How are DNAs woven into chromosomes?

See allHide authors and affiliations

Science  03 Nov 2017:
Vol. 358, Issue 6363, pp. 589-590
DOI: 10.1126/science.aap8729

It was not recognized that chromosomes contain one immensely long DNA molecule until Kleinschmidt et al. published electron micrographs of a lysed bacterial cell with DNA spilling out that contained no apparent breaks and, indeed, not even any free ends (1). We now know that before replication, each of our chromosomes contains a single DNA molecule of immense length. If the DNA of an average human chromatid were a wire with a diameter of 2 mm, then this wire would be 50 km long. The vast length of chromosomal DNAs poses a number of fundamental problems. For example, how are they packaged during cell division into cylindrical threadlike chromatids? On page 672 of this issue, Terakawa et al. (2) describe an activity associated with a protein complex called condensin that has the potential to answer this. Remarkably, it might also explain chromosome segregation in bacteria as well as the mechanism that regulates enhancer-promoter interactions during development.

The first step in solving complex questions is to articulate precisely what needs explaining. In this case, there are numerous questions. How are chromosomes packaged into threads as opposed to balls of chromatin, which would be the natural product of a condensation process? How is DNA belonging to the same molecule invariably packaged into the same chromatid whereas that of its sister (replicated) chromatid is packaged into an adjacent chromatid? A related question is, what facilitates resolution of the sister DNA intertwining (catenation) that arises after DNA replication? DNA catenations are removed by topoisomerase II, which functions by passing DNAs through a transient break in their sister molecules. This reaction could cause reentwining, but it usually does not. Lastly, what is responsible for the rigidity and elasticity of the longitudinal axes of chromosomes that prevents them from being unraveled when pulled on by the mitotic spindle in the process of sister chromatid separation?

We now know that all these features depend on members of a new class of enzymes that comprise structural maintenance of chromosomes (SMC)–kleisin complexes. These are thought to associate with DNA molecules by entrapping them inside a “ring” structure created by the binding of N- and C-terminal domains of kleisin subunits to adenosine triphosphatases (ATPases) at the apices of V-shaped SMC dimers (see the figure). The organization of DNA into chromatids during mitosis depends on a version of SMC-kleisin complexes called condensin, whereas holding sister chromatids together depends on a version called cohesin. In both complexes, related hook-shaped proteins composed of HEAT (huntingtin/EF3/PP2A/tor1) repeats (called hawks, HEAT proteins associated with kleisins) bind their kleisin subunits and regulate their ATPase activity and transactions on DNA (3). Similar Smc-kleisin devices promote chromosome segregation in bacteria, but with kite (kleisin-interacting winged-helix tandem elements) regulatory proteins. Regulating higher-order nucleosome structures cannot therefore be their modus operandi, because bacteria lack histones. Indeed, condensin has recently been shown to organize DNA into chromatidlike threads in Xenopus extracts in the complete absence of nucleosomes (4).

Cohesin is thought to hold sister DNAs together by entrapping them inside its ring, but how condensin functions is more mysterious. By analogy, it has been suggested that condensin entraps two segments from the same DNA together and thereby “crosslinks” chromosomal DNA together. A variation on this theme is that condensin rings that have entrapped DNAs interact with each other through weak multivalent interactions between their hawk subunits (5). An alternative model postulates that condensin creates chromatids by supercoiling DNA (6). However, none of these provide an explanation for how chromatids emerge from these postulated activities. The nub of the problem is how to avoid interactions brought about by three-dimensional diffusion, which would otherwise entangle DNA molecules together.

If important phenomena cannot be explained in terms of what we already know, then one must invent something. As Francois Jacob put it, “The process of experimental science does not consist in explaining the unknown by the known, as in certain mathematical proofs, it aims to give an account of what is observed by the properties of what is imagined.” Remarkably, most of the key features of chromosome morphogenesis can be explained by postulating that condensin possesses the ability to trap small bights of DNA and then extrude these in a processive manner into longer and longer loops. According to this loop-extrusion (LE) theory (7, 8), individual loop-extruding condensins eventually bump into each other, whereupon the loops associated with each complex are arranged into a helical spiral, with the result being that condensins line the longitudinal axis of the chromosome around which loops radiate (see the figure). Whether the encroachment of loop-extruding condensins leads to specific, albeit weak, interactions between them that create the narrow helical axis along which they accumulate is a key unanswered question. Interestingly, most plant and animal cells possess two types of condensin: Condensin II is thought to build loops earlier, eventually building loops up to 400 kilobases (kb) in length, whereas condensin I has been postulated to extrude shorter loops within those created by condensin II (9).

LE explains why chromatids are cylindrical and what constitutes their longitudinal axis. According to LE, the chromatid axis is not an independent proteinaceous scaffold, as previously postulated (10), but instead includes what may be very short segments of DNA between extruded loops (linker DNAs). LE also explains how sister DNAs assemble into separate chromatids and how all DNAs extruded into loops lack catenation with their sisters. Catenation will accumulate within linker DNAs on the axis, as neighboring loop extruders move toward each other. The concentration of topoisomerase II (11) and catenation on the axes facilitates sister decatenation. It also helps ensure that looped DNA is not recatenated.

There has been a fundamental problem with the LE hypothesis. Hitherto, there has been no evidence that condensin can actually extrude loops of DNA—hence, the importance of the findings from Terakawa et al. showing that condensin translocates in a highly processive manner along DNAs stretched out on a glass slide in vitro. Crucially, the movement of condensin depends on ATP hydrolysis. The complexes translocate at 3.5 kb/min and do so for tens of kilobases. Unlike other DNA translocases, which move in 1 base pair (bp) increments, the step size of condensin is about 30 bp. An important issue that was not addressed is whether the complexes observed in the act of translocation contained single or multiple condensin rings. This is a highly important issue, as one way of imagining how translocation could be turned into LE is if two different complexes were to translocate in opposite directions while somehow joined together.

Structure of condensin and its role in loop extrusion

The condensin-kleisin-hawks complex forms a ring that binds DNA and contains ATPase activity that is required to move along DNA. The condensin complexes extrude loops of DNA. As separate complexes approach each other, chromosome loops are arranged around the longitudinal axis to form threadlike structures.


If LE is the primary function of condensin, why was it observed to translocate and not to extrude DNA into loops? The answer may lie in the design of the experiments. The DNAs were stretched out on slides with each end tethered. Thus, even if an individual complex tried to extrude DNAs, the tension in the DNA would not have permitted it to do so, and translocation instead of LE ensued. Interestingly, when DNAs are not tethered at both ends, condensin causes them to fold up in a manner that might be due to LE. The discovery that condensin is a DNA translocase is certainly consistent with the idea that it functions as a loop extruder but by no means proves it. The challenge will be to observe extrusion as well as translocation, to establish whether it is a property of individual or multimeric complexes, and to elucidate the molecular mechanism.

The organization of DNAs into chromatidlike threads by LE is also thought to be a property shared by cohesin as well as the kite-containing bacterial complexes. It has been suggested that enhancers and distant promoters are brought into proximity by cohesin-mediated LE. In this case, a site-specific DNA binding protein, CCCTC-binding factor (CTCF), prevents illegitimate interactions between enhancers and the promoters of neighboring genes, known as insulation. CTCF has been postulated to halt the extrusion process (12), presumably by inhibiting the cohesin ATPase (13). Remarkably, lengthening its residence time on chromosomes enables cohesin eventually to overcome insulation by CTCF, leading to the formation of chromatidlike threads during interphase (14). Thus, cohesin and condensin seem to be interphase- and mitosis-specific loop extruders, respectively. Meanwhile, bacteria are thought to zip up the two arms of their circular chromosomes during replication by extruding them through Smc-kleisin complexes (15). It will be important to establish whether cohesin and bacterial Smc-kleisin complexes also share the ability to translocate along DNA.

In describing the discovery of atoms, the great French chemist Jean Perrin said, “We certainly observe these visible pieces as closely as we can, but at the same time we seek to divine the hidden gears and parts that explain its apparent motions. Our task is to explain the complications of the visible in terms of invisible simplicity.” The invisible simplicity responsible for creating complex chromosomes out of immense tangled skeins of DNA are SMC-kleisin complexes. The findings of Terakawa et al. are just the beginning of the story. Seeing this happen inside cells, as well as in a test tube, represents another huge challenge.


View Abstract

Stay Connected to Science

Navigate This Article