I wonder why life uses the particular proteins that it does, about 10^6 different proteins, I think? Evolution cannot explain it because the number of possible proteins is far far too large to ever come into existence anywhere in the visible universe, let alone on Earth. And evolutionary selection only works on stuff that exists.

Is there some chemical or other explanation, or is this an unresolved paradox in biology?

The median number of amino acids in a human protein is 375. Most are longer than 200 and maybe all are longer than 100. The 20 different kinds of amino acids can be combined to 20^375 = 10^487 different proteins which have the length 375.

If the entire crust of the Earth consisted of proteins which all were 375 long and each protein mutated once every nanosecond during 5 billion years and each mutation resulted in a protein which is unique in time and space, and there exists 100 such "protein Earths" around every star and 1000 billion stars in each of 1000 billion galaxies in the visible universe, it still only adds up to something like 10^100 proteins. Adding 10^37 years until the last star remnant has cooled down to less than 1 Kelvin, doesn't change this picture, almost all of the possible proteins will literally never exist anywhere. Even if I'm off by a factor a million on each of those numbers, only one in 10^300 of all possible median length proteins could ever exist.

Since our mind-bendingly small subset of possible proteins supports life, shouldn't we expect that many many (as in 10^300) other potential subsets of different proteins would do so too? And can we conclude that if DNA/protein based life has originated somewhere else, it will not have even one single protein in common with us or with any other independent origin of life? Or is this focus on proteins somehow invalid?

  • 4
    $\begingroup$ Sentences "...evolutionary selection only works on stuff that exists." and "What has caused life to choose this..." Look like "Life suddenly started to use arbitrary set of proteins, like guy who buys some tools in nigh-infinite shop" and it's not true - it has everything to do with evolution. $\endgroup$ – Mithoron Apr 20 '15 at 14:51
  • $\begingroup$ @Mithoron If mutations are random, then the median protein can mutate to 10^487 different new proteins. Less than one in 10^300 will ever happen anywhere, and still that works fine for life! Maybe 10^200 other mutations would have worked fine too? Or are we that lucky? Evolution has nothing to do with selecting proteins which never exist. Evolution only kicks in given this ridiculously small subset. I wonder if any known mechanism can explain this selection of protein varieties. $\endgroup$ – LocalFluff Apr 20 '15 at 15:03
  • 1
    $\begingroup$ This question is a bit like shuffling a deck of cards and asking why they ended up in that particular order. They just are by chance. $\endgroup$ – C_Z_ Apr 20 '15 at 19:57
  • 2
    $\begingroup$ @LocalFluff Proteins don't mutate. Genes mutate. Individuals are selected. Species evolve. Even when genes mutate they could not do so every nanosecond. The phenomenally large amount of radiation that could cause every gene on earth to mutate every nanosecond--I've got that correct? That is akin to your gedanken experiment?--would obliterate life on earth in less than a second. No life, no endless pool of 375 amino acid long proteins. When we say that there are 10^6 different protein sequences on the planet--is that your assertion(?)--that is misleading, they don't have 10^6 functions. $\endgroup$ – mdperry Apr 21 '15 at 5:22
  • 1
    $\begingroup$ "Maybe 10^200 other mutations would have worked fine too?" Yes they would have. Luisi (I strongly suggest you to ree his work if you are really interested in this topic) made a good point on that. There are plenty of studies showing that purely random proteins have high probability of having function, it turned out, for example, that for a given environment (buffer condition), a consistent percentage of random protein will collapse into a folded structure and have some biochemical reactivity. $\endgroup$ – alec_djinn Apr 21 '15 at 8:19

The answer is chance or, even better, contingency.

About your calculations, it is true that the theoretical sequences are almost unlimited, but the basic scaffolds are not. Very different sequences can fold into the same basic scaffold and have a similar reactivity/function. So, even if not all the sequences have been explored on this planet, most of the scaffolds and functionalities have been. It is true that protein-based life could be evolved somewhere else, and it is true that the overall sequence space evolved somewhere else would be different from the one evolved on this planet. However, I would expect to find almost the same palette of biochemical reactions that we have discovered on our proteins.

I would suggest reading the work of P.L. Luisi here. He makes really good points on this issue.

A new paper just came out, I think it is a nice addendum to this answer.

  • $\begingroup$ +1 for the sequence - scaffold comparison. Good insight, $\endgroup$ – Nandor Poka Apr 20 '15 at 16:11
  • 1
    $\begingroup$ Would chunk copying of DNA also bias against protein diversity in terms of what is easy to generate (first mover advantage) and perhaps in terms of error containment? $\endgroup$ – Paul A. Clayton Apr 20 '15 at 18:20
  • $\begingroup$ That is not a logical answer. Evolution has no relation to which "scaffolding" would be optimal, since evolution never ever can test and select any protein which is better than 10^100 or so times worse the actual theoretical optimum, for a protein which exactly consists of a medium number of amino acids. "Chance" is not the answer. "Chance" is precisely what is utterly defeated by these grotesque monster numbers! $\endgroup$ – LocalFluff Apr 20 '15 at 18:28
  • $\begingroup$ Evolution doesn't know indeed which scaffold will be optimal and it will never try all the possibilities, but this is not the point. My point is, that given an infinite number of sequences you do not end up with an infinite number of functions nor domains. Most of the infinite sequences will fold in few main scaffolds, with more or less the same shape and will end up having very similar biological functions. $\endgroup$ – alec_djinn Apr 21 '15 at 7:54
  • $\begingroup$ Let's consider a very easy protein to make an example, the GFP. The only function it has is to be fluorescent. Many variants of this fluorescent protein are known, all with a different sequence. The sequence may vary a lot, however all the proteins ends up having the same main structure and function. Now, all the variants differ for absorption/emission spectrum and for stability under different conditions. These factors will be selected by evolution but it will depend from case to case whether a red fluorescent protein will be more advantageous of a green one for example. $\endgroup$ – alec_djinn Apr 21 '15 at 8:02

I'm not sure I understand the question. You've elegantly demonstrated that only a tiny fraction of all protein sequences could possibly exist, but then asked why only a tiny fraction of all protein sequences do exist. Your conclusions about independent origins of life having no proteins in common are accurate, but also consider that you as a human being have not even a single protein in common with the bacteria in your yogurt (I mean you do, but they have slightly different sequences).

As far as which protein subsets make life, there are demonstrably thousands of sets of proteins that support life. Each genetically distinct individual has a slightly different set of proteins, most of which support life.

As far as how many of these subsets exist, that's unknown. Maybe aliens use only ribozymes based on RNA molecules. Maybe they have a different set of amino acids, or a silicon backbone to their proteins, or what have you. Until we find alien life(or perfect abiogenesis in the lab) the total diversity of life-like things is unknowable.

  • $\begingroup$ I thought all life on Earth shared some genes, even me and the bacterial yogurt I eat at breakfast. But maybe none of that is expressed as proteins? And is it really "demonstrable" that thousands of sets of proteins support life? And what has "thousands" got to do with the orders of magnitude involved here? This is not about thousands, it's about one in a trillion of a trillion of a trillion of a trillion of a trillion of a trillion of a trillion of a trillion of a trillion. At the very least. For the median length protein which keeps you and me alive and happy. $\endgroup$ – LocalFluff Apr 20 '15 at 18:14
  • 1
    $\begingroup$ again, it is not the sequence that matter in this case but the "protein domain". To make an easy analogy, think about a car, has 4 wheels and an engine and have one main function. Now you can think of trillion trillion trillion different cars, each of a different colour, a different shape, a different size, different setups... do the math and you will find that the numbers are huge. However the function will be always more or less the same. So in other planets, with other combinations of pieces, always the same kind of machines will arise. Proteins are molecular machines. $\endgroup$ – alec_djinn Apr 20 '15 at 22:23
  • $\begingroup$ You and a yogurt share many genes, but the sequences are not identical. Metaphorically you both have a steering wheel but it is a different steering wheel. Maybe slightly smaller or with a leather inset or whatever. Does the same thing but isn't identical. $\endgroup$ – Resonating Apr 21 '15 at 10:31
  • $\begingroup$ @alec_djinn Are you sure that all 'life' has precisely the same set of molecular functions? 'RNA world' life would be similar to life that we recognize, but the vast majority of the proteins would have completely different functions, or not be proteins at all. 'RNA world' life is also the most similar to familiar life we can conceive of, and there's a statistical certainty alien life would be more alien than we can conceive of. 'The same kind of machines' needn't necessarily arise. $\endgroup$ – Resonating Apr 21 '15 at 10:40
  • $\begingroup$ @ Resonating Well, if we talk about "chemistry based life" then yes. Whatever set of molecules you choose, there will be common problems to solve to have life, like how to store information, get energy flow running etc... In a molecular scenario these problems are solved by coupling chemical reactions (oxidations, reductions, bonds formation etc...) in larger pathways. If those pathways are catalyzed by proteins, nucleic acids or other molecules it doesn't matter and may vary from life form to life form. Don't you think? $\endgroup$ – alec_djinn Apr 21 '15 at 11:34

You ask why living things use the proteins they use, and not something different.

It's mainly because they must do something that works, and they have found something that works OK.

Each protein serves one or more functions for its cell. If it changes at random, probably it will change into something which does those functions less well. Because it has been selected for thousands of generations to get something that works well, and so most of the alternatives are worse.

Many mutations have no effect. But if you have 300 proteins and you have a change in each of them, chances are you will not live so well. And if your son has another change in each protein, he will probably do worse than you. It won't take many generations before your descendants can't compete and die out.

So each change is a gamble. If you don't make any changes, then eventually something which does change will do better and will drive you extinct. But if you make too many changes your descendants will do worse because of them. Ideally you would make just the right number, to balance the weighted odds of damage from bad changes versus benefit from good changes.

For many organisms, the rate of change is down somewhere around 1% -- 99% of the time they get no change when they reproduce.

Since our mind-bendingly small subset of possible proteins supports life, shouldn't we expect that many many (as in 10^300) other potential subsets of different proteins would do so too?

Sure. And we have tended to find some that we could get to from where we were before. Point mutations tend to happen in small steps. You change an amino acid to a similar one. The genetic code is a Gray Code, often mutations have no effect, or hydrophobic amino acids are replaced by other hydrophobic ones, amino acids are replaced by others of similar size, etc.

Some of the best enzymes might be things that have nothing else good similar to them, so that they are very hard to evolve to.

Science fiction writers have sometimes written about life that has evolved for different times, and they usually suppose that the more-evolved one will simply and easily outcompete the ones that have not had so long to discover better ways. That isn't necessarily so. Human beings transported back to the Silurian Period might eat well (once they figured out which seafood is poisonous) and might not be eaten by anything (but they might get mobbed by centipedes if they sleep, and be unable to make hammocks out of cooksonia, and have no trees to hang the hammocks from either). They still may not thrive. Half a billion years evolving to fit into a different ecosystem might not be that useful after all.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.