Each of our cells contains a meter of DNA, scrunched up to microscopic proportions. This DNA is particularly visible during mitosis, when it is neatly condensed into chromosome brushes, which get pulled and segregated in the remarkable ballet of cell division. We also know that topoisomerases, which can cut this DNA, either nicking on one strand to allow unwinding, or on both strands to allow strand passage (to uncatenate or untangle as needed), are central players in keeping this mess under control, at least in a passive sense. But obviously, to make neat chromosomes, and for countless other tasks, some of which will be described below, more than topoisomerases are required. Another class of proteins called SMC (structural maintenance of chromosomes) supplies some of the lariats, knots, and other rope tricks that are needed to keep our nuclear DNA going in the right direction, and has been the focus of quite a bit of recent progress.
Bacteria generally have one SMC protein, but we have six, paired into three functional ring-shaped complexes: condensin, cohesin, and SMC5/6 (no clever name, unfortunately). They were originally found to function, respectively, in the condensation of mitotic chromosomes, in the cohesion of sister chromatids, and in DNA repair. The SMC proteins pair up to form large rings, of about 50 nm, have an ATPase activity which has been recently found to function as a motor along DNA, and associate with a bunch of other proteins to regulate their activities. It is generally believed that the rings they form can encircle one or two strands of DNA, to provide the pairing and looping functions to be described.
 |
| Cartoon of how cohesin encircles both daughter strands of replicated DNA, to keep them paired until the critical separation point in mitosis. |
For example, cohesins glue together daughter strands after DNA synthesis, so they do not float apart, but can be carried along as pairs into mitosis, and then separated at the right time (anaphase) with a dose of protease which cleaves the cohesin ring. Mutant cohesins cause DNA tangling and loss of proper segregation at cell division, leading to mutations and thence to death, cancer, etc. But having sister DNAs close together is convenient for other reasons as well. If one gets a mutation, homologous repair can use one strand of the good copy to directly invade the bad one, and after excising the bad portion, encode the repair. In fact, the various SMC complexes have somewhat overlapping functions, so some can fill in for defects in the others.
 |
| Electron micrographs of purified condensin, showing the structure of a ring with large blobs (containing ATPase and other accessory proteins) at one end (customarily the bottom) and a smaller blob on the other side where the two SMC proteins also dimerize (called the "hinge" region). Bar is 100 nm. |
Condensin is the main driver of chromosome compaction, looping, and transcriptional domain establishment. Ultrastructural studies of mitotic chromosomes have long shown that mitotic chromosomes are composed of loops- variably sized and perhaps of multiple levels. Condensin has perhaps been the best studied of the SMC family, with beautiful recent work (also here) showing that it forms loops by pumping DNA through its ring structure. In the experiment shown below, a single molecule of DNA, fluorescently labeled, was attached to a surface at both ends. Then a flow was set up in the ambient fluid, towards the top right. When cohesin protein (purified from yeast cells) was added, along with ATP, it selected a site on the DNA, then started forming a collar and pumping out the free portion of the DNA, forming a little loop. At the end of the experiment, several minutes down the line, this cohesin complex spontaneously let go, allowing the DNA back into its original state, waving freely in the flow.
 |
| A single DNA molecule in a fluid flow cell, anchored at two points (red circles), shown through time as it is bound by one codensin molecule, which forms a little pumped loop within a few seconds. |
Such looping is not only relevant for mitotic chromosome structure, but also for transcription. Genes are driven by regulatory protein binding sites, "enhancers", that can be very far from the core coding portion of the gene- often tens of thousands of base pairs away. How does such an enhancer know which gene it is supposed to enhance? It has gradually become clear that genes are surrounded by a zone of isolation with "insulator" DNA sites on the boundaries. Cohesins have recently been shown to be key creators of these zones, binding to the boundary sites and pumping out the intervening DNA, isolating one loop from its neighbors, at least with respect to processively scanning searches by DNA binding proteins, and also with respect to mega-complex fomation by the enhancers of each zone.
 |
| The SMC5/6 complex is proposed to facilitate replication by keeping the daughter strands close so that that topoisomerase II can come in and relieve topological tangles. |
The SMC5/6 proteins perform yet another function, of facilitating replication. Like cohesins, this complex forms rings around recently replicated DNA. The replication fork is itself enormously complex, but as it works, (probably stationary, being fed in sewing machine fashion), the DNA going in and coming out is continuously writhing about to accommodate its helical twist. Imagine a sewing machine working on fabric with an intense twist of one full turn per ten stitches- it would be quite a challenge to operate. Most of the stress can be accommodated on both sides, incoming and outgoing, by continuously nicking and relaxing by topoisomerase I. Yet it is thought be helpful to keep the two daughter strands in close proximity, to allow stand-passing topoisomerase II to undo more serious tangles, and also to prepare for the long-term use of cohesin which keeps the daughter strands together through interphase and into mitosis.
What is known about how the SMC proteins actually operate? That is still a work in progress. In order for condensin to form DNA loops, it needs at least two functions- an anchor to hold on to one region of DNA, and an ATP-using pump that scrunches the neighboring DNA and feeds it through its ring structure. The anchor is quite well characterized. It needs to be relatively agnostic about the sequence it binds to, but once attached, it must stay put while the rest of the molecule does its work. This is accomplished with a special knotting portion of one of the accessory proteins attached near the ATPase portion of the complex. These proteins form a positive charged groove, ready to bind DNA. Once bound, there is also an unstructured extension of one of these proteins (in green below) that comes down to lock the DNA in place, prompting the authors to call it a "safety belt". This structural shift does not require ATP, but is required before the ATPase nearby can become active.
 |
| Structural cartoon of two accessory proteins of condensing (yellow and beige), which form the anchor. This binds DNA non-specifically, and once bound, gets locked in place by a unstructured protein extension and also licenses the ATPase to begin operating to pull neighboring DNA through the ring. The topology of the shown DNA/chromatin is likely wrong. |
What is the ATPase doing? It has just recently been shown that it really is a DNA translocase, (partly as shown above), after some years of doubt. It is also remarkably efficient, traversing at a rate of ~70 basepairs per second, and only using two ATP per second, thus covering about 35 bp per ATP. It must be using the length of its ring to make jumps of some kind- a mechanism more reminiscent of actin or kinesin than of typical DNA/RNA translocating enzymes. Researchers working in this field have proposed a couple of models. One is that the ring separates at the base (right, below) to allow the two ATPases of the paired SMC protein complexes to "walk" alternately along the DNA, taking long strides of up to 50 nm. This obviously risks losing whatever is being enclosed in the ring, so is problematic. The second idea (left, below), is that the extended, coiled portions of the SMC proteins somehow fold and unfold in response to ATP hydrolysis at the end, allowing the complex to take half-steps while rigorously keeping the ring closed. It would be difficult to envision how this mechanism works in detail. It may be that more than one ring cooperate, to resolve some of these coordination issues.
 |
| Rings composed of two SMC proteins (red, blue) are proposed to walk along DNA using their ATPase activity by two alternative mechnisms. Note that the SMC molecules with blobs on the ends are not symmetrical. The ATPase and other significant accessory proteins and activities of the complex are all on the lower (striped) blob of each SMC protein, while the other blob represents a much simpler "hinge" region of the proteins which dimerize but have no other known functions. |
