Tuesday, December 9, 2008

PT eye evolution flare up

I was notified that there was a little flare up at Panda's Thumb about our recent article on eye evolution, entitled:

Opening the "Black Box": The Genetic and Biochemical Basis of Eye Evolution by Todd H. Oakley and M. Sabrina Pankey (PDF) published in Evolution: Education and Outreach.

Warning: Long Post. If you only read one thing, read this:
One might argue that we are just pushing back the origins, changing the question of "phototransduction" origin to the question of "transduction" origins. In a way this is true, but it is also a fundamental insight about how evolution works. New features are not breathed into organisms by some unknown force, they evolve by duplication/divergence or recombination of existing features. Trace a feature like phototransduction back far enough in evolutionary time, component by component, and it grades into something else altogether.

Here is the story:
A week or two ago, I was contacted by an undergraduate from a university in the UK (I see no reason to reveal her name). She had questions about the article, writing:

I've just been reading your 2008 paper 'Opening the Black Box: the genetic and biochemical basis of eye evolution' and found it really interesting! I'm just writing an essay on eye evolution atm and am trying to get to the crux of the issue, and find out how the phototransduction cascade itself evolved. After explaining that opsin probably arose by a mutation in a serpentine gene/protein, you mention in your paper that:

"In yeast...these receptors [GPCR's - the serpentine proteins] are sensitive to pheremones, and they even direct a signal through proteins homologous to non-opsin phototransduction proteins."

What I'm wondering is, is it this whole yeast pathway that has been modified for the metazoan phototransduction cascade? Or is it only the opsin which has been derived from it? (With the subsequent molecules involved in the phototransduction cascade being co-opted from other proteins not involved in the yeast signalling pathway).
Based on this email, it seemed the student had a pretty good grasp of the issues, and I replied:
Thanks for you questions.  I found out after writing the paper that the yeast pheromone proteins are not the "rhodopsin type" GPCR, so they are distantly related at best to opsins.  So they should not be considered anything like direct ancestors of opsin.

As for other components of the yeast pheromone cascade, these are different than phototransduction. Yeast pheromones activate a MAP kinase cascade. So, I think all that is similar is the GPCR and G-protein.

So, it really is an open question as to what the ancestral function was of some of the genes of phototransduction, although some of these genes do function in other sensory transduction pathways...

Unfortunately, she misinterpreted this email, understanding it to mean (written over at PT, using the pseudonym "Green"):

"Yeah I knew that co-option thing would be coming. Turns out there was a mistake in that Oakley and Pankey paper.

The yeast intracellular signalling cascade turns out not to be homologous in any way to the metazoan signalling cascade. It just got published before the authors realised."

First of all, this is an incorrect interpretation of what I wrote. In fact, I wrote "So, I think all that is similar is the GPCR and G-protein." This is far from "not homologous in any way", as claimed by the student. The G-protein is undeniably homologous, and the yeast pheromone receptor is a 7-transmembrane protein, at least conformationally like opsin (whether or not the opsin and pheromone receptor sequences are homologous is a trickier issue).

But the larger issue is I think an issue of "linear thinking", which I address quite often on this blog. The student seems to think that if we cannot identify in yeast (taken as a linear ancestor of animals) a cascade identical to phototransduction except for opsin, then the origin of phototransduction requires numerous simultaneous mutations. This is not the case. First of all, yeast is a more distant relative of animals with phototransduction than is sponges. I just mentioned the yeast pheromone photoreceptor in the paper as a well studied example of a pathway outside of animals with partial homology (some components homologous, some not) to photoreception. There are closer "relatives" of phototransduction in sponges (poorly studied functionally, but the genes are known) and in other animals (better studied).

Another issue is a difficulty that people have with thinking about partial homology - that some components can be homologous and some not, depending on the time scale of the comparison (see my post The Red Herring of Eye Evolution).

Partial homology is a pattern that indicates a mechanism of co-option in the evolution of features. Co-option is the combination of existing things in a new way (analogy: dijonaisse = dijon mustard plus mayonaisse). All of the components of phototransduction pre-date animals, except opsin. And if we consider opsin to be a GPCR, which it is, then all of the components of phototransduction pre-date animals. This may be considered a pattern of co-option, or exaptation. Signaling pathways were already present before phototransduction. Some of the phototransduction components function together as far back as the yeast + human common ancestor (GPCR + G-protein). Other components of phototransdcution function together in non-phototransduction cascades of other animals. This indicates that phototransduction did not assemble all at once, but built incrementally upon an existing scaffold.

One might argue that we are just pushing back the origins, changing the question of "phototransduction" origin to the question "transduction" origins. In a way this is true, but it is also a fundamental insight about how evolution works. New features are not breathed into organisms by some unknown force, they evolve by duplication/divergence or recombination of existing features. Trace a feature like phototransduction back far enough in evolutionary time, and it grades into something else, component by component.

There were other comments, too. Again at PT, she also commented:

Yeah I read Oakley and Gregory’s articles on eye evolution a couple of weeks ago. Unfortuantely neither address the crux of the issue: namely the origin of the biochemical phototransduction cascade.

To be fair, Oakley’s article (the ‘Black Box’ one) at least tries to give some biochemical details. But it only scratches the surface by suggesting a potential origin of the opsin protein. Unfortunately the origin of a new opsin protein is not equivalent to the origin of an entire phototransduction cascade.

So it seems the Darwinian account still falls quite far short of any satisfactory biochemical explanation. Descriptions of morphological change, comparisons of genes, crystallins, etc. all skirt the issue if it cannot be shown how the phototransduction cascade itself arose

The difficulty with this comment is that the origin of opsin defines the origin of phototransduction. The other components of the cascade were already there, they all predate opsin, as described above.

This is also a "God in the Gaps" argument, or maybe, a "God under the surface" argument, stating that describing the origin of the keystone molecule of phototransduction (opsin) "only scratches the surface".

Also, I don't understand what the difference is between "comparisons of genes" and "biochemical explanation". What would a biochemical explanation be for the evolutionary origins of things that doesn't involve "comparisons of genes". The genes of the phototransduction pathway have biochemical interactions with each other, many mediated by interactions between specific amino acids.


NickM said...

All of her puzzling comments are channeling Behe in a very direct way. Amongst his favorite assertions are that eye morphology evolution is irrelevant because you have to explain the origin of phototransduction, and that molecular homology is "just sequence comparison" with no implications about how something evolved.

You've got a live one! Send her this:

Todd Oakley said...

Nick - Yeah, that is what the original article was about - opening Behe's black box. Here is one juicy quote from the paper:

"Unfortunately, instead of pointing to the molecular
evolution of multicomponent systems as a rich area for
new scientific research and synthesis, Behe chose to
commit scientific suicide by incorrectly claiming that the
only way for multistep biochemistry to arise is by
intelligent design."

Thanks for the link to the Nat Immun. paper, I hadn't seen that one....

Anonymous said...

Todd, I hadn't seen your blog before - it looks like an excellent read. Thanks for the work you've put in against creationism.

Anyway, have you seen this paper:

Jékely, G. et al. (2008) Mechanism of phototaxis in marine zooplankton. Nature 456, 395-399.

It is really interesting. There's a summary here for anyone who can't access the Nature paper:


Returning to the issue at hand, the commentor is particularly struck (following Behe) about the simultaneous mutation issue. For example:

A minimum no. of components are needed simultaneously for this cascade, otherwise no light signal is transduced and no selection pressure is exerted. For example, Oakley and Pankey say that even the opsin protein itself - just one protein in a whole cascade - can’t become light sensitive until it gets another mutation to make it associate with retinal. So this is at least two co-ordinated mutations that have to happen before opsin is any use to the cell.

Anyway, the above example pales in comparison to the whole phototransduction cascade, as well as the proteins that are needed to restore opsin to original state after the cascade has been activated. It’s quite clear you need *numerous* *simultaneous* mutations before any functional advantage is conferred. Doesn’t sound too Darwinian to me.

You address this kind of thinking here a little. I was wondering if there is anywhere with a larger treatment of the simultaneous mutation argument. It seems to be an increasing ID theme. If Nick is reading then maybe a PandasThumb article on the subject might be appropriate (if there isn't one already).


Todd Oakley said...

mammuthus: Check out my post here:


Most notably, check into Joe Thornton's work. My post links a presentation of his. Also, in the comments on my post, Ryan Gregory cites a Science paper by Thornton, and Ryan's own paper in the Eye Evolution EEO issue that also summarizes Thornton's work.

He has reconstructed the molecular steps, one at a time, of an interaction that appears to require 2 simultaneous mutations.

Anonymous said...

Thanks for the reply, I had read the Thornton paper but not for some time. I'll have to check it out again. I'll also have a watch of the presentation when I've got a bit more time.

It's probably worth mentioning Behe's response to the Thornton paper. It can be read here:


He argues that the paper doesn't actually deal with irreducible complexity. Ian Musgrave at Pandas Thumb has an excellent (as usual) counter to Behe in this thread (see also his comments to the thread, on the second page):


Thornton has also contributed to other interesting work. For example:

Ortlund, E.A. et al. (2008) Crystal structure of an ancient protein: evolution by conformational epistasis. Science, 317, 1544-1548.

The key quote here is:

“Permissive” mutation—substitutions of no immediate consequence, which stabilize specific elements of the protein and allow it to tolerate subsequent function-switching changes—played a major role in determining GR’s evolutionary trajectory.

because it neatly illustrates the overly simplistic thinking that creationists have.

Returning to the topic at hand, I'd appreciate any further thoughts you had on "green's" most recent comment:


Thanks again for your efforts, much appreciated!


Anonymous said...

Whoops, the Ortlund et al paper should be 2007 not 2008.