Thursday, July 26, 2012

The Evolution of Enzymes from Promiscuous Precursors

New enzymes often evolve from old ones via a gene duplication event. This gives rise to gene families encoding evolutionarily related enzymes with similar, but different enzymatic activities.

There are two principle ways to evolve related modern enzymes with specific activities. The first way postulates an ancestral enzyme that catalyzes a specific reaction (e.g. A1 → B1). Following a gene duplication event, one of the copies evolves an enzyme that catalyzes a completely different (but chemically related) reaction: A2 → B2.

Most people think that this is by far the most common pathway. That's because they are used to thinking that enzymes are highly specific—that's what's emphasized in most biochemistry courses. They think that it should be possible to reverse engineer the evolutionary pathway on the right-hand branch and thus transform one enzyme to another.

This works in some cases. For example, the enzymes shown in the figure are all members of the lactate dehydrogenase/malate dehydrogenase family. It's possible to take a modern lactate dehydrogenase enzyme and convert it to a malate dehydrogenase by changing only one amino acid [see Evolution and Variation in Folded Proteins, and Monday's Molecule #178]. It's likely that the ancestral enzyme was a malate dehydrogenase.

There's another way to evolve related enzymes.

Imagine that the ancestral enzyme was promiscuous—it catalyzed a bunch of related reactions. This is not as far-fetched as you might imagine since most modern enzymes catalyze a variety of similar reactions even though the rate for one of them may be several orders of magnitude greater than that of the side reactions. Enzymes make "mistakes" quite often and their specificity is exaggerated. There are some classic examples: Fixing Carbon: Building a Better Rubisco.

When the gene for a promiscuous enzyme is duplicated, both of the daughter enzymes can each catalyze several reactions. However, now that there are two enzymes, each of them can evolve to become more specific without the cell losing the other function. The end result is two related enzymes with different specificities just as in the first scenario.

The difference is that in this scheme there never was a time when an enzyme with one specificity was transformed into an enzyme with a different specificity. The common ancestor could catalyze both reactions. This has implications for protein engineering and for those studies that try to mimic the evolutionary in the laboratory. You can't transform one enzyme into another without going through an intermediate with relaxed specificity [Evolution of a New Enzyme] [Evolution by Gene Duplication].

We'll return to this topic in a later post when I discuss the claims of creationists Douglas Axe and Ann Gauger.

It's difficult to decide between the two scenarios when dealing with gene duplication events that likely occurred two or three billion years ago. However, it should be much easier if you look at recent events.

Plants make a variety of obnoxious, sometimes toxic, compounds in order to discourage herbivores (i.e. vegetarians). Cilantro is a good example—no sane animal would eat a plant that makes cilantro [I Hate Cilantro/Coriander!]. I'm sure you're all aware of the most poisonous plants but you probably don't realize that almost all plants are trying to kill you (e.g. jalapeños).

Much of the chemical warfare being carried out by plants involves the relatively recent evolution of new enzymes with strange specificities. As you might expect, these enzymes evolve from related enzymes following a gene duplication. In many cases, the ancestral enzyme is promiscuous.

A recent paper in Science discusses plant metabolism and gives three examples (Weng et al., 2012).

Most plants have an enzyme called HCT that catalyzes the fusion of two molecules of shikimic acid to produce p-coumaroyl shikimate (blue). Many herbs produce rosmarinic acid. One of the steps involves formation of p-coumaroyl-4′-OH-phenyllacate. This reaction (green) is a minor side reaction with HCT but the herbs have a related enzyme, RAS (rosmarinic acid synthase), that catalyzes it very efficiently. This is an example of the evolution of specificity from a promiscuous ancestor.

There are two other examples in the paper but I think you get the point. The evolution of enzymes with specific activities often involves a much less specific promiscuous ancestor.

Weng, J.K., Philippe, R.N., and Noel, J.P. (2012) The rise of chemodiversity in plants. Science 336:1667-1670. [PubMed] [DOI: 10.1126/science.1217411]


  1. Prof. Moran, would it be possible for you to write a blogpost about what, exactly, is meant by "completely different but chemically related" reactions at some point? I know creatIDiots usually waffle around endlessly on this subject, in order to deny the evolution of distinct enzymes. From those words alone it's sufficiently vague to hide all sorts of bullshit in.
    They'll usually bring up an argument along the lines that "yeah sure, Enzyme1 elvolved into Ezyme1A, but we never see it evolve into Enzyme2 or 3, and so we have all these unique enzymes that would have to derive from a large set of ancestors which all would have had to be created from scratch because they couldn't themselves have evolved from common ancestors, the chemistry they catalyze is just too different".

    1. Follow this link to see an example of completely different but chemically related reactions: Monday's Molecule #178.

      Another example is actually shown in the post.

    2. Thank you, though I was hoping for something a little more in-depth like what makes enzymes, in general, chemically related and is there some kind of cut-off line where they're not? Do you know of any good websites about stuff like this?

    3. Dr. Moran would be way more knowledgeable than I on the topic, but the best I can think of is the cytochrome P450s. They're a family of super-promiscuous enzymes involved in metabolism of drugs, toxins, and other compounds in the liver.

      Their reactions are "chemically related" in that they all oxidize their target molecule. There's a heme group in the active site, which and in combination with the surrounding amino acids, causes the formation of a highly reactive oxygen species. This ultimately allows for a wide variety of reactions, such as the addition of a hydroxyl or epoxide group to the target substrate. You could call these reactions "chemically related" in that they involve oxidation of a target by a highly reactive oxygen species.

      The substrates of a specific P450 might also be "chemically related" in that they might have a similar size, shape, and distribution of charge and electronegative atoms, so that they fit the same active site.

      I think this is an extreme example of what Dr. Moran was discussing.

    4. Rumraket, Larry is misleading you. What he describes in the figures is an example of a promiscuous bi-functional enzyme "subfunctionalizing" following a duplication event. This entails the differential partitioning of ancestral functions rather than the gain of new ones. Larry will deceive many, but he's not fooling me.

    5. This entails the differential partitioning of ancestral functions rather than the gain of new ones. Larry will deceive many, but he's not fooling me


      Oh Well, you can't blame me for trying. :-)

    6. @clast -

      What he describes in the figures is an example of a promiscuous bi-functional enzyme "subfunctionalizing" following a duplication event.

      That is a gain-of-information mutation according to creationist Lee Spetner in Not By Chance. Creationists have different definitions of "information."

      For Spetner, if a protein becomes more specific (loses an interacting partner) that is gain of information.

      For Behe, if a protein gains an interacting partner, that is gain of information.

      Between these two definitions, every conceivable biological change will be designated loss of information by at least one creationist.

      But conversely, every conceivable biological change will be designated gain of information by at least one creationist.

      At least it will be called gain of information before it's observed. After it's observed, all the creationists go switcharoo and call it loss of information.

  2. In general, I find the promiscuous enzyme route far more appealing as a likely explanation as it does not rely so much on luck and timing. If an enzyme is very good at A1->B1 and not so good at A2->B2 then an increased need for A2->B2 might create a selective advantage for gene amplification. This actively maintains the duplicated copies until one of them becomes sufficiently good at A2->B2 that the other can lose this ability without detriment.

    For the pure "neofunctionalisation" route, there is a limited window of opportunity to stumble across the new specificity before pseudogenisation and gene loss.

    My query here is how necessarily true is the statement: "The difference is that in this scheme there never was a time when an enzyme with one specificity was transformed into an enzyme with a different specificity."

    Even under the promiscuous enzyme route there could still be a time when the enzyme has no promiscuity and ONLY catalysed A1->B1. As A2->B2 became beneficial, the specificity could shift to make the enzyme promiscuous. (I believe this has been done in the lab.) Once promiscuous, gene duplication and subfunctionalisation can take care of the rest.

  3. The gene amplification to overcome sub-sufficient activity concept would be more enticing if there were current examples of multi-gene-copy enzymes that didn't have particularly good activity. So, for example rubisco, is it found in multi-copy? Any other examples?

    1. There are lots of examples where something particularly highly expressed is present in multiple copies. For example, I think antibiotic resistance genes are quite often amplified in number to increase activity. I am not sure whether anyone has explicitly linked this to examples of something "that didn't have particularly good activity". (I don't think rubisco qualifies here, does it?)

      The problem is that apart from things that are present at really high levels (such as all the copies of tRNA genes) or transiently need a boost (such as antibiotic resistance) the long-term expectation is probably not for amplification as there are other ways to stably boost gene expression for the long term. We are only just beginning to get to grips with the extent and relevance of gene copy number variation, so I would be surprised if some of these scenarios are not "caught in the act" over the coming years.

  4. Here is a nice paper on this subject:

    I especially like this figure:

  5. If you are using gene duplication then you have left the modern synthesis behind- read "Not By Chance" by Lee Spetner

    1. From the wiki for Lee Spetner:

      "Spetner first became interested in evolution in 1970 after moving to Israel. In Israel he indulged in searching for evidence which contradicted the modern evolutionary synthesis. Spetner was inspired by the rabbi David Luria (1798 - 1855), who calculated that according to Talmudic sources that there was 365 originally created species of beasts and 365 of birds. Spetner developed what he called his "nonrandom evolutionary hypothesis", which (in common with Christian young earth creationists) accepted microevolution (which he attributed to Lamarckian-like inheritance), but rejected macroevolution.[5]"

      Solid stuff, no doubt.

  6. This is not a recent idea, although the details may be. As a biology undergrad or beginning grad student in 1970, I remember buying and reading a slim volume "Evolution by Gene Duplication" by Susumo Ohno, who spent much of his career at City of Hope Medical Center northeast of Los Angeles.

    1. Right. Intelligent scientists have understood evolution by gene duplication for over forty years.

      It's only a problem for Intelligent Design Creationists, otherwise known as IDiots.