17 May 2008

How the bat got its wing

Nothing can be more hopeless than to attempt to explain this similarity of pattern in members of the same class, by utility or by the doctrine of final causes. The hopelessness of the attempt has been expressly admitted by Owen in his most interesting work on the 'Nature of Limbs.' On the ordinary view of the independent creation of each being, we can only say that so it is;—that it has so pleased the Creator to construct each animal and plant.

The explanation is manifest on the theory of the natural selection of successive slight modifications,—each modification being profitable in some way to the modified form, but often affecting by correlation of growth other parts of the organisation. In changes of this nature, there will be little or no tendency to modify the original pattern, or to transpose parts. The bones of a limb might be shortened and widened to any extent, and become gradually enveloped in thick membrane, so as to serve as a fin; or a webbed foot might have all its bones, or certain bones, lengthened to any extent, and the membrane connecting them increased to any extent, so as to serve as a wing: yet in all this great amount of modification there will be no tendency to alter the framework of bones or the relative connexion of the several parts.

– from On the Origin of Species, 1st Edition (1859), Charles Darwin
The wing of a bat is an amazing thing. It's not just a wing; it's clearly a modified mammalian limb. A bat looks like a lot like a rodent with really long, webbed fingers on elongated arms.

Image from Animal Diversity Web at the University of Michigan.

Recent genetic analyses have yielded a fairly solid outline of the evolutionary history of bats, which have left a somewhat poor fossil record in which the earliest fossil bats look pretty much like modern bats. ResearchBlogging.orgIt seems that bats arose relatively quickly during evolution, acquiring their distinctive feature – powered flight – in a few million years. No transitional forms have yet been found, which is a shame, because this particular evolutionary transition is the kind that is otherwise reasonably approachable for the detailed study of how changes in form come about.

The fossils can't yet show us how paws gave rise to wings, but that doesn't mean we can't test specific hypotheses regarding the paths that evolution could have taken. In fact, developmental biologists have enormous resources that can be brought to bear on the question, by virtue of decades of research on the development and genetics of the wingless terrestrial bat better known as the mouse. A few months ago, an interesting new report described one kind of genetic change that can lead to bat-like bodies, and the findings put some new wind in the sails of evo-devo.

Two of the more remarkable aspects of bat wing structure are the forelimbs and the forelimb digits, what humans would call the arms and the fingers. Both are dramatically elongated in the adult animal, despite getting off to a very typical start during early development. Check it out: in the picture below, bat and mouse limbs are compared with the image scaled so that body lengths are comparable.

Image from Figure 1 of Cretekos et al., cited below.

Developmental biologists have some pretty good ideas about how this might arise physiologically: certain growth factors (called bone morphogenetic proteins, or BMPs) are known to control limb growth, and some BMPs seem to be turned up in developing bat fingers. But the genetic mechanisms underlying these processes are unknown.

Enter Chris Cretekos and colleagues, then working in a group in Houston headed by Richard Behringer. They set out to examine the genetic underpinnings of the elongation of the forelimbs (arms) of bats, using the formidable tools of mouse developmental genetics. And, clearly, they also sought to directly test one of the central hypotheses of evo-devo: that changes in regulatory DNA sequences (as opposed to changes within the genes themselves) are a potent source of variation in evolution. Consider the beginning of their abstract:
Natural selection acts on variation within populations, resulting in modified organ morphology, physiology, and ultimately the formation of new species. Although variation in orthologous proteins can contribute to these modifications, differences in DNA sequences regulating gene expression may be a primary source of variation.

– From C.J. Cretekos et al., "Regulatory divergence modifies limb length between mammals, Genes & Development 22:141-151, 15 Jan. 2008
Besides their expertise in mouse genetics, the authors brought two major assets to their study: 1) they had already carefully mapped the development of the short-tailed fruit bat (Carollia perspicillata, "our model Chiropteran"); and 2) they knew a lot about the genetic control of limb length in other mammals. In particular, they knew that the protein Prx1 is known to influence limb elongation, by controlling the expression of other genes. So they hypothesized that changes in the activity or level of Prx1 might underlie the difference in limb length between bats and mice, and they were well-equipped to do the experiments.

First, the authors examined the Prx1 gene in the two species, and found that the overall structure of the gene is very similar in both mice and bats, and that the actual coding sequences of the two genes are almost completely identical. (Aligning the coding sequences showed that more than 99% of the amino acids are the same in both species.) In other words, the part of the Prx1 gene that codes for protein is almost certainly not a source of variation between mice and bats. This could mean that Prx1 doesn't have anything to do with the difference between forelimb length in these two species, or it could mean the the difference is generated, at least in part, by variation in the regulation of the gene. Cretekos et al. postulated that altered Prx1 regulation might be involved, and designed a cool experiment to address this possibility.

They already knew that the Prx1 gene in mice contains known regulatory elements in particular locations within the gene. (Such elements are often located in the DNA sequences that precede the coding region.) When they looked at the bat gene, they found similar elements in the same location, but these elements showed some intriguing variation: when the two regions were aligned, they shared only 67% identity, meaning that a third of the DNA bases were different in mouse and bat. They did some nifty cell biology to show that this region did function as a regulator of the expression of Prx1, then did something that biologists could only dream about before the genomic era: they altered the mouse genome by replacing the mouse regulatory region with the corresponding region from the bat genome. In other words, they gave a mouse a piece of a bat's genome, without actually changing the coding sequence of any gene.

The result was dramatic, although it won't sound that way at first. The mice with the bat DNA displayed forelimbs that were 6% longer than normal. Why is this a dramatic result? Well, first of all, think about a 6% change in a major structural attribute. If adult males in a certain country average 5'10" in height, a 6% increase would mean an increase of more than 4 inches. But more importantly, the Prx1 gene is known to account for about 12% of forelimb length – mice that lack the gene altogether show a 12% reduction in forelimb length. That 6% change reflects a huge change in Prx1 activity, a change that was completely due to alterations in regulatory DNA sequences without any change in coding sequence.

If that's not impressive enough, the authors went on to examine the importance of this regulatory region in mice, by deleting it altogether. The result was very surprising, but very interesting: limb length in mice was completely unaffected by the loss of this chunk of regulatory DNA. (The region we're discussing is 1000 bases in length.) This means that the Prx1 gene of both bats and mice contains a regulatory region that is completely dispensable for normal development but that can be altered to generate significant changes in limb length, which points to significant evolutionary potential in genetic regions that seem unimportant. Here's how the authors say it:
Maintenance of redundant enhancers for essential developmental control genes would allow changes in expression pattern to arise from mutations that alter regulatory activity while preserving the required gene function.
So, why is this significant? Here are two aspects of the story that are worth highlighting.

1. The results provide strong (and rare) experimental support for the ideas of the evo-devo school. The currently-heated debate over the merits of evo-devo is focused on the central evo-devo claim that morphological evolution (i.e., evolutionary changes in form) is driven to a large extent by changes in the regulation of gene expression, and less so by changes in the structures of the proteins that are encoded. To simplify, evo-devo postulates that significant evolutionary change – like that discussed here – is more likely a result of the varied use of a protein toolkit than a result of modification of the toolkit itself. Cretekos et al. have presented a case in point, and one that is considered outstanding in that it documents a morphological gain; many previous examples showed only losses.

2. The results provide a sharp picture of what Darwin's vision of "successive slight modifications" means in terms of developmental biology. In this case, the modifications (of a redundant regulatory region) can yield significant anatomical remodeling without altering protein structure at all.

The article was a notable advance for evo-devo and for evolutionary science, but soon there will surely be many others like it. Desperate or ignorant creationists will always find a way to avoid facing the explanatory power of common descent, but scientists are just plugging away, and for every blog post by a creationist ignoramus, there are 30 unheralded publications in the biological literature that advance our understanding of common descent and the mechanisms that generate biological novelty. And they're fun to read.
Article(s) discussed in this post:


Karen James said...

What an excellent post. You not only gave a good overview of the paper but explained how it related to the evo-creo debate. As a big fan of evo-devo myself (and author of one of those unheralded papers in the literature you mentioned), I thank you for writing this post.

AIGBusted said...

Very Neat! I do have one thing I'd like to add to the article: It states that "The fossils can't yet show us how paws gave rise to wings". This is true, there is not a sequence of fossils that show this transition (Bats are too fragile to fossilize well). However, there was a fossil reported a few months ago which definitely represented a transitional stage between tree climbers and bats.

I wrote about this on my blog:



hey there, i thought your post was great, but i just don't really like the article you chose.

i went over it in journal club and thought it was a lemon. if you read closely (p146, end of the second left para), they couldn't find any significant limb length difference in the BatE/BatE adults. a phenotype that is only expressed prenatally doesn't seem like one that selection might act on... but then again who says the path to wings can't be paved with neutral mutations?

i personally think that it's a bit hasty to conclude that the regulatory sequence the authors found is responsible for any significant part of the morphological difference between birds and bats.

the authors begin their abstract referring to natural selection's action on variation within a population. but since their sequence maintains creates no significant variation, how can they say that cis-regulatory regions like theirs are responsible for the greater part of heritable variation?

i mean, sure, they're probably right. but i don't think that their evidence is strong enough to support their sweeping conclusions.

all that aside, i think it's an awesome proof technique and a stunning example of what molecular techniques can do these days.

i agree with you entirely that for every sheet of creationist dribble there are at least 30 unheralded scientific studies advancing our understanding of how evolution works. this one just isn't my favorite of the bunch.

Anonymous said...

I am alarmed by this quote "Maintenance of redundant enhancers for essential developmental control genes would allow changes in expression pattern to arise ..." As I read it, it implies some sort of planning (genome manager A: "maybe we should chuck this junk"; genome manager B "nah, let's keep it; it might be useful someday") Evolution has no foresight!

Anonymous said...

Redundancy in this case is a technical term. The regulatory element is redundant in the mouse because removing it has no discernible effect. That does not mean it has no function, only that its function is dispensable, perhaps the slack taken up by other mechanisms.

Many whole gene knockouts have no discernible effect on their own. The point is that such elements are the most plastic bits that can be played with when the variations are being selected against.

Alabastah said...

The maintentance of these redundant enhancers can be interpreted as the diversity reservoir of a particular individual. All organims change themselves in responce to pressures of all kinds, some are capable of more drastic changes then others. While we can't breathe methane or simply not breath like some little microscopic beasties, one of the ways we can respond is to have these backup sections of useful but not essentail information. The backup is there in case something goes wrong, like a toxin or radiation damage, or as a reserve bit of diversity to throw at the coming selective pressure. I am imagining some kind of analogue to the inate immune responce. But in this case, the immunity is to extinction, heh.
This same sort of redundancy can be seen in kinases, where elimination of many kinase enzymes from an organism can show no phenotype. The concept the Evo-Devos are selling is the molecular tool box; a powerful evolutionary tool that would certainly have significant benefits for a species who had one. No foresight is needed, just natural selection!

Sarah S. said...

John: That's a tricky case to make, because it relies on secondary selection. If a trait (redundant regulatory areas) is only useful when a mutation occurs somewhere else (e.g., by the radiation damage you suggest), then its selective advantage will be apparently only rarely. If it's only useful in rare cases, then it would have to be *very* useful to avoid being lost by drift. This scenario is possible, it just requires extreme parameters.

A simpler explanation is that the redundant bits are harmless, rather than useful. They're not selected for, they just aren't selected out, and because these kinds of duplications happen fairly often, some of them become common in the population by chance (drift). The vast quantity of apparently useless-but-harmless noncoding DNA in mammalian genomes supports the idea that this is the process going on. In this scenario, the redundant bits are free to undergo mutations and subfunctionalize without damaging the fitness of the organism, and that's how they can be around to provide new functionality.

Anonymous said...

It seems to me that such a fact renders the set of possible animal morphologies completely predictable. Is not that annoying?

Anonymous said...

@if you read closely (p146, end of the second left para), they couldn't find any significant limb length difference in the BatE/BatE adults.

also the difference between the control and the experiment in the paper is only a 6% difference. Between the mouse and the bat, how much is the difference 1000% or so, no?
The review of the article was "Mouse given bat like forelimb". Is not that fraudulent?

Reminds me the Speeman's organizer graft experiment, supposedly giving a mirror frog, which never was.

Stephen Matheson said...

Hey Karen (nunatak)--
Thanks for the kudos. I just looked at your most recent unheralded papers, and I'd like to herald it sometime soon. :-) I'll be in touch.

That recent bat fossil is cool, but most people have interpreted it to show that echolocation arose after flight. I agree it seems somewhat transitional, but it doesn't help much with the limbs-to-wings issue. It was a good month or two for bat evolution, though, eh?

Hmmm. I did read that part of the paper, but I think I see it as much less of a problem than you do. First, I agree with the authors' interpretation: that limb development regulates, and that the effects during mid- and late gestation were overridden by later processes. After all, neither the authors nor any of the commenters (including myself) have claimed that Prx1 is the major player in limb length, and any subsequent process that can regulate could lead to compensation. But more importantly, I think that the significance of the paper derives from the finding that the swapping of an enhancer alone, in the absence of structural differences in coding regions, can generate significant morphological change, and change that is both "positive" and correlated with the known evolutionary trajectory. Of course you're right that changes in mid- and late gestation aren't visible to selection, and it's certainly disappointing that the adults aren't different (this probably answers the "Why Genes & Development and not Nature or Science" question). But I'm not sure that the authors oversold their results. The emphasis has been on the evo-devo connection, specifically on the regulatory element-morphology connection, and not on some overarching claim that we found the gene that makes wings. Good point, though, and thanks for the comment.

Anonymous: not sure where you read "mouse given bat like forelimb" but that's at worst a minor exaggeration and is not something the authors themselves claimed. Your hilarious claim of fraud suggests to me that you should continue posting anonymously till you get some basic concepts of ethics straight.

Anonymous said...

after all, you are right, it is not fraud, because the data are honnest. Let me withdraw the word.
The report I mentionned was in


Nevertheless, it seems to me the finding is extremely modest. I would expect a 6% change of mouse limb extension by many other genetic mutations, for example in fgf's. as there is no change in the adult, and no change in the mouse when prx1 is deleted, I find the overal rationale rather weak, and the claims disproportionate.

Stephen Matheson said...

Your criticism is noted, and thanks for backing off on the "fraud" claim. I think it's appropriate for you and others to note the potential for overstating the importance of the study, especially if/when it is interpreted as the discovery of the explanation for mouse-bat limb length divergence. One thing though: you say "no change in the mouse when prx1 is deleted" but that's wrong. There was no change when the 1-kb regulatory element was deleted; deletion of the gene leads to significant reductions in forelimb length, as shown by the authors (Fig. 3A) and by others in 1995.

Claire said...

Great blog, most informative. Having taken a (very extended!) maternity leave break from teaching/research this gives the old grey matter a good work out!

But is it helpful to label those who believe in special creation as creationist ignorami? They're not going to be uplifted by such titles or encouraged to investigate the truth further. I just think as Christians we should leave the mud-slinging to the world.

Stephen Matheson said...

Hi Claire--

Thanks for the kind words, and for the criticism. If your aim is to steer me away from "mud-slinging," I welcome the feedback, and I'll pay attention. I should note, though, that I do not refer to anyone as an ignoramus merely for holding to special creation, or even for believing odd or silly things. (See my strong support of the work of the BSG, a group of young-earth creationist biologists.) Those who make inexcusably inaccurate claims – i.e., those who are badly wrong and should know better – are the ones who earn the title of "ignoramus." If you follow the link in the blog post, you'll read a remarkably bad series of blog posts in which, among other simplistic and laughable dismissals of evolutionary theory, you will see evolution referred to as "myth" as opposed to science. When people with Ph.D.'s write stuff like that, something other than mere disagreement is called for.

So I'll keep working at tempering the language. But truth-telling will always trump manners here, and there's a place for exposing the misconduct of Christians who abuse science (and theology) in desperate and disreputable attacks on evolutionary science, all because they falsely believe that common descent is an obstacle to faith in Christ.

Anonymous said...

There is something that puzzles me and should be shared by scientists and creationists. If I read well Darwin, and this article, there exists a genetic cursor that does nothing else but extend the limb. Therefore, from mouse to bat, the spectrum of limb shapes is completely obvious (what is called a one degree of freedom problem in physics), and in order for a bat to exist, all shapes in between should be somewhat good.
Therefore there exists an arrow of evolution, from mouse to bat, it is just the one way shift of the cursor, and all animals in between are predictable, in terms of shape.
There is even no need to look for them in quarries, it suffices to morph down or up the bat limb into the mouse limb.
The problem is that mathematics tell you that a one degree of freedom only evolution will necessary happen, eventually. Therefore, the bat was actually
"already present" in the mouse, and bound to appear.

Anonymous said...

A few thoughts:

1. They deleted the regulatory sequence from the mouse genome and found no obvious change in phenotype, but did they also delete it from the *bat* genome to see what would happen? I'd think that would be an obvious test.

2. It seems to me that mice have pretty stubby forelimbs compared to the average mammal. Perhaps this regulatory sequence in mice is "dialed down" all (or most of) the way, thus removing it results in little change. If this is the case, this sequence (in mice) might be a minimized version of what was, in the mouse's longer-limbed ancestors, a sequence with a more functional use.

Stephen Matheson said...

When techniques for ES cell propagation and gene targeting are developed in chiropterans, some decades from now, I would bet that the authors will be interested in performing that currently-impossible experiment. Regarding your second idea, it might be interesting to examine the relevant region in other mammals.