Transcription factories in a Hela cell [from Cook PR (1999) Science 284, 1790]

Nuclear Structure and Function Research Group

Our Science / An essay
Stationary Traveller by Damian Counsell

The machinery copying our genes doesn't get around as much as it used to...

There is one diagram you are sure to find in most modern textbooks of biology. It shows the copying of DNA - the genetic material - by a polymerase enzyme in the cell nucleus. This duplication of genes takes place in all cell-based living things, from the simplest bacteria to human beings. It is the essential function of life and consequently the central theme in all cell biological studies since Watson and Crick's description of the structure of DNA in 1953 and its accompanying "possible copying mechanism". In every case the classic illustration will show one of two variations on the theme - replication or transcription. In every case it will be misleading.

Replication is rather like photocopying, for posterity, the entire "book of instructions" for a living thing, while transcription is more akin to writing out relevant paragraphs for day-to-day use. Recently cell biologists have produced results that make classroom assumptions about replication (and possibly transcription) seem ever more unlikely. We no longer imagine the complicated and tightly-regulated gene-copying "engine" chugging along a seemingly endless single DNA molecule "track". Instead, a better picture may be one of an extended loop of DNA "recording tape" sliding past the fixed "playback head" of the polymerase complex.

Replication is the copying of DNA into DNA. Replication is a massive and demanding task as any errors in the huge new library of genetic information are potentially fatal. Once duplicated, one copy is segregated to each of the two daughter cells when the parent cell divides.

Easy Assumptions

Molecular biology - the study and manipulation DNA - has changed the way biological and medical researchers work. What is more important, perhaps, is that it has changed the way they think about life. They can "cut and paste" genetic material almost at whim in microbes and lab-grown cells, changing their characteristics as they change the instructions. They can "read out" (or sequence) the "manual" for a given organism. They can multiply a million-fold (amplify) a short passage of code from a single copy. Polymerases synthesised in bulk by genetically manipulated organisms are essential tools in this engineering. These are used in solution along with cocktails of various salts and additives in a routine way in thousands of labs world-wide. This use of dissolved polymerases has become so commonplace that cell and molecular biologists often forget that in real, living cells the reactions they catalyse take place with vastly greater efficiencies. It's easy to be dazzled by the power of our man-made methods of genetic manipulation in the same way a mediaeval horseman would be amazed by the power and speed of an internal combustion engine stuck permanently in second gear. If we travelled back in time to explain to him that we could show him how little of its power he was using he would protest that his wonderous machine "worked just fine". He would insist that his way of using it was the only way. Another seductive belief underpinning the textbook view of replication is one derived from everyday experience. When we see two structures in relative movement we tend to assume that the smaller object is mobile - trains travel along tracks and not vice versa. Perhaps the view of biologists is cluttered by the beautiful computer images they see so often in molecular biology journals, that represent models of components of the replication machinery. Undirected by arrows to indicate the progress of the copying apparatus, readers nevertheless interpret the sight of a stretch of DNA bearing an intimately attached polymerase as being a snapshot of the enzyme's journey along the rest of the packed coil of life. A further distraction is that the polymerase is the active partner in the interaction: since it is doing the work it ought to do the moving as well.

A more sophisticated charm of railway-engine models is the appeal of "Occam's razor": when presented with a phenomenon to explain, never call on any more elements than are absolutely necessary. Whether the philosopher's name is invoked or not, this idea is drummed into every young scientist early in his or her career. Simpler models with fewer bits are more likely to hold together under testing. You could say that the longer an idea holds together in the rough house of science, the "truer" it is. This may not be quite so safe a weapon in biology as it is in physics. Living things are, by the nature of evolution, messy. Models in physics can be beautifully simple. Cell biological models, in contrast, reflect the nature of a fairly sound structure - the cell - built up by endless compromise - evolution - and covered with the ad hoc adjustments - successful mutations - necessary to cope with the changing conditions that challenge life. The new icon of biology - the elegant structure of the double helix of DNA - encourages this kind of "physics-envy".

If we are to believe that polymerases are fixed, what are they attached to? If they are anchored in the swamp of the nucleus, unable to whiz around in search of a suitable foothold, we must explain how the enzyme can reach the DNA. How can we account for these extra "elements"?

Getting knotted

In the standard replication diagram, the DNA molecules are depicted as straight lines, ignoring the helical nature of DNA, but helicity poses several important problems. Viewed from above, a child on a helter-skelter rotates in circles, changing the direction she faces continuously as she progresses. In the same way a "free" polymerase puffing along a DNA molecule must rotate if it is to maintain the same attitude to the genetic template. The cell nucleus is more complicated. The fairground slide is an open helix, but transcription and replication take place on short, uncoiled stretches of DNA. The ends of such loops are attached to the rest of the chromosome and cannot rotate freely to release any extra coiling, or supercoiling, we wind into them. In transcription there is at least one free-ended molecule: a freshly- made copy of the DNA molecule - the transcript.

If our imaginary infant were trailing a reel of ribbon it would wrap around the tower of the helter-skelter as she slid to the bottom. The "ribbon" carried by a polymerase is a transcript. In transcription the analogous process would be the tangling of this transcript with its gene. We escape this tangling problem in the world of the fixed polymerase. Our spiralling infant is frozen in space while the slide rotates upward and past her. She can calmly play her ribbon out over the wall of the slide without it becoming mixed up with the structure. There are no free ends in replication. Somehow the nucleus has devised a way to tease apart the anti-parallel strands of chromosomal DNA at many points along each chromosome and duplicate the content and helicity of each strand until all the segments, or bubbles, fuse, leaving two untangled twin daughter molecules - each carrying exactly the same amount of supercoiling. Analogies in this case are far harder to come by. What is important is that a fixed polymerase model helps us to deal with some of the problems this must involve.

We can think of each assembly of replication components arranged onto a framework in the nucleus in teams, responsible for a specific short stretch of the genome, or replicon. Joining the polymerase molecules there would be speciality players such as helicases and topoisomerases. Helicase enzymes unwind parental DNA so that the other participants can get at it. Topoisomerases cut, unwind and rejoin coiled DNA, rather as you might briefly unplug a telephone receiver to uncoil the lead to the telephone. They may help to untangle mixed up daughter molecules or remove interlocks between strands.

As mentioned earlier, replication is rigorously controlled in time. In our model we could envisage waves of activity spreading across a population of clusters of these sites so they kick into action in a synchronised way. Every loop affected by such a signal would spool into the ends of its specific controlling assembly and emerge from its centre as rabbit-ears of replicated daughter loops, each containing two strands of DNA - one original, one copy. These assemblies could also include a processing site where the short lengths of RNA used to start up each duplicate strand might be removed and any gaps could be sealed up with ligase enzymes.

By arranging specific teams of enzymes like this the cell can more easily ensure that replication of a particular stretch of DNA takes place once only - a team acts on its region in a single event and then attaches to a specific daughter strand region to wait for its next signal. This arrangement might also help to deal with the question of why, within the lifetime of a single individual, a population of skin cells remains stably skin and liver cells stably liver, despite containing exactly the same genes from generation to generation. The specificity of these attachments may be inherited in a similar way to the genetic code, contributing to the way the information is expressed.

It is likely to be far easier to detect unwanted intertwisting in a short region of DNA. In such assemblies the topoisomerases would be readily available to iron out entanglements locally.

Getting in a tangle

Various cell biology studies seem to show that the site of replication may be some fixed structure in the nucleus. The problem with many of these experiments is one common to much work on cells - to study some aspect of theirfunction you may have to treat cells very badly indeed. Typically their delicate, fatty outer membranes will be broken open by shearing them in a lab version of a domestic blender (homogenisation) to get to their nuclei - where all the action takes place. These nuclei have to be separated from the rest of the debris by straining the resulting gloop and spinning it at accelerations equivalent to thousands of times the force of gravity (ultracentrifugation). The nuclei settle out into a distinct layer in a spinning tube and can then be fractioned out and examined. Unfortunately DNA is very sticky, especially in association with the highly-charged proteins that go to make up chromatin. Coiling coiled DNA into this packaged form with such protein components stops it from noodling out uncontrollably all over the nucleus.

To get around the tendency of sticky DNA to aggregate into an "unworkable" mess all sorts of unphysiological hocus-pocus may be used. An unnaturally low or high concentration of salts may be added - essential salts like magnesium are crucial to many kinds of molecular tackiness. To keep dubious nuclear structures like "matrices", "scaffolds" and "ghosts" stable, temperature treatments may be applied. This kind of tinkering with real-life, or in vivo, conditions has tended to give the field a bad name. There is even scepticism about the nucleoskeletons obtained under more plausibly life-like conditions. How is it possible, in a realistically physiological soup, to subject cells to the kind of treatments necessary to strip the nucleoskeleton bare, without creating a useless tangle?

One neat technique is to cage cells in tiny microbeads of agarose gel. Agarose is a sugar obtained from seaweed. Cells - for example, human cells from a lab-grown cancer line - can be taken straight from their growing medium into an isotonic solution and shaken vigorously in a mixture of paraffin and cooling, melted agarose. In the resulting emulsion, pockets of suspended cells are encapsulated by agarose bubbles as they set. These microbeads protect the cells as they are manipulated. It is then possible to perforate their membranes, to reveal their nuclei, without worrying about the chromatin contents spooling out and tangling together into a mass. The network of protein fibres that makes the nucleoskeleton reaches throughout the nucleus like the cytoskeleton ramifies throughout the "clear", jelly-like cytoplasm that fills the rest of the cell. In fact, in electron micrographs the thickness of the filaments and the underlying repeat length of the protein rods that make them up are similar to those of a well-known class of cytoskeletal elements called intermediate filaments. One group of cytoplasmic intermediate filaments, for example, the cytokeratins, helps to give strength to nails, hair and (in sheep) wool.

Test-tube truths

Picture yourself in the cell nucleus. Imagine immense coiled reels of replicating DNA dotted with active polymerases. According to the traditional view, these copying machines are tracking along (replicating) genes or free-floating (dissolved) in the nucleoplasm. Suppose we were to hack through the coils of DNA with a metaphorical pair of shears, cutting at random, and then wash the debris away through holes we had punched in the "wall" of the nucleus to leave only a tiny fraction of the DNA. We would expect little of the copying activity to remain with this residue as most would be washed away with the DNA to which the enzymes were attached. We measure activity by feeding the remnants with fresh DNA building blocks, or nucleotides. It would be shocking if we found these nucleotides becoming incorporated into old strands of DNA as rapidly as before.

When this experiment is done in the lab by full-sized humans the result is completely different. If we chop DNA with enzymes called endonucleases and wash it away by putting nuclei in a powerful electric field that draws the charged chromatin away (electrophoresis), we find that labelled nucleotides can be incorporated just as readily afterwards as they were before our vandalism. This is a far less surprising result if we suppose that the polymerases were anchored to the nucleoskeleton all along . It's also worth noting that the rate at which replication takes place in these experiments on gutted nuclei is far higher than in the absence of any nuclear structure. When biologists use purified, soluble polymerases and DNA in a tube things happen far more slowly. In these encapsulated husks the rates of nucleotide incorporation are comparable with those measured in complete, living cells. Even when we add all sorts of unphysiological chemicals to the mix the recipe behaves as efficiently as our mediaeval's motor car.

Looking and learning

People, even scientists, tend to believe what they see. Some of the most immediately convincing evidence for fixed sites of replication comes from microscopy. Active sites of replication in encapsulated cells can be "lit-up" chemically. Living cells are supplied with analogues of the precursors of DNA which can be added by the cell to growing DNA molecules almost as easily as their naturally-occurring counterparts. They differ in that they contain inert labels. Living cells are allowed to incorporate these analogues for limited times and then fixed. Antibodies - molecules which will bind specifically to the labels - are added. If these antibodies in turn carry fluorescent tags, sites where they have been incorporated into DNA can be seen using a microscope. Fluorescent sites of replication appear as bright speckles or foci. Similar experiments can be performed with the electron microscope. The visible labels are gold particles. In a section of a cell nucleus, an electron fired from the gun of this microscope finds it much harder to penetrate gold than tissue. Particles of precious metal conjugated to specific antibodies show up as black dots on electron-micrographs. The dots cluster around egg-shaped aggregations on the nucleoskeleton called factories - the sites of active replication. Each factory represents a cluster of the assemblies of replication machinery described in our model. If replication is allowed to continue for longer in the presence of labelled analogues the gold particles stuck to freshly copied DNA are found further away from the factories. It seems that the freshly-minted genetic material is extruded from the factories as it is produced.

New models

When asked about Albert Einstein's contribution to scientific thought a typical dinner-party answer would be "everything is relative". Our dinner-party guest might reasonably ask, "what difference does it make whether the train or the track moves? Surely it's their relative movement that matters?". One possible reply seems almost trivial - the other is potentially very important.

If you want build a model to predict the movements of the planets in the sky it's easier to assume that the sun is the centre of the solar system and the Earth and the others orbit around it, though your only real concern is with relative positions. Similarly, some strange phenomena involving polymerases and other "gene-handlers" are more easily explained by picturing the participants in replication and transcription in a different way.

To a researcher, any information that allows him to recreate the conditions of in vivo replication more accurately could improve the efficiency of routine genetic manipulation. When we fix the polymerase in our minds we require in our picture the involvement of components that previously may have been ignored. We could use these chemicals or conditions from life in experiments in the lab to put our labouring genetic engines into higher gear. If gene expression is intimately involved with insoluble activities attached to a protein framework there may be many new targets available for drug action waiting to be investigated. There may also be many other possible explanations for unexplained genetic aberrations, including perhaps cancer. Equally, these new targets may provide novel ways for us to interfere with replication and transcription in chaotically dividing tumour cells and stop the growth of the disease.

© Damian Counsell 1994

Top | Home | Maintained by Peter Cook |