Issue Date: November 29, 2010
Picture yourself in a subway car. When you’re alone or with only a few other passengers, raising your arm to grab an overhead strap or to look at your watch is easy. But in a packed train at rush hour, such simple motions aren’t so easy. Inside cells, proteins face crowded conditions similar to those in that rush-hour train, but scientists have traditionally studied proteins in dilute solutions, as if they had the car to themselves.
Although a given protein or other macromolecule may be present at only a small concentration in a cell, collectively macromolecules occupy more than 30% of a cell’s volume, with total concentrations reaching as high as 300 g/L. These densely packed conditions bring about a situation known as macromolecular crowding. While studying the polymerization of sickle hemoglobin during the mid- and late 1970s, Allen P. Minton, a physical chemist at the National Institutes of Health, recognized that in order to understand the properties of proteins and other biological macromolecules in concentrated solution, nonspecific solute-solute interactions had to be taken into account explicitly. This recognition led to the general concept of macromolecular crowding introduced in 1981. But even though he and others recognized the importance of crowding on protein behavior nearly three decades ago, the idea has only started gaining momentum in the past few years.
In the early years, the few people interested in crowding tended to be physical chemists or biophysicists, Minton says. It wasn’t until 2001 that the biological community began to take notice. Since then, the crowding field has gradually picked up steam as experimental and modeling tools have made such studies possible.
“If you think about it, studying macromolecular crowding is antibiochemistry,” says Gary J. Pielak of the University of North Carolina, Chapel Hill. “Biochemistry is purifying a biological molecule and looking at its properties.”
Now, people are starting to acknowledge that understanding protein properties requires understanding them in cell-like dense conditions, because crowding affects a number of protein characteristics, including structure, function, and activity.
The best-understood effect is the increase in protein stability with concentration. Under jam-packed conditions, proteins tend to favor compact rather than extended conformations. In most cases, this means that the folded, native state is even more stable in a crowded environment than in solution.
This and other effects of crowding can be understood through what are called excluded-volume effects, things that happen simply because molecules occupy space. The more macromolecules there are in a solution, the harder it is for any particular one to find empty space. As a result, making room for a molecule has an entropy cost, Minton says.
Scientists approximate crowded conditions in the lab by adding macromolecular agents such as Ficoll (a commercially available branched polysaccharide), dextran, or polyethylene glycol to a solution of isolated protein. An ideal crowder would be spherical, inert toward proteins, and compatible with all proteins so that any difference between protein behavior in dilute solution and that in a crowded solution would derive solely from excluded-volume effects.
But that’s not always the case. In at least some systems, Minton says, these synthetic agents can have weakly attractive interactions with proteins.
Synthetic crowders make modeling easier, but they don’t realistically mimic the cellular environment, says Huan-Xiang Zhou of Florida State University. “We should abandon the ideal model for theory’s sake; rather, mimetics should be toward the goal of being as realistic as possible.”
To increase realism, scientists have started using proteins as crowding agents, but that tack can be experimentally challenging. “You’re trying to detect and manipulate your teeny-tiny bit of test protein in 300 g/L of other protein,” Pielak says. Tricks that scientists use to probe a test protein—for example, increasing temperature or adding urea to trigger unfolding—cause similar, but undesired, changes in a protein-crowding agent.
And with protein crowding agents, crowding effects become difficult to explain with just excluded volume because the amino acid sequences contribute to interactions between the test protein and protein crowders. “I’m really surprised by how important the actual chemistry is,” Pielak says.
Modelers are now grappling with how to introduce those chemical interactions in their calculations as new types of crowding agents are used. “We need to worry about other types of ‘softer’ interactions,” such as electrostatic and hydrophobic interactions, says Zhou, whose group is developing computational methods to account for these additional interactions.
For proteins with significantly different conformations in the native state, the populations of these conformations can shift because of crowding. From their modeling, Zhou’s group has predicted that crowding increases the population of closed conformations relative to open conformations and affects transition rates between conformations (PLoS Comput. Biol., DOI: 10.1371/journal.pcbi.1000833).
Theoretical physicist Margaret S. Cheung of the University of Houston and her experimental collaborators are showing that crowding affects not just protein structure and stability but also function and activity. She and Martin Gruebele, a chemist at the University of Illinois, Urbana-Champaign, have shown via modeling and experiment that crowding perturbs the function and folding kinetics of phosphoglycerate kinase (PGK) (Proc. Natl. Acad. Sci. USA, DOI: 10.1073/pnas.1006760107).
PGK is an enzyme with two subunits connected by a flexible hinge region. The enzyme, which makes adenosine triphosphate, has binding sites for glycerate and adenosine diphosphate that are widely separated in the crystal structure. Researchers have postulated that a hinge motion brings the two binding sites together when these substrates bind.
When Gruebele looked at the protein under dilute and crowded conditions by fluorescence resonance energy transfer, which reports the proximity of two labels on the protein, he found that the protein becomes more compact and closes up under crowded conditions, even without the substrates. “Instead of being like a dumbbell that has two halves that are separated, each with half of the active site, it may turn out that in this compact state, all of the pieces are stuck together that need to be together,” he says. Cheung’s calculations confirm that in this compact state of PGK, the two halves of the active site are closed up.
Crowding agents paradoxically accelerate the rate at which PGK folds into this compact state even as they slow down the motions of the protein. This phenomenon is in contrast to what happens when smaller molecules such as sucrose are added to the solution. Sucrose makes the solution more viscous and slows down the folding rate. In fact, Gruebele uses sucrose as a check to show that his results are from crowding rather than just an increase in viscosity.
“Crowding in a very natural way enhances the folding of the protein into the native state, as opposed to keeping the unfolded chain,” Gruebele says. “Even though the motion is slowed down, it still takes the protein less time to find the folded state, because it doesn’t have to search extensively through as many different conformations.”
In the case of PGK, the folding rate speeds up as the conditions become more crowded, but only to a certain point. Cheung’s model suggests that above an optimal crowder concentration, PGK is localized in a small enough volume that the structural fluctuations that may be required to cross the folding free-energy barrier are restricted. When concentrations climb beyond this optimum, viscosity begins to dominate and folding rates drop.
Another protein property that is greatly affected by crowding is diffusion. Perhaps not surprisingly, proteins move much more slowly in crowded conditions than they do in dilute solution.
But not all kinds of diffusion are affected equally. For example, Gruebele found that PGK’s folding diffusion coefficient is only a factor of two lower inside living cells than in dilute solution, whereas its translation diffusion coefficient is much more severely reduced (Biophys. J., DOI: 10.1016/j.bpj.2010.08.066).
In hindsight, he says, the results aren’t so surprising. “When a protein is folding, it’s really just sitting in one spot in the cell. As long as the pocket it’s in still looks more or less like a water pocket, the protein only has to wrap itself up to fold. There’s no reason why the folding diffusion should be enormously different from in vitro measurements. It turns out that the local water environment in the cell doesn’t have a viscosity that’s all that different from that in the test tube.” Only when the protein moves around over greater distances does a smaller diffusion coefficient kick in.
Jeffrey Skolnick of Georgia Institute of Technology is focusing on the more crowding conscious translational diffusion, with a goal of simulating dynamic cellular processes. “If you can’t simulate translational diffusion,” he says, “hoping to simulate a cellular pathway is an exercise in futility.”
His work with green fluorescent protein (GFP) indicates that hydrodynamic interactions are as important as steric effects in simulating translational diffusion. For GFP, experiments show that translational diffusion in vivo and that in dilute solution differ by about a factor of 10. But when Skolnick ran a simulation with a mixture of 15 types of macromolecules as crowding agents, the GFP diffusion constant decreased by only a factor of three. Only after he added hydrodynamic interactions could he replicate the slowdown seen experimentally (Proc. Natl. Acad. Sci. USA, DOI: 10.1073/pnas.1011354107).
Adrian H. Elcock and graduate student Sean R. McGuffee of the University of Iowa have done simulations with an even broader array of crowding agents. They constructed an atomic-level model of the cytoplasm in an Escherichia coli cell, using the 50 most abundant macromolecules, and then used that model to simulate translational diffusion of GFP (PLoS Comp. Biol., DOI: 10.1371/journal.pcbi.1000694). By including parameters in addition to steric effects, they were able to reproduce experimental values for GFP’s translational and rotational diffusion constants.
“The work we’ve done is really only a first step,” Elcock says. “It’s a baby step to show in principle we can start building molecular models of intracellular environments.”
But a model with 50 types of macromolecules is still only an approximation of the real system. Lila Gierasch, a biochemist at the University of Massachusetts, Amherst, is leapfrogging over synthetic and protein crowders and going straight to the real system. “We’re not so interested in how a spherical polymer affects folding as we are in how the nature of the cytoplasm affects folding,” she says.
She and Pielak have both tried to work with cell lysates to provide a more realistic crowded environment. Pielak’s group uses freeze-dried cytoplasm. “We have instant cytoplasm that we can redissolve to any protein concentration we want,” he says.
Meanwhile, Gierasch has also been developing techniques to measure folding energetics, kinetics, and intermediates inside cells, a task she calls “horrifically difficult.” In one method, she uses Roger Tsien’s FlAsH biarsenical fluorophore, which binds to a genetically encoded motif in the protein of interest, to reveal whether a protein is folded. Her results show that the protein-folding energy landscape in the cell looks very different from that in vitro.
Many people had hoped that recently developed in-cell nuclear magnetic resonance methods might also offer a way to look at protein behavior in living cells (C&EN, Dec. 21, 2009, page 35), but the methods haven’t yet panned out. The quality of in-cell NMR spectra of small proteins in cells is tantalizing, Gierasch says. “We were all very excited and hoped we could do more with it, but it’s tough.” For example, when one tries to analyze proteins in cells using NMR, the signals of all but the smallest proteins become very broad, she says. “Somehow or other, rotational motion is impeded,” she says. “It appears that dipolar relaxation is just too efficient,” making signals so broad that the in-cell spectra can’t be interpreted.
Shape is another property that people believe will not be affected equally by crowding. For example, globular proteins—ones that are approximately spherical—will be more stable under crowded conditions, but their structures are unlikely to change. The same is true for inflexible nonglobular proteins.
Many metalloproteins, such as cytochrome c, are “very stubborn to crowding,” says Pernilla Wittung-Stafshede of Umeå University in Sweden (Biochemistry, DOI: 10.1021/bi100578x). “Those proteins are very robust. They need to be very rigid, because they hold a metal that’s going to transfer an electron. They’re intrinsically so stable that there’s really not much more you can do.”
The most dramatic example of shape change in a nonspherical protein that Wittung-Stafshede has studied, in collaboration with Cheung, is the borrelia protein VlsE, which is involved in Lyme disease. Under crowded conditions, this protein changes its shape from elongated to spherical, exposing an antigen in the process (Proc. Natl. Acad. Sci. USA, DOI: 10.1073/pnas.0803672105).
The sometimes-dramatic changes that proteins undergo in crowded conditions raise the question of just how relevant are all those years of work in dilute solutions.
“I don’t want to scare the populace,” Pielak says. “What we learn about metabolism in undergraduate biochemistry is all from dilute solution studies of purified enzymes. It turns out that matches pretty much the metabolism in living things, which occurs under crowded conditions.”
Indeed, the people who study crowding defend work with isolated purified proteins as a necessary baseline for understanding the effects of crowding. “It’s good to have done so much in buffer,” Wittung-Stafshede says, “because we need that to build on when we add crowding.”
The experiments in dilute solutions are necessary to make sense of highly crowded conditions, Gruebele agrees. “Having the simpler system allows us to build up toward the more complex,” he says. “I don’t think it’s an accident that a number of people who have spent plenty of time looking at individual proteins under very isolated conditions in vitro are moving toward looking at them in live cells or under crowded conditions of various kinds. We feel that we have gleaned a basic understanding of what’s going on in vitro. We can use that base of knowledge to extrapolate how crowding really affects proteins in the cell.”
Although theoreticians and experimentalists are collaborating to understand crowding and its effects on proteins in real situations, some people would like to see an even closer partnership.
“We’re still at a stage where you do one set of experiments and you make a theory that is tailored to that particular experiment,” Zhou says. “We need a critical assessment of theories and simulations, so that the theories and simulations are put on much more solid ground. These theories have to be established beforehand with all the parameters well determined so they can pass the test of multiple experiments.”
Even though the proteins in cells may be crowded, the field itself is not. Only a handful of researchers focus on these issues, but it is picking up steam. “I think it’s about to explode,” Elcock says.
“Eventually, people will expect that a complete protein data set includes in-cell or crowded studies of various kinds, in addition to the current aqueous buffer data,” Gruebele says. “This is clearly where things will shift in the next five to 10 years.”
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society