Systematic: How Systems Biology Is Transforming Modern Medicine

SYSTEMATIC

For my family, my teachers, and

the original Broadway cast of Hamilton

Content

Author’s Note

Preface: The Big Idea

PART I:  THE BASICS

Chapter 1:    Seeing the Systems in Biology:

Technological Advances Are Letting Scientists Understand Living Things in a New Way

Chapter 2:    Déjà Vu All Over Again:

The Common Patterns and Principles of Natural Systems

Chapter 3:    America’s Next Top Mathematical Model:

Understanding Complex Systems Sometimes Requires Math

Chapter 4:    Ignoring the Devil in the Details:

Robustness, Prediction, Noise, and the General Properties of Systems

PART II:    CELLS, ORGANISMS, AND ECOSYSTEMS

Chapter 5:    Beyond Tom Hanks’s Nose:

Sequencing Technology Is Enabling Scientists to Study All of a Cell’s Genes at Once

Chapter 6:    The Smells of the Father:

RNA, DNA Margin Notes, and the Other Missing Parts of the Cellular System

Chapter 7:    Growing Pains:

How Cells and Tissues Coordinate Development, from Egg to Adulthood

Chapter 8:    No Organism Is an Island:

The Interactions between Individuals and Species that Shape Ecosystems

PART III:    APPLICATIONS

Chapter 9:    Build Me a Buttercup:

Using Synthetic Biology to Make Diesel Fuel, Programmable Cells, and Malaria Medicine

Chapter 10:  More Than Just 86 Billion Neurons:

The Science of the Brain, and How Connections among Neurons Make It Work

Chapter 11:  Death and Taxes:

Aging Is Governed by an Organism-Wide System that We Might Be Able to Manipulate

Chapter 12:  Your Microbiome and You:

The Body Is Host to Trillions of Microbes that Affect Human Health

Chapter 13:  This Is Your System on Drugs:

Tweaking Biological Systems to Produce Better Medical Treatments

Epilogue

Acknowledgments

Notes

Index

A Note on the Author

Author’s Note

In introducing general readers to systems biology, I am including interesting research that is guided by the systems idea rather than pretending to provide a comprehensive or even fully representative overview of the field.

The challenge to responsibly communicate new ideas that may be revised or even discarded with future research is one I take very seriously. The reader should keep in mind that this book covers many areas of research that are rapidly changing. In some of these cases, the work is still very preliminary. I have attempted to present our best current understanding of what’s happening, and I hope readers will pair enthusiasm for the progress scientists have made so far with a realization that there is still much more work to be done. I encourage the reader to check this book’s website at www.systematicbook.com for updates on the topics discussed herein.

Preface: The Big Idea

Neurons are the cells that do most of the heavy lifting in your brain, but the most exciting thing a single neuron can do is fire an electrical pulse. Big deal. So can my toaster. But 86 billion neurons connected together in just the right way form your brain, which enables you to think, feel, imagine, and wonder. That’s pretty impressive for a three-pound hunk of cells.

The difference between one neuron and your brain is more than a matter of scale. Those 86 billion neurons must be connected together, and then something fundamental changes. This book is about the magic that happens when many proteins, cells, or other biological pieces are connected into a system.

In biology, connected systems underlie many complex behaviors that have historically been hard to understand. Microbes in your gut, for instance, interact closely with your body, and some evidence suggests this system might affect your weight, alter your risk of developing cancer, or even influence your mood. Bacteria can use simple systems of proteins, DNA, and other biological material to anticipate sunrise, to decide what to eat, and even to predict what kind of food they are likely to encounter next. Systems in other living things affect aging, development, and overall health. And thinking in terms of systems can help explain interactions between organisms, such as the spread of plague through gerbil populations in the deserts of Kazakhstan.

Unfortunately for the impatient scientist, the systems found in biology are enormous and complex. Each person has about 40 trillion cells,¹ and each cell is made of billions of tiny components.² If you bought a gumball for every cell in your body, you could fill Fenway Park in Boston about 1,000 times over,³ and those gumballs would cost you the entire GDP of Russia if you ordered them from Amazon.com at current bulk rates.⁴ When this many parts are connected together in one big biological system, astoundingly complex behavior results. Even simple systems of just three or four components can produce sophisticated behaviors, and it is precisely these webs of interacting components that make life possible.

Systems biology—the study of connected groups of biological parts that all work together—is relatively new. For example, the Harvard Systems Biology Ph.D. Program matriculated its first class in 2005. There is even still some disagreement about what it means to be a systems biologist. Some think it’s all about bringing the math back to biology research; others say it’s about working with huge amounts of data. But at its core, systems biology is simply the recognition that life is complex because it’s connected. Systems biologists want to understand how systems make life—and all of its weirdness—possible.

This is a book about how understanding natural systems is helping us unravel some of the biggest mysteries in science. We will explore biological systems that range in size from microscopic proteins to entire ecosystems. Despite superficial differences, these systems can be studied with similar approaches, and all of them have implications for our understanding of life or our ability to treat diseases.

PART I

THE BASICS

CHAPTER ONE

Seeing the Systems in Biology

Technological Advances Are Letting Scientists Understand Living Things in a New Way

Biologists were studying systems long before the term systems biology existed. Indeed, all living things are made of systems. The human body, for example, is a system made of organs—such as the heart, skin, and brain—that work together to make life possible. In turn, each of these organs is itself a system made of specialized cells that coordinate to pump blood, heal wounds, or sense pain. And each of those cells is a system made of proteins, DNA, and other microscopic biological parts.

But while scientists have always studied systems, their capacity for fully understanding them has been limited historically. A complete accounting of how a system works requires not only knowledge of each of its parts, but also an understanding of how those parts interact to make the system work. In many cases, we haven’t even had the technology to measure all of the system’s components; it was like doing a jigsaw puzzle with half of the pieces missing.

To make progress, researchers started by breaking systems down into little pieces and studying each part individually—doing biology from the bottom up. Some of today’s scientists devoted their entire doctoral dissertations to trying to understand a single gene or protein, and for good reason: deconstructing systems into their constituent components has taught us much of what we know about biology today. But we now also have the capability to start putting those pieces back together to understand how they make life work. That’s systems biology.

To finally have the ability to grapple with systems in earnest took great effort by many scientists, such as Dr. Eric Wieschaus.¹ As a young researcher at the European Molecular Biology Laboratory in Heidelberg in the early 1970s, Wieschaus was focused on understanding the biological machinery that governs fruit fly development. Fruit flies, like most complex life, start out as a fertilized egg, and Wieschaus wanted to know how the complex adult fly can come from such a simple beginning.

At the time, we didn’t know which genes were involved in the system that governs fly development, so Wieschaus and his colleague Dr. Christiane Nüsslein-Volhard decided to find out. They mutated a bunch of flies using chemicals that damage DNA, hoping that damage would, by random chance, happen to disrupt a gene that was important for development. Wieschaus and Nüsslein-Volhard then painstakingly examined the progeny of these mutated flies, looking for those with developmental defects. They repeated this process over and over, hoping to mutate enough flies that they would be likely to disrupt most of the genes in the fly at some point. The researchers looked at thousands and thousands of mutant flies—about 27,000 genetic variants in total—and meticulously documented how each embryo’s particular genetic mutation messed up its development. Wieschaus estimates that it took them six to eight months of long shifts looking at fly embryos under a microscope, but ultimately their patience was rewarded: the experiment revealed the vast majority of genes that control fly development. Their work won them the 1995 Nobel Prize in Physiology or Medicine along with another geneticist, Dr. Edward B. Lewis.

Wieschaus and Nüsslein-Volhard’s experiment gave them a first draft of the list of parts that compose the fly development system. They also had some good luck: it turned out that many of the mutations they were making had easily interpretable consequences, so they could guess the function of some genes by seeing how a change in that gene affected development. For example, they found that mutations in one gene affected the development of a set of regular segments that form early in the fly’s development. These mutated flies had only developed the odd-numbered segments and were missing the even-numbered segments. Based on these observations, Wieschaus and Nüsslein-Volhard could guess that this gene, now called even-skipped, helped create those even-numbered segments in normal flies. This type of information allowed scientists to start piecing together how genes might interact in order to produce the complex patterns of an adult fly.

With Wieschaus and Nüsslein-Volhard’s foundational work and additional research by many other scientists, the challenge of understanding fly development shifted. We no longer had to ask which biological parts were involved in this system. Instead, we could ask: How do these parts work together to produce the patterns we see? And what can that answer tell us about how humans develop? These are the kinds of questions that interest systems biologists.

In other cases, scientists did systems biology by making detailed observations of a system’s behavior and using brilliant mathematical intuition to deduce how the system works. For example, two decades before Wieschaus and Nüsslein-Volhard were poring over their flies, Alan Hodgkin and Andrew Huxley, biophysicists at the University of Cambridge, were carefully inserting silver wires into one of a squid’s neurons to try to understand how it fires an electrical pulse. The neuron Hodgkin and Huxley used is unusually large—large enough to allow direct measurement of its electrical activity using tiny wires and a device called a voltage clamp that allowed the researchers to manipulate the flow of electricity in the cell.

At the time, scientists knew that the electrical activity produced by a neuron was caused by a flow of ions—electrically charged atoms, in this case mostly sodium and potassium—into and out of the cell. The exact details of the process and the cellular components that controlled this flow were still a mystery, however. Rather than trying to isolate all of the components of the system individually, Hodgkin and Huxley focused on gathering an enormous amount of data about how the system as a whole behaved. They measured the neuron’s electrical activity in dozens of different conditions as they altered the voltage across the neuron’s cell membrane using their electrodes. By thinking of the neuron as an electrical circuit and applying what they knew from physics, they were able to deduce how the neuron fires an electric pulse, known as an action potential. Hodgkin and Huxley described this process using mathematical equations—which they solved using a calculating machine that had to be cranked by hand—that predicted electrical spikes similar to those observed in real neurons. Their equations even foreshadowed some properties of then-hypothetical structures called ion channels—which we now know to be proteins that open and close to let ions pass through the cell membrane—long before these channels were probed by other means. Their mathematical description of this process, known as the Hodgkin-Huxley model, is still in use today, and they won the 1963 Nobel Prize in Physiology or Medicine.

Hodgkin and Huxley’s work is an example of what systems biology often aspires to be—a concrete, mathematical understanding of a complicated biological process—but successes like theirs were relatively rare even as recently as the 1990s. Scientists were often stymied by systems that had more parts or were more complex than the firing of a neuron’s action potential.

Things began to change around the year 2000 when a combination of exponentially increasing computing power and new experimental techniques made systems biology more practical. Faster computers allowed scientists to look for patterns in large data sets that were too big for humans to work with. At the same time, scientists and engineers were inventing new tools, such as DNA and RNA sequencing, that could produce data at that scale. Scientists have long been able to measure how strongly a single gene is turned on, for example, but measuring how every gene is behaving in many different cells simultaneously was impossible until recently. These days, the technology to get large-scale information about every gene or every protein in cells is commonplace, and scientists have made similar breakthroughs in their abilities to measure other aspects of systems. We’re increasingly limited not by our technical capabilities but by our ingenuity.

The advent of these new tools made modern systems biology possible, but it also changed the kinds of skills biologists need to succeed. Since math and computer power are so important to many modern laboratories, there has been a flood of talent from physics and mathematics.

David Botstein, then the director of the Lewis-Sigler Institute for Integrative Genomics at Princeton University, was one of the scientists who recognized early on that the same technologies that made systems biology possible were moving all biology in a more quantitative direction. Aware that college biology students were not getting the math, physics, and computer skills that they would need, Botstein created a new approach to teaching budding scientists in the early 2000s. His course, launched with physics professor William Bialek, is called Integrated Science. This program is an evolution of earlier courses at Stanford and the Massachusetts Institute of Technology, and it combines physics, biology, chemistry, and computer science into one intense course that teaches students how to think like a research scientist.

As a freshman at Princeton in 2008, I became one of those students. In an Integrated Science introductory session, I listened with rapt attention as Botstein articulated how scientists were using physics, math, chemistry, and computer science to solve the most interesting problems in biology. This interdisciplinary approach enabled researchers to study fascinating phenomena, and they used mathematical methods to help them understand what was really going on. For systems biologists, those tools were illuminating systems that had not previously been accessible.

CHAPTER TWO

Déjà Vu All Over Again

The Common Patterns and Principles of Natural Systems

When I lived in Germany in 2009 for a summer doing research at the European Molecular Biology Laboratory, I had a tiny problem: I didn’t speak German. The science all happened in English, so that was easy. But if I wanted to make friends, navigate the transit system, or buy food? That was a bit harder. I’m ashamed to say I never really did get good at German, but the experience taught me a valuable lesson.

A few weeks into the summer, my coworkers invited me to see Terminator Salvation, the sequel starring Christian Bale. The film was in German, without subtitles. I was afraid I would be totally lost, but a Hollywood movie isn’t too hard to understand on the macro-level. The good guys look like good guys. The bad guys look like evil robots.¹ Even though the details were different, I knew the formula from seeing similar movies.

Just like Hollywood movies, systems biology has a common language. Biologists study processes that seem very different on the surface: how neurons link up to form the brain, how simple molecules assemble to form a cell’s skeleton, how proteins interact to warn a cell that its environment is too salty, and how infectious bacteria coordinate to attack only when they have sufficient numbers to overwhelm their host. But even though these processes operate at wildly different scales and depend on unrelated parts, they often contain common patterns and operate according to similar principles.

In the early 2000s, Dr. Uri Alon’s laboratory at the Weizmann Institute of Science in Israel was searching for these types of patterns in naturally occurring networks.² The researchers collected a half-dozen examples of biological systems—networks of neurons, proteins, and genes—but also some examples of nonbiological networks, such as web pages connected by links, transistors connected by wires in computer chips, and even kindergarteners connected by friendship. (Alon explained, You go to a kindergarten and ask children: who are your friends? One of the tragedies of life is if X likes Y, Y doesn’t always like X.) They then applied a simple mathematical analysis to search their collection of seemingly dissimilar networks for patterns.

The researchers first needed to write down these networks in a common language. Alon’s group used a simple representation of their networks that consists of components and arrows. In the case of a network of genes where the activity of gene X turns on gene Y, they would draw an arrow from X to Y:

Figure 1: X activates Y.

Similarly, for a network of neurons, an arrow from X to Y means that neuron X sends signals to neuron Y. And for kindergarteners, an arrow represents a one-way friendship: the fact that child X likes child Y. Each network was ultimately represented by a large web of components connected by arrows, something like a much larger version of this hypothetical network:

Figure 2: A hypothetical network.

Next, the researchers looked for patterns in these networks by examining groups of three or four components at a time. To start, they listed all of the possible ways that these sets of three or four components could be connected together. For example, a set of three components can be connected in any of 13 distinct ways. Here are three:

Figure 3: Different ways of connecting three components.

Alon’s group then asked whether any of these ways of connecting the components shows up in real systems more often than one would expect by chance. Indeed, they found that certain sets of three or four components appeared far more often in the real networks than in randomly generated networks³—much in the same way that the three letters c, a, and t show up together in written English (cat) much more often than they would in a sequence of random letters. Alon and his colleagues call these repeated small patterns network motifs.

In biological systems and some types of computer chips, for example, Alon and his collaborators often saw a network motif called the feed-forward loop. In this motif, X turns on both Y and Z, but Y also turns on Z. Schematically, it looks something like this:

Figure 4: A feed-forward loop.

Alon’s team found that the feed-forward loop motif showed up frequently in networks of genes and proteins in bacteria and yeast, in networks of neurons in worms, and in certain kinds of electronic circuits. It seemed that the feed-forward loop often made an appearance in systems that process information—whether that was a bacterial cell decoding signals from its environment in order to move toward food, or a laptop interpreting a user’s keystrokes. Indeed, Alon thinks that evolution and human engineers keep using these motifs because they work well: he says they may represent the simplest, most efficient solutions to the shared problems that cells have. If that’s true, the feed-forward loop would be the Michael Jordan of networks—the all-star who plays as much as possible because he gets the job done.

In the case of the feed-forward loop, one possible use for the motif is to help the network deal with noise—noise being, here, random fluctuations that could cause the network to make a mistake. If both X and Y must be turned on in order to activate Z, for example, the feed-forward loop could be used to turn on Z only after X has been active for some time—since the signal needs time to trickle down through Y—and to turn off Z quickly when X disappears. In a biological system, this behavior might be useful when dealing with noisy inputs, Alon suggests—if the organism might not want to respond with Z until it’s sure the signal X is for real.

Long before Alon’s group was analyzing kindergarten friendships and internet links, scientists had been prying apart biological systems to look for simple modules they could understand. In the 1960s, they began to find some simple circuits that seemed tractable, many of which were in bacteria. Even a single bacterial cell is exceedingly complicated, and we don’t yet have the ability to understand the full dynamics of all of the stuff that goes on in the cell. But these early researchers made progress by finding smaller chunks of the bacterial system that were more manageable.

One early example of such a subsystem is the small circuit that allows the bacterium Escherichia coli to decide what to eat. In a pinch, they’ll eat pretty much anything that has caloric value, but E. coli are actually picky eaters when food is plentiful—like teenagers who choose pepperoni pizza over a salad. Specifically, the bacterium sometimes must make a choice between eating one of two types of sugar that are commonly available to it.

The bacteria’s first choice is glucose, the pepperoni pizza of sugars. Glucose is easy to break down and turn into energy, so it’s ideal for a bacterium that wants, above all else, to get energy as quickly as possible: this helps it to grow quickly and outcompete other nearby microbes. But E. coli also has the ability to eat lactose, a kind of sugar found in milk. E. coli has to do a little extra work to break down lactose in order to use it as an energy source, and it makes a special enzyme⁴—a protein that speeds up a chemical reaction—called β-gal that helps break down lactose. It doesn’t make sense for E. coli to be constantly producing this enzyme if there’s no lactose around, or if the bacterium has plenty of glucose that it’s using instead.

So what’s a picky bacterium to do to make sure it only makes β-gal at the appropriate time? It uses a very simple regulatory system to test for the presence of lactose and glucose and to make a decision based on that information.

Tests for glucose or makes a decision are just anthropomorphic ways of conveying what is actually a pretty simple idea on the molecular level. E. coli senses lactose through a protein that normally grabs on to the DNA at a specific spot. This part of the DNA contains the gene that tells the cell how to make the lactose-digesting enzyme, β-gal. When this protein is attached to this bit of DNA, the cell can’t make β-gal:

Figure 5: The blocking protein prevents