It’s another Dawkins question!

Why do cave-dwellers lose their eyes? They’re useless, but are they harmful? Costly to make? Or eroded by rain of uncorrected mutations?

I thought I’d already addressed this in a blog post long ago, but I searched, and I didn’t — it was my inaugural column in sadly defunct Seed magazine, way back in the paleolithic, I think. Fortunately, I still have the copy I sent in to the editor, so I resurrect it here.

---

Degeneration and development
It’s not disuse that leads to loss of organs in evolution, but competitive genetic interactions

Reduced or degenerate organs, such as the evolution of flightless birds or eyeless cave dwelling animals, were a problem that Charles Darwin considered; his answer was that disuse would lead to their progressive reduction over time (we do not believe this is correct any more). Darwin’s confusion is shared by many even today, and Stephen Jay Gould, in his magnum opus, The Structure of Evolutionary Theory, listed three things that his readers found most confusing, as measured by the correspondence he received:

“I can testify that three items top the list of puzzlement: (1) evolution seen as anagenesis rather than branching (“if humans evolved from apes, why are apes still around”); (2) panselectionism (“what is the adaptive significance of male nipples”); and (3) Lamarckism and the failure of natural selection (“doesn’t the blindness of cave fishes imply a necessary space for Lamarckian evolution by disuse”).”

While all three are interesting questions, let’s consider just the third, which Darwin failed to answer. Why should animals living in total darkness lose their eyes? Gould described two good answers in his book, but recent work by W.R. Jeffery and his colleagues on the Mexican blind cavefish has supported a third. It’s an answer that highlights the importance of developmental biology in explaining some evolutionary phenomena…and it’s also an excellent way to introduce this new column, in which I’ll regularly be discussing the evo-devo way of thinking.

One possible answer is that it is an economical adaptation. It requires energy and effort to build something as intricate and fragile as an eye, so shutting off that pathway would be a sensible strategy in the embryo: the energy that would be used in constructing and maintaining the eye could instead be diverted to other growing organs. One analogy would be in building a house that will have one wall facing a brick wall; it would be sensible to not bother building a large and expensive picture window there. For the cave fish, those embryos that did not bother to build an eye that would never be used acquired some slight advantage over their fellows that did bear the burden of an eye, and so gradually came to dominate the cave population.

In the case of the Mexican blind cavefish, though, there is a striking observation against this explanation: the embryos make eyes! They initially develop, they form an eye cup, they develop the beginnings of neural circuitry, neurons proliferate…and then they stop. The rest of the skull continues to grow, overwhelming the budding eye with new tissue. It’s as if one paid to have that picture window built into the house, and then halfway through construction had it ripped out and a wall put in. This is hardly economical.

Another possible answer that loss of an eye in a cave fish does not impose any cost, and the eyes disappear in the population by random chance, and that without selection for sightedness, there is nothing to prevent the blind variants from competing equally with their sighted counterparts. This is a neutral theory of the loss of unused characters, which suggests that mutations that knocked out genes needed for development of the eye wouldn’t necessarily have any advantage, but they’d also have no cost. The eye is lost by harmless attrition, and would be represented by broken genes in the animal’s genome.

This explanation doesn’t seem to fit the facts, either. The genes involved in generating the eye all seem to be present and functional in the blind cave fish. Transplanting a lens from a cavefish species with eyes to the blind cavefish embryo is enough to rescue the eye—it then develops into a perfect and functional visual organ. The problem isn’t caused by outright broken genes, but by genes that are being regulated in a different way. Something is actively switching off lens formation and thereby removing a signal for eye development, and in fact, analysis of gene expression in the developing blind cavefish eye reveals that many genes are more active than they are in the sighted fish.

If neither the economical hypothesis or the neutral hypothesis adequately explain how cavefish lose their eyes, what is the answer? The evidence suggests a third alternative, an explanation based on pleiotropy and developmental interactions.

Pleiotropy is a common phenomenon in genetics: all it means is that a single gene may have many different effects on the organism. Two genes that interact with one another in this system are pax6, a ‘master gene’ controlling the development of the eye, and hedgehog, a signaling molecule that plays an important role in setting up the midline of the animal. pax6 is a transcription factor (a gene that regulates the expression of other genes) and is active in the region of the embryonic head where eyes will form, is expressed in the eye cup and lens, and regulates the development of the eye. hedgehog is a protein that is secreted at the midline and diffuses laterally to regulate many other processes. It’s function is complex, but one role is to inhibit and separate structures; one effect of mutations in hedgehog is midline defects, such as cyclopia, where the eyes fuse together in the absence of a hedgehog boundary. Among those effects is an inhibition of pax6.

hedgehog is also expressed in teeth, tastebuds, and the jaw. This is what we mean by pleiotropy — the gene is a midline gene that suppresses the eye gene pax6, but it’s also a jaw gene and a tooth gene and a tastebud gene. Now imagine a population of fish feeding and swimming in total darkness; which individuals will be most adapt at finding food, and therefore most likely to pass on their genes to the next generation? Those that are best at using senses other than vision, that can use the tactile senses of their lower jaw to probe the environment for a meal, and that can use taste instead of sight to discriminate among food choices. What this suggests is that animals with expanded hedgehog function would thrive best, and selection would work to increase the frequency of greater hedgehog expression in the population.

There is a side effect — pleiotropy at work — in that hedgehog also inhibits pax6 expression, which means that expanding jaws and tastebuds will lead to a concomitant reduction of the eyes. What we have is a perfect example of an evolutionary tradeoff. Because hedgehog and pax6 are negatively coupled to one another, one can be expanded only at the expense of the other, and what is going on in the blind cavefish is not selection for an economical reduction of the eyes, nor the accidental loss of an organ that has no effect: it is positive selection for a feature that is only indirectly related to the eyes.

Gould would have appreciated this discovery. The cause of the blindness is the interrelationship between two genes which have complementary roles in development in establishing the architecture of the face. This is not a necessary relationship — not all genes have to be coupled to one another in this way — but a result of the evolutionary history of the organism, and a different kind of economy. The lower part of the face can be expanded, but only at the expense of the upper half.

The general lesson from this analysis is that understanding selection is not enough. We also need to understand the developmental interactions present in an organism to understand how selection for one feature might lead to a surprising pleiotropic change in something completely different.

---

Just for completeness, I’ll include some bits that weren’t published, but passed along for the editor’s edification.

(Chris: for your fact checker, the reference to Gould is from page 204 of The Structure of Evolutionary Theory.

Darwin’s explanation for the eyelessness was this:

“By the time that an animal had reached, after numberless generations, the deepest recesses, disuse will on this view have more or less perfectly obliterated its eyes, and natural selection will often have effected other changes, such as an increase in the length of the antennae or palpi, as a compensation for blindness.”

That’s from Chapter 5, “Laws of Variation,” in his Origin of Species.

I don’t know how you want to include formal bibliographic citations. A good summary of the cavefish work is here:

Jeffery WR (2005) Adaptive evolution of eye degeneration in the Mexican blind cavefish. J Hered 96(3):185-196.

I notice these kinds of things weren’t explicitly listed in previous examples of the column.)

Here’s another of those Dawkins questions.

Stuart Kauffman’s thought experiment: If evolution could be re-run 1000 times, would certain patterns predictably recur? Humanoids?

Only I already answered it a few weeks ago! I know that Dawkins is more sympathetic to the idea of evolutionary convergence than I am, but I don’t think you’d get recurrence of humanoids. If we could reset everything to the precise state the world was in during the Cambrian, half a billion years ago, I think it would be safe to say we’d get a planet full of molluscs and arthropods, but that’s about all we could say — tetrapods are not inevitable. If we could rewind to a few billion years ago, before the evolution of eukaryotes, we could talk about a planet of algae and bacteria, but other derived forms would be so contingent on chance that no prediction is possible.

The best examples of actually doing this experiment come from the Lenski lab and their work on bacteria, where generations could be frozen and restarted at will, and the answer is…no, it isn’t inevitable that a lineage will emerge that carries even an expected optimal simple biochemical pathway.

Ophelia has summarized a series of science questions Richard Dawkins asked on Twitter. Hey, I thought, I have answers to lots of these — he probably does, too — so I thought I’d address one of them. Maybe I can take a stab at some of the others another time.

I like this one, anyway:

Why do cells have the complete genome instead of just the part that’s needed for their function? Liver cells have muscle-making genes etc.

My short answer: because excising bits of the genome has a high cost and little benefit, and because essentially all of the key exaptations for multicellularity evolved in single-celled organisms, where modifying the DNA archive would have serious consequences for all the daughter cells.

This is an interesting issue, though: different kinds of cells in the same organism express genes that are qualitatively and quantitatively different. Here’s a set of nice graphs in which the relative fraction of different classes of genes in gene transcripts in different cell types were measured. Notice in the list of biological processes that a lot of them, such as the genes involved in transcription and translation and metabolism, are going to be used in all cells, but some, such as neuron-specific or testis-specific genes, are only going to be expressed in some cells.

It also gets complicated because some genes are found in very different forms: there is a kind of universal myosin, myosin I, for instance, that is expressed in all cells as part of the intracellular transport machinery, and then there is a myosin variant, myosin II, that is expressed only as a part of the contractile machinery in muscle. So you might think that it would be more efficient for a skin cell to simply cut out and throw away Myosin II, since it’ll never use it, and keep Myosin I.

But how does the cell determine which genes it will never use? Where does it draw the line? All those testis development genes, for example — I never used many of them until I hit puberty. Wouldn’t it have been terrible if my young toddler testicles threw out a set of unused genes, and then a dozen years later discovered that they had a use, after all? There are a great many genes regulated by timing and signals, and as can be seen in that figure above, every cell has a different expression profile. There are a variety of cells in my skin that are busy replicating and making keratin proteins as a matter of course, but they only switch on cellular repair mechanisms if I cut myself. There are also many genes that get reused in complicated ways, too: the gene even-skipped is first switched on as part of the segment forming process in flies, but it later is switched on again in making neuroblasts, and later still is expressed in axons during pathfinding. Cells would rather recycle genes than throw them away.

These properties are not unique to us mammals, either. Bacteria regulate which genes are turned off and on, too — they change their biochemical behavior in response to signals in their environment. The ability to switch on and switch off genes, without eliminating the DNA, is a solved problem. Life figured that one out a few billion years ago. Key molecules required for multicellular patterns of gene expression first evolved in bacteria — they worked out how to have a cell with the same genetic material behave differently in different circumstances. We came about ready made with a toolkit equipped to have one set of genes turned on in livers, and a different set turned on in muscles, easy.

But, you might think, wouldn’t it be so much more cost-efficient if cells in multicellular organisms just got rid of genes they’d never turn on in their lifetime, once they’ve committed to a certain tissue type? Muscle cells will never make sperm recognition proteins, and liver cells won’t ever have to lift weights, and you could probably cut the amount of DNA in differentiated cells in half with no effect on function.

But that’s penny-wise accounting. In bacteria, only about 2% of the cell’s energy budget is invested in replication — so removing a bit of DNA here and there is only going to shave a tiny amount off the cost of cell division. On the other hand, an amazing 75% is spent on transcribing and translating genes, so efficient mechanisms of simply turning off unused genes reaps huge savings for the cell. Evolving a complex process to pare away unused DNA in terminally differentiated cells simply does not make sense energetically, while simply taking advantage of an already fully implemented and refined process for regulating gene expression…heck, that’s what evolution does best, reusing what’s already there.

By the way, not all cells carry the complete genome: there are also a few cases where the DNA of an organism is modified — the CRISPR system in bacteria, and the somatic recombination system used in vertebrates to generate diverse immunoglobulins. In both of those cases, though, it’s not a mechanism to cut away unused DNA. It’s a specialized process to create variation during an organism’s lifetime to cope with environmental challenges.

---

Lane N, Martin W. (2010) The energetics of genome complexity. Nature 467(7318):929-34.

Ramsköld D1, Wang ET, Burge CB, Sandberg R (2009) An abundance of ubiquitously expressed genes revealed by tissue transcriptome sequence data. PLoS Comput Biol 5(12):e1000598. doi: 10.1371/journal.pcbi.1000598.