Life's Ratchet: How Molecular Machines Extract Order from Chaos (41 page)

Living is programmed molecular dancing. We cannot allow molecular motors to move random cargo to random places. Specific cargo needs to go to specific places at specific times. The same is true for every activity of our cells: What proteins should be produced? When? How many? When should a molecular motor take on cargo, and when should the cargo be released? In
Chapter 6
, we glimpsed how cells regulate such decisions. Most complex proteins can be controlled through allostery—the change in structure and activity when the protein binds a control molecule. In our cells, proteins regulate other proteins, but also the transcription of DNA. In turn, proteins are made by the instructions contained in DNA and are controlled by other molecules, including sugars, ions, and lipids. The complicated program of “living” emerges from complicated feedback loops between all these molecules, linking them together in complex networks. The idea of a dynamic state of complex feedback loops is difficult to fathom. But that is what is going on in a cell. The complex molecules of our cells are marvels of evolutionary engineering. But the cell only becomes a cell when these molecules cooperate in a rich network of regulated interactions. This cooperative, self-sustaining, regulated activity is what we call
living
.

How does regulation work? When we talked about how molecular machines work, we never mentioned how they know when to do their work. Do they work all the time, grabbing cargo, moving it, pumping ions in and out of the cell, or producing more ATP, even if it is not needed? That would be a recipe for disaster. Molecular machines are enzymes, first and foremost, and most enzymes in our bodies are regulated by inhibitory binding or allosteric interaction. Inhibitory binding is a direct way to regulate an enzyme or a motor. If a molecule other than the enzyme’s substrate binds to the enzyme’s binding pocket, and this molecule cannot be transformed by the enzyme, then this molecule “gums up” the enzyme’s function. This is inhibition. Allostery, as we have already learned, is the control of an enzyme’s activity by the binding of a control molecule to a separate binding pocket on the enzyme, which then through some conformational change controls the binding of the substrate, either enhancing or inhibiting it.

Inhibitory binding is how most drugs (and many poisons) work. The antacid pantoprazole is a proton-pump inhibitor. The particular proton pump that this drug inhibits is a molecular machine in the membranes of the cells that line the stomach. The machine’s “job” is to take energy from ATP and use it to pump protons (hydrogen ions) into the stomach. Protons make the stomach acidic. Pantoprazole binds to the machine and blocks it temporarily, inhibiting an increase of acidity in the stomach. Inhibitor binding strength, or how long the inhibitor will remain bound before it again frees up the enzyme, depends on the height of the activation barrier. This allows nature to tailor inhibitors to control reactions over different time scales. The same is true of allostery: The binding of a control molecule is controlled by the binding strength (the
affinity
), the rate at which the molecule dissociates (which depends on the activation barrier), and the concentration of the control molecule in the cell (at low concentrations, the probability that an enzyme and a control molecule will meet is low).

Kinesins are a good example of how molecular machines are regulated in the body. Kinesins consist of one or two similar motor domains, which process ATP and bind to microtubules. In addition to the motor domain, kinesins can contain a number of other domains to bind cargo, to bind to specific locations, or for regulation. In the cargo-carrying motor kinesin-1 (as well as in kinesin-2, kinesin-3, and kinesin-7), the cargo-binding domain also serves a regulatory function, as it would be wasteful to have motor proteins running around in the absence of cargo. After all, they use up ATP. How is kinesin regulated? When no cargo is around, the kinesin molecule folds up, such that the cargo-binding domain can bind loosely to the motor domains. In effect, the molecule puts on its parking brake. If the cell wants to activate the motor, it sends two control molecules, which bind to the cargo domain and release it from the motor domains. The process essentially takes off the parking brake. This is not the only way kinesin-1 is regulated. For instance, the cell can control which microtubule a kinesin walks on using various proteins that bind to the microtubule. Some will “attract” kinesin, and some will inhibit its motion on the track. Another set of regulatory proteins controls the binding and release of cargo. It is quite typical for enzymes and molecular machines to be regulated in various ways to make sure they do the right job at the right time in the right location.

The very existence of molecular motors is also regulated by various molecular feedback loops and control molecules. For example, certain different kinesins are needed only during a specific phase of cell division (mitosis). In this phase, these kinesins are manufactured at a higher rate. Once the next phase begins, other proteins direct the breakdown of these kinesins, which are then recycled. Mitosis is a complicated, highly choreographed process. Not only the presence of kinesins (which help to separate the chromosomes), but also their location must be regulated. Control proteins make sure the various kinesins do their work at the right place.

Understanding the roles and regulation of molecular machines has been a boon for the pharmaceutical industry. Medical drugs, with few exceptions, work by inhibiting enzymes or molecular machines and are artificial control molecules. The task of R&D personnel at major drug companies is to come up with chemicals that specifically bind to target proteins, blocking their activity, and not binding to anything else, as this would cause side effects. Drugs need to be specific.

Kinesins are a target for cancer drugs. Cancer cells divide prodigiously, creating tumors. Eventually, the cells spread, causing metastasis, the main cause of cancer deaths. A drug called monastrol targets kinesin-5, which plays a major role during mitosis. As described in
Chapter 7
, kinesin-5 is a double motor that can bind to two microtubules at the same time. Kinesin-5 controls tension in the spindle, which separates chromosomes during the cell division. When monastrol binds to kinesin-5, the drug causes a change in structure of its ATP-binding site. The kinesin can still bind ATP, but it can no longer release ADP after hydrolysis. With ADP stuck in its ATPase pocket, the motor cannot obtain energy and falls dead. The spindle falls slack, and the cell cannot divide. The cancer cell, stuck in the middle of dividing, commits cell suicide.

At Wayne State University, the lab of my colleague Rafi Fridman studies the interaction of the collagen-eating, membrane-anchored enzyme, MMP-14 (metalloproteinase 14), and its inhibitor TIMP-2. My lab collaborates with Rafi, trying to measure the affinity of TIMP-2 for MMP-14 on living cells. This requires measurements at the single-molecule level.
To do this, we attach TIMP-2 to an AFM tip and let it interact with MMP-14 on the surface of a living cell. Then we move the lever up to pull on the bond between MMP-14 and TIMP-2 and record the force needed to break the bond. After countless measurements and applying the proper statistics, we can determine the average lifetime of the bond between the two proteins.

The interaction of MMP-14 and TIMP-2 is of special interest because scientists previously discovered that TIMP-2 not only inhibits MMP-14, but also primes the enzyme to
activate
another, free-floating MMP, called MMP-2. In other words, the so-called inhibitor of MMP-14 inhibits the collagen-destroying activity of MMP-14, but at the same time
activates
another collagen-destroying machine, MMP-2. The inhibitor is not really an inhibitor, but rather switches from one method of destroying collagen to another. Why? This is a question of ongoing research.

As mentioned in
Chapter 7
, MMPs play an important role in the motion of cells. In cancer cells, MMPs are often produced in high numbers and allow cancer cells to spread throughout the body. Because of this implication for cancer, MMPs have been a major target for drug development. Like monastrol, drugs that target MMPs are artificial inhibitors that stop MMP from doing its work. When the first artificial inhibitors were developed, they worked very well—in a test tube. However, after lengthy approval processes, these drugs were used on terminally ill cancer patients, but did not work very well. Moreover, they caused serious side effects, especially excruciating joint pain. What happened? MMPs are not only used by cancer cells, but also by regular cells, including those that maintain the cartilage in our joints. Indiscriminately shutting down the activity of an important molecular machine is not the best way to battle cancer.

Incidentally, the kinesin-5-targeting drug, monastrol, also ran into trouble. Besides causing a number of side effects (noncancer cells like to divide, too), it was ineffective in many types of cancer. It is not always clear why certain drugs do not work as hoped from laboratory experiments. Shutting down a single type of enzyme or molecular motor may not always be the key to finding a cure for cancer or other diseases. The molecules in our cells interact with each other in complicated ways that we are just beginning to understand. Our cells are well-regulated machines. Deciphering their regulation has proved difficult, as their complexity is staggering. This
lack of understanding of the full complexity of our cells is the main reason why medical drugs often fall short of producing the desired results.

Regulation in biological systems proceeds on many levels. DNA contains information to make proteins. The types of proteins and the timing of their production are regulated by special DNA-binding proteins. Transcription and translation are regulated by control molecules. All of these processes involve positive and negative feedback loops. Enzymes and molecular machines, as we have seen, are regulated in a variety of ways, from the auto regulation seen in the “parking brake” of kinesin-1 to the complex, sometimes contradictory regulation of molecular machines involving multiple, co-interacting control molecules. On top of this, the cell surface contains numerous specialized receptors, which are controlled by external chemicals. Once a receptor binds to a chemical target, it undergoes a conformational change, which releases or binds a control molecule, setting off a cascade of feedback loops inside the cell, leading to a “macroscopic” response of the entire cell. These signaling pathways are a large part of what biochemists and cell biologists study today.

The first regulatory pathway that was deciphered is the so-called lac-operon in
E. coli
bacteria, for which Jacques Monod and François Jacob received the 1965 Nobel Prize.
E. coli
can live off of a variety of “foods,” one of which is lactose (milk sugar). To break down lactose, three enzymes are needed. One of these enzymes is a molecular machine that pumps lactose into the cell. Since it takes resources to make these enzymes, it would not make much sense to produce them if there were no lactose present. In addition, the pump uses precious ATP. But how does DNA know if lactose is present? If lactose is present in the “broth” surrounding the bacteria, a lactose receptor on the cell’s surface becomes activated (it binds to lactose and sets off a chemical signal inside the cell through allostery). This activates the few lactose pumps present at the cell’s surface, and they begin to pump lactose into the cell. However, these few lactose pumps are not enough to take in all the lactose that is floating by. What to do? Make more pumps!

The gene that encodes the enzymes needed for lactose digestion is preceded by a DNA sequence called the operon. The operon is a DNA patch
to which a control protein (a
repressor protein
) can bind. When the repressor binds to the operon, the RNA polymerase, which transcribes DNA into RNA, is blocked, and transcription cannot proceed. No lactose-digesting enzymes are manufactured.

Lactose, however, can bind to the repressor protein, causing it to change shape (via an allosteric interaction) so that the protein can no longer bind to the DNA operon. Now the RNA polymerase is free to transcribe the lactose genes, and lactose-digesting enzymes and lactose pumps are produced in large numbers. Thus, the interaction between lactose, repressor, and the DNA operon makes sure that lactose-digesting enzymes are only produced when lactose is present. This is how the “computer logic” of our cells works. The manufacture of just about every protein is regulated by similar feedback loops.

The study of how these feedback loops work is called
systems biology
. Where molecular biology takes chemical bonding for granted, systems biology takes molecular biology for granted and treats protein and DNA sequences as interacting mathematical entities—players in the computer program of our cells. In this way, scientists work their way up from atoms to molecules to proteins to networks to systems, and finally to an entire cell. Around the world, there are a number of groups trying to develop
virtual cells
—complete simulations of the regulatory networks of simple cells—to understand in detail how they operate.

An important finding from such studies is that as the complexity of simulated networks increases, surprising new properties emerge. In a 1999 paper in the journal
Science
, Upinder S. Bhalla and Ravi Iyengar, two researchers from the Mount Sinai School of Medicine in New York City, simulated interacting signaling networks from their experimental studies of a variety of such networks operating in cells. Bhalla and Iyengar found that by linking different networks together (for example, one network produces a control molecule, which controls an enzyme in another network), new properties emerge that were not part of the individual networks. One such property is
persistent activation
. This property is the persistent production of a protein or control molecule, even after the initial stimulus that caused the production in the first place is long gone. Why would persistent activation be useful? Some processes in our cells take a long time—but at the same time, they may be triggered by a short-lived stimulus. Examples
include development, where stem cells need to transform into blood, kidney, or brain cells. Another process where persistent activation is important is the formation of memory. To remember something, our brains must make physical changes to the structure and interaction between brain cells. These changes are triggered by sensory impressions, which become translated into chemical signals. Sensory impressions do not last forever, and neither do the chemical signals derived from them. Yet, we need to remember. Persistent activation makes this possible.