Dennis Bray takes a video from the shelf and sticks it into the VCR. On the TV screen, some bacteria flicker into view. To be honest, they are not much to look at. Single-celled and rod-like, they鈥檙e bald as coots save for a half-dozen flagella sticking out proudly from the surface. But we are not here to admire bacterial appearances. Bray, a biologist at the University of Cambridge, is demonstrating their remarkable talent for navigation.
Initially, we watch as the bacteria 鈥 Escherichia coli 鈥 move about almost aimlessly. They swim in lines, then tumble. After a moment, the tumbling ceases, and they swim off again in a new and random direction. But when someone adds to the setting some bacterial food 鈥 a few aspartate molecules 鈥 tumbling stops as they lock on to the scent and head for a molecular feast. If intelligence is the ability to sniff out a decent meal, then bacteria are among the intellectual elite.
Chemotaxis, as this bacterial food finding is called, illustrates how bacteria can detect and respond to information in their environments. But Bray isn鈥檛 so much interested in bacteria, as in what their workings might tell us about this 鈥渃ell signalling鈥 in other creatures 鈥 from crickets to human beings. In every one of your cells, something similar is going on: millions of molecules are working together in biochemical networks to gather information from outside the cell, and to trigger appropriate action within.
Advertisement
In more complicated cells, of course, the signalling is more intricate, which is the attraction of a bacterium. The cell signalling that carries it to its next meal is about as simple as it can get. 鈥淲ith chemotaxis,鈥 says Bray, 鈥測ou can build detailed models on the computer and use them to test your understanding. That鈥檚 what we鈥檝e been doing.鈥 And on the strength of these mathematical models, he and other scientists are beginning to uncover the sort of tricks that all cells use to handle the flow of information essential for life.
Despite its simplicity, chemotaxis has always contained something of a puzzle for biologists. On the one hand, bacteria are remarkably sensitive to environmental signals 鈥 they can detect and respond to just a few molecules. But they can also maintain this sensitivity, this ability to sense tiny gradients in concentration, even when the concentration of the attractant soars by as much as 100 000 times. Our eyes do something similar: the light-sensitive cells of the retina that can detect as little as a single photon of light can also distinguish tiny visual contrasts over a huge range of light intensities. In engineering terms, eyes and bacteria have a wide 鈥渄ynamic range鈥: they can deal with weak or strong signals alike.
You might imagine a bacterium as a sausage-shaped balloon with 2000 or so miniature egg cups protruding through its surface. These egg cups are the protein receptors 鈥 the business end of the biochemical network 鈥 which carry signals through the cell membrane. Each receptor is a switch, and has two possible shapes corresponding to 鈥渙n鈥 or 鈥渙ff鈥.
When on, each receptor stimulates an internal network of chemical reactions that rotate the flagella 鈥 hair-like helical structures that work like a ship鈥檚 propeller 鈥 clockwise. This sends a bacterium into its tumbling routine. When off, the receptors tend to make the flagella rotate the other way, propelling the bacterium in a straight line. Of course, what happens depends on the overall activity of the system 鈥 the sum total of receptors in the on state. Ordinarily, only about half the receptors are on, which is what gives the bacterium its weird exploration strategy 鈥 sometimes it swims, at others it tumbles.
So how does a bacterium find food? In this simple view, if a food molecule binds to a receptor, it tends to flip that receptor off. This alters the overall balance of receptors, and swimming wins out over tumbling. If the bacterium is swimming in the right direction when it detects food, it will just keep going. If it is tumbling and starts swimming in the wrong direction, food concentrations will fall, receptors will turn on and it will start tumbling again. By trial and error it will eventually end up swimming in food.
Sensitive swimmers
This is a tidy picture, simple and straightforward, and, to some degree, correct. Trouble is, it falls to pieces when pushed for quantitative answers. It is far from clear, for example, how a bacterium can be sensitive to only a few food molecules. Individual receptors exert only a minuscule influence over the flagella. To have much effect, you would need many switched to the off setting. But that ought to require lots of food molecules, and high concentrations. Low concentrations shouldn鈥檛 trigger enough receptors to make any difference 鈥 but they do.
A few years ago, Bray and his colleagues tried to understand this sensitivity in terms of some mathematical models for receptor activity. 鈥淲e tried various contortions to explain it,鈥 he says, but with little success. Then, in 1993, Janine Maddock of the University of Michigan discovered something odd about the way in which the protein receptors were distributed on the bacterial membrane. Instead of being randomly scattered, most of the receptors seemed to aggregate into clusters at each end of the bacterium. 鈥淭hat was unexpected,鈥 says Bray. 鈥淣obody knew what it meant, why it was there, or what the possible consequences were.鈥
Dominoes
Clues to its possible significance came when Mike Manson of Texas A&M University discovered that nearby receptors can influence one another. That may seem like a minor finding, but to Bray this immediately suggested an answer. His idea was this. Suppose that a receptor binds an attractant molecule. This receptor turns off and, by 鈥渢alking鈥 to nearby receptors, also turns off its neighbours (see Diagram). So although a single receptor cannot influence the flagella enough to alter a bacterium鈥檚 behaviour, by enlisting other receptors in its cause, it can. Cooperation is the key.
Bray embodied this 鈥渄omino effect鈥 in a mathematical model by supposing that each receptor, when triggered by an attractant, would respond and take with it a fixed number of nearby receptors. This simple assumption elegantly explained the high sensitivity of bacteria. But it made a mess of the other half of the puzzle-their broad dynamic range. 鈥淚t was easy to see how you could get a low threshold of response,鈥 says Bray. 鈥淏ut if one attractant molecule affects, say, 100 receptors, then it doesn鈥檛 take long before everything is affected, so that you can鈥檛 detect any more.鈥
So to get high sensitivity, many receptors have to respond to just one attractant molecule. But if a large number respond to just one molecule, then not too many more will saturate the entire system. It seemed an irreconcilable conflict. Yet after much head scratching, earlier this year Bray and his colleagues found a way to resolve the mystery (Nature, vol 393, p 85). Their hunch was that both sensitivity and dynamic range might be achieved if the 鈥渋nfectivity鈥 of the receptors could decrease as the bacteria encounter ever higher concentrations of attractants.
How might that help? In one scenario, all the receptors would, at low concentrations, form one giant cluster. In this case, because of the domino effect, even a few attractant molecules could trigger a response from all the receptors. At higher concentrations, to preserve the sensitivity, the cluster might break up into a set of smaller clusters. This would prevent a response from spreading to all the receptors and saturating the system. At very high concentrations, there might be no clustering at all-each receptor would behave independently.
The idea that clusters might change in this way becomes plausible when you appreciate how mobile protein receptors really are. The surface of a bacterium is more of a liquid skin than a solid one, and the protein receptors diffuse freely over its entirety. 鈥淧eople tend not to have an intuitive feel for how fast diffusion is; a freely diffusing protein would be visiting the entire surface of a bacterium in a tenth of a second,鈥 says Bray. So clusters form and dissolve all the time. What we鈥檝e been describing so far is more like a snapshot of what鈥檚 really going on. If increasing concentrations of attractant tend to inhibit the formation of larger clusters, then this explanation might be right.
Another possibility is that changing cluster size has nothing to do with it. The chemical interactions between receptors might simply grow weaker as attractant concentration grows. 鈥淓ither mechanism,鈥 Bray says, 鈥渃ould be used to produce both the range of response and the sensitivity.鈥 But the 鈥渃ould be鈥 is revealing. Nobody has a clue whether receptors really cluster and disperse depending on the concentrations of attractant, nor if altered concentrations can influence the interactions between receptors.
Bray is the first to point out the limitations of his model, and he is almost downbeat about his breakthrough. It was, in his view, only an initial stab at the problem. Now he鈥檚 more enthusiastic about the work of two biophysicists, Yu Shi and Tom Duke from the Cavendish Laboratory at Cambridge University, who have taken his ideas a step further. 鈥淒ennis [Bray] asked us whether there were any systems in physics that we could use to look at this problem,鈥 says Duke. Their answer was yes.
In Bray鈥檚 model, when an attractant molecule binds to a receptor in a cluster, the signal spreads to a fixed number of nearby receptors-say 100 or so. 鈥淚t is difficult to see,鈥 says Duke, 鈥渉ow that could come about on a molecular level.鈥 Receptors ought to influence their nearest neighbours, but shouldn鈥檛 be able to 鈥渞each out鈥 to others further away. So Duke and Shi have come up with a similar but subtly different style of infectivity, by considering the coupling between neighbouring receptors only.
They have also included the effects of noise in an explicit way. Bacteria live in a world made 鈥渘oisy鈥 by the ordinary thermal banging about of molecules. This noise rattles receptors, making them flicker randomly and rapidly between the on and off states. This doesn鈥檛 stop them interacting, however. Amid the haze of random flipping, off receptors have a tendency to turn other receptors off, whereas on receptors turn their neighbours on.
Like a magnet
Shi and Duke鈥檚 model is grounded in biological reality, but it has a direct counterpart in physics. 鈥淭he model naturally corresponds to the Ising model of statistical mechanics,鈥 says Duke, which is the model that physicists use to describe the behaviour of systems made of many interacting elements, especially magnetic systems. A piece of iron, for example, contains tiny magnets or 鈥渄omains鈥, which can point in different directions. Adjacent domains tend to line up, and their interactions determine the state of the iron. If the domains succeed in overcoming the disruptive jostling of thermal noise, then the order can spread across the piece of iron and magnetise the entire system.
For Shi and Duke, the receptors play the role of the magnetic domains. Receptors can be on or off, and nearby receptors try to agree with one another (see Diagram). The great thing about the Ising model is that physicists have studied it in exhaustive detail. Shi and Duke believe it can illuminate chemotaxis and cell signalling as well. Here鈥檚 how chemotaxis seems to work, based on the Ising model.
In the system鈥檚 resting state, receptors flip between on and off at random, and on average, half are on. If a molecule binds to a receptor, this doesn鈥檛 turn it off permanently, but merely makes it more likely to be off than on. So the receptor spends slightly more than half the time being off. Because it is coupled with its neighbours, these receptors will also tend to be off more frequently. 鈥淭he extent of this effect depends on the coupling between the nearest neighbours,鈥 says Duke. In the Ising model, making the interactions between neighbouring domains stronger lets order move across the system more readily. 鈥淚f you make the coupling very strong, then all the receptors will be switched to the same state,鈥 he adds. 鈥淔or high sensitivity, all you need is the coupling strength of the system to have evolved to the right level.鈥
That鈥檚 the sensitivity taken care of. What about the dynamic range? Here another aspect of cell chemistry comes naturally to the rescue. When a receptor turns off, its properties become slowly modified by a reversible chemical process known as methylation. Biologists have known for years that methylation was somehow involved in enabling bacteria to adapt to concentration changes. Bray and his colleagues suspected that it might help explain how bacteria maintain sensitivity over a broad dynamic range, but they couldn鈥檛 see quite how.
Duke and Shi鈥檚 model gives a clue. If a bacterium enters a high concentration region, many attractant molecules bind to its receptors. As a result, these receptors will be biased towards being off, the very thing that tells the flagella to move accordingly. Methylation fits into the model as an influence that can undo this bias, by slowly altering the chemical properties of receptors, tuning them back towards having an equal probability of being on or off. Every receptor in the network is slowly 鈥渄esensitised鈥, which brings it back to a fluctuating state evenly balanced between on and off. When that happens, the system is again able to respond to changes in concentration (See Diagram).
Given the modest tools that bacteria have, they seem remarkably successful at searching for scents. But perhaps we should not be too surprised. Shi and Duke鈥檚 simple model manages to balance on the knife鈥檚 edge between sensitivity and dynamic range. Signalling can be simple. There may not be any magnets in cells, but simple models stolen from physics might be ideal for understanding the more complex biochemical networks that lie at the heart of cell signalling. If Shi and Duke are right, then cell it is nothing but a flux of flipping proteins.
