Posted: May 22, 2014 11:15 am
DavidMcC wrote:Rumraket wrote:DavidMcC wrote:One question for Rumraket: This may not be known, but which genetic mechanisms that allowed cell specialization might have been available to the earliest multicellular animals? (If they did not have the modern panoply of mechanisms for preventing accidental gene expression, then maybe it would have been necessary for them to use only non-genic DNA for suppressing accidental gene expression.)
First off it's important to note that accidental gene expression still happens, but at very low levels.
This is an argument for saying that the many other mechanisms for suppressing aren't quite sufficient after all!
Sufficient for what - completely removing noisy transcription? Then sure, extant mechanisms aren't sufficient for that. But natural selection isn't concerned with perfection, only sufficient to ensure a viable organism. If a byproduct of the transcription system is low level noisy transcription, but this doesn't normally (or sufficiently) influence normal organismal function in a negative way, then there's not enough of a selective pressure to get rid of it.
DavidMcC wrote:Rumraket wrote:The primary protection against accidental transcription is for the binding site to have a unique sequence not found anywhere else in the genome (goes for both single celled and multicellular organisms). With a large genome with many transcription sites (whether that be protein coding or RNA genes), you obviously need a certain minimum activation site length (a stretch of sequence usually 10 bases long) to ensure that no two sequences are identical. This way, a transcription factor (usually a protein) can reckognize a single specific sequence only.
That means the protein will only attach to the DNA and initiate transcription in places where it "fits" with the specific matching DNA sequence. It's a structural thing of course, a certain stretch of DNA will have a certain 3 dimensional structure, and the protein has a structure that fits only this sequence of DNA bases.
But the problem is that the junk DNA is randomly mutating without constraint. That means eventually some stretch of junk will simply end up being somewhat similar to one of those binding sites. They're usually not totally identical, which is why the protein will only attach to the DNA weakly/for a short time and stick around for a single or a few transcripts before letting go again (think brownian motion and molecules bumping into the transcription factor, making it let go again). Sometimes of course, the junk site will be similar enough to the "intended" binding site that the protein sticks around longer and produces many transcripts, and as you correctly noted earlier this can potentially cause problems if that transcript turns out to interfere with normal cellular processes.
On the other hand, such accidental transcripts can also turn out to have useful functions, so this is how many ORFan genes originate, as GFL's op post explains. ...
A. I thought I had just admitted that I was wrong about the importance of chromatin in suppressing random gene expression.
In so far as chromatin unpacking functions as a kind of expression control, it's subject to the same basic biophysical constraints as transcription from DNA to RNA. Think about it. As GFL points out, genes are associated with their own specific histones that unpack the chromatin in specific locations to allow transcription of those specific genes. How would the histones achieve this kind of selectivity and not just unbiasedly attach to any chromatin complex and start unpacking? It obviously has to reckognise a unique structural signature that is only associated with a specific piece of DNA sequence. So we're basically back to the same problem facing the actual transcription itself: A specific protein has to associate with a specific sequence in order to bind efficiently and initiate a process. That means the system would have the same basic vulnerability - if the junk regions are randomly mutating, sooner or later and simply through chance, an area will start to look similar to an area containing genes, which means you will have a potential for something like noisy histone activity.
DavidMcC wrote:B. GFL 's point about novel genes coming out of accidental transcripts misses the point that these are nevertheless bad for the organism in the short term.
Why? Some times they are, some times they aren't. They're evidently not bad enough for natural selection to have gotten entirely rid of the the process.
DavidMcC wrote:NS may be good for the long term survival of the species, but the price is paid by the dead individuals!
Sure, but how often does this happen really? We actually don't know. And if the carriers manage to reproduce before such kinds of accidental transcripts can lead to cancers or whatever, then the price of the individual has no effect on the long-term evolution of the species.
You're welcome.
