Telomeres are protein–DNA structures protecting the ends of chromosomes. The crystal structure of a four-stranded stretch of human telomere DNA, bound to K+ ions, has implications for the design of anticancer drugs.
The ends of the linear chromosomes of eukaryotes are capped by telomeres — protective structures composed of repetitive DNA bound to proteins. Every time a cell divides it replicates its genetic material, but the replicating enzymes cannot copy the extreme ends of chromosomes. If genes were present right up to the chromosome ends, they would gradually be eroded with each cell division. Telomeres prevent that from happening and, although they are themselves eroded, they can, under some circumstances, be renewed by the enzyme telomerase1. Among their other functions, telomeres also stop chromosomes from fusing end to end.
The DNA of human telomeres consists of repeats of the nucleotide sequence TTAGGG, ending in a single-stranded segment that overhangs at the end of the double-stranded DNA helix. On page 876 of this issue, Parkinson and colleagues2 describe striking crystal structures of single-strand sequences — consisting of two or four TTAGGG repeats — in the presence of K+ ions, revealing topologies that could readily be incorporated into a higher-order DNA architecture.
In these structures the single-stranded repeats fold up into four-stranded (quadruplex) topologies2. The basic building block of any quadruplex is a G●G●G●G tetrad, composed of four hydrogen-bonded guanine nucleotides in a horizontal planar arrangement3 (Fig. 1a). Individual tetrads stack up on each other, with monovalent cations (Na+ or K+) sandwiched between them4. With TTAGGG repeats, the GGG sequence forms a vertical strand (column) of the quadruplex; the TTA sequence loops over, joining into the GGG sequence of the next repeat, and so on until there are four vertical GGG columns, three horizontal tetrads deep. The GGG columns can adopt either parallel or anti-parallel alignments (they are oriented in the same or opposite directions), depending on how the connecting TTA loops are oriented (Fig. 1b–d). Edgewise loops connect adjacent antiparallel strands; diagonal loops connect diagonally opposite antiparallel strands; and double-chain-reversal loops connect adjacent parallel strands.
So how do Parkinson et al.'s crystal structures look2? In the four-repeat structure, the GGG columns are all parallel to each other, and the three connecting loops are of the double-chain-reversal type2 (Fig. 2a). Moreover, all three loops are splayed out like a propeller from the main body of the quadruplex (see Fig. 3b on page 878). They can thus be recognized by proteins that are either components of the telomere complex or part of the telomere-associated nuclear membrane and matrix.
Parkinson et al. also observed the same parallel-stranded alignment in crystals of the two-repeat quadruplex in the presence of K+ ions. This quadruplex is formed when one two-repeat sequence pairs up with another. It seems that this requires the formation of A●T●A●T tetrad planes, thus expanding the tetrad alphabet beyond G●G●G●G tetrads. Such mixed tetrads might serve a valuable role in cells: they could direct the pairing of homologous chromosomes to enable the chromosomes to 'recombine', swapping segments of DNA.
So the two-repeat structure sheds light on how homologous chromosomes might bind to each other during recombination. The four-repeat structure shows how the basic topology could protect the ends of individual chromosomes. Interestingly, the same four-repeat human telomere sequence adopts a completely different quadruplex architecture in a solution of Na+ ions5 (Fig. 2b). Here, the opposing GGG columns are antiparallel, and there are one diagonal and two edgewise TTA loops. Different monovalent cations can therefore alter the four-repeat quadruplex topology. This might be because K+ cations (which have an ionic radius of 1.51 Å) are invariably sandwiched between adjacent guanine tetrads6, whereas Na+ cations (ionic radius 1.18 Å) can sometimes be coordinated within a tetrad7. Single-nucleotide changes within the telomere repeats might also have a say in cation-mediated folding topology. In mammalian cells, however, telomere repeats are more likely to form K+-coordinated quadruplexes as the intracellular K+ concentration greatly exceeds that of Na+ (140 mM versus 5–15 mM).
The distinct topologies adopted by the K+-coordinated and Na+-coordinated four-repeat sequences could have implications for higher-order packing. The diagonal and edgewise loops on the top and bottom of the Na+-coordinated structure5, together with the fact that the two ends of the sequence are located on the same face, would stop quadruplexes from stacking up on top of each other.
But there is no such barrier to the stacking and subsequent packaging of the compact, disc-like quadruplexes in the K+-coordinated structure2: the double-chain-reversal loops are directed outwards in a radial orientation and the chain ends are located on opposite faces of the quadruplex. Double-chain-reversal loops could also facilitate the necessary folding and unfolding of stacked quadruplexes during chromosome replication, without potential complications from the formation of knots in the structure.
Could the quadruplex structure that is described by Parkinson et al. help to protect chromosomes from fusing or recombining inappropriately, and stop telomeres from being mistakenly recognized by the cellular DNA-repair machinery as broken ends? In general terms such events are prevented by telomeres forming a complex with specific proteins8, or by insertion of an overhanging single-strand telomeric sequence into an adjacent telomeric double-stranded segment through a process known as 't-loop' formation9. It remains to be seen whether quadruplex formation can contribute to this process.
Beyond fundamental chromosome organization, the new results2 might also have clinical implications. The enzyme telomerase is needed for complete duplication of telomeric DNA during cell division1. In humans, it is highly expressed only in tumour cells, enabling them to carry on replicating their chromosomes almost indefinitely. (In non-tumour cells, by contrast, telomerase is not expressed and telomeres gradually erode to the point at which cells cannot duplicate their DNA safely, and so no longer divide.) So telomerase is a promising drug target. It binds to single-stranded telomere ends, and could potentially be inhibited by drugs that compete for these ends (in their quadruplex form). The structure-based design of such drugs10 necessitates a molecular understanding of the range of quadruplex topologies. So the new structure2, together with the previous Na+-coordinated structure5, is valuable in providing such information.
Finally, the quadruplex scaffold adopted by guanine-rich sequences has been observed in other contexts. For instance, quadruplexes have been associated with genomic regions that control gene transcription, and have been identified in vivo by antibody staining in the nuclei of certain protozoans11. Moreover, RNA quadruplexes have been identified in mammalian brain messenger RNAs that are recognized by the fragile X syndrome protein12. So the architecture of quadruplexes, and their interactions with proteins and other DNA and RNA sequences, should continue to be an active area of research with broad therapeutic implications.