Peripheral myelinated axons (arrowheads) recognized from 56.4 × 56.4 × 60 nm3 per voxel imaging of nerves (E) and the ear (F). (G, H) Neuronal processes including myelinated fibers can be segmented at 18.8 × 18.8 × 60 nm3 per voxel resolution. (I–K) Targeted re-imaging to distinguish finer neuronal structures and their connections. Scale bars: (B, C) 10 µm; (D) 50 µm; (E, F) 5 µm; (G, H) 1 µm; (I–K) 0.5 µm. SOURCE: Reprinted by permission from Springer: D.G.C. Hildebrand, M. Cicconet, R.M. Torres, W. Choi, T.M. Quan, J. Moon, A.W. Wetzel, et al., 2017, Whole-brain serial-section electron microscopy in larval zebrafish, Nature 545:345, copyright 2017.

in theory and in experiment, and these are continuing. The number of neurons that we can monitor simultaneously, and the quality of these recordings, continues to grow. Complete maps of the connections among tens of thousands of neurons have been achieved in the fly brain, and there is intense effort in other systems. Different organisms, from hydra to octopus, are emerging rapidly as model systems in which to make coordinated explorations of neural networks and behavior. There remains, however, a gap between theory and experiment. New, larger data sets need

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

new tools for analysis, but these analysis methods should be grounded in deeper theoretical ideas. Many quantities that theory points to as being crucial still are not so easy to measure. The hope for the coming decade is that there will be not only continued, parallel progress in theory and experiment, but new ideas about how to build bridges between the two.

COLLECTIVE BEHAVIOR

Collective behaviors in animal groups provide some of the most familiar examples of emergent phenomena in living systems. Most of us have seen the ordered patterns of geese flying in formation, the more fluid flocking behaviors of other species (see Figure I.3), the analogous schooling behaviors of fish, and the apparently chaotic motions in swarms of insects. But animals can do more than just move together; they can also organize themselves to accomplish large construction projects, such as the termite nest in Figure 3.15. All of these phenomena are emergent: There is no blueprint for the nest, no commander broadcasting a common movement direction to all the individuals in a flock, school, or swarm; instead, order arises out of the interactions among individuals in the group. More subtly, the very existence of ecology is an emergent phenomenon, since it is not obvious why a large number of different species can coexist, stably, in a single environment (Chapter 5). The focus of this section is on the collective behavior of multicellular organisms; collective behaviors in unicellular organisms are discussed in relation to communication (Chapter 2) and active matter (Chapter 5), although one can hope for unifying principles. The theme of our discussion is the interplay between theory and experiment: Qualitative observations inspire theories, new and more quantitative observations test these theories and show how living systems have found unexpected regimes of order and fluctuation, and theorists are sent back to search for new mechanisms that can generate the observed behaviors. It now is inescapable that collective behavior provides examples of new physics, beyond the examples from emergent phenomena in the inanimate world.

Flocks and Swarms

The emergence of a common movement direction in flocks, schools, and swarms is tantalizingly close to familiar ordering phenomena in the inanimate world. Could it be that all the birds in a flock agreeing to fly in the same direction is like all the spins in a ferromagnet agreeing to point in the same direction? In the 1990s, physicists began to explore models that embody this intuition. These models could be expressed as rules by which individuals in the group adjust their movements in relation to their neighbors, or equations of motion for animals acted upon by what had been called social forces in the earlier biological literature.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Image
FIGURE 3.15 Collective behaviors in animal groups, such as the large construction projects of termite nests, provide examples of emergent phenomena in living systems. Three-dimensional structure, reconstructed via X-ray tomography of a Cubitermes nest, from a set collected in equatorial forest regions of the Central African Republic and Cameroon. (A) A conventional photograph of the nest. (B) A virtual slice through the middle of the nest, constructed from X-ray tomography, showing the different chambers. (C) A virtual “cast” of the nest, illustrating the three-dimensional chambers and galleries. (D) A representation of chambers and galleries as a network, where each node corresponds to a chamber and each edge to a corridor. The color of the nodes reflects their degree, i.e. the number of corridors connected to that chamber, as shown in the legend. SOURCE: Reprinted by permission from Springer: A. Perna, C. Jost, E. Couturier, S. Valverde, A. Douady, and G. Theraulaz, 2008, The structure of gallery networks in the nests of termite Cubitermes spp. revealed by X-ray tomography, Naturwissenschaften 95:877, copyright 2008.

These equations could be simulated, or coarse-grained to derive the analog of fluid mechanics for a flock, using renormalization group ideas. Both simulation and analysis agreed on a first striking result, that these systems could break symmetry and produce ordered motion in a particular direction even for a hypothetical flock or swarm confined to two dimensions. Such symmetry breaking in two dimensions is not possible for a system in thermal equilibrium with local interactions, and thus is a harbinger of the qualitatively new physics that is possible in living systems.

The early work on theories of flocks and swarms became a foundation for what is now the lively field of active matter physics, as described in Chapter 5. This theoretical work also prompted efforts to collect more compelling quantitative data on collective behavior. One approach is to bring the behavior into the laboratory, studying schooling fish in a tank or swarming insects in a box. The other approach is to bring the laboratory into the field, setting up multi-camera video to reconstruct the trajectories of all the organisms in the group as they move through

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

their natural environment. In both approaches, a central role is played by analyzing the observed fluctuations in movement velocities around the mean, and statistical physics gives us a natural language of correlation functions to use in this analysis.

Analysis of correlation functions in flocks of European starlings showed, surprisingly, that correlations among velocity fluctuations are independent of scale (see Figure I.3). Scale invariance for fluctuations in flight direction can be understood by realizing that this is a system with local interactions, and the system breaks a continuous symmetry, so there will be “massless” modes. But there is no generic expectation for scale invariance of correlations in speed fluctuations. Although swarms of midges exhibit no overall velocity ordering, they too show scale invariant fluctuations in velocity, and analysis of correlations in both space and time reveals dynamic scaling, with an exponent closer to ballistic propagation rather than diffusion of information through the swarm (see Figure 3.16). Ballistic propagation is seen also in flocks, during events where the entire flock turns. All of these correlation structures are outside the predictions of the generic

Image
FIGURE 3.16 Swarms of midges exhibit scale invariant fluctuations in velocity. (A) A system of three synchronized high-speed cameras is used to collect video sequences of midge swarms in their natural environment. (B) A swarm of approximately 300 midges. (C) Two trajectories within the swarm. (D) Spatiotemporal correlation functions of the velocity, as a function of time t (seconds) and the Fourier variable k conjugate to distance. Upper panel: normalized correlation function in one natural swarm at various values of k. Bottom panels: correlation as a function of the rescaled time, t/τk, in various attempts to rescale the data. (left) Rescaling by a k-dependent time for all k in one swarm. (center) Comparing many swarms at the same k. (right) Measuring the static correlation length ξ for each swarm, and choosing ∼1, then rescaling time. This is evidence for dynamic scaling, Ĉ(k,t) = Ĉ(t/τk,), with τc = kzg(kξ); further analysis shows that z = 1.12 ± 0.16. SOURCE: Reprinted by permission from Springer: A. Cavagna, D. Conti, C. Creato, L. Del Castello, I. Giardina, T.S. Grigera, S. Melillo, L. Parisi, and M. Viale, 2017, Dynamic scaling in natural swarms, Nature Physics 13:914, copyright 2017.
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

active matter models, and there are continuing efforts to find a theory that captures these behaviors. Methods range from the renormalization group analysis of a wider range of microscopic theories to direct inference of statistical descriptions from the data using maximum entropy methods. This last approach connects the study of collective animal behaviors to the study of sequence variation in protein families (Chapter 3) and patterns of activity in networks of neurons (Chapter 3).

Flocks of birds, swarms of insects, and schools of fish function in unconfined environments. In contrast, communities of ants and termites that construct tunnels and structures in soft materials such as soil must move collectively in their confined and crowded nests. Such densely packed and disordered conditions in non-living systems lead to a breakdown of flow, through glassiness and jamming (Chapter 5). Physicists studying the traffic of confined fire ant colonies have revealed that they routinely move through foraging tunnels that are comparable in dimension to the individual ants. Movement is hindered not only by the spatial restrictions, but by social interactions when ants moving in opposite directions encounter one another and pause to touch antennae (“attenation”), presumably to exchange information. Models that incorporate these interactions exhibit a phase transition as a function of the attenation time, similar to the fragile glass or jamming transition. Longer attenation times likely allow for more effective information flow through the colony, but this is useless if the colony is jammed. Real colonies appear to function close to the transition.

Social Insects and Superorganisms

Flocks of birds, swarms of insects, and schools of fish are undeniably emergent phenomena. These also are social behaviors, but evidently this term covers a much wider range of possibilities. In some cases, the collective behavior is so compelling that what emerges is a “superorganism,” as with social insects such as termites, ants, and social wasps. These superorganisms breathe, feed, grow, breed, and modify their environments, as with the termite nests in Figure 3.15.

To appreciate the analogy between a superorganismal insect colony and individual organism, consider a colony of fire ants composed of hundreds of thousands of individuals. The superorganism is composed of individual colony members in the same way that an organism is built of cells. Instead of specialized organs, superorganisms consist of specialized castes responsible for different functions, all of which contribute to the survival and reproduction of the whole group. Thus, superorganisms display tremendous cooperation and integration in roughly the same way that cells of a single organism work together to help the individual succeed. Importantly, from a physicist’s perspective, the processes which define the behavior of the superorganism are emergent, such that the rules by which the higher levels function are often unknown and might not be easily predictable from the behavior of a single organism.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

Ants and other invertebrates have external skeletons (exoskeletons) that give the individual organisms shape, structure, and strength. The superorganism also has a nest that provides protection and a location where food is returned and where young animals are reared. Nests promote division of labor among individuals, modulate communication and information distribution, and regulate the physical environment. Indeed, social insect nests are often referred to as part of the “extended phenotype” of the colony. Importantly, no single insect has a conception of how the nest should be built or what it should look like when it is complete; insects do not have managers directing the construction process. Instead, social insects use micro-scale rules that lead to the formation of the complex colony exoskeleton. One important aspect of colony self-organization is the concept of stigmergy. Stigmergy is a process of indirect coordination and activation of behaviors through environmental signals. For example, rapidly growing structures within a nest may act as strong stimuli for additional construction and grow quickly until a positive stimulus plateau is reached, triggering negative feedback leading to the reduction in construction. Stigmergic models are often “agent based” and consider insects (agents) to be engaged in sets of limited behaviors when encountering particular environmental stimuli.

The biological physics community’s understanding of flocks and swarms began with somewhat complicated agent-based models from the biological literature and went through phases of simplification and deeper theoretical analysis, followed by dramatic improvements in quantitative measurement that exposed new statistical physics problems. The understanding of social insects seems somewhere near the beginning of this process, and it is encouraging to see new experiments probing the collective behaviors of honeybees, ants, and others using modern physics-based approaches. There are significant technical challenges in tracking individuals through much denser groups, and in some cases having to work in opaque environments, and it is reasonable to expect progress on these experiments over the next decade.

Perspective

In a flock, as in a fluid, natural macroscopic variables are spatial averages over the velocities of the component parts, be they birds or molecules. Still, finding the correct effective theory for these coarse-grained variables is an unfinished project—these collective behaviors belong to a universality class beyond what we have understood from conventional physics problems. In contrast, it does not seem likely that spatial averages over the behavior of individual termites will capture their contribution to nest building. A statistical physics of social insects will require development of approaches in which coarse-grained variables change their character as we change the scale of our observations, leading to more new physics. As in neural networks (Chapter 3), the distinction between systems that

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×

can search for a quasistatic equilibrium and systems that generate spontaneous and self-sustaining dynamics is blurring. In flocks and swarms, and with social insects, the search for theories proceeds in parallel with dramatic improvements in experimental observations, and there are opportunities for substantial leaps forward in the coming decade. The world of collective behaviors is much larger than described in this section, and it is possible that the deepest insights will come from taming an example that currently is only barely explored.

Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 119
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 120
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 121
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 122
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 123
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 124
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 125
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 126
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 127
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 128
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 129
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 130
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 131
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 132
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 133
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 134
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 135
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 136
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 137
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 138
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 139
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 140
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 141
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 142
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 143
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 144
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 145
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 146
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 147
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 148
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 149
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 150
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 151
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 152
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 153
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 154
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 155
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 156
Suggested Citation:"3 How Do Macroscopic Functions of Life Emerge from Interactions Among Many Microscopic Constituents?." National Academies of Sciences, Engineering, and Medicine. 2022. Physics of Life. Washington, DC: The National Academies Press. doi: 10.17226/26403.
×
Page 157
Next: 4 How Do Living Systems Navigate Parameter Space? »
Physics of Life Get This Book
×
Buy Paperback | $60.00 Buy Ebook | $48.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Biological physics, or the physics of living systems, has emerged fully as a field of physics, alongside more traditional fields of astrophysics and cosmology, atomic, molecular and optical physics, condensed matter physics, nuclear physics, particle physics, and plasma physics. This new field brings the physicist's style of inquiry to bear on the beautiful phenomena of life. The enormous range of phenomena encountered in living systems - phenomena that often have no analog or precedent in the inanimate world - means that the intellectual agenda of biological physics is exceptionally broad, even by the ambitious standards of physics.

Physics of Life is the first decadal survey of this field, as part of a broader decadal survey of physics. This report communicates the importance of biological physics research; addresses what must be done to realize the promise of this new field; and provides guidance for informed decisions about funding, workforce, and research directions.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!