Skip to Content

The Blind Watchmaker’s Workshop

Did it have to be this way? I mean all of it – biochemistry, the molecules of life. More specifically, as proteins evolve and change, how many paths could they have taken that would have taken them to the same sorts of function?

That’s a pretty hard question to answer, since we’re looking at a billion years of evolutionary tinkering, driven by all sorts of random craziness and a willingness to run with anything at all that works. But here’s a new paper from a team at the University of Chicago that tries to address the question empirically. They’d previously worked out a plausible sequence for the ancestral steroid receptor, from whence all the others have descended. It’s selective for the estrogen-response-element in DNA, so that was the original mode of action. Way back when, three particular mutations in the DNA recognition region of the protein sent some of them down the track for selectivity to the steroid-response-element (including androgen response elements and others), which is still the major divide in the mechanistic details of the whole family.

In this paper, the group systematically examines mutations at those three key sites (and one adjacent one that also varies in the superfamily). A library of all 160,000 mutants was prepared, and they used yeast fluorescent reporter assays to figure out the DNA binding and transcriptional competency of the whole shebang. As an aside, I really love modern molecular biology and chemical biology, because you can actually do experiments like this – just make every single one of the mutant proteins, test them all, sequence the ones that you find interesting, and nail down the answer. It’s not easy, and it’s not an afternoon’s work, but I well remember the days when a proposal like this would have been considered dangerously wasteful and stupid.

So starting from ancestral ERE selectivity, they found 828 variants that were specific for SRE DNA sequences, working as well as the current steroid receptors or better, with no affinity for the ERE sequences. Interestingly, they get to the selectivity in different ways, with different amino acid/DNA contacts than are seen in nature.  Graphing these out to see what the relationships were, it appears that virtually all of the functional mutants (1351 in total, including unnatural ERE selective ones) can be reached by single-point mutations from each other without going through inactive mutant proteins. The new SRE-specific protein sequences are scattered all around the graph; there’s no one bottleneck mutation that they particularly have to pass through.

And this persists under the constraints of many evolutionary models. Even if you insist that every step has to yield better affinity for SRE sequences, there are still many ways to get to them. In fact, the SRE sequences that we have now represent neither the shortest mutational path nor the highest-affinity final results. Over 90% of the SRE-selective mutants can get there by a path that’s no longer than the historical one. The evidence is that they just happened to happen that way, they worked, and here we are. It could have been many other ways, and if you reset they clock, odds are that something else would have fit instead. Another result from the graph is that if you start from different ERE-selective points, you have different paths to new SRE-selective ones. So the starting ancestral sequence we happened to have also was an influence; if you wound the clock back even further before the evolution of the ancestral receptor itself, you’d have a whole possible landscape of both forms.

There’s a further complication, though. There are eleven mutations (11P) that also differ between the ERE- and SRE-selective proteins, although they’re not directly at the DNA binding site. If you take those 11P mutations out and make the 160,000 variants again at the four mutation sites around the DNA-binding helix, things are more constrained. Now there are only 43 ERE-specific ones and 41 SRE-specific ones in the whole batch: 99.92% of the mutants are just nonfunctional, and what’s worse, the functional ones are almost entirely connected through nonfunctional nodes. So the ground had to be prepared – the relatively permissive protein landscape isn’t always there, and what we’re seeing is very likely the result of something like that being in place for the DNA-binding mutational landscape to be explored productively in the first place.

So how and why did the 11P permissive mutations take place? Looking at the mutational landscape, it fits with the hypothesis that these nonspecifically increased the affinity of the protein(s) for both ERE and SRE DNA sequences, and seem to have been part of a general improvement in the function of the steroid receptors. These general increases in affinity opened the landscape up to new functional mutations. Here’s the authors’ summary:

Our results shed light on the roles of determinism and chance in protein evolution. The primary deterministic force is natural selection, which drives the evolution of forms that optimize fitness. Chance appears in two non-exclusive ways: as historical contingency, when the accessibility of some outcome depends on prior events that cannot be driven by selection for that outcome; and as stochasticity, when there are paths to numerous possible genotypes of similar function, and which one is realized is random. . .Our results point to strong stochasticity and contingency in the many histories by which SRE specificity could have evolved. Hundreds of genotypes encoding SRE specificity were accessible from AncSR1, but selection for that function alone could not have deterministically driven evolution down any of those paths, because all were contingent on permissive mutations. . .

So the role of chance in evolution is there to see: chance built on chance, actually, with all the time in the world (well, all the time in our own world) to try things out. Things may look like they evolved deterministically, because we tell ourselves stories and our brains themselves seem to have evolved to look for them. But in reality, it’s likely that a lot of molecular-scale evolution just happened for no particular reasons at all. . .


Note: All opinions, choices of topic, etc. are strictly my own – I don’t in any way speak for my employer

20 comments on “The Blind Watchmaker’s Workshop”

  1. “I well remember the days when a proposal like this would have been considered dangerously wasteful and stupid.”

    Realistically, such an approach wasn’t even possible 20 years ago and would have been insanely expensive even a decade ago.

  2. EP says:

    So, how much is this an argument against intelligent design? As in “only a smart designer could have put this together” getting met by “actually, experiments suggest that designer could’ve done a rather better job here”?

    1. AR says:

      That’s always been the case, you ever bit the inside of your cheek while chewing? Who the hell designed that

      1. Chemsufferer says:

        I take it, both of you guys are male. Ever thought of the whole childbearing and giving birth business from the intelligent design standpoint? The designer must’ve actively hated women…

        1. Pennpenn says:

          Well I mean based on a lot of how many religions treat women that’s not entirely surprising…

  3. Li Zhi says:

    Until they do the experiment where they start with 75% H and 25% He (optionally with a trace of Li), and run it forward a couple of dozen times and only get DNA/AA worlds, I’ll wait before jumping to any sort of global conclusions. How confident should we objectively be that all life on Earth is DNA/RNA based? If the global genome is actually mostly located below the Earth’s surface, as many believe, are we able to argue that we’ve analyzed a representative random sample? (and of course, random sampling of the genome as it existed 3-4 billion years ago is going to be a bit more of a problem). I’m unlikely to be around to hear about the Martian drill hole results, which are likely to significantly enlighten us. (Once they actually rise above the horizon of what is feasible)

    1. Uncle Al says:

      Primordial “75% H and 25% He” is by mass; 11.8:1 by atom abundance. Intelligent Design had best take a high school course in general chemistry re Avocado’s number and guacamole.

      The universe is slimed with star tar life. High order mechanisms thereafter aside, is it all D-sugars and protein L-amino acids? Really interesting exo-life pays taxes. One doubts it is within our spatial or temporal reach.

  4. Joe Blo says:

    In an interesting twist of today’s web reading, passing through an information bottle neck may be important for both gene development (genetic algorithms?) and artificial neural networks (human learning?) in a well defined mathematical way.

    “The network starts to shed information about the input data, keeping track of only the strongest features — those correlations that are most relevant to the output label. This happens because, in each iteration of stochastic gradient descent, more or less accidental correlations in the training data tell the network to do different things, dialing the strengths of its neural connections up and down in a random walk.”

  5. steve says:

    So I’m an atheist but there’s always a religious response to such studies. It’s the story of the scientist who tells God that we don’t need him anymore, we know the secret of life and can proceed without him. God says, “Really? Can you make a human being like I did?”. “Sure thing,” says the scientist. “Then show me, says God.” So the scientists reaches down and picks up a handful of dirt to start. To which God responds, “No, no. Get your own dirt.”

    1. Joe Blo says:

      To which the the scientist responds… “That’s ok, I’ll just have the Flying Spaghetti Monster make another god… that one’ll create the dirt for me”. Universes are a dime a dozen.

      It’s not like an all-powerful all-knowing deity just popped out of nowhere! You need the FSM to do that, right?

      1. steve says:

        Someone who’s religious – and isn’t insulted by those who denigrate things they don’t believe or understand – would simply say that God’s always been here. Bottom line is that believing in the Big Bang is just as much a leap of faith as believing in God and there’s simply no way to logically prove otherwise. Much smarter people than you have tried. In an infinite universe your ability to explain everything with a very finite mind is extremely limited.

        1. asdf says:

          “Bottom line is that believing in the Big Bang is just as much a leap of faith as believing in God and there’s simply no way to logically prove otherwise.” <- How is it the same leap of faith to believe in something that's scientifically corroborated and believing in the (I assume) Christian God?
          I'll agree that bashing religion is pointless and the people doing it are often totally blind to their own ideological biases.

          1. steve says:

            The Big Bang is no more scientifically corroborated than faith in God, whether it’s Jewish, Christian, Hindu, Buddhist or otherwise. We have a number of facts and have extrapolated that there was a Big Bang. It’s only a theory that we think fits the limited facts we have. People of faith (again, I’m not one of them but I don’t disrespect them) have a set of facts as well. They see extreme complexity, an infinite universe that humans can’t possibly understand any more than an insect understands that its part of a solar system and extrapolate that there must be a God. Point is, you’ll never prove either with absolute certainty so both as some level are simply a matter of faith. Science has faith that what we perceive as falsifiable facts with our limited experimental data represent the truth while others believe that science, being finite, will never be able to comprehend something as incredibly complicated as the universe. Which one you think is true is simply a matter of your starting principles, otherwise known as faith.

          2. Andrew says:

            steve, I admire your balanced view and props to you for seeing the faith requirements both for atheism and religion. It seems very difficult for most atheists to grasp that for some reason.

          3. Pennpenn says:

            @Steve: I’d say the main difference between the Big Bang hypothesis and “God arguments” is that the Big Bang is meant to be provable (or disprovable) but God isn’t. You could undo the Big Bang hypothesis with a single fact , but no matter how many facts you throw at any kind of God hypothesis it’s proponents will pretty much never admit that they could be wrong. It feels like the “everything-proof shield” of a schoolyard playfight rather than any attempt to further the discussion or learn or understand anything.

            Science doesn’t have faith, it tests and retests, results are checked and rechecked by peers, and anything can be overturned, whereas with faith the object of faith can never be questioned, never tested, and the only peers are those who believe the same thing so nothing gets checked. And when some “inconsiderate jerk” starts testing and checking and it turns out the object of faith wasn’t as advertised, then we get gods retreating (the gods lived on mountains until we climbed them, then among the stars until we studied them, now they’ve run off to some other plane of reality we can’t get to), or the faithful use solipsism like squid ink to avoid any admission.

            And really, the whole “it’s so complicated a God must have made it” falls over when you consider that complex things can come from sets of simple rules and that if a complex God is required to create a complex universe and complex things can’t come into being autonomously then it raises the question of where did the God come from…

          4. Zach W. says:

            @PennPenn, I always liked this quote. “God is an ever-decreasing pocket of scientific ignorance.”

        2. RarelyComments says:

          If you have not read it I would highly recommend “The Big Bang” by Simon Singh. It is a great historical account of how the theory evolved (for lack of a better word) based on the observable evidence. As a chemist it gave me a much better appreciation for the scientific basis of the theory – there is a great deal more science as opposed to faith than I had appreciated.

  6. Insilicoconsulting says:

    Brilliant! Made monday morning bearable..I would rather read this blog and that paper than endure another one of Richard Dawkins’s books. After the first book, it’s all all been rehashes.

    1. Jenks says:

      Id go further and say in retrospect, selfish gene was dumb too. A massive diatribe against ‘ group selection ‘ with no scientific baisis. Group selection is just darwinism applied to a different system ( groups ). Duh….

  7. James says:

    Ahh…Monday Morning. Any other PIs out there hating your grad students right now for not sending them data yet? This generation frankly sucks.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.