Quintessence of Dust

If it's not natural selection, then it must be...

2011-09-28T18:17:00.000-04:00

The folks at the Discovery Institute (DI) are engaged in an extensive attempt to rebut my friend Dennis Venema's critiques of Stephen Meyer's surprisingly lame ID manifesto, Signature in the Cell. There are several aspects of this conversation that I hope to address in the coming days and weeks, but one jumped out at me today: the consistent confusion about natural selection in depictions of evolutionary theory by design advocates.

Consider this excerpt from a recent blog post by a writer at the Discovery Institute:

...we need a brief primer in fundamental evolutionary theory. Natural selection preserves randomly arising variations only if those variations cause functional differences affecting reproductive output.

A few sentences later, the same claim is repeated:

Indeed, given that natural selection favors only functionally advantageous variations, ...

Those claims were first made in a piece written by unnamed DI "fellows" mocking the work and conclusions of Joe Thornton, an evolutionary biologist at the University of Oregon. And the claims are badly misleading.

For one thing, in my view the DI commentators imply that Thornton and colleagues relied exclusively on natural selection in their analysis of the evolutionary trajectory in question. That's false. (Read Joe Thornton addressing these criticisms himself at Carl Zimmer's blog.)

But more importantly, the DI commentators falsely claim "that natural selection favors only functionally advantageous variations," referring to this as "fundamental evolutionary theory." As I regularly emphasize, that simplistic summary looks reasonable at first blush, but it morphs into a seriously misleading error when it is presented the way that design advocates persistently do. These are very, very basic concepts, but they're worth emphasizing. Here's why the DI's portrayal of "fundamental evolutionary theory" above is badly wrong.

1. Natural selection is not the only force that can yield evolutionary change. In some situations, it is not even the most prominent force yielding evolutionary change: random genetic drift is known to strongly influence evolution. And the relative contributions of natural selection versus genetic drift are constantly debated among biologists, both in general terms and on a case-by-case basis.

Friends, you should be suspicious anytime you read a depiction of evolution that focuses solely on natural selection, especially if the writer is addressing function or design. It's hard to overstate the seriousness of that error.

2. Even when natural selection is acting strongly, it most certainly can favor variations that are not "functionally advantageous." This is one major motivation for the new "Harmful mutations" series here at Quintessence of Dust. Not only can natural selection fail to remove variants that we may judge to be disadvantageous, it can actually favor them in some situations. It's simply not true that "natural selection favors only functionally advantageous variations," and again I urge you to be suspicious of writers who disparage the work of professional scientists like Joe Thornton with slogans like that one.

Questions and theories of design and purpose need not rely on such careless errors. ID thinkers need to do better.

Common ancestry, bottlenecks, and human evolution

2011-09-26T18:24:00.000-04:00

Human evolution has been in the news quite a lot recently.

New genetic data suggest that ancient humans included both Neanderthals and Denisovans, which colonized different parts of the world but subsequently interbred with so-called modern humans and left telltale traces of this history in the genomes of living humans.
New analysis of current genetic diversity suggests that human population size underwent interesting fluctuations throughout the history of our species, but concludes that the population never dipped below a few thousand reproducing individuals.

Unsurprisingly, these findings have been discussed in the context of Christian views of human origins. In the context of some of these discussions (among Catholics, for example), I have noticed some confusion regarding the implications of common ancestry. I will illustrate the error with a stylized example, then explain why it is an error.

Consider an important and human-specific genetic feature. For example, consider the human-specific version of the FoxP2 gene. This gene is thought to play an important role in human speech, and the data indicate that the gene was mutated at some point to create the human version, since the gene itself is not specific to humans. In other words, at a key juncture in human evolution, the human-specific FoxP2 gene form came to be.

Since that time, that gene form became the only gene form in humans. There might have been more than one occurrence of the mutation, but it can't be that the mutation is terribly common, since it isn't found in any other mammal or in other primates. Therefore, the existence of the human-specific FoxP2 gene is overwhelming evidence that all humans (past and present) trace their ancestry through one or just a few ancestors who first acquired the mutation.

So far, so good.

Now, if we all trace our ancestry to just one or two ancestors, then it must be that the human population must have gone far below a few thousand. It must be that the human population declined to near zero, and specifically it must have declined to the number of those common ancestors. That's the only way that all humans could currently have that genetic mutation.

That's completely wrong. It might make sense superficially, but it's wrong, and some careful thinking about how inheritance works should make that obvious.

Let's assume that the mutation provided some advantage to the first animals who bore it. In fact, let's assume that the benefit was huge. And let's assume that it occurred in a small population of 100 individuals. (In reality, those are probably unrealistic assumptions; mutations rarely confer an instantaneously huge benefit, and the human population isn't thought to have gone as low as 100.) The outcome of such a scenario is this: the next generation will include more individuals bearing the mutation. Considering human reproduction rates, let's say that the next generation includes 5 of those individuals, and let's allow those individuals to interbreed. The generation following that one would include, say, 15 individuals (3 females with 5 kids each). And the following generation might include 40-ish, and so on. Soon, every member of the population would have the mutation, and every one of those individuals and all of their descendants would share ancestry with the first animal with the mutation. But the population never shrank; in fact, it could have grown during the process and the march of the mutation would have occurred just fine.

This is almost certainly what occurred hundreds or thousands of times in the evolution of our species, and the result is that we have hundreds or thousands of human-specific features that we all inherited from one or a few common ancestors. That fact alone does not mean that our entire population ever contracted to include only those common ancestors.

Evidence for common ancestry is not evidence for genetic bottlenecking. Think about it.

Harmful genes, and sneaky, too: Genetic hitchhiking in the human genome

2011-09-23T08:41:00.000-04:00

Genetic hitchhiking is thought to be an inevitable result of strong positive selection in a population. The basic idea is that if a particular gene is strongly selected for (as opposed to selected against), then the chunk of the genome that carries that gene will become very common in the population. The result is a local loss of genetic diversity: all (or nearly all) of the individuals in the population will have that same chunk of genetic information, whereas before the selection process acted, there might have been a lot of variation in that chunk throughout the population. And this means that areas of the human genome that are less variable between people are suspected sites of recent positive selection. Within that chunk, there are potentially many genes and genetic elements that became more common in the population by virtue of their placement near the gene that was actually selected for. Those other genes are the hitchhikers. And it's likely that some hitchhikers are bad news – they're harmful mutations that would normally become rare or extinct in the population, but instead have become common by hitchhiking.

In the last few years, large amounts of genetic information have become available that have enabled biologists to look for evidence of such phenomena in the human genome. Specifically, two major projects have collected genetic data for the purpose of analyzing genetic variation among humans. One project, the International HapMap Project, mapped and quantified sites in the human genome that are known to vary among humans by a single genetic letter. These sites are called single nucleotide polymorphisms, or SNPs (pronounced "snips"). The project has mapped millions of these sites in a group of 270 humans representing various lineages. Another project that has made the news recently is the 1000 Genomes Project, which also seeks to provide a picture of human genetic variation using more people (more than 1000 at present) and slightly different technology. Efforts like these have taken analysis of the human genome to a new level. No longer do we merely wonder what "the" human genome is like – we can begin to learn about how genetic differences give rise to biological differences such as susceptibility to particular diseases.

And we begin to look for evidence of positive selection in the human genome. One goal is to identify those genetic changes that might underlie recent evolutionary changes in our species, whether they are changes that affected the whole species or changes that are specific to particular human subpopulations; the HapMap Project spawned a prominent article in Nature in 2007, and more recent work has expanded on those initial findings. A different goal is to look for evidence that harmful genes can indeed hitchhike into our genome, riding the coattails of the good stuff that natural selection picked out. A recent paper in PloS Genetics ("Evidence for Hitchhiking of Deleterious Mutations within the Human Genome" by Sung Chun and Justin Fay) reports on a very interesting attempt to address that question, using data from the 1000 Genomes Project.

Before we tackle the authors' data, let's carefully frame their question and their assumptions. Picture a random chunk of the genome, large enough to contain a number of genes. Like any chunk of human DNA, it should contain SNPs; in other words, there should be variation in that chunk that would be apparent when it was compared among numerous people. Now, most of the variation will be neutral, meaning that it has no detectable effect on fitness. I have an A at that site, you have a G, and we're both fine. But some of the variation might not be neutral. There might be a harmful SNP in that chunk. I have an A at that site, you have a G, but my A makes me more likely to die before I reproduce. Now, all things being equal, the neutral SNPs should be a lot more common than the harmful ones, because natural selection will weed out the harmful variants but have no effect on the frequency of the neutral SNPs. And this leads to a prediction: hitchhiking should favor the harmful variants compared to the neutral variants, because hitchhiking shields the harmful SNPs from natural selection. Such shielding doesn't benefit the neutral variants at all, because they were never subject to selection in the first place. And so, hitchhiking should increase the frequency of harmful SNPs compared to neutral SNPs, specifically in the chunks of the genome that have been subjected to recent positive selection. It's important to understand this prediction, because the purpose of the authors' analysis was to test the prediction as it applies to humans.

In order to perform their analysis, the authors needed a way to identify whether a given SNP is neutral or harmful. Ideally, this would be done empirically or experimentally. But we know so very little about the effects of particular genetic variants in humans. So Chun and Fay classified SNPs as neutral or harmful by doing a comparison across 32 species using a procedure they had previously developed. Their reasoning was basically as follows: if a particular change is also found in other animals, it's likely to be neutral. While there's no way to precisely know the accuracy of the classification test, it's reassuring to learn that 72% of known human disease genes were classified as harmful using their method.

Let's focus on three of the figures in their paper to see how they reached their conclusions.

Their most basic prediction was that in regions showing evidence of hitchhiking (i.e., regions with reduced variation among humans), harmful SNPs would be increased in frequency compared to neutral SNPs. Figure 3D (below) shows that this is indeed the case, and it applies no matter how the hitchhiking is detected. (There are at least 9 different ways to identify genomic regions that are likely to have experienced hitchhiking.) In the graph, a ratio of 1 indicates that the region of interest is no different than non-hitchhiking regions. A ratio of greater than 1 indicates that there are more harmful SNPs per neutral SNP than in non-hitchhiking regions. Note that the increase in the ratio varies depending on the methods employed, but that the ratio is statistically different no matter what.

A related prediction is this one: the ratio of harmful SNPs to neutral SNPs should be higher when closest to the hitchhiking region, and should decline with distance. This is because the shielding effect of selection depends on proximity to the positively-selected gene; hitchhikers that are further away are more likely to be separated from their ride by recombination. And this is again the case, as illustrated in Figure 4 (below; click to enlarge). Distance away from the epicenter (green dot) of the hitchhiking region increases from left to right, and the harmful-to-neutral ratio declines with distance. The two panels depict results using two different methods of identifying hitchhiking regions, and the authors explain why the method on the right shows a less-dramatic decline (see the last paragraph on page 3 of the PDF).

The preceding data was from analysis of lots of likely hitchhiking regions, identified in various ways. But we know of some specific sites of likely recent positive selection in the human genome – sites where we even know which gene is involved and can infer the reason why that variant is beneficial. So Chun and Fay zoomed in on 10 of those regions, and looked for that telltale sign of harmful freeloaders. Sure enough, harmful SNPs were strongly enriched right around those positively-selected genes. Have a look at Table 2 below (click to enlarge); the harmful-to-neutral ratios are 1.83 inside the target gene (the positively-selected gene) and 0.69 in the neighborhood immediately surrounding it. The ratio in non-hitchhiking regions is 0.41.

One further finding that is worth mentioning: Chun and Fay showed that known disease-causing mutations are enriched in hitchhiking regions compared to non-hitchhiking regions. Now for some closing comments.

By providing evidence for enrichment of harmful variants in hitchhiking regions, Chun and Fay have also provided further evidence that such regions are indicative of the influence of positive selection.

The authors suggest that "positive selection has had a significant impact on deleterious polymorphism and may be partly responsible for the high frequency of certain human disease alleles." And they propose a nice way to further test their hypothesis: perform similar analyses of the genomes of domesticated animals, where there has been strong recent positive selection, typically on known traits.

Hitchhiking is just one way that harmful genetic variants can be maintained in a population, and its existence is an expected result in sexually reproducing organisms like us. There are many reasons to distrust highly simplified depictions of evolutionary genetics, and I think this is one of the most important.

The paper is worth spending some time to read. Some sections are technically demanding, but the Discussion section is approachable, I think, and it is a thorough and cautious overview of the findings and their significance and context.

Chun, S. and Fay, J.C. (2011). Evidence for Hitchhiking of Deleterious Mutations within the Human Genome. PLoS Genetics, 7 (8) DOI: 10.1371/journal.pgen.1002240

New limbs from old fins, part 3

2011-09-22T11:03:00.000-04:00

The third installment of my series at BioLogos is now up. It discusses the developmental mechanisms that underlie the construction of limbs, and the striking fact that these mechanisms are the same ones used to construct fish fins. Watch for an appearance by Sonic Hedgehog.

Genetic hitchhiking in English

2011-09-19T21:02:00.000-04:00

The next post will discuss recent evidence for genetic hitchhiking in humans. So, what do we mean when we say that genes can hitchhike? To make sense of this phenomenon, we first need to review chromosomes and sexual reproduction.

Most people know that sexual reproduction creates offspring that are genetically distinct from both of the their parents. That's true, but the genetic scrambling that occurs is more significant than is sometimes reported. Let's start by looking at chromosomes.

Like every other animal (or plant or pretty much any other organism), your genetic endowment is carried in chunks of DNA called chromosomes. You have 23 of these chunks, which are rather like volumes in a set of encyclopedias. More completely, you have 23 pairs of these volumes; one set was contributed by your mother and the other by your father. Each of your parents had a complete set, also consisting of a set from Mom and a set from Dad. When your mother made the egg that became the zygote that became you, she provided you with one copy of each volume in the set, and she chose those copies randomly. For example, she may have chosen her dad's copy of chromosome 1, but her mom's copy of chromosome 2. Just by virtue of this random picking process, she made an egg with a shuffled version of her own genetic cards. Dad did the same when he made his sperm, and so your genetic complement is an amalgamation of your parents' genomes which were amalgamations of your grandparents' genomes, and so on.

But before the random picking process occurred in the steps leading up to the final egg/sperm, something remarkable happened to further shuffle the genetic deck. For each chromosome, the different copies lined up with each other and exchanged contents. In other words, a new chromosome 1 was made that was an amalgamation of the maternal chromosome 1 and the paternal chromosome 1. The two new versions were chromosomes unlike any in your mom or dad; they were new creations, clearly designed to maximize the diversity in your genetic inheritance. This process, which is illustrated in the diagram below, is called crossing over.

The figure shows two instances of crossing over, creating the amalgamations that are part white, part black. In the real process, crossing over can occur at multiple sites along the chromosome, so that the resulting amalgamations are black-white-black-white and so on.

What this means is that you received, from each of your parents, a set of chromosomes that included at least some which were shuffled versions of their own chromosomes. And more importantly, this means that the units of genetic material that you received were much bigger than individual genes (which can barely be visibly represented on a diagram like the one above) but typically smaller than an entire chromosome. It's as if you were given a set of encyclopedias in which individual volumes had chapters from one version of that volume and chapters from another. Individual genes would be merely pages. The basic lesson here is that you received your genes from your parents in chunks, like chapters, and not one by one, like pages.

What does this have to do with hitchhiking? Well, suppose that in one of those chapters, meaning in one section of one chromosome, there appeared a beneficial mutation of some kind, and suppose that this mutation conferred an advantage on every individual who carried it. Over a relatively short time (evolutionarily speaking), that chapter could become a lot more common in the population. It may even become so common that it's the norm, in which case it would be considered to be fixed in the population. (The process is then called fixation.) Notice, importantly, that we said the chapter will become fixed. Why not just the gene? Because the pieces of DNA that are passed down in each generation are a lot bigger than that, as we just saw.

The basic message, then, is this: when an organism inherits some new and beneficial gene, it inherits everything in the vicinity of that gene as well. If that new and beneficial gene becomes fixed in the population, then everything in the vicinity will be fixed as well. The result is that when a strong selection process acts, and drives a new gene to fixation in a relatively short time, it leaves a mark on the genome: one chapter in the set of encyclopedias will be oddly the same in everyone. That chapter will display a lot less genetic diversity than other chapters. That's the signature of recent positive selection, resulting from a so-called selective sweep. And it results in the fixation of a lot of stuff, most of which is just along for the ride by virtue of being located in the same chapter as the beneficial gene. All that other stuff got there by hitchhiking.

Image credit: Wikipedia. Image is from T.H. Morgan, A Critique of the theory of evolution (1916).

New limbs from old fins, part 2

2011-09-16T12:32:00.001-04:00

The second post in my series on limb evolution is now up at the BioLogos site. This installment reviews the fossil evidence on fin-to-limb evolution, introducing the famous Tiktaalik. Next up: evidence from developmental biology.

The first post at BioLogos outlined limb structure and some historical background. The series at BioLogos was spawned by an idea here at QoD, which aimed to discuss some new findings in the fins-to-limbs story. Those new findings will be discussed in the fifth installment of the series at BioLogos.

Please comment! You can leave comments here or at BioLogos.

"The stamp of one defect": an endless series on harmful mutations

2011-09-13T17:07:00.001-04:00

Not surprisingly, Hamlet weighed in on the nature vs. nurture question, at least once.

So, oft it chances in particular men,
That for some vicious mole of nature in them,
As, in their birth,―wherein they are not guilty,
Since nature cannot choose his origin,―
By the o’ergrowth of some complexion,
Oft breaking down the pales and forts of reason,
Or by some habit that too much o’er-leavens
The form of plausive manners; that these men,
Carrying, I say, the stamp of one defect,
Being nature’s livery, or fortune’s star,
Their virtues else, be they as pure as grace,
As infinite as man may undergo,
Shall in the general censure take corruption
From that particular fault: the dram of eale
Doth all the noble substance of a doubt,
To his own scandal.

– Hamlet, Act I, Scene IV, The Oxford Shakespeare

It is certainly true that "the stamp of one defect" can wreak havoc on the scale that Hamlet describes, and whether the result is a debilitating physical limitation or damage to "the pales and forts of reason," the outcome is tragic by any measure.

Reflecting on the reality of inherited dysfunction, we might be tempted to assume that a "vicious mole of nature" is something seen only "in particular men," and that those who are not so characterized (let's call them "normal people") have been dealt a genetic hand that lacks such devilish cards. Normal people don't have bad genes.

Okay, so in the real world I suspect that most people are not so naïve; if you're reading this blog, then you probably know that bad genes can be carried by normal, healthy people. Nevertheless, when we think about bad genes – or more technically, deleterious mutations – we are likely to think that they are not very common.

There is at least one good reason to assume that deleterious mutations are uncommon in the human gene pool: good old-fashioned natural selection. Deleterious mutations are eliminated from every gene pool by natural selection, and everyone knows that. Given that this fact is both obvious and widely known, it is likely to be a surprise to many people to discover deleterious mutations in any significant frequency in any population anywhere.

But in fact, deleterious mutations are ubiquitous in populations of all sorts, and they don't necessarily go away just because they're harmful. To assert or imply (as many regularly do) that a deleterious mutation cannot be maintained in a population is to oversimplify a basic principle to the point that it becomes badly misleading.

But how? How can a harmful mutation remain in a population? Consider this list of possible explanations, which are not mutually exclusive.

The mutation is only harmful in certain circumstances (e.g., certain environmental conditions or certain combinations of other genes) and is invisible or even beneficial in others.
The mutation is only harmful when present in two copies, and is invisible or even beneficial when present in one copy.
The mutation exerts its harmful effects after the organism has reproduced.

Now, that list doesn't seem to take the problem seriously enough. It's a list of reasons why the mutation isn't really harmful; instead, the gene is conditionally harmful or its harmful effect is irrelevant from an evolutionary perspective.

Here are some more interesting possibilities.

The mutation is truly harmful, but the population is (or recently was) quite small and so natural selection acts less intensely. Put metaphorically, the population "tolerates" the deleterious mutation because it doesn't have a "choice."
The mutation is truly harmful, but in previous generations it was beneficial.
The mutation is truly harmful, but it comes as a package deal with something else that is beneficial. The net effect is positive.

I think that list covers just about everything. Explanation number 3 isn't very interesting, so we probably won't come back to it very often. Explanation number 5 is a special case of number 1, in which the environment has changed permanently.

That's the overview. This post serves as the introduction to an open-ended series on deleterious mutations, and the debut for two new tags: "deleterious mutations" and "bad genes." The next post in the endless series will discuss a new paper in PLoS Genetics, which presents evidence for "hitchhiking" of deleterious mutations in humans. That sounds like a risky business, now doesn't it?

----
Image: Kemble in the role of Hamlet, standing on a grassy bank wearing a long cape edged with fur and a large feathered hat on his head; he is holding a skull in his hand. Engraving after Sir Thomas Lawrence. Courtesy of Wellcome Images.

New limbs from old fins, part 1

2011-09-08T13:02:00.000-04:00

Last month, I started a series on the topic of limb evolution, here at Quintessence of Dust. That series has been transformed (through a series of intermediates) into a series of posts at the BioLogos site. The first installment is now up, and it provides an expanded introduction to the topic and a little historical context. Subsequent posts will tackle fossils, developmental biology, genetics, the explanatory role of design, and related themes.

So go check out the introduction, and feel free to contribute comments, questions and suggestions here. And enjoy the image below, from Wellcome Images, which is featured in the post at BioLogos. Cool, huh?

Molecular evolution: improve a protein by weakening it

2011-08-08T02:14:00.015-04:00

In the cartoon version of evolution that is often employed by critics of the theory, a new protein (B) can arise from an ancestral version (A) by stepwise evolution only if each of the intermediates between A and B are functional in some way (or at least not harmful). This sounds reasonable enough, and it's a good starting point for basic evolutionary reasoning.

But that simple version can lead one to believe that only those mutations that help a protein, or leave it mostly the same, can be proposed as intermediates in some postulated evolutionary trajectory. There are several reasons why that is a misleading simplification – there are in fact many ways in which a mutant gene or protein that seems to be partially disabled might nevertheless persist in a population or lineage. Here are two possibilities:

1. The partially disabled protein might be beneficial precisely because it's partially disabled. In other words, sometimes it can be valuable to turn down a protein's function.

2. The effects of the disabling mutations might be masked, partially or completely, by other mutations in the protein or its functional partners. In other words, some mutations can be crippling in one setting but not in another.

In work just published by Joe Thornton's lab at the University of Oregon, reconstruction of the likely evolutionary trajectory of a protein family (i.e., the steps that were probably followed during an evolutionary change) points to both of those explanations, and illustrates the increasing power of experimental analyses in molecular evolution.

The main focus of the Thornton lab is the reconstruction of the evolutionary pathways that gave us various families of steroid receptors. By combining phylogenetic inference (i.e., examining molecular "pedigrees" to infer the nature of an ancestral protein) and hard-core biochemistry and biophysics, Thornton and his colleagues can identify the likely ancestral proteins that gave rise to the modern receptors, then "resurrect" them in the lab and study their properties.

The reconstructions of receptor pedigrees have led to the following basic history of the receptors.

1. An ancestral receptor, present about 450 million years ago in a vertebrate, responded to two different kinds of steroid hormone at high sensitivity. This means that the receptor wasn't very selective, but it was sensitive, so it didn't take a lot of hormone to get a response.

2. The gene encoding that receptor was duplicated at some point, and the two resulting genes diverged to become specialized. One receptor (we'll just call it the MR) retained many of the ancestral features: low selectivity, high sensitivity. The other receptor (the glucocorticoid receptor, which we'll call the GR) became selective for one type of hormone (glucocorticoids, the kind of steroid that includes cortisol) but also became a lot less sensitive. And that is the current situation in all vertebrates, as far as we know.

Previous studies in Thornton's lab outlined a likely trajectory through which the different receptors probably changed specificity. But the interesting loss of sensitivity remained unexplained. In the June 2011 issue of PLoS Genetics, Sean Michael Carroll and his colleagues in the Thornton lab tackle this question, uncovering an interesting type of genetic interaction that is a hot topic in evolutionary biology right now. Their paper is titled "Mechanisms for the Evolution of a Derived Function in the Ancestral Glucocorticoid Receptor."

The authors already knew that all known GRs have low sensitivity. That includes GRs from the main types of vertebrates: tetrapods like us, bony fish like salmon and piranhas, and non-bony fish (called cartilaginous fish) like sharks and rays. They suspected that this meant that hormone sensitivity was first reduced in the common ancestor of all of those animals. To test this hypothesis, they first needed to examine GRs from animals throughout the vertebrate family tree, so they could infer the nature of the GR in the common ancestor. You might want to take a few minutes to examine the simple family tree of the receptors and of the various animals, presented in Figure 1 and reproduced below.

So the authors obtained GR gene sequences from four different species of shark (representatives of the cartilaginous fishes), then created the proteins in the lab and asked whether low sensitivity was a universal feature of cartilaginous fishes. Figure 2 illustrates the simple answer: yes, it is.

Each of the points on the graph represents the amount of hormone necessary to activate the receptor. The different colors represent different kinds of hormone, and we can focus on just one, let's say the dark blue. The different animals are indicated across the bottom; on the left are ancestral receptors, inferred from looking at the family tree, and on the right are each of five cartilaginous fishes: a skate (previously known) and the four new contestants (the sharks). Notice that the amount of hormone needed to activate the receptor goes up with each step in the left-hand box, and is even higher in each animal in the right-hand box. It takes more hormone to activate the receptor – it became less sensitive, and seems to have acquired this characteristic a long time ago.

How long ago, and how would did the authors infer that? Well, they created a postulated reconstruction of the ancestral receptor – the receptor in the common ancestor – by combining knowledge of rates of change in proteins with the new knowledge gained by looking at four new animals (the sharks). The graph in Figure 2 shows the hormone sensitivity of this new reconstruction (it's version 1.1) compared to an older reconstruction based on just one cartilaginous fish (that was version 1.0). The new "resurrected" ancestral receptor (AncGR1.1) has much lower sensitivity than its much more ancient ancestor (AncCR), the one with high sensitivity but no selectivity. So it seems that the change in sensitivity happened after the duplication event but before any of the various kinds of vertebrates diverged, something less than 450 million years ago.

But how did the change come about? Let's look back at the family tree in Figure 1 to get oriented. The grandparent of all the receptors is called AncCR and it was sensitive but not selective. After the duplication, there were two parents, if you will: the MR parent which we're not discussing, and the GR parent, called the AncGR. AncGR, the parent of all the GRs, had reduced sensitivity. Carroll and colleagues addressed the how question by first looking at the types of changes that occurred between the grandparent and the parent.

There were 36 changes that accrued during that time. The authors used some straightforward reasoning to narrow the list of suspects down to six. In other words, six different changes in the protein, together or separately, were likely to account for the change in sensitivity. They went into the lab and resurrected each of those mutant proteins, and measured their sensitivity. And their data tell a very interesting story, presented as a graph in Figure 3.

The gray bar is the grandparent. The yellow bar is the parent. (The scale is logarithmic, so the change in sensitivity from grandparent to parent is at least 100-fold.) The other bars represent the sensitivity of some of the mutations that must have generated the low-sensitivity parent. So, the first white bar is mutant 43. (The number represents a particular location of the mutation, but we're not interested in that here.) That mutation drops sensitivity to near-parental levels. The next bar is mutant 116. It also reduces sensitivity to the parental level. Sounds good. But wait: both of those mutations are present in the parent. What happens when you put them together? Disaster. Look at the next white bar: it's the combination of 43 and 116, and the receptor is effectively dead. As the authors put it in the abstract, "the degenerative effect of these two mutations is extremely strong."

Think about what this means. There are two mutations in this transition that account for the loss of sensitivity to the hormone. Both are present in the final product (the parent receptor) but when they are introduced together, they drastically disable the protein. How, then, could this protein have come about?

Well, there were six mutations in the candidate pool. One of them creates mutant 71, the last white bar on the far right. By itself, mutant 71 doesn't affect sensitivity. But when mutant 71 is present with the deadly 43/116 combination (see the next-to-last bar), the result is a receptor with the parent's characteristics. Lower sensitivity, but not complete loss of function.

Carroll et al. dissected the biophysics of these resurrected receptors, and provided a clear explanation for the influence of each mutation in terms of effects on stability of the hormone-receptor complex. Here's how they summarize the results:

We found that the shift to reduced GR sensitivity was driven by two large-effect mutations that destabilized the receptor-hormone complex. The combined effect of these mutations is so strong that a third mutation, apparently neutral in the ancestral background, evolved to buffer their degenerative effects.

That third mutation almost certainly occurred first. Recall that it has no apparent effect on function by itself. Carroll et al. refer to its effect as "buffering" because its presence reverses the destabilization caused by the other two mutations.

Here are some closing comments on the paper.

1. It's nice to keep in mind that we can design experiments to test hypotheses concerning evolutionary trajectories. We can move beyond simplistic and distorted models (if we want to understand evolution, that is).

2. The reduction in function of a protein is not necessarily a "degenerative" change. In the case of the GR protein, lower sensitivity to hormone is likely to create the opportunity for different responses to different levels of steroids.

3. Mutations that are damaging – even devastating – in some circumstances can be harmless or even beneficial in others. On reflection, that should be obvious, but it's a simple fact that seems easy to overlook.

Several recent papers have converged on this basic point: interactions between different proteins, and between different mutations in one protein, exert significant influences on the nature and direction of evolutionary change. The key word here is 'epistasis,' and it's a topic we should return to.

Carroll SM, Ortlund EA, and Thornton JW (2011). Mechanisms for the Evolution of a Derived Function in the Ancestral Glucocorticoid Receptor. PLoS Genetics, 7 (6) : 10.1371/journal.pgen.1002117

Let's see a show of autopods. Part 1.

2011-08-03T18:53:00.001-04:00

The discovery of deep homology was a milestone in the history of evolutionary thought. Anatomical structures in distantly related organisms, structures with only the barest of functional similarities, were found to be constructed under the influence of remarkably similar genetic pathways. The original and classic example from 1989 involves genes controlling pattern in both insects and mammals – the famous Hox genes. Another great example emerged from the study of limb development and evolution in vertebrates, work beautifully described by Neil Shubin in Your Inner Fish.

The idea that the limbs of various animals are homologous – meaning that they are variations on a theme inherited from common ancestors – is certainly not new, with roots in the exploration of 'archetypes' by the great Sir Richard Owen. But deep homology goes, well, deeper, suggesting that even basic themes like 'limb' or 'eye' or even just 'thing-sticking-out-of-the-body-wall' can be identified and seen to be conserved throughout the biological world. And, importantly, deep homology points to genetic mechanisms that underlie basic themes, structural concepts so distinct that they would not be judged to be related by structural criteria alone. Consider, for example, limb development in vertebrates.

Tetrapod vertebrates, like humans and whales and birds, sport limbs that are recognizably homologous, built according to a plan that Neil Shubin memorably summarized as "one bone, followed by two bones, then little blobs, then fingers or toes." The parts are recognizably homologous, and so are the genetic mechanisms that assemble them. The genetic circuitry is basically the same throughout the tetrapods, with subtle distinctions that create structural differences. Bat wings and mouse forelimbs are homologous, both structurally and genetically, because both were modified from common ancestors.

So, what about "limbs" in fish? After all, fish and mice descended from a vertebrate common ancestor, and those fins in fish sure look like they could be modified into limbs. They were, of course, and that's the great story of Your Inner Fish and the wondrous fossil Tiktaalik. Get the book and/or visit the Field Museum in Chicago to learn all about it.

The development of tetrapod limbs and fish fins displays deep homology. Despite the fact that fish fins and human limbs seem structurally distinct, the developmental pathways that sculpt them are actually quite similar. A tetrapod limb (like, say, a human arm) can be divided into three anatomical segments:

Segment (technical name)	Simplified by Shubin	What we call it in humans
Stylopod	One bone	Humerus
Zeugopod	Two bones	Radius and ulna
Autopod	Little blobs and digits	Wrist and fingers

Examination of vertebrate fossils shows that fish fins and tetrapod limbs really are built on a common scaffold, and molecular developmental studies have shown that the first two segments are conserved across all vertebrates. So, a fish pectoral fin has a stylopod extending from the body, with a zeugopod connected to its end. And, amazingly, the genetic systems that control the development of those segments are homologous as well.

But the autopod is a different story. Some data suggest that the autopod is a tetrapod invention, something that developed since fish and tetrapods went their separate ways. But there are some very interesting indications that the ends of fish fins and the hands of humans develop under the control of the same genes. This would indicate that the deep homology of fins and limbs extends all the way to the end, such that human hands and rodent paws are somehow homologous to the margins of fish fins.

How could we address this question? The best approach would go like this. Take the genetic element that controls the development of mammalian digits (e.g., fingers), and look for a similar element in a fish. If you find one, then do a "molecular transplant" experiment in which you swap the two elements and see if they do the same thing. In other words, put the fish element into a mouse, and the mouse element into a fish, and see if they really have the same function.

The results of those experiments were published by Schneider and colleagues in the 2 August 2011 issue of PNAS. The article is open access, and the work was done in Neil Shubin's lab at the University of Chicago. We'll look at their data in Part 2.

[Image is adapted from Figure 4 of Schneider et al., and shows fish fins at the top, mammalian hands at the bottom, and postulated intermediates in between.]

What a selfish little piece of...

2011-08-02T13:32:00.000-04:00

"The Selfish Gene." "Selfish DNA." Oh, how such phrases can get people bent out of shape. Stephen Jay Gould hated such talk (see a little book called The Panda's Thumb), and Richard Dawkins devoted more time to answering critics of his use of the term 'selfish' than should have been necessary. Dawkins' thesis was pretty straightforward, and he provided real examples of "selfish" behavior of genes in both The Selfish Gene and its superior sequel, The Extended Phenotype. But there have always been critics who can't abide the notion of a gene behaving badly.

Leaving aside silly bickering about the attribution of selfishness or moral competence to little pieces of DNA, let's consider what we might mean if we tried to imagine a really selfish piece of DNA. I mean a completely self-centered, utterly narcissistic little piece of DNA, one that not only seeks its own interest but does so with rampant disregard for other pieces of DNA and even for the organism in which it travels. Can we imagine, for example, a piece of DNA that deliberately harms its host in order to propagate itself?

Sure, we might picture genes acting in naked self-interest, perhaps colluding to create an organism that can fly and mate but can't eat. We can picture genes driving organisms to take outrageous risks in order to reproduce. And we can picture millions and millions of "jumping genes" that don't seem to care at all about the host's welfare while they hop about in bloated mammalian genomes. (If you are one who prefers to think of these transposable elements as beautifully-designed marvels of information transfer and storage, you can have a pass on that last one for now, because you won't like where we're going with this.) But can we picture a gene that actively harms its host in order to get ahead?

At first, this might seem ridiculous. How can harming the host help a gene propagate itself? We can talk about the examples above, and explain each through some reproductive benefit or trade-off. But I'm not talking about negligence here; I'm talking about harm. Well, okay. I'm talking about killing babies.

I'm talking about a gene that kills the embryo in which it's expressed, unless the embryo promises to propagate the gene. The most famous example of such an outrageously selfish gene is the Medea element, found in certain beetles. ('Medea' is both an acronym and a deliciously evil description of the effect of the element.) Here's the basic idea: a female that carries the Medea element has some offspring. Some of those embryos will have the Medea element in their genomic endowment and others won't. But all of the embryos will be exposed to the Medea effect, because it comes into the embryo through the egg, which was created by the Medea-carrying mother. The Medea effect kills any embryo that doesn't carry its own copy of the Medea element. The survivors are the ones that carry the element. Pretty smart, huh?

How this works, exactly, is not well understood. But Medea isn't the only selfish little piece of DNA that stoops to infanticide. Another example was described just a few years ago in the nematode C. elegans, that workhorse of developmental genetics. Called the peel-zeel element, it's just a little different from Medea: in the peel-zeel system, the embryo-killing curse comes from the dad. (Selfish elements like this are quite rare, and this paternally-acting system is the only known element of that kind.) But the sick story is otherwise the same: only those embryos that carry their own copy of the peel-zeel element can avoid sperm-carried destruction. Now some new results, published in this month's PLoS Biology, are revealing how this evil plan is carried out. The article, "A Novel Sperm-Delivered Toxin Causes Late-Stage Embryo Lethality and Transmission Ratio Distortion in C. elegans," was authored by Hannah Seidel and colleagues.

The group had previously shown that the paternal genetic element would kill embryos that didn't have an "antidote," and had explained the peculiar genetic arrangement that keeps this element from being driven completely to fixation in the population. (An element that kills everyone but itself would be expected to quickly infest the entire population, but this doesn't occur in the case of the peel-zeel element.) Although the authors knew a bit about the antidote gene (called zeel-1), they knew nothing about the killer gene or how it worked; they knew only that it was probably very close to the antidote gene. They did have one particularly useful tool, especially valuable in the experimental wonderland of genetics that is C. elegans: they had some mutants with perfectly good antidote function but no killing ability. So they used those mutants to do some very nice genetic mapping experiments, and discovered the precise locations of the mutations that abolished the lethal effect. Interestingly, those mutations were in an "intergenic interval" in the fully-sequenced C. elegans genome, right next to zeel-1. In other words, the killing activity seemed to be right next to the antidote, in a part of the genome that contained no known genes. Or, more accurately, it contained no annotated genes. It turns out that we're still discovering new genes in fully-sequenced genomes. (It's actually not that easy to identify a bona fide gene in a gigantic DNA sequence.) And Seidel et al. had just discovered a new gene – the peel-1 gene. It makes a protein somewhat similar to zeel-1.

Once they had the actual gene in hand, the authors could probe the protein's function. They showed that it is packed into a particular type of delivery vehicle inside sperm, which are the only cells that express it. The delivery vehicles ensure that each embryo is provided with an adequate dose of the toxin. Oddly, the lethal protein acts somewhat late in development, in skin and muscle cells, and the embryo dies a grisly death unless it carries the antidote. The image on the right (from the cover of the July 2011 issue of PLoS Biology) shows two affected embryos (the blobs on the left and right) and one happily normal worm.

In another cool experiment, the authors turned on the death gene artificially in adult animals, and it killed them just fine. They could save those otherwise-doomed worms by turning on the antidote artificially.

The peel-zeel element, then, is a great example of a truly ruthless selfish genetic element. The toxin and the antidote are side-by-side in the genome, so that an animal with the antidote will almost certainly also receive the toxin. (Think about how different things would look if the antidote gene were separate from the toxin; the toxin could quickly lose its ability to propagate itself through the generations.) And the toxin is sperm-delivered to all embryos. This combination of traits allows the paternally-carried element to kill any embryo without a copy of the element.

As far as we know, the peel-zeel system serves only its own interests. It offers no fitness advantage to its host, and is likely instead to exact a cost. Its presence in the nematode genome is easy to explain in a biosphere teeming with "selfish" DNA that admits no evident "purpose" beyond its own propagation. That's not to say it can't be useful; as an accompanying commentary notes, DNA-encoded toxin/antidote systems could be employed by well-meaning humans to seemingly benevolent ends. But whether or not one chooses to see the peel-zeel system as a product of "design," the pattern of "selfish" propagation is hard to miss. And, surely, hard to restrain.

Seidel, H., Ailion, M., Li, J., van Oudenaarden, A., Rockman, M., & Kruglyak, L. (2011). A Novel Sperm-Delivered Toxin Causes Late-Stage Embryo Lethality and Transmission Ratio Distortion in C. elegans. PLoS Biology, 9 (7) DOI: 10.1371/journal.pbio.1001115

Evolution cheats, or how to get an old enzyme to do new tricks

2011-07-31T06:34:00.006-04:00

It is of course a cliche to state that eukaryotic cells (i.e., cells that are not bacteria) are complex. In the case of an animal, tens of thousands of proteins engage in fantastically elaborate interactions that somehow coax a single cell into generating a unique and magnificent organism. These interactions are often portrayed as exquisitely precise, using metaphorical images such as 'lock-and-key' and employing diagrams that resemble subway maps.

Many of these interacting proteins are enzymes that modify other proteins, and many of those enzymes are of a particular type called kinases. Kinases do just one thing: they attach phosphate groups to other molecules. This kind of modification is centrally important in cell biology, and one way to tell is to look at how many kinases there are: the human genome contains about 500 kinase genes.

Now, kinases tend to be pretty picky about who they stick phosphate onto, and this specificity is known to involve the business end of the kinase, called the active site. The active site is (generally) the part of the kinase that physically interacts with the target and transfers the phosphate. You might think that this interaction, between kinase and target, through the active site, would be by far the most important factor in determining the specificity of kinase function. But that's probably not the case.

It turns out that many kinases have other ways of choosing mates. Some use 'docking' sites that are outside the active site. And some rely on a different set of proteins called scaffold proteins to fix them up with their partners. And this suggests that a kinase signaling system can be rewired by messing around with these alternative modes of specificity, without changing the active site itself. This raises a very interesting question, one with implications for evolution and for synthetic biology: just how flexible are kinase signaling systems? Can they be rewired at all? Or perhaps at will?

Wendell Lim's lab at UCSF is famous for tackling questions like this. (Their excellent lab website includes most of their papers, including a very recent review article on scaffold proteins.) In previous work, they showed that kinase pathways can in fact be rewired: in a widely-discussed 2003 Science paper, Sang-Hyun Park and colleagues converted one yeast kinase pathway (controlling mating) into another (controlling osmotic stress responses) by altering the scaffolds without changing the enzyme active sites at all. It was a remarkable result, suggesting that one could get completely new kinase signaling systems without any need to make new catalytic machinery; i.e., without any monkeying with the touchy active sites.

But in that experiment, only one kinase was involved. The rewiring involved differential deployment of the same kinase for distinct purposes and, importantly, that kinase does normally control both mating and osmotic stress responses. So the 2003 experiment left open this question: can a kinase be forced to perform a completely new function, without changing its basic catalytic machinery? To address that question, Angela Won and colleagues in the Lim lab turned to an interesting set of kinases in yeast: the MAP kinase kinases, or MAPKKs. (The nomenclature is awful, I know.) Their work was published in an open-access article in PNAS last month, titled "Recruitment interactions can override catalytic interactions in determining the functional identity of a protein kinase." Figure 1A illustrates their experimental system:

All those ovals are kinases. The MAPKKKs act on the MAPKKs, which act on the MAPKs, which further elicit the biological responses listed across the bottom (the triggers for these responses are listed across the top). Won et al. focused on the MAPKKs, in the second row. Note that there are four yeast MAPKKs: Mkk1, Mkk2, Pbs2, and Ste7. The Mkks are both involved in cell wall remodeling, whereas Ste7 controls both mating and filamentation (i.e.,growing as a mold). Notice that each stimulus-response pair (e.g., pheromone and mating) is controlled by a cassette of kinases (e.g., Ste11-Ste7-Fus3 in the case of mating). And notice the box thingy on the far right, which enfolds the mating cassette. It represents Ste5, which is a scaffold protein. Ste5 assembles the cassette, bringing the players into proximity, and that scaffolding function is critical: when Park et al. mutated Ste5 to abolish its binding to Ste7, the mating response disappeared. Ste7 also bears two docking sites for the kinase Fus3, and those too are known to be important for mating function.

So, we have Ste7 at the center of a specific kinase signaling cassette. It relies on two docking sites and a scaffold protein to get into an effective signaling complex. How important are those interactions in determining the specific effect of Ste7 on mating? In other words, could we force Pbs2 to control mating, by giving it a Ste5-like scaffold? Or those two docking sites? Or all of the above? Or is Pbs2 catalytically designed to only act on Hog1, and thus to only control osmotic stress responses?

The first experiment showed that Ste7 does need both the docking sites and the scaffold to fully function. The authors showed this by making a mutant of Ste5 that can't stick to Ste7: that mutant kills the pathway. (Second bar in the graph below, which is Figure 2C. The top bar shows normal mating with normal components.) When they took this dead mutant and tethered it to Ste7, poof, mating was fine (third bar), unless they removed the docking sites from the Ste7 beforehand, in which case mating was reduced by 100-fold.

The next experiment was the big one: the authors attempted to force the other MAPKKs to induce mating by giving them the scaffold (by tethering to Ste5) and/or the docking sites (by adding in the relevant piece of Ste7). With just one or the other, the system was still dead. (Check out Figure 3B.) But with both a scaffold and docking sites, two of the three other kinases activated mating. For example, Pbs2, which normally controls osmotic stress responses, induced mating when it was tethered to Ste5 and to the two docking sites in Ste7. And, interestingly, that rewired Pbs2 variant was now incompetent at its normal job.

What this means is that kinases can acquire most or all of their legendary specificity from the company that they keep, such that their functional roles can be completely changed without altering their catalytic machinery. It means that signaling systems in cells are a lot more flexible than we used to think.

For synthetic biologists, this is good news: creating completely new signaling systems need not mean engineering new enzymes, as previous work from the Lim lab and elsewhere had already demonstrated. And the result strengthens the notion of evolvability and concepts central to evo-devo (especially modularity); in fact, Lim and his colleagues have compared modularity of signaling systems with that of gene expression:

The ability to recombine pathways and regulate signaling behaviors with scaffolds can be likened to how the modular architecture of promoters gives rise to the diverse transcriptional responses that differentiate cell and tissue types.

Results like this point to flexibility in system architecture, but also demonstrate how enzyme specificity – something thought to be particularly difficult to evolve – can be altered through changes in recruitment and location. Remarkably, changes in these interactions, known to be labile under normal conditions, can lead to new biochemical functions. It doesn't happen every time; there are no "universal laws of cellular evolution" being proposed here. But such flexible wiring is something to keep in mind every time you read how impossibly hard evolutionary changes must be.

I'll let Won et al. provide the closing summary:

A growing body of work supports the idea that sophisticated cellular signaling networks in complex eukaryotes have arisen through the generation of new circuitry using a limited toolbox of parts rather than the evolution of novel proteins. Here, we examine the specificity of MAPKKs through attempts to use existing connections to convert MAPKK identity. We have successfully converted two alternative MAPKKs to Ste7 functionality, showing that we can use these simple protein interaction elements to redefine kinase behavior. However, we were unable to find an absolutely generic formula for converting kinases, as we find that some intrinsic contributions to specificity cannot be overcome by recruitment interactions.

Won, A., Garbarino, J., and Lim, W. (2011). Recruitment interactions can override catalytic interactions in determining the functional identity of a protein kinase. Proceedings of the National Academy of Sciences, 108 (24), 9809-9814 DOI: 10.1073/pnas.1016337108

Design and falsifiability

2011-07-29T17:34:00.002-04:00

Last month I had an interesting conversation with Casey Luskin of the Discovery Institute (DI), at Evolution News and Views (ENV), a DI blog/site that recently opened some articles to comments. The topic of the original post was common ancestry in humans and other primates, but Casey and I discussed various aspects of design thought.

One subject that came up was the falsifiability of design. I maintain that design arguments, whenever they also postulate the existence of an omnipotent deity (or any super-powerful being, for that matter), are inherently unfalsifiable. And I want some feedback on my argument.

Here's what I wrote on ENV:

Design is unfalsifiable to whatever extent the postulated designer is capable of acting in the world. If the designer (like the Creator God) is omnipotent, then it is impossible to rule out deliberate design in any place at any time. This is a necessary conclusion that can only be avoided by restricting the expected actions/motives of the designer. You claim that "shared non-functional similarities" can falsify "common design," and that's true only if you have defined "common design" in a fairly restricted way. What such similarities don't do – cannot do – is rule out the action of a designer. (That designer could have other reasons for doing things the way she does, meaning that "shared non-functional similarities" could evince design just as strongly as any other genomic feature.) That's what I mean when I say that design is unfalsifiable, and I hope that clarifies things.

Casey's response focuses on "the theory of intelligent design," which he claims is solely concerned with positive evidence for intelligent design, which is assumed to be detectable in the world. He concedes that yes, the theory could fail to detect design when/if the designer has acted in ways indistinguishable from "secondary material causes." He illustrates this using a standard type of example of design (in his case, flowers that spell out "Welcome to Disneyland").

He's right about all that. But I think he's wrong about the falsifiability of design, and he himself has told us why. Consider his flower-based message example. He's quite right that a person (let's call him Steve) looking at a bed of flowers that spells out a message in English can and should conclude that the flower bed is the product of design. But Steve can't point at any other collection of flowers and claim that it is not the product of design.

In order to make that claim, Steve would first need to stipulate some of the characteristics of the designer (we'll call her Coco). Specifically, Steve would need to tell us whether Coco is thought to – or known to – design flower beds that don't look designed (to Steve). And this is where my argument gets specific: I maintain that once Steve postulates Coco's omnipotence, then he has acknowledged Coco's ability to design flower beds of every possible configuration, few of which Steve would identify as "designed." Thus any designation of a flower bed as "designed" is unfalsifiable, since all flower beds are potentially designed regardless of their appearance. If Steve wants his design argument to be falsifiable, he needs to further specify Coco's characteristics (limitations, preferences, and so on) as a designer and explain how such characteristics can enable him to rule out design of a particular flower bed.

If Steve takes Casey's line and claims not to know anything about Coco, then Steve cannot under any conditions point to anything that Coco didn't design. And so his claim that the flower-based message is designed is unfalsifiable.

We can add that this doesn't mean Steve is wrong. In fact, in the case of the hideous "Welcome to Disneyland" flower bed, we'd all agree that he's right. It just means that his design claim can't be falsified.

Now, I don't think this means that design thought is therefore nonsense, or that attempts to identify evidence of design are therefore invalid. Not at all. But I do think it points to a vast difference between "the scientific theory of intelligent design" and common descent. Common descent is falsifiable, at least on a case-by-case basis, meaning that there are observations we can imagine that could not be explained in principle by common ancestry. But it seems to me that there is no such observation vis-a-vis intelligent design, especially when/if the designer is taken to be super-powerful or even omnipotent.

[Cross-posted at Panda's Thumb]

Conversing with Casey Luskin

2011-07-25T00:16:00.001-04:00

Last month I wandered over to Evolution News and Views (ENV), a Discovery Institute (DI) blog, and read a piece by Casey Luskin on the topic of human/chimp common ancestry. I saw some stuff I didn't like, and left a comment, and an interesting exchange ensued. You can read it yourself, but here are some of my comments.

1. I think I was able to communicate the nature of my disagreement with Casey, namely that I object to misleading portrayals of science but not to efforts to emphasize design perspectives and theories. I've said all that before, but I really want to emphasize it in future discussions with design theorists. Unlike many critics of ID, I don't think that design concepts are ridiculous and I don't believe that wondering about design is the same as advancing "creationism." But I also have serious criticisms of some of what is said and written by DI thinkers.

2. Casey identified mistakes on my part (in some of my characterizations of his writing and his positions), corrected one error of his (and I applaud that action), and acknowledged the explanatory power of common descent. Those were all constructive aspects of our conversation.

3. We both said we'd like to have lunch sometime. I think it will happen; I'm not sure where or when.

4. Casey wrote something really important that summarizes what I hope will become the central theme of all future discussions between the thinkers of the DI and me:

Everyone makes errors sometimes. Isn’t it better to ‘judge not’ and simply rebut the arguments of one’s opponent rather than making constant accusations of general incompetence? Let’s take a more civil approach where we just critique one another’s arguments and not constantly allege intellectual or moral failings of our opponents.

I wholeheartedly agree, and in fact I'll go a little further and make it a commitment. Let me close with one caveat and then point to another example of how the conversation should go.

The caveat is this, and I know it sounds trite: it will be difficult at times to avoid the appearance of alleging intellectual or moral failure. For example, when/if I suggest that a DI commentator has significantly misunderstood a scientific concept or a report in the literature, it may seem that I am accusing that person of an "intellectual failing." (And the same applies to my critics when/if they suggest that I don't understand something.) If/when I do seem to make that accusation, I will apologize. But I think we will all have to tolerate that kind of criticism, then rebut it (or correct our own errors) without objecting to imputations of intellectual incompetence. Specifically, alleging that someone is badly mistaken, or even suggesting that they have failed to adequately study an issue on which they are writing/speaking, is not the same as accusing that person of stupidity or indecency. After all, as Casey rightly notes, we all make errors sometimes.

I hope that was clear. I will seek to err on the generous side, and will be quick to apologize and move on. I'm sure there will be problems, but I'm also sure that Casey Luskin really does want to have a productive conversation, and it sure does seem that we can disagree politely and honorably.

Finally, I will credit Reed Cartwright – an evolutionary biologist and fellow blogger at Panda's Thumb – with opening another productive and respectful exchange with Casey Luskin, also at ENV. The topic was statistics and modeling, and while I think Reed was right and Casey was mistaken, I also think that the discussion focused on rebuttal without ad hominem.

Let's shoot for that. It won't be easy, and there will be mistakes. But the goal is a worthy one.

Genetics, evolution, and sexual orientation: the gay extinction hypothesis

2011-07-22T01:38:00.002-04:00

Three weeks ago, I went to the Cornerstone Music Festival with my two oldest kids. For the second year, I was an invited speaker in the festival's excellent seminar program. This year, my two series were entitled "Alien Worlds" and "Zombies on Jeopardy" – exploring extreme biology and human nature, respectively. It was fun, if a little too hot for a day or so.

At one point, I was discussing human intelligence and its genetic underpinnings. And I got a loaded question, paraphrased thus: "What happens when you substitute 'sexual orientation' for intelligence? Is homosexuality 'genetic' and if so, what does that mean for Christian views of sexuality?" (The Cornerstone Festival is a Christian music festival, known for embracing music at the 'fringes' while remaining consistent with most mainstream evangelical sensibilities, including a typically evangelical view of homosexuality.) I answered that sexual orientation also has a fairly significant heritable component, meaning that some of the variation in sexual orientation is accounted for by genetics. Then I got a followup question/comment, delivered with intriguing smugness, and paraphrased as follows: "Homosexuality can't be genetic, because homosexuals don't have kids and so the trait will be eliminated from the population." Without going into the complexity of sexual orientation as a biological phenomenon, I will critique this person's claim, since I hear it from Christians with disheartening frequency.

Let's call this claim the gay extinction hypothesis. It can be broken down into two basic assertions:
1. The trait reduces fitness (to near zero) by inhibiting (almost completely) reproduction.
2. Because the trait reduces fitness to near zero, the gene that causes it should have been eliminated from the human population.

Both assertions, when applied to sexual orientation, are incorrect.

Homosexuality does reduce reproduction, at least in men, but not to zero. According to a widely-known study from the 1990's, gay men have about 1/10 the number of children as straight men. That's surely an approximation, and it's a dramatic effect. But it's clearly not the case that homosexuals can't or don't reproduce, and we haven't even explored all the other ways (e.g., reproductive technologies) in which the assertion doesn't work.

Still, let's grant the reproductive disadvantage for the sake of argument and move to the second assertion. Shouldn't we expect any gene that reduces reproduction by 90% to be extremely rare in a population? No, not necessarily. This is the bigger error in the gay extinction hypothesis.

The main error, in my view, is the assumption that the gene that causes the trait is always expressed. (I'm simplifying things to a large extent by referring to one "gene" and one "trait." Correcting these simplifications makes the gay extinction hypothesis even more problematic.) And this would imply that the trait is dominant. (A dominant trait is one in which the individual will express the trait if he or she possesses just one copy of the dominant form of the gene.) But we have no reason to suppose that sexual orientation is a dominant trait. On the contrary, sexual orientation is the kind of complex trait that is expected to be influenced by multiple genes interacting with other factors. While we might postulate that certain gene forms can influence the trait to a much bigger extent than others, we would also expect that gene forms associated with homosexuality can be carried by heterosexual individuals. In other words, it is not the case that a "gay gene" (or genes) is expected to lead to homosexuality whenever it is present in an individual.

To see what I mean by all this, consider the example of cystic fibrosis (CF). This awful disease is as purely genetic as any human disease we know. It shortens life expectancy and, more to the point for our gay extinction hypothesis, leads to significant infertility. (At least 98% of males with CF have no vas deferens and are therefore infertile; the disease also devastates female fertility.) Homosexuality looks tame by comparison. And yet, the gene form that causes CF is distressingly common in certain human populations (namely in those people of European descent). How can this be? How can such a gene form be present in the population at all? Note that this is exactly the challenge posed by the gay extinction hypothesis.

The basic answer is this: a person with just one of those lethal gene forms is unaffected. We call the person a "carrier." Before having an affected child (or a genetic test), such a person would never know that he or she is a carrier. One copy of that wrecked gene form would have no influence on that person's reproductive ability, a.k.a. the person's fitness.* My basic point is one that should be taught in any introductory biology course: there is nothing surprising about the presence of gene forms that lower fitness, and that should be clear with a single visit to the OMIM database. To be sure, we might wonder how they got into our genomes in the first place, and how they are maintained at particular frequencies in our population. We really needn't wonder at their existence, or even their ubiquity.

There is much, much more that could be said about sexual orientation, its genetic components, and evolutionary influences thereupon. But there should be no place for simplistic and bogus claims about the evolutionary impossibility of homosexuality. The extent to which sexual orientation is a "choice" is, I think, an open question. The validity of the gay extinction hypothesis is not.

* Interestingly, one model suggests that possession of one of the wrecked CF gene forms provides a small fitness advantage in the form of protection against tuberculosis. A conceptually identical model has been proposed to explain relatively high rates of human homosexuality.

Mapping fitness: ribozymes, landscapes, and Seattle

2011-05-30T01:09:00.005-04:00

A few months ago, we were looking at the concept of a fitness landscape and how new technologies are creating opportunities for biologists to look in detail at relationships between genetics and fitness. The first post discussed the concepts of a fitness landscapes and adaptive walks, with some focus on the limitations of the metaphor. The second post summarized some recent work on bacterial fitness and mutation rates, with the concept of a fitness landscape as a theme, and the third post reviewed another recent paper, one that described techniques for studying fitness landscapes in detail by linking protein function (which can be screened and/or selected) and genetic information. Here we'll look at yet another approach to the problem, in which the subject of the analysis is not an organism (as in the first paper) or a protein (as in the second paper) but an RNA molecule.

Recall that Fowler et al. set out to design a system in which one can study a protein's function (its “fitness”) as it varies in sequence. The idea is to look at all (or at least nearly all) of the variants of a particular protein to see how well each one works, and then to map this measure of fitness onto the sequence space of the protein. Such a map would be a form of fitness landscape. Fowler and colleagues (henceforth called the UW group) used a previously-known technique (protein display) to link each variant of the protein to its gene sequence, then used next-generation gene-sequencing technology to rapidly determine the gene sequences of millions of variants.

Last October, a group at the Fred Hutchinson Cancer Research Center in Seattle reported the results of a somewhat similar experimental effort. Jason Pitt and Adrian Ferré-D'Amaré co-authored the paper in Science, and their title identified their research objective: “Rapid Construction of Empirical RNA Fitness Landscapes.” The first couple of sentences of the abstract should sound familiar by now. (The genotype is the gene sequence. The phenotype is the function.)

Evolution is an adaptive walk through a hypothetical fitness landscape, which depicts the relationship between genotypes and the fitness of each corresponding phenotype. We constructed an empirical fitness landscape for a catalytic RNA by combining next-generation sequencing, computational analysis, and “serial depletion,” an in vitro selection protocol.

And they identify two major challenges, both of which we have already discussed:

First, even for macromolecules of modest length, the sequence space is vast; a 20-mer RNA or protein has ~10¹² or ~10²⁶ possible sequences, respectively. Second, to characterize the landscape, the phenotypic fitness of each individual genotype needs to be measured, or an indirect measure of fitness needs to be validated.

The authors tackled the challenges using a strategy very similar to that of the UW group: first they designed a functional screen, a way to subject an enormous population of variants to a gauntlet of selection, so that the population would be altered in structure after each round of selection. Think of it as evolution in a test tube. But the UW group had a problem that Pitt and Ferré-D'Amaré didn't have to worry about: the linkage of protein function with the underlying gene sequence. Why the difference? Pitt and Ferré-D'Amaré didn't study protein. They studied RNA – specifically, they analyzed the function of a ribozyme, which is a molecule of RNA that is capable of altering chemical reactions the way protein enzymes do. This means that there was no translation problem for them, since the gene sequence (the base sequence of the RNA) also comprises the structure of the molecule that is being functionally assessed.

So, like the UW group, they took a known molecule and made zillions of variants, through the use of random mutation. Then they assessed the function of the variants by putting them into pools (huge groups) and forcing them to compete with each other. (The competition involved binding to a specific target; the UW group used a similar approach.) Each round of competition (selection) led to the pool being enriched for “functional” molecules. And, importantly, they demonstrated that the binding competition really does select for function; that is, the selection process is enriching for higher “fitness.” After selection, they saw the enrichment that they expected: random sequences (added as a control) were depleted, whereas sequences very similar to the known “normal” sequence were enriched. And, interestingly, lots of other sequences were intermediate between those. Now, how can we graphically depict this? Pitt and Ferré-D'Amaré decided to plot the rate of change in frequency over time for each genotype (i.e., for each variant as identified by sequencing) against a representation of genotype space. The challenge of representing genotype space, or sequence space, is daunting: it will hardly do to put every sequence onto the axis of a graph. So the authors devised a similarity score as an indicator of sequence space, with the known normal sequence as the standard for comparison. And here is their empirical fitness landscape, from Figure 2B:

Each dot is a single sequence. (Actually, each dot is a whole set of sequences that have the same similarity to the reference sequence. In Figures 1C and 1E the authors introduce another dimension to show the spread that each dot represents.) The green dots show enrichment of sequences after 1 minute of competition; the reference sequence is on the far right, such that the steeply-sloping peak on the far right represents sequences that are similar to that reference sequence. As we might expect, the more similar a sequence is to the reference sequence, the more “fit” it is (in general). Fitness is indicated by extent of enrichment, which the authors term “fecundity.” The magenta dots represent not enrichment, but depletion; in a reciprocal experiment, the investigators removed the most fit molecules from the pool by subtracting the best-binding population from the pool. Notice that the depletion landscape is basically a mirror image of the enrichment landscape, as we would expect if the process is truly selecting based on binding activity.

There's a lot of data in that graph. Here's how the authors describe the result:

...the fecundity of any individual sequence provides a metric of its fitness, and we can create an experimental fitness landscape composed of ~10⁷ different RNA genotypes in a single experiment.

And yet the picture is a vast oversimplification of that huge data set. For one thing, the graph provides no specific sequence information even though the sequence of every one of those 10 million variants is known. Pitt and Ferré-D'Amaré write:

The empirical fitness landscape we generated is a high-dimensional object. We visualized it by computing the information content per residue of the master sequence, in essence projecting the landscape onto the ribozyme sequence.

The resulting visualization (in Figure 4A) is a heat map of the actual structure of the catalytic RNA. It's simpler than it seems: each base in the RNA is colored according to information content as indicated on the color scale. More information means more diversity at that position; low information content means that there is little diversity at that position, indicating strong conservation due to functional constraint. The graph seems utterly unlike the topographical landscape that Sewall Wright sketched, but it's a fitness landscape nonetheless, made possible by the creativity of Pitt and Ferré-D'Amaré and by the power of next-generation sequencing.

So, we've looked at three significant articles in the last year or so on fitness landscapes, in which talented scientists explored the relationships between genotype and phenotype, on scales barely imaginable just a decade ago. All three studies were carried out in Seattle, Washington, within just a few miles of Biologic Institute, where the scientists of the intelligent design movement work on questions of the same ilk. If those scientists really want to be taken seriously, if they really seek to understand how structure and function and evolution are related, they'll have to understand fitness landscapes and their experimental applications. Fortunately, they can find some of the world's experts on that very subject right in their own backyard. Whether that amounts to tragic irony or a golden opportunity is a choice for the intelligent design apologists of the Seattle area. May they choose wisely.

[Cross-posted at Panda's Thumb.]
-----
Pitt, J., & Ferre-D'Amare, A. (2010). Rapid Construction of Empirical RNA Fitness Landscapes. Science, 330 (6002), 376-379 DOI: 10.1126/science.1192001

New reading on "junk DNA"

2011-05-21T21:47:00.001-04:00

John Farrell runs an interesting blog at Forbes.com, and he regularly discusses genetics, design, and other topics of interest around here. His latest points to work by Larry Moran and Ryan Gregory, both of whom have debunked some of the "junk DNA" misinformation concocted by design theorists, then looks at some interesting new blogging from one Stanley Rice. It's interesting stuff.

Casey Luskin shows up in the comments. Nothing new there. Run over and check it out.

Alu need to know about parasitic DNA: telling the whole story about Alu elements and "design"

2011-05-15T18:37:00.002-04:00

So, Alu elements are mobile DNA modules that can exert diverse influences on genomes and the organisms harboring them. They can affect genome function in constructive ways, by altering gene expression or supporting chromosome structure. And they can be damaging, even deadly. There are more than a million of them in the human genome, and we don't know what each one does. But, as I explained in the first post in this series, we do know that they can play both helpful and harmful roles, in the same way that other kinds of parasites can be good, bad, or indifferent.

Alu elements and other genome-wide repeats are a big problem for intelligent design (ID) theorists of some stripes. Any ID proponent who claims that genomes are carefully-designed, well-optimized systems must deal with the reality of the enormous numbers of mobile elements in (for example) the human genome. Now, I can think of various ways such an ID theorist might discuss Alu elements. She could propose that all of their characteristics (including their mobility) are part of their design, such that they can bring new design features quickly into being; she could propose that their mobility is a "bug" rather than a "feature," and perhaps speculate on how things went wrong; she could postulate that the damage caused by their expression and their mobility is being misattributed to the genome when it is instead caused by some other external process. (Or she could say, "We're still working on that one.")

Well, sadly, there's one other option for an ID apologist. She could tell only part of the story, claiming that Alu elements are one of God's gifts to humankind, while omitting the most important facts about them.

Consider what we know about Alu elements. Here is a list of what we might call their "positive" attributes:

They can participate in the control of gene expression.
They can participate in structural functions in the genome.
They can be converted into functional genes.

That list should not be ignored or underestimated. Alu elements, like other genome-wide repeats, can be put to use by the genome.

And here is a list of what we might call their "negative" attributes:

They can cause harmful mutations by hopping into the middle of essential genes.
They can destabilize the genome by facilitating damaging large-scale physical interactions.
They can generate toxic RNA molecules that must be controlled by other cellular systems.

And so, while we continue to learn more about these fascinating genomic components, we must keep both of those lists in mind. Recent reviews by experts on the topic have strongly emphasized that balance.

Troublingly, one ID advocate at Reasons To Believe (RTB) has chosen a different approach. Dr. Patricia Fanning is a "visiting scholar" at RTB, and she recently completed a six-part series on Alu elements at RTB's blog. Her articles are clear summaries of the first bullet list. She describes recent work on Alu elements and their structure, and her overviews of their effects on gene expression are very nicely written. I don't much care for her claim that non-function is a key assertion of evolution biologists with respect to genome-wide repeats (that's just not true), but we've been over that before, and I think that her mistake there is an easy one to make. What I find so discouraging and disappointing is her complete omission of the best-known facts about Alu elements.

First, she omits the entire second bullet list. She makes no mention at all of the documented cases of genetic illness caused by Alu-mediated mutation. She makes no mention at all of the well-known phenomenon of damaging recombination caused by genome-wide repeats. And although the macular degeneration article was published in Nature on 17 March, two weeks before she began her series, she makes no mention of this important and influential new finding.

But much worse, she wrote six blog posts, including one that emphasizes the structural components in each Alu element, without mentioning the one thing that is best understood about Alu elements: the fact that they are mobile. It is a systematic omission that makes the entire series profoundly misleading. Once that central fact about Alu elements is re-introduced, many of Dr. Fanning's claims about their form and function and placement are seen to be well-explained by common ancestry and purifying selection.

It would be like advertising "dwelling for sale: one room, windows, air conditioning, heat, furnishings," omitting key facts that would be readily apparent upon seeing the advertised "dwelling." That's how serious I consider Dr. Fanning's omission to be.

Dr. Fanning may be right that Alu elements were designed by God and situated in the human genome according to design or a plan. Her enthusiasm for design-based explanation of genomic structure is legitimate, and I don't wish to criticize her or RTB for expressing that preference. But I am troubled by the choice to tell an incomplete, inaccurate, and misleading story about Alu elements. In my opinion, RTB should consider an addendum to the series, noting that Alu elements are mobile and that their activity is known to be damaging as well as beneficial. This would not require abandoning the pro-design position, and it would be a welcome step toward a more responsible and credible RTB.

Comments are most welcome.

Exploring the protein universe: a response to Doug Axe

2011-05-13T13:30:00.003-04:00

One of the goals of the intelligent design (ID) movement is to show that evolution cannot be random and/or unguided, and one way to demonstrate this is to show that an evolutionary transition is impossibly unlikely without guidance or intervention. Michael Behe has attempted to do this, without success. And Doug Axe, the director of Biologic Institute, is working on a similar problem. Axe's work (most recently with a colleague, Ann Gauger) aims (in part, at least) to show that evolutionary transitions at the level of protein structure and function are so fantastically improbable that they could not have occurred "randomly."

Recently, Axe has been writing on this issue. First, he and Gauger just published some experimental results in the ID journal BIO-Complexity. Second, Axe wrote a blog post at the Biologic site in which he defends his approach against critics like Art Hunt and me. Here are some comments on both.

1. Like my friend Todd Wood, I am encouraged by the fact that Biologic Institute is doing good scientific work and generating publishable data. Axe and Gauger seem to be smart and capable scientists, and they are asking good questions. May their Institute and its scientific work live long and prosper.

2. Axe is primarily interested in the evolution of protein folds. That question is both intensely interesting and important. And difficult.

3. Like Todd, I found the BIO-Complexity paper to be interesting technically but badly flawed in its theoretical approach and conclusions. Specifically, I note what I think any evolutionary biologist would immediately see: that Axe and Gauger did not test an evolutionary hypothesis. Todd explains this very well, but here's the basic problem. To test an evolutionary hypothesis, as I mentioned above, one must study an evolutionary transition. In other words, one must study a change or transition from an ancestral state to a current (or later) state. Joe Thornton's work is a great example: his group examined protein function in a reconstruction of an evolutionary transition. What Axe and Gauger did was study a "transition" that has never been proposed to have happened. They examined a transition from one currently-existing protein to another currently-existing protein. It's as though they analyzed the "transition" from a cat to a dog, when they should have analyzed the transition from ancestral mammals to dogs and/or cats. Their conclusions tell us something about protein structure and function but, crucially, not about the evolution of those proteins.

This does not mean that Axe and Gauger are incorrect in their hypothesis, namely that different proteins are separated by vast evolutionary wastelands that can only be traversed with the help of "design." That may be the case. But the newly-published work in BIO-Complexity gets them no closer to establishing that hypothesis as reasonable or even likely.

4. In his blog post, Axe continues to insist that evidence for rarity of function in the protein universe is evidence for isolation of individual functions in the protein universe. His arguments from probability, which have been used so many times before, simply do not convince me because, as I wrote before: isolation and rarity are not the same thing. I don't happen to think that Axe's data tell us much about the rarity of function (more on this below), but even if I did, I would find that insufficient to undermine the proposal that proteins are linked in a phylogenetic tree the way species are. Again, this is not to say that I know that Axe is wrong. I'm saying that his arguments are unconvincing to me, and that the experiments needed to test his conjecture have yet to be done.

5. Axe claims that I was wrong to describe his 2004 experiments as "whopping mutations on crippled proteins." But that's what they were. He nicely explains why that was the best way to do his experiment, and I think he's right about that. But the fact remains that his analysis doesn't help us understand evolution if his experiment involved a barely-functioning enzyme subjected to mutagenesis that changed ten amino acids at a time. As I think Art Hunt tries to make clear, this doesn't mean that his experiment was stupid or poorly designed. It does mean, clearly in my view, that the experiment tells us little about evolutionary change. And Axe himself seems to agree: he explains that he wasn't attempting to simulate evolution, only to estimate the rarity of protein function in the protein universe (or the protein-fold universe).

6. In my opinion, Axe significantly overstates his findings on the topic of "function." So for example, in both the 2004 paper and the new BIO-Complexity paper, the experiments involve measuring a single function for each enzyme. It seems to me (and I could be wrong) that when the authors see that a particular variant (mutant) of the protein stops performing that one function, they conclude that the protein "has no function." (In the BIO-Complexity paper, it's two proteins and two functions, but the point is the same.) But of course we don't know that, and evolutionary explanations would propose that new functions frequently arise when an enzyme has more than one function (or is broad-based in its function, or is modular in its structure and function). This is why I think that Axe and colleagues can't make any headway in their efforts to understand the evolution of protein function until they focus intentionally on evolutionary transitions. Instead of showing us that mutated proteins no longer do what they used to do, they should invert their reasoning to look like something like this:

Here are the proteins in a postulated evolutionary trajectory. What can we learn about the functions of the intermediates during the transition?

Those would be extensive and demanding experiments, to be sure, but they're the only kinds of experiments that can address the difficult questions that Axe wants to ask. This, by the way, is the same critique I gave Mike Behe in response to his erroneous claims in his most recent book.

7. I'm not so sure that function is as rare as Axe (and others) think. It turns out that completely novel (and foreign) protein sequences can be shown to have function, in living bacterial cells. We may be mistaken in our assumption that islands of function in the protein universe are fantastically rare.

8. Axe and his colleagues do good work, and they're asking important questions. I hope they are in close contact with scientists working on similar questions. There are many strong labs working hard on protein evolution, from various angles, and I'm sure that the scientists at Biologic Institute would profit immensely from regular interactions with the scientific community. (Consider, for example, the authors of a 2010 PLoS ONE paper on "Evolutionary Innovations and the Organization of Protein Functions in Genotype Space.") Perhaps this is happening, and if so, great. But it needs to be emphasized.

So, kudos to the scientists of Biologic Institute for working hard in the lab, and for tackling an important and formidable problem. They haven't shown us anything important about evolution yet, but I hope they keep at it, with a little more careful thought and a lot more input from colleagues.

How do fish adapt to life in hydrogen sulfide?

2011-05-01T16:59:00.002-04:00

To find out, and to read some of the best recent blogging on evolution, visit the new Carnival of Evolution, 35th Edition, at Lab Rat. And go to the official carnival page to learn more about the Carnival of Evolution and perhaps to sign up as a future host.

Alu need to know about parasitic DNA: Alu elements and blindness

2011-04-28T10:53:00.004-04:00

Age-related macular degeneration (AMD) is a leading cause of blindness in humans, and the leading cause of visual impairment during advanced age. The condition comes in two basic forms, the most severe of which is untreatable. Called geographic atrophy (GA), this condition involves the steady destruction of the retinal pigment epithelium, a layer of tissue in the eye that is essential for the health and maintenance of the photoreceptors in the retina. Loss of the pigment epithelium means certain death for the photoreceptors, and that means visual impairment and then blindness for the affected person.

A major publication in Nature last month (Kaneko et al., "DICER1 deficit induces Alu RNA toxicity in age-related macular degeneration," Nature 17 March 2011) now points to one likely cause of AMD, and in the process provides a chilling example of what can happen when the parasitic Alu elements in our genomes (see the previous post for an introduction) are left unrestrained.

The story begins with experiments that showed that a very interesting enzyme, DICER1 (which we'll just call Dicer) was markedly depleted in the retinas of people with the GA form of AMD. Dicer (as its name is meant to convey) specializes in chopping things up. Specifically, it chops up microRNAs, which are small (of course) pieces of RNA that cells make. This is a more precise and important job than it seems: Dicer carefully trims the microRNA into a version that is fully active in the control of gene expression. The resulting pieces of RNA are tiny (21 letters long) but potent, able to substantially reduce the expression of the genes they target. (The phenomenon is called RNA interference and its discovery revolutionized cell biology by giving biologists a simple way to manipulate gene expression.) The authors were probably examining Dicer in the context of AMD because several previous reports had shown that loss of Dicer led to problems in the development of the retina.

So, having found that Dicer was reduced in diseased retinas, the authors showed that this deficit can lead to the disease process. (The mere correlation they started with need not mean that the Dicer problem was causative in any way.) They genetically engineered mice that lacked Dicer in their retinas, and the mice got a nasty GA. And so, after only the first illustration of the article, the biologists had strong evidence that depletion of Dicer could lead to AMD, and that alone is a significant finding.

But why Dicer and AMD? The first and most reasonable hypothesis was that the loss of Dicer led to a failure to trim microRNAs and thus to an overall problem with the microRNA-based gene control system. To test this hypothesis, the authors deleted gene after gene in the microRNA processing system (Dicer is one of several enzymes in that system) and failed to see any retinal problems in any of the resulting mice. The surprising conclusion: the problem that leads to AMD when Dicer is depleted is not a problem with microRNA processing. Dicer's critical role lies elsewhere.

But where? Well, Dicer specializes in chopping up RNA, and specific types of RNA (double-stranded RNA, to be exact). So the authors looked first to see if there was excess double-stranded RNA in retinas of people with GA. Sure enough, there was a big difference between diseased and normal retinas. So they did some nifty biochemistry to grab those excess double-stranded RNAs and identify them. And something jumped out: they were getting Alu RNAs. Click.

Recall that Alu elements are the most abundant type of repeated DNA sequence in the human genome. The human genome contains more than a million of these things, and they account for more than 10% of the genome by themselves. They are mobile genetic elements, meaning that they love to hop from place to place in the genome, and they do this by making RNA copies of themselves. This means that we can expect, at least sometimes, to see Alu RNAs in cells.

Back to the article. What followed was a series of compelling experiments that showed that Alu RNAs are specifically enriched in the doomed cells of the RPE of diseased retinas, and that Dicer does indeed destroy Alu RNAs. The critical next step is an experiment you've probably already identified: to see whether adding Alu RNA to normal cells can kill them. It can. First the authors showed that generic Alu elements can kill the RPE cells when artificially introduced. Then they did something really cool: they took one of the Alu elements that they had fished out of a diseased human retina and introduced it into a mouse retina. This is what they saw (Figure 4c):

On the left are the retinas that got the Alu element. The devastation is most apparent in the red pictures in the bottom row, which show the outlines of happy normal cells on the right and a nightmare of degeneration on the left. Alu elements, when left unchecked, destroy the retinal pigment epithelium and lead to AMD.

But there's one more experiment, a coup de grace, that would really nail this. Here's how it goes. Dicer depletion leads to degeneration and to AMD. Check. It also leads to increased Alu element RNA, and that leads to degeneration and to AMD. Check and check. Now, if those observations add up to a coherent explanation, then we can make the following prediction: the degenerative effect of Dicer depletion should be negated by erasing the runaway Alu elements. The authors took their engineered mouse, which normally gets nasty GA, and then depleted the Alu elements using a simple technique that mimics RNA interference. In short, they induced AMD by depleting Dicer, then attempted to prevent the AMD by killing the Alu elements. And it worked (Figure 5c).

Focus again on the red pictures in the bottom row. Both retinas depicted are lacking Dicer. The trashed retina on the right got no help; the one on the left was rescued by the erasure of the most widely-expressed Alu elements.

The implications of this work are very significant. For one thing, the authors have identified a target for further work aimed at reversing or preventing AMD. But we've also learned something important about Alu elements. As we might expect while considering a parasite that is the most abundant mobile genetic element in the human genome, Alu elements do not tend to have our best interests in mind, and thus their activity must be regulated, and even opposed. We already knew that they can wreak havoc by jumping indiscriminately or by destabilizing genome structure; now Kaneko et al. have shown us another dark side of the Alu world. They write:

This also is, to our knowledge, the first example of how Alu could cause a human disease via direct RNA cytotoxicity rather than by inducing chromosomal DNA rearrangements or insertional mutagenesis through retrotransposition, which have been implicated in diseases such as α-thalassaemia, colon cancer, hypercholesterolemia, and neurofibromatosis.

And ominously, they point out that it's possible that other experiments (and diseases) in which Dicer is lost or depleted may be explained by Alu toxicity rather than by problems in microRNA processing. Until now, biologists didn't know just how precious Dicer was.

In the next post I'll conclude by discussing these findings in light of various creationist claims on the topic, including an ongoing series at Reasons To Believe.

----
Kaneko, H., Dridi, S., Tarallo, V., Gelfand, B., Fowler, B., Cho, W., Kleinman, M., Ponicsan, S., Hauswirth, W., Chiodo, V., Karikó, K., Yoo, J., Lee, D., Hadziahmetovic, M., Song, Y., Misra, S., Chaudhuri, G., Buaas, F., Braun, R., Hinton, D., Zhang, Q., Grossniklaus, H., Provis, J., Madigan, M., Milam, A., Justice, N., Albuquerque, R., Blandford, A., Bogdanovich, S., Hirano, Y., Witta, J., Fuchs, E., Littman, D., Ambati, B., Rudin, C., Chong, M., Provost, P., Kugel, J., Goodrich, J., Dunaief, J., Baffi, J., & Ambati, J. (2011). DICER1 deficit induces Alu RNA toxicity in age-related macular degeneration. Nature, 471 (7338), 325-330 DOI: 10.1038/nature09830

Alu need to know about parasitic DNA: Introduction to Alu elements

2011-04-23T21:21:00.005-04:00

Defenders of intelligent design theory often dwell on the topic of "junk DNA," which has been molded into a masterpiece of folk science. The ID approach to "junk DNA" involves a fictional story about "Darwinism" discouraging its study, and a contorted and simplistic picture of a "debate" about whether "junk DNA" has "function." The fictional story is ubiquitous despite being repeatedly debunked. But the picture of an ongoing "debate" about "function" is harder to sort out. Like most propaganda, that picture contains enough truth to sound plausible. (Browse my "Junk DNA" posts, and work by Ryan Gregory and Larry Moran, for more information on errors and folk science associated with these topics.)

There is, in fact, some scientific disagreement about functions of various elements in genomes, but it's not the crude standoff that ID apologists depict, and it has very little to do with "Darwinism." The debate, if we must call it that, is about at least two matters: 1) the extent to which certain genomic elements contribute to normal function or development of organisms; and 2) the means by which we might determine this. The debate is not about whether non-coding DNA can have function, or even about whether some segments of non-coding DNA do have function. That debate was invented by anti-evolution propagandists.

Now, one thing that is often overlooked in discussions of non-coding DNA is the fact that we know quite a bit about most of it. In other words, it's not the case that scientists look at the human genome and say, "Oh dear, what is all that extra DNA?" Instead, they look at the human genome and say, "Wow, look at all those mobile elements." While this is not to say that there aren't a lot of things in genomes that we don't yet understand, it's important to note that a substantial fraction of the human genome is made up of things we understand pretty well: pieces of DNA that, virus-like, are capable of copying themselves and/or moving to new locations in the genome. More than 40% of the human genome is composed of these mobile elements.

Before we talk about "functions" of these elements, we should face the magnitude of their presence. Also known collectively as genome-wide repeats, they fall into four categories: SINEs, LINEs, LTR elements, and DNA transposons. Consider the SINEs (short interspersed nuclear elements), just one of the four families of mobile elements. Together, SINEs make up a staggering 13% of the human genome. In raw numbers, this is 420 megabases out of 3.2 gigabases of DNA sequence in the human genome. (A base is one "letter" in the genetic code.) Those 420 million letters of code are accounted for by about 1.6 million individual elements. And 1.1 million of those SINEs are of a very interesting type: they are Alu elements.

Alu elements are the most abundant mobile genetic elements in the human genome. They are primate-specific, meaning that they are only found in monkeys and apes and their close relatives. It would take a separate post to fully discuss their characteristics and theories regarding their origins, and I'll come back to that soon. For now, let's start here: an Alu element is a piece of DNA that resembles a virus in that it is mobile and relies on the cell's machinery for its activity. Alu elements are retroelements, meaning that they first copy themselves into RNA in order to "jump" elsewhere in the genome.

Now, how do we know this about Alu elements? Surely we haven't examined each of the 1.1 million Alu elements in the human genome, much less the zillions of them in other primates. No, but here are the two main sources of evidence that Alu elements are mobile genetic elements.

1. We've seen them jump, and we know a bit about how they do it. Simply put, Alu elements are known to move, and they move by using known mechanisms. In fact, biologists have estimated the likelihood that a new "jumping" event will occur in a newly-conceived human embryo to be about 5%, meaning that roughly one in twenty persons are born with a new Alu element insertion somewhere in her/his genome. The process by which Alu elements move is very similar to the processes used by other mobile elements.

2. They're highly conserved, meaning that one Alu element looks a whole lot like all the others. (There are five or six subtypes of Alu elements; the similarity is even more pronounced within those groups.) So we're not talking about a vague category of things that look sort of like a jumping gene. We're talking about a family of DNA elements with very specific features. Remember that they're also called "repeats," because even in the early days of genomic analysis (before we had the actual sequences of whole genomes) biologists knew that huge stretches of the human genome were made of chunks of DNA that were highly similar -- often identical -- and repeated over and over and over.

Taken together, then, in the Alu elements we have a huge family of closely-related DNA elements with structural features that are known to mediate movement within the genome. If all that sounds a little too technical, don't worry; what matters is that you grasp the basic notion (human genomes harbor so-called "jumping genes" that can move about within those genomes) and its magnitude (Alu elements are just one type of mobile element, and they alone make up more than 10% of the human genome).

So, do these things have a "function"? That's a tricky question. (Just the kind of question preferred by ID propagandists.) Alu elements and their kin are currently viewed by biologists as parasites, and if you know anything about parasitism then you know it's a bit too simplistic to ask whether a parasite is "good" or "bad" for its host. In many parasitic relationships, the host organism incurs some cost (or risk) by hosting the parasite, but also enjoys some benefit. You might think of the bacteria in your gut in this way; they're good to have around, but they can cause problems if they get out of bounds. It's like they've been domesticated: they're still potentially harmful, but if kept in control they're useful. Or at least, if they obey the rules, they're not too big a burden.

Considered in this way, Alu elements make sense. They can be useful. For one thing, their sheer bulk enlarges the genome, and genome size affects things like cell size. Alu elements can introduce new genetic diversity into a species, far more quickly than other kinds of mutation, and so they can be drivers of evolutionary change, by altering the genetic landscape literally overnight. Some Alu elements are known to influence the expression of genes (i.e., when those genes are on or off). And, notably, Alu elements can sometimes be converted into useful genes. In other words, like conventional biological parasites, they can be good to have around.

But, like conventional biological parasites, they can be dangerous. Alu elements can destroy critical genes by hopping into them. (The Alu element is about 300 DNA "letters" in length, and if those letters are added to the protein-coding part of a gene, the nearly-certain outcome is the conversion of the gene to gibberish.) Such events are known to underlie instances of devastating human genetic diseases. Because the Alu elements are so numerous and because the various types all look almost completely alike, they foster damaging interactions between parts of the genome and thereby facilitate large-scale genetic damage. And so humans (and other animals) pay a significant price for hosting mobile genetic elements, and the risks are exactly what we would expect from "jumping genes" that move without regard to the potential harm their relocations can cause.

Those facts alone should lead us to predict that humans (and other animals) would employ defenses against these parasites, if not to eradicate them then at least to keep them from overrunning the place. And these facts should make readers of most ID writing on this topic think a lot differently about ID claims. For one thing, we should be suspicious of any argument tackling the strawman of whether or not Alu elements are "functional elements" as opposed to "junk DNA."

But look again at the risks that I discussed. I mentioned two big ones: the risk that an Alu element would hop into a gene and thereby damage the gene, and the related risk that Alu elements would cause other structural damage to genomes. But can these elements be damaging in other ways? If they function as parasites, and if they insist on making RNA in hopes of hopping to another genomic neighborhood, mightn't they pose risks more immediate than mutation? We now know that they do, and we know a little about how humans and other mammals fight back. That's for Part II.

34th Carnival of Evolution

2011-04-01T23:14:00.008-04:00

Welcome to the 34th Edition (1 April 2011) of the Carnival of Evolution, and welcome to Quintessence of Dust. It's nice to be hosting this fine carnival, and to see that it's still going strong.

I've organized the carnival under some chapter and section headings that I got from some old Victorian's magnum opus, but I think you'll find the topics require no further creative embellishment.

On the Imperfection of the Geological Record

Before you read anything else, check out this major new fossil find, discussed over at Panda's Thumb. This one's really appropriate for today. Tomorrow, not so much.

Miscellaneous Objections to the Theory of Natural Selection

Larry Moran over at Sandwalk has begun a series explaining, again, why creationist writings on so-called "junk DNA" are misleading (deliberately so, in my view). Larry takes apart a recent "inverview" of Jonathan Wells on the subject of Wells' forthcoming book, which Larry calls "the upcoming train wreck." Train wrecks, of course, are often caused by human error. This is not the case with Wells' works of propaganda. Ick. Anyway, keep tabs on Larry's series, called "Junk & Jonathan."

Another favorite subject among ID creationists is the vague notion of "Complex Specified Information." Vacuous drivel? Largely yes, but I think that in some sense it's a useful concept. Joe Felsenstein, in a guest post ("Uncommon Dissent") at Panda's Thumb, explains why he thinks so, too. Further antidotes to creationism can be found at Playing Chess with Pigeons; Troy answers the silly question "Does being the 'fittest' mean eliminating the less fit?" The cartoon is pretty funny.

At BioLogos, you'll find some seriously smart scientists (who happen to be Christians) working hard to dispel ID myths. Dennis Venema is a friend of mine and a superb science writer, and he's in the midst of a series on "Evolution and the Origin of Biological Information." In "Part 2: E. Coli vs. ID" he discusses Richard Lenski's work in the context of ID misinformation. Go.

Instinct

At Epiphenom, Tom Rees explores kin selection and hypotheses concerning altruism and even religious behavior. He discusses a recent article on "Sixteen common misconceptions about the evolution of cooperation in humans" in a post called "The evolution of nice." It was kind of him to do that for us. At Pleiotropy, Bjørn is unimpressed by a recent paper on a similar subject. Writing on "The trouble over inclusive fitness theory and eusociality," he expresses agnosticism regarding kin selection in the evolution of eusociality. I share his discomfort with projects that mix science and "theology," even though I think I qualify as an "infamous accommodationist."

Organs of Extreme Perfection

The Department of Systems Biology at Harvard Medical School runs a cool blog called "It Takes 30." (And yes, they explain the strange name.) One great recent post tackled the interesting question of how chemotaxis might have evolved in bacteria. Chemotaxis is simply movement toward a chemical attractant, sort of like following the smell of brownies. Because it involves two main components – sensing the attractant and moving through space – it presents one of those chicken-and-egg problems that creationists love to crow about. The post summarizes a fascinating new paper which concludes that motility arose before chemical attraction, and suggests that this early motility would have been random (i.e., undirected) but strongly beneficial. An excellent post, earning bonus points for a sly title: "The random walk of evolution."

Where, oh where have all the stromatolites gone? I won't tell you what they are, or where they've gone, or even why you should care. Greg Laden takes care of all of that, with some excellent photos. Key words: biofilms and snails. Seriously, I won't tell you anything more. Read about Lester Park Stromatolites at Greg's place. And at Denim and Tweed, Jeremy discusses recent work showing the cascading effects of bird near-extinction on plant communities: think pollination and co-evolution. Jeremy's blog is full of similarly great stuff.

Organs of Small Importance

Yoda said "size matters not," but I'm told that women often claim otherwise. If you think I'm being witty as I introduce a post on, say, Devonian megafauna, guess again. Human penises are the subject of an extensively-documented post at Observations of a Nerd, where Christie surrenders to her "inner scientist" and asks the one burning question we all secretly wish someone would answer: "does penis size even matter from an evolutionary perspective?!" The map that inspired her quest is fascinating. It's a great post, called (of course) "Is Bigger Really Better?"

Variation under Nature

Okay, so what about brains? Does size matter in that department? Zen Faulkes at NeuroDojo reviews (critically) some recent studies of the importance of brain size in behavioral complexity, studies done in orb-weaver spiders, which make cool webs but vary in size over three orders of magnitude. "Oh, what a tangled web we weave"... because of small brains?"

Don't like spiders? Ah, then maybe you'll prefer Carl Zimmer's interesting tale about a certain kind of animal with saliva that has yielded biomedical bounty for decades. The animals are pit vipers, their saliva is venom, and the stuff you can get from that venom will amaze you. Gene duplication, alternative splicing, anticoagulants... oh yeah. Collecting the samples looks like fun, too. It's "How a pit viper saved millions of lives" and it's only at The Loom. If you'd prefer worms, go to Genome Engineering to read a little about "The genetics of wriggliness." But come on. Don't be a wuss. Check out the vipers, too.

Struggle for Existence

Now here's a remarkable fact: "of the hundreds of disease-causing microbes or pathogens that we know of, none are archaea." That's a quote from "Why are there no (or almost no) disease-causing Archaea?" The post at Byte Size Biology discusses a recent paper that offers a potential explanation. A nice bonus: one of the authors of the appears in the comments.

Natural Selection; or the Survival of the Fittest

"Everything's just the same, unless it isn't." Is evolution about change? See The Mermaid's Tale for a thoughtful discussion of concepts of stasis, change and drift.

Michael Scott Long provides a brief summary of some new work that suggests that dispersal of individual organisms as a significant evolutionary influence, alongside natural selection (for example), in his post on Cumulative Spatial Sorting.

Recapitulation and Conclusion

You HAVE to check out some of the delightful posters that have been created for Darwin Day at the Essig Museum of Entomology at UC Berkeley. They're included in "Evilution" at The Dispersal of Darwin, and created by Ainsley Seago, a "beetle biologist" according to Michael. My favorite is over on the right.

Thanks for attending our carnival. Next month's edition will appear on or about 1 May at Lab Rat. See you there, and drop us a line if you'd like to host in June.

Mapping fitness: protein display, fitness, and Seattle

2011-02-05T23:21:00.004-05:00

A couple of months ago we started looking at the concept of fitness landscapes and at some new papers that have significantly expanded our knowledge of the maps of these hypothetical spaces. Recall that a fitness landscape, basically speaking, is a representation of the relative fitness of a biological entity, mapped with respect to some measure of genetic change or diversity. The entity in question could be a protein or an organism or a population, mapped onto specific genetic sequences (a DNA or protein sequence) or onto genetic makeup of whole organisms. The purpose of the map is to depict the effects of genetic variation on fitness.

Suppose we want to examine the fitness landscape represented by the structure of a single protein. Our map would show the fitness of the protein (its function, measured somehow) and how fitness is affected by variations in the structure of the protein (its sequence, varied somehow). It's hard enough to explain or read such a map. Even more daunting is the task of creating a detailed map of such a widely-varying space. Two particular sets of challenges come to mind.

1. To make any kind of map at all, we need to match the identity of each of the variants with its function.

2. To create a detailed map, we need to examine many thousands -- or millions -- of variants. This means we need to be able to make thousands of variants of the protein.

So let's take the second challenge first: how do we make a zillion variants of a protein? Well, we can introduce mutations, randomly, into the gene sequence for the protein and use huge collections of those random variants in our analysis. The collection is called a library, and believe it or not, the creation of the library isn't our biggest challenge. Because if the library only contains gene sequences, then it's no use in an experiment on protein fitness. We need our library of gene sequences to be translated into a library of proteins. How are we going to do that? And remember the first challenge: we need to be able to identify each variant. So even if we can get our gene sequences made into protein, how will we be able to identify the sequences after we've mapped the fitness of all the variants?

Or, in simpler terms, here's the problem. It's pretty straightforward to make a library of DNA sequences. And it's pretty straightforward to study the function of a protein. (Note to hard-working molecular biologists and protein biochemists: no, I'm not saying it's easy.) The problem is getting the two together so that we can study the function of the proteins with biochemistry but then identify the interesting variants using the powerful tools of molecular biology. What we need is a bridge between the two.

The bridge most commonly used in such experiments is a technique called protein display. There are a few different ways to do it, but the basic idea is that the DNA sequence is encapsulated so that it remains linked to the protein it creates. One cool way to do this is to hijack a virus and force it to make itself using your library. The virus will use a DNA sequence from your library, dutifully make the protein that is encoded by that DNA sequence, and displaying that protein on its surface. There's our bridge: a virus, with the protein on the surface ready for analysis and the DNA sequence stored inside the same virus. Brilliant, don't you think?

Yes, but there's one more problem to be solved. We said we want to do this millions of times. That means we have to grab the viruses of interest, get the DNA out of them, and read off the sequence of that DNA. (That's how we can identify the nature of the variation.) Millions of times. Methods of protein display provided the bridge, but until very recently a crippling bottleneck remained: the sequencing of the DNA was too time-consuming to allow the identification of more than a few thousand variants at a time.

That was then. This is now: the era of next-generation sequencing, in which DNA sequences can be read at blinding speed and at moderate cost. (A currently popular technology is Illumina sequencing.) These techniques have given us unprecedented capacity to decode entire genomes and to assess genetic variation on genome-wide scales. A few months ago, the same methods were used to eliminate that last bottleneck in the use of protein display, demonstrating how a protein fitness map can be generated simply and at very high resolution. The article is "High-resolution mapping of protein sequence-function relationships" (doi) from Nature Methods, by Douglas Fowler and colleagues in Stan Fields' lab at the University of Washington.

The experiment focused on one interesting segment of one protein. The segment is called a WW domain and it's an interesting building block which is found in various proteins and which mediates interactions between different proteins. (A sort of docking site.) The authors chose the WW domain both for its interesting functions and because it has been used in protein display experiments of the type they performed. Then they created their tools.

1) They generated a library of more than 600,000 variants of the domain, displayed on the surface of their chosen bridge -- the T7 bacteriophage (a virus that targets bacteria).

2) They designed a means to assess the function of the variants. Because the function of the WW domain is docking, they used docking as their functional criterion, and then devised a straightforward system to detect the strength of the binding of the variants to a typical docking partner. (For the biochemically inclined, they used a simple peptide affinity binding assay on beads.)

Then the key experimental step: the authors used that system to select the variants that can still bind. In other words, they selected the functional variants. The selection step was moderate in strength, and the idea is that variants that bind really well will be enriched at the expense of variants that bind less well. Variants that don't bind at all will be removed from the library.

They repeated the selection step six times in succession. So, the original library was subjected to selection, generating a new library, which was subjected to selection again, and so on, until the experimenters had six new libraries. Why the repetition? It's one of the really smart aspects of the experiment and it has to do with the strength of selection. If selection were quite strong, such that only the strongest-binding variants survive, then the analysis will just yield a few strong-binding variants. That's a simple yes-or-no question, providing no information about the spectrum of binding that can be exhibited by the variants. Instead, the authors tuned the system so that selection is moderate, leading to enrichment but not complete dominance of the stronger-binding variants. Recall that binding represents fitness in this experiment; this means that the authors subjected their population to a moderate level of selection in order to map the fitness of a large number of variants. By repeating the selection, they could watch as some variants gradually increased in frequency. Sounds kind of like evolution, doesn't it?

Finally, the scientists subjected those libraries to Illumina sequencing, thus closing the loop between function and sequence. (In genetic terms, we would say that they closed the loop between phenotype and genotype.) And at that point they were able to draw fitness landscapes of unprecedented resolution, shown in the graphs below. The top graph shows the original library. The height of each peak represents frequency in the library, and the two horizontal axes represent each possible sequence of that WW domain. Notice that the original library is complex and diverse, as indicated by numerous peaks on the graph. The second and third graphs show the library after three and six rounds of selection. Note the change in the number of peaks and in their relative sizes: selection has reduced the complexity of the library, removing variants that are far less fit and altering the relative amounts of the survivors. The first three rounds of selection reduced the library to 1/4 the original size, and after six rounds it was down to 1/6 original size, but still contained almost 100,000 variants.

The bottom graph, then, is a fitness landscape, of a segment of a protein, at very high resolution. More technically, it depicts the raw data (relative amounts of surviving variants) that the authors used to determine relative fitness; to make that assessment, they calculated "enrichment ratios" to account for the fact that the initial library didn't contain equal amounts of each variant. These enrichment data enabled them to calculate the extent to which each point in the sequence is amenable to change, and then to identify the particular changes at those points that led to changes in fitness. Now that's high resolution.

The power of approaches like this should be obvious: disease-related mutations can be identified in candidate genes, and the same approach can be used to map the landscape of resistance to drugs in pathogens or cancer cells. And, of course, evolutionary questions of various kinds are much more tractable when tackled with methods like this. The authors expect the payoff to be immediate:

Because the key ingredients for this approach -- protein display, low-intensity selection and highly accurate, high througput sequencing -- are simple and are becoming widely available, this approach is readily applicable to many in vitro and in vivo questions in which the activity of a protein is known and can be quantitatively assessed.

Now, given these vast opportunities now available to scientists interested in protein evolution, wouldn't you think that design theorists who write on the topic will be eager to get involved in such studies? I sure would, especially since the lab that did this work is within a short drive of the epicenter of intelligent design research, a research insitute headed by a scientist whose professional expertise and interest lies in the analysis of protein sequence-function relationships. As I've repeated throughout this series, there's something strange about a bunch of scientists who want to change the world but who can't be bothered to interact with the rest of the scientific community, a community that in this case is well-represented in active laboratories right down the road. (I'm eager to be proven wrong on this point, by learning that ID scientists have interacted with the Loeb lab or the Fields lab.)

More to the point, there's something tragically ironic about the fact that the ID movement is headquartered in Seattle, inveighing against "Darwinism" while obliviously amidst a world-class gathering of scientists who are busy tackling the very questions that ID claims to value.

(Cross-posted at Panda's Thumb.)
-------
Fowler, D., Araya, C., Fleishman, S., Kellogg, E., Stephany, J., Baker, D. and Fields, S. (2010). High-resolution mapping of protein sequence-function relationships. Nature Methods, 7 (9), 741-746 DOI: 10.1038/nmeth.1492

It's just a stage. A phylotypic stage. Part III: Fish and more

2010-12-18T03:03:00.002-05:00

Given that disputes over the existence and meaning of the phylotypic stage and the hourglass model have simmered in various forms for a century and a half, the remarkable correspondence between the hourglass model and gene expression divergence discovered by Kalinka and Varga and colleagues would be big news all by itself. But amazingly, that issue of Nature included two distinct reports on the underpinnings of the phylotypic stage. The other article involved work in another venerable model system in genetics, the zebrafish.

The report is titled "A phylogenetically based transcriptome age index mirrors ontogenetic divergence patterns" and is co-authored by Tomislav Domazet-Loso and Diethard Tautz. To understand how their work has shed light on the phylotypic stage and the evolution of development, we'll need to look first at an approach to the analysis of evolutionary genetics that these two scientists pioneered: phylostratigraphy.

The authors first described phylostratigraphy in 2007 and have since used the approach to examine genes that cause human genetic disease and cancer. They define it as:

a statistical approach for reconstruction of macroevolutionary trends based on the principle of founder gene formation and punctuated emergence of protein families.

The idea is that every gene has a birthday, a point at which it is first identifiable in evolutionary history. Some genes are ancient, having arisen before there were even complex cells, and others are relative juveniles, having arisen much more recently. Genes present today, then, can (in principle) be assigned an "age." Domazet-Loso and Tautz represent the "age" of a gene by the evolutionary "epoch" in which it appeared, by analogy with the identification of the appearance of biological lineages with stratigraphic epochs in earth's history. So for example, some genes appear with the development of true animals (metazoa), and so these genes are assigned to that "stratum" of biological history. In fact, the authors call each epoch a 'phylostratum' to reinforce that metaphor.

So how does this work? To do phylostratigraphic analysis, you need two major sets of tools. First, you need a pretty solid phylogeny, or family tree, of your organism(s) of interest. Second, you need complete or nearly-complete genome sequences of the organism of interest and of organisms that can represent the major branch points (or nodes) in the family tree. The procedure from there seems clear enough: using a well-known alignment program, you search through the family tree for each of the genes in your organism of interest, to see where it is first recognizable in the phylogeny. That point is the phylostratum from which that gene arises. With that data, you could look at the contributions of various phylostrata to various body parts or processes. Or conversely, you could look at the relative age of the sets of genes associated with those body parts or processes. Or you could look at the relative age of the sets of genes associated with different stages of development. And that's what Domazet-Loso and Tautz did in their Nature paper on the hourglass model.

Specifically, the authors took their phylostratigraphic data and merged it with expression data at various stages of zebrafish development; they called the resulting parameter the transcriptome age index (TAI). Basically, they calculated a relative age of the genes that are turned on at each stage of development, corrected for the extent to which particular genes are being used at those stages. Then they mapped the TAI onto the timeline of zebrafish development. And this is what they saw (click to enlarge):

Does that look familiar? Like, say, half an hourglass? In the earliest stages of development, active genes are young-ish, as they are in the juvenile and the adult. In between, the genes that are active are older -- a lot older. And the low point, where genes are oldest? It's the end of segmentation and the beginning of the pharyngula stage. That's the stage that is considered the phylotypic stage in vertebrates. (What this has to do with godless liberalism, I have no idea.) And so we see that hourglass again, this time traced out by the evolutionary age of the genes that are active during the phylotypic stage.

As you look at the graph, you might notice some other interesting periods in the life of a fish. There's a prominent peak of gene youthfulness at 6 hours of development; this corresponds to gastrulation, that wonderful time in your life when you established yourself as a three-layered animal. That peak is due to the activation of a lot of animal-specific genes, namely those that date to the metazoan phylostratum. This includes genes that control cell-cell interactions, certainly a hallmark of animal-building. Those might seem like incredibly basic functions, but they're relatively young compared to even more basic cellular processes, and the genes that control those processes are the ones that predominate during the later phylotypic stage. (The authors showed, in fact, that extremely ancient genes are active uniformly throughout development, whereas the younger gene sets display the hourglass pattern: high-low-high.)

And notice that gene youthfulness declines during aging (after adulthood). Now why would that be? The authors propose that the most recent innovations (facilitated by relatively young genes) are likely to have resulted from adaptation, and so:

The fact that ageing animals revert to older transcriptomes is in line with the notion that animals beyond the reproductive age are not 'visible' to natural selection and can therefore not be subject to specific adaptations any more.

There's a lot more: the study found differences between males and females (look at the dotted lines in the figure), for example. But they also extended their analysis to other animals with known genomes: fruit fly, roundworm and mosquito. In every case they saw the same pattern: young-old-young. Their fly graph displays a pattern strikingly similar to that in the fish, and nicely dovetails with the distinct analysis done by Pavel Tomancak's group (click to enlarge):

Look at the low point, where the genes are the oldest. It's the germband elongation stage -- the recognized phylotypic stage for insects, and the same point singled out in the fly paper. Remarkable.

So to summarize, the two papers, reported separately but simultaneously, strongly support the hourglass model of development, in which embryos are seen to converge on an evolutionarily-ancient form, after diverse beginnings and followed by radical divergence into the wonderful variety of animals seen today and in the past. Domazet-Loso and Tautz explain how these new results make sense of the hourglass:

These consistent overall patterns across phyla, as well as the detailed analysis within zebrafish, suggest that there is a link between evolutionary innovations and the emergence of novel genes. Adaptations are expected to occur primarily in response to altered ecological conditions. Juvenile and adults interact much more with ecological factors than embryos, which may even be a cause for fast postzygotic isolation. Similarly, the zygote may also react to environmental constraints, for example, via the amount of yolk provided in the egg. In contrast, mid-embryonic stages around the phylotypic phase are normally not in direct contact with the environment and are therefore less likely to be subject to ecological adaptations and evolutionary change.

And as they note, Darwin himself made this connection, reflecting on von Baer's earlier observations. Ideas, like genes, can have a long and productive history.

[Cross-posted at Panda's Thumb.]
----------
Domazet-Lošo, T., & Tautz, D. (2010). A phylogenetically based transcriptome age index mirrors ontogenetic divergence patterns. Nature, 468 (7325), 815-818. DOI: 10.1038/nature09632