Location of DNA target sites by proteins and nucleic acids

Summary MOLECULES IN MOTION. Down in the world of the very small / of atoms and molecules / there is madcap motion. Molecules whizz around, and do so very quickly, even at room temperature. This was first discovered nearly 200 years ago by the English botanist Robert Brown. He noticed that tiny grains of pollen dust jiggled about in water, following a random zigzag path. Movement of particles in this chaotic manner became known as 'Brownian' motion, and it was Albert Einstein who developed the first mathematical description of particle diffusion in 1905 (his first Nobel Prize). In air, molecules can zoom several centimetres (a long way) before hitting another molecule. In liquids, molecules jostle around each other very quickly, but don't travel very far at all in a 'straight' line - they always bump into neighbouring molecules. A useful way to think of how molecules diffuse is to imagine a game of football: the players on pitch run around in a lot of space, occasionally bumping into each other trying to get the ball / this is like the behaviour of molecules in air. Now consider a football fan up in the packed stadium trying to run through the crowd / impossible! He would collide endlessly with the other fans, and it would take a long time indeed to move any appreciable distance / this is like the behaviour of molecules in liquid: jostling, but constrained by the crowd. The slowness of diffusion in solution is easy to observe: a sugar lump in water takes a long time to dissolve. One way to speed up diffusion is heating: in hot water the sugar dissolves more quickly. Another way to facilitate (speed up) diffusion is of course to mix the molecules - why we stir our tea! The process of facilitated diffusion also happens in another quite different system, but without involving heating or stirring. The first thing to know is that the molecule DNA (deoxyribonucleic acid) is very good at binding certain proteins that help it carry out its cellular roles: processes like replication and gene expression. The second thing to know is that these DNA-binding proteins don't work properly unless seated on a few very specific patches of the DNA chain. So here's the problem: the protein and its binding site on DNA have to diffuse together in order to function, but since DNA is an extremely long chain, this would take a very long time to happen by chance: a bit like hunting for a single loose railway sleeper on the London to Edinburgh line - blindfolded! So how does site location occur? The answer is: facilitated diffusion. The protein and DNA first of all encounter one another in solution by way of random Brownian motion. Odds are that the protein will not hit the specific sequence straight off; the chances of that are extraordinarily low. Though the protein is best shaped to fit its specific target sequence, it will bind non-specific DNA, forming a loose complex in which the protein can slide up and down the DNA a bit. In a short time however, this complex falls apart. But since diffusion prevents the protein moving very far away (remember the football fan in the crowd), it remains very close to the DNA and is likely to be recaptured at the same or a slightly different position on the DNA chain. Now continue this process repeatedly, and it is possible to imagine the protein hopping, jumping and sliding its way through the DNA, until it eventually arrives at its preferred sequence, where it binds more tightly and can carry out its biological function. This facilitated search process / limiting the protein to hunt around inside the tiny volume of solution occupied by the DNA - is much more efficient than random Brownian motion. There are many different types of DNA binding protein, and work in this laboratory looks at how these proteins use facilitated diffusion to locate their DNA targets. Such work will contribute to our understanding of how molecules move, and the dynamics of macromolecular motion in a cell nucleus.

Technical Summary The location of DNA target sites occurs in every aspect of genome metabolism. Sequence-specific binding of a protein to its recognition site is the most familiar form of DNA site location: a host of examples of this kind exist in transcription, replication, recombination and restriction-modification systems. Structure-selective binding is another form of site location: for example, repair systems have to scan DNA molecules for particular configurations indicative of damage, such as backbone nicks or mispaired bases. DNA is also recognised by molecules other than proteins: an example of sequence-specific binding is triple helix formation, in which one DNA strand binds to a specific duplex sequence. These site-specific binding molecules, whether protein or nucleic acid, still face the same problem: how to search a DNA molecule as efficiently as possible in order to find their preferred target site. It is extremely unlikely that their initial encounter with a long DNA chain is at the exact site. Rather, it will be at a non-specific sequence, followed by a transfer process that conveys it to its site. This is known as facilitated diffusion. Various schemes have been proposed for facilitated diffusion: 'Sliding', 'Hopping/Jumping' and 'Intersegment transfer'. During 1D sliding (linear diffusion), the protein remains bound to the DNA and undergoes a series of repetitious bidirectional steps that allow it to scan a relatively short section of DNA very efficiently. During 3D hopping/jumping, the protein moves within the domain of the DNA chain, stepping on and off the DNA chain repeatedly. Usually, rebinding occurs at or near its departure point (a hop), but occasionally further away (a jump). Intersegment transfer applies only to proteins that possess two DNA binding surfaces, such as lac repressor or SfiI endonuclease. In the simplest form of this scheme, the protein transfers from one segment of non-specific DNA to another by transiently bridging between the two using both binding clefts. This type of 'Tarzan' motion is limited to large step sizes (>400 bp) due to the rod-like nature of DNA chains. It is likely that each of the binding clefts also undergoes some form of local sliding or jumping motion, thereby allowing the protein to take smaller steps. The starting point for experiments on DNA site location are usually length-dependency studies. These test whether a DNA binding molecule uses non-specific sequences on the pathway to its target site. A series of DNA molecules with one site for the protein of interest are constructed, with different lengths of non-specific DNA surrounding the site. The rate at which the protein associates or dissociates from its target is measured on these substrates. However, such experiments don't reveal the intramolecular transfer mechanism. Sliding is often invoked, when hopping could provide an equally convincing explanation of the data. Until the turn of the millennium, there were no assays that could unambiguously distinguish between 1D and 3D behaviour of DNA-binding proteins. The first assay that could do so correlated restriction enzyme processivity with site location mechanism. The second assay, on plasmid, catenane and minicircle substrates, used differences in the topology of the non-specific DNA to distinguish 1D from 3D events. Together, length-dependency studies, processivity measurements and catenane experiments are decisive for determining the existence of a transfer pathway and the nature of the transfer mechanism - whether primarily 1D or 3D steps are taken. The objectives of this project are to use similar experimental strategies to investigate how proteins and nucleic acids locate specific target sites in long DNA molecules. Specifically, to find out how DNA ligase scans for nicks in the DNA backbone, how triplex-forming oligonucleotides (TFO's) find their homopurine targets, and how the looping enzyme SfiI finds two sites in the same DNA chain.