It’s been pointed out that the concept of “Junk DNA” came not just from the observation of apparently non-coding genetic elements and their interpretation as “parasitic”, but from a theoretical prediction by the noted evolutionary biologist Susumu Ohno in 1972. Ohno said that in mammals, natural selection could only cope with a limited number of harmful mutations without being swamped, with deterioration and extinction as the result. He estimated that, given known rates of mutation, a maximum of 30,000 genes could be subject to selection. This makes intuitive sense – even under the best circumstances how could the environment select the best combination of hundreds of thousands of finely varying traits? It’s even rather astonishing that, in the real world of predation and harsh weather, as many as 30,000 independent variables could be under the control of natural selection.
The writer of the Wikipedia article on “Noncoding DNA”, though not unbiased, says Ohno’s prediction remains robust today, and fits well with the 20,000 known coding genes in human DNA. It seems this writer takes into account the possibility of some extra “selected loci”, as shown in the ENCODE project. But it’s equally clear that, since those 20K genes represent only 2% of the genome, he’s not expecting the idea of Junk DNA to be overturned any time soon, or Ohno’s predictions would be knocked into a cocked hat – requiring some radical new calculations.
This picture is mitigated to some extent by the neutral theory of Kimura, which agrees with Ohno’s maths and argues that most of the change we see in organisms does indeed occur beyond the reach of natural selection. It seldom seems to be stressed that in effect, this now dominant, anti-adaptationist position is a theory about “the appearance of design” arising not through random variation and natural selection, but predominantly by random variation alone. It’s as if it describes Jackson Pollock’s method of painting and then attributes the Mona Lisa to him. But it does at least leave room for some extra coding genes to be discovered beyond Ohno’s numerical limit.
Now Cornelius Hunter has drawn attention to a new paper, which raises some interesting questions. But first I’ll need to explain in elementary terms, and probably wrongly, some more technicalities.
For years one way of showing what genes are “under selection” has been to compare mutations that change the proteins being manufactured, with mutations that leave the proteins unchanged. This works because the genetic code has some redundancy, ie variant DNA codes make the same amino acids – and obviously natural selection won’t notice the difference in these, because the resulting proteins are unchanged.
The new paper points to the discovery that, within the sequence of protein coding genes, there are also signals that affect mRNA used for other purposes. So that instead of mRNA, as was once thought, being only what transfers the message of the DNA code to the ribosome for protein manufacture, it is now known to have all kinds of functions like controlling protein folding, the relative amounts of each protein and a host of others – and the codes controlling all this are on the same DNA as the amino acid sequences we’ve always known. That is, they’re on the same stretches of DNA, not just on the same spool of tape.
One could view this like one of those military cyphers, where within the message “Ask Aunt Agnatha to collect the groceries” is another, non-verbal code giving troop movements. Not that the mRNA coding is secret – though it was unknown by us until relatively recently. But it is, in effect, like having the same message conveying two separate meanings at once. And that’s very clever, because one assumes that in the military cypher, so long as the surface message looks plausible and is in English (or Dutch, or whatever), what it actually means is irrelevant, and can be altered to suit the necessities of the coded message. If “Johannes says he’s remembered your birthday” is better for the code, so be it.
But imagine having to work out how to vary the message so that both the superficial and deep messages carry useful meanings that can be changed. One way might be to use the redundancies of English to vary the code, eg “Ask Auntie Agnatha if she will collect all the groceries.” And in fact it appears that something similar happens in the cell. Mutations that change mRNA control functions can happen without altering the protein coding by using DNA’s redundancy, which is dashed clever for a blind process. But of course the changes are actually controlled by natural selection on the organism like any other mutations.
Hunter points out that this must raise doubts about the method of showing which genes are under selection, because changes in the redundant parts of the code are no longer to be seen as non-selected: they are coding for the cell’s immensely complex control systems and are as beneficial or deleterious as any other mutations, even though the proteins they produce at the same time are unchanged. That criticism seems to me to makes sense.
But another aspect is that these findings of new complexities in cell regulation due to mRNA variants expand hugely the number of traits under selection. Bear in mind that one of mRNA’s functions is alternative splicing, ie the selection of different segments of coding DNA to form multiple proteins. It now seems that the average piece of genetic DNA codes for maybe half a dozen separate proteins – which means half a dozen separate traits needing to be supervised and honed by hardworking natural selection for each gene. Then there are the myriad of other mRNA functions, each of which has to be finely tuned. On the face of it it would seem almost certain that Ohno’s limit of 30K loci under selection has been exceeded by several orders of magnitude. So something must be wrong somewhere.
The simplest answer, I suppose, is to shrug and transfer all these mutations on to the pile marked “neutral”. If selection can’t cope, we must just do without it. All it takes after all, to conceive of random mutations, under no guiding mechanism, simultaneously modifying two separate semiotic coding systems on the same piece of DNA to the organism’s advantage, is a bit of imagination.