Telomeres! They're the aglets on the shoelaces of eukaryotic chromosomes. They cause cancer, and prevent cancer, and probably have a whole lot to do with "aging". Can't live with 'em, can't live without 'em. They're the poster children for dynamic equilibrium.
DNA is made up of four bases — adenine, cytosine, guanine and thymine — on a sugar-phosphate backbone. The backbone is directional. One end terminates in a phosphate group (the 5' end, for the 5th carbon of the nucleotide to which the phosphate group is attached), and the other (the 3' end) in a hydroxyl group. By convention, DNA sequences are written 5' to 3', so a single strand of DNA looks something like this:
5'-agtcgatcgatagccagtgagact-3'
Single strands bind — A to T, G to C — to make the familiar double helix, and they do this in opposite ("antiparallel") directions:
5'-agtcgatcgatagccagtgagact-3'
||||||||||||||||||||||||
3'-tcagctagctatcggtcactctga-5'
A given run of DNA that actually codes for some protein might be on either strand, and might have associated regulating sequences on the opposing strand. It's all biological and messy, but it works, so hey.
The directionality is thus insignificant as far as "genes" are concerned, but it's a big deal for cell division. Chromosomes are replicated by unzipping the two strands and synthesizing new strands that complement each:
/gccagtgagact-3'
5'-agtcgatcgata
|||||||||||| yoink!
3'-tcagctagctat
\cggtcactctga-5'
/gccagtgagact-3'
5'-agtcgatcgata cggtcactctga-5'
||||||||||||
3'-tcagctagctat gccagtgagact-3'
\cggtcactctga-5'
DNA polymerases — the enzymes that "read" the original strands and synthesize the new ones — can only do so by attaching nucleotides to an exposed 3'-OH. In other words, DNA is always read 3' to 5', and always extended 5' to 3'.
This has the inconvenient side effect that a single-strand of DNA, all by itself, is never enough for a cell to reconstitute a double-stranded chromosome. All the information is there, but the process can't get started without at least one complementary nucleotide with a 3'-OH for DNA polymerase to latch on to.
3'-tcagctagctatcggtcactctga-5' (frickin' useless!)
5'-a-3'-OH (good to go)
3'-tcagctagctatcggtcactctga-5'
How does it get that little bit of "primer" sequence? Using very short sequences of RNA instead of DNA. RNA has the same exposed -OH, but unlike DNA polymerase, RNA polymerase can start a chain from nothing.
5'-agtcgatcgata-3'
||||||||||||
3'-tcagctagctat gccag-3'-OH ← ta-da!
\cggtcactctga-5'
Replication starts at various places all over chromosomes, with the "unzipping" headed in either direction. Replicating the strand that happens to go 3' to 5' in the same direction the unzipping is headed is easy; only one RNA primer is needed, after which DNA can be synthesized non-stop for as long as the "replication bubble" — the section of unzipped single strands — travels down the chromosome. That's the "leading" strand.
The "lagging" strand is a little trickier. As the replication bubble moves along the chromosome, the leading strand is synthesized continuously, but the lagging strand has to be filled in as short fragments ("Okazaki fragments") extending in the opposite direction. These all have to be started by RNA primers, which are later removed (RNA isn't very stable), leaving short gaps. At the interior of the chromosome, these gaps can then be filled with DNA by DNA polymerase attaching to the -OH from the next fragment down.
The ends of the chromosomes, however, are a problem. There's no next fragment. The RNA primer gets chewed up, as RNA is wont to do, and that's it. There's just a bit of single-stranded DNA sticking out, with no way to add a complementary sequence. This 3' "overhang" is necessarily present on both ends of linear chromosomes, but on opposite strands:
5'-atcgatagccagtgagact-3' (world's shortest chromosome!)
3'-tcagctagctatcggtca-5'
More disturbingly, this happens with every replication. Every time your cells divide, your chromosomes get a little shorter; they lose a little more information.
Telomeres — finally, the good part! — are evolution's answer to the "end replication problem", and an elegant hack indeed. They're a short sequence — ttaggg in vertebrates — repeated many times at the ends of chromosomes, acting as a buffer for the rest of the DNA. Since they're repeated, no information need be lost when they fail to completely replicate. Instead, they're replaced by telomerase, an enzyme that includes its own copy of the telomeric sequence as an RNA fragment — 3'-caaucccaauc-5' in humans, matching the 5'-ttaggg-3' repeat (RNA substitutes uracil for thymine). The other component of telomerase is a reverse transcriptase — a special class of DNA polymerase that takes RNA instead of DNA for the source sequence. It still needs a 3'-OH to get started, but, of course, the overhang has one waiting.
ttagggttagggttaggg-3' end of a chromsome
aatccc-5'
ttagggttagggttaggg-3' telomerase RNA binds to telomere DNA
aatccc-5' caaucccaauc
ttagggttagggttagggttag-3' 3' overhang DNA is extended,
aatccc-5' caaucccaauc complementary to telomerase RNA
ttagggttagggttagggttag-3' telomerase proceeds down the
aatccc-5' caaucccaauc chromosome
ttagggttagggttagggttaggg-3' overhang is extended by a full
aatccc-5' caaucccaauc ttaggg repeat
ttagggttagggttagggttaggg-3' now there's room for a regular RNA
aatccc-5' 3-aauccc-5' primer to attach...
ttagggttagggttagggttaggg-3' and regular DNA polymerase can
aatcccaatcccaatcccaauccc-5' extend the 5' strand
ttagggttagggttagggttaggg-3' the RNA gets chewed up, leaving a
aatcccaatcccaatccc-5' fresh, short overhang and
everything has shifted up by one
telomeric repeat.
(Let's all stop a moment to marvel that biology works. I mean, damn.)
As long as telomerase keeps this up, the normal DNA replication process won't lose any information that can't easily be recovered. The telomerase caaucccaauc is essentially a backup of what the ends of your chromosomes are supposed to look like. With telomerase expression turned off, you lose ttaggg buffer until you're within a few replication cycles of losing other, unpreserved sequence, and then cellular machinery that notices short telomeres (by, e.g., binding with lower affinity to proteins that normally bind to telomeres — so if they're available, you know there's not enough telomere to go around) kicks in and shuts down replication. Bam. Cellular senescence.
Or, at least, that's how it goes in healthy cells. If something goes wrong with the senescence pathway, a cell might keep on dividing and start losing coding DNA. Then you've got two problems: you're missing chunks of genes, and, possibly worse, you've got chromosome ends that don't look like chromosome ends. Chromosomes occasionally break, but we've got repair mechanisms that ligate free-floating DNA ends together. They know not to do this to legitimate chromosomes ends by recognizing the telomeric repeat. Once you lose that recognition sequence, your own repair systems try to attach your chromosomes to each other. That causes physical strain, which causes more breaks, and more random ligations, and pretty soon you've got a completely rearranged genome.
Most unfortunate cells just stop there. The chromosomes-in-a-blender game tends to not produce viable results. But every so often, you wind up with something that works just enough to activate telomerase. The ends stabilize, replication resumes, but the cell's normal checks and balances are in ruins and it just goes on replicating the new crazy chromosomal arrangement as long as it can. That's a tumor cell.
So telomerase "causes" cancer, in the sense that tumor cells rely on it. But it may also "prevent" cancer, in the sense that if it'd been active all along, the telomeres would never have shortened enough to start the whole mess. Would you have more mildly fucked up cells, but fewer really fucked up cells? Who knows. More funding, more research, plz.