By selecting and rearranging several different coding sequences from three different chromosomes, each lymphocyte develops its own antibody "design", permanently changing its own genome in the process. This is what allows us to develop immunity to a broad range of foreign antigens with a relatively small amount of genetic code. Think of it as an "immune alphabet", if you will. You certainly don't want to interfere with that!
Other rearrangements also occur, some harmful, others not so much. Even if you wanted to target just the bad ones, though, a virus really isn't the tool for it. Your cytoxic T lymphocytes already do that. In fact, what you're really describing isn't all that far off from what we already have. I think we even have mechanisms for genetic error correction and repair, but I'll have to hit the books again to double-check that.
Obviously it would be more complex than
if hash(nuclear DNA) != clean_DNA_hash) die;
but I wonder to what extent the immune system could be enhanced. I wonder if there are some techniques that could be adapted from the computer virus/malware detection field back into biology.A recent article on creating false-positives for a virus scanner (http://lock.cmpxchg8b.com/aids8064.html) by analysing the signatures makes me think of creating 'virus pre-images' for vaccination.
Other than the specialised cells which manipulate their own genome, would there be value in positively checksumming {D,R}NA, rather than adaptively pattern-matching for the bad ones, which is (I think) how it mostly works now?
Granted, you'd be hindering the evolutionary process by preventing mutation, but if we ever intend to start messing around inside ourselves, the first step would probably be to make sure that whatever we create, it's going to stay that way, or die.
The problem is
a) there's a LOT of DNA in a cell, and a lot of a cells, so checksumming would logistically be pretty much impossible. This is assuming there is some mechanism by which you can "scan" the DNA, which in itself has a whole host of steric/timing issues (different chromatin states, how do you deal with cell division? histone methylation and DNA accessibility, DNA binding proteins inhibiting interaction with certain regions at different times)
b) the more conceptual issue is that DNA in cells have natural variation anyway. My DNA is different to yours, but DNA in some of my cells is going to be subtly different to DNA in other parts of me.
The necessary internal variation sounds like it'd be the fatal flaw here, unless there's some way to determine 'good' changes and 'bad' changes.
As you said above, there are already repair/watchdog proteins for specific sections, so maybe that could be the basis for additional checking in other high-risk areas (Telomerase?).
You can think of it as like a "count-down timers". Every time a cell divides, the DNA gets doubled, so one cell doubles its DNA, creating two cells with identical DNA. When this happens, each of the telomeres in the new cell get a little bit shorter by a constant amount, and with this happening in a recursive way at some point they reach a critical length (Hayflick limit). At this point the cell knows it's undergone a certain number of divisions (~40 I think) meaning it's an old cell, so statistically it's likely to have picked up a few mutations. These mutations could make the cell less effective, or even worse, lead to some kind of disease (i.e. cancer), so as a precaution the cell initiates apoptosis. It's pretty clever - there may not be anything wrong, but there's a significant chance there is, the cell isn't perfect at detecting problems, so to avoid the risk to the rest of the body it kills itself.
Telomerase rebuilds telomeres, so is crucial in cells where lots of division is going on rapidly (such as embryonic cells) so cells can divide many more times by adding on the bit of teleomere that's lost. The problem is in cancer cells, telomerase is very often mutated into an “always on” position as it gives cells a way of being immortal. In normal cells telomerase isn't usually active, although it may have some other roles in pseduo-related areas. I was lucky enough to see Elizabeth Blackburn (Nobel prize winner for telomerase's discovery) speak a few years ago on this topic – it's really fascinating.
Proteins involved in DNA checking/repair are ones like ATM, Chk1, Chk2 , BRCA1, BRCA2, Rad51. There are many more, and they tend to work cooperativly together, though I can't remember the others off hand.
For more information regarding this kind of thing I'd really recommend Weinberg & Hanahan's The Hallmarks of Cancer [1]. Although published 11 years ago it's pretty much a “classic”, and gives a good overview of DNA damage and telomerase, albeit from a cancer viewpoint as opposed to generally.
[1] - http://www.weizmann.ac.il/home/fedomany/Bioinfo05/lecture6_H...
I tried to find some specifics on wikipedia, but apparently the system is incredibly complicated!
There are also a host of proteins that check for DNA breaks, damage from UV or other carcinogens and just generally keep an eye on things.
Would it be possible to devise some sort of virus which could somehow 'checksum' a given cell's DNA, and trigger cell-death if it doesn't match?
I've no idea if that is even possible, or whether it'd require a custom virus per-person (or whether you could add some sort of 'training stage' by introducing it to clean host DNA first)
Another tricky problem would be making sure it hits every/enough cells to kill the other virus, but slowly enough that there's time for the body to replace them. And of course, that the virus checksums itself regularly and self-terminates on mutation.