No genes were lost in the making of this whale


I just learned (via John Lynch) about a paper on cetacean limbs that combines developmental biology and paleontology, and makes a lovely argument about the mechanisms behind the evolution of whale morphology. It is an analysis of the molecular determinants of limb formation in modern dolphins, coupled to a comparison of fossil whale limbs, and a reasonable inference about the pattern of change that was responsible for their evolution.

One important point I’d like to make is that even though what we see in the morphology is a pattern of loss—whale hindlimbs show a historical progression over tens of millions of years of steady loss, followed by a near-complete disappearance—the molecular story is very different. The main players in limb formation, the genes Sonic hedgehog (Shh), the Fgfs, and the transcription factor Hand2, are all still present and fully functional in these animals. What has happened, though, is that there have been novel changes to their regulation. Even loss of structures is a consequence of changes and additions to regulatory pathways.


This retention of major genetic pathways should be obvious just looking at a whale. They evolved from four-limbed tetrapods, and lost their hindlimbs as more and more locomotor function was committed to the tail and flukes, yet they still retain forelimbs. It is the same set of genes that operate in the hind- and fore-limbs, so of course you can’t just get rid of them—this is a case of selective limb loss. In addition, the genes have multiple functions making simple gene loss untenable. Shh, for instance, is a critical signaling molecule involved in the specification of midline structures in early development, and loss of the gene as a whole is lethal. What evolution did was to modify the domains of expression, selectively inactivating limb genes in the hindlimb region.

One other curious feature of cetacean development is that they start by making perfectly respectable hindlimb buds, at about the fifth week of gestation. As is typical, they go through a period of phylotypy where their embryos resemble the embryos of other vertebrates, and they initiate the formation of the full four limbs. What happens next, though, is that the hindlimbs regress and their remnants become imbedded in the body wall. This gives us a clue about the change: the molecules involved in limb initiation are still active, but the ones responsible for limb maintenance in early development have been shut down.

If you’ve taken any developmental biology courses at all, you’ve already been exposed to these well-known TLAs: AER and ZPA. The AER, or apical ectodermal ridge, is one of the earliest signs of limb formation. It is a ridge of thickened ectodermal tissue that demarcates the distal margin of the forming limb, and is a signaling center, with a whole family of molecules, Fgf4, 8, 9, and 17, emanating from it and triggering the growth of the structure. The ZPA is the zone of polarizing activity. It’s another signaling center that forms on the posterior margin of the limb, and as you might guess from the name, is important in setting up the polarity of the limb, but it’s also important in maintaining the tissue. It is defined early by a domain of expression of the Hand2 transcription factor, and cells in the ZPA then turn on the Shh gene.

Analysis of the development of limbs in a small number (four—dolphin embryos are not easy to come by, or casually used) of river dolphin (Stenella attenuata) embryos was sufficient to come up with a straightforward picture of the differences in molecular development.

Cetaceans form the AER for both the fore- and hind-limb. They express Fgf8. This is the normal tetrapod pattern.

Cetaceans form a ZPA for the forelimb. Hand2 is expressed broadly at first, and then is restricted to just the posterior part of the fore-limb; Shh is expressed in a perfectly ordinary fore-limb ZPA.

Hand2 is not expressed in the hind limb region. Shh is never activated. No ZPA forms for the hind limb, and the structure arrests and ultimately regresses.

Looking at the evolutionary history of whales, the authors think they can pin down when this downregulation of Hand2 occurred. Shutting off that gene causes a complete loss of the limb, so older fossils that show a gradual diminution of the hind limbs must have retained an active Hand2/Shh combination; the complete loss occurred about 34 million years ago, so that would have been the ‘moment’ when this restriction would have caused the final disappearance of the whale’s posterior limbs.

Simplified phylogeny of cetaceans discussed here with evolutionary events indicated. Hind limbs represent fossil ambulocetid Ambulocetus, fossil basilosauroid Basilosaurus, and two modern mysticetes
(Bowhead Whale and Sei Whale, respectively). In most odontocetes, the only hind-limb element preserved is the innominate, as in the Sei Whale. Labeled bony elements of the hind limb are innominate (inn.), femur (fem.), and tibia (tib).

A promising correlation in the fossil morphology is that there was a concurrent reorganization of the vertebral skeleton at the same time that the hind limbs were lost. The distinct identity of the sacral vertebrae was lost, and the caudal vertebrae became more homogeneous. This implies a change in the expression pattern of the Hox genes, which are responsible for anterior-posterior positional information. That suggests that we ought to look upstream of Hand2, and ask what’s going on with Hox gene expression in cetaceans—the changes that streamlined the vertebral column may have simultaneously induced the changes in Hand2.

It is tempting to speculate that modulation of Hox
gene expression along the craniocaudal axis underpins the
altered expression of Hand2 in the hind limb and posterior
flank of Stenella. This hypothesis is supported by work showing
that ZPA position in the fore limb is specified by the anterior
boundary of Hoxb8 and by the recent finding that modulation
of Hoxd gene expression in mice can shift the boundary of
Hand2 in the early limb bud. This finding provides a
mechanistic link between hind-limb reduction and homogenization of the posterior axial skeleton in whale evolution and
can be tested by studies of Hoxd expression.

How do you make a whale? Clearly, you don’t just “lose” the genes required to make hind limbs. You have to revise and add to the control information for existing banks of regulatory genes involved in limb formation.

Thewissen JGM, Cohn MJ, Stevens LS, Bajpai S, Heyning J, Horton WE (2006) Developmental basis for hind-limb loss in dolphins and the origin of the cetacean body plan. Proc.Nat.Acad.Sci. USA 103(22):8414-8418.


  1. says

    I wish I had this last night when I was teaching. Perhaps I’ll just send my student the link to this post – get them hooked on science blog won’t hurt them a bit…

  2. says

    Yes. That’s its name. Apparently geneticists are as cute as the astronomers who nicknamed a Mars rock “Stimpy.”

    Think it also played a role in the formation of feathers or wings or something. PZ can and most probably has rambled on at length about it.

    Here’s one mention of his.

  3. says

    If we only had a cetacean genome sequence we could look at the non-coding sequence flanking the Hand2 gene for missing highly conserved sequences (using the other vertebrates to identify the conserved sequences). We could also look for changes in the transcription factors that regulate Hand2 expression, the cis-regulatory regions of those transcription factors, the transcription factors that regulate those transcription factors, the cis regulatory regions of the second group of transcription factors, etc, etc. Seriously, though, sequence data would be pretty cool.

  4. Whitey says

    “Sonic hedgehog gene??? Is this its real name?”

    Yes. The name comes from the appearance of the Drosophila Hedgehog mutant embryo. Because there are three homologs in vertebrates, they were given the prefixes sonic, indian, and desert.

    This is one of the few instances in which the mouse geneticists didn’t turn a cool fly gene name into something boring and nondescriptive.

  5. G. Tingey says

    What happens if you get hold of a just=fertilised whale egg, and re-express the genes?

    ( Thought experiment )

  6. says

    Far be it from me, a non-scientist, to nit pick real biologists but having done a bit of field work with dolphins over the years I have to point out that Stenella attenuata is actually a pelagic species, not a river dolphin. Not that it matters. Still a very interesting post.

  7. beervolcano says

    What has happened, though, is that there have been novel changes to their regulation. Even loss of structures is a consequence of changes and additions to regulatory pathways.

    What happens if you get hold of a just=fertilised whale egg, and re-express the genes?

    That’s what I thought. Could you grow a whale with some sort of limbs by ‘unblocking’ their expression?

  8. speedwell says

    Hi, beervolcano. I’m not a scientist… I’m “in class” with you :) but let’s see if I’ve paid enough attention.

    It depends on what is blocking the expression in the first place. If it is a hormone or other chemical, then it can sometimes be neutralized. In that case, the limbs may grow provided the embryo can develop without the chemical. But such interference may have other consequences and the whale embryo may need the chemical for other things. In short, I think the answer to your question is “maybe, provided that the interference has no other effect besides unblocking the gene expression.”

    Professor? participation grade? :)

  9. GRadice says

    Just read the paper and noticed one other cool thing: “All embryonic specimens of S. attenuata were from the museum collection of the Los Angeles Museum of Natural History (LCAM). Embryos were immersion-fixed and and preserved in 70% ethanol. Specimens were stored without refrigeration for a periods ranging from 15 to 32 years.”

    Gotta love museums.

  10. typekey pseudonym says

    PZ wrote: Shh, for instance, is a critical signaling molecule involved in the specification of midline structures in early development, and loss of the gene as a whole is lethal

    And I always thought that Shh was the gene whose function had to be suppressed so that whales could vocalize. Live and learn.

  11. says

    By administering thyroid hormone, thyroid stimulating hormone, or stimulation of hypothalamic neurons[1], you can induce an axolotl to grow legs. Although that’s somewhat different (axolotls stay neotenous by default), I don’t know why in principle you couldn’t do some similar kind of intervention to get a whale with limbs.

    1. Rosenkilde P, Ussing AP. What mechanisms control neoteny and regulate induced metamorphosis in urodeles? Int J Dev Biol. 1996 Aug;40(4):665-73.

  12. Owlmirror says

    Regarding limbs on whales –

    Given that whales will occasionally naturally grow vestigal limbs (, it seems likely that it wouldn’t be too out of the question to find a simple way to “turn off” the entire sequence of developmental apoptosis without otherwise harming the whale’s development.

    Of course, I don’t think it would be a very kind thing to do, since full legs would probably badly spoil the whale’s hydrodynamic efficiency, not to mention being a shocking surprise to the mother.

    I recommend Zimmer’s At the Water’s Edge for a splendid description of whale evolution, as well as explorations of why some marine mammals developed small tails and large fore- and hind- flippers (seals, etc), and others developed large tails and small or non-existent hindlimbs (ceteceans, manatees).

    Also useful:

  13. says

    Why do men have nipples? (I havnt read the book) It seems that its easier/safer to Down-Regulate rather than eliminate a structure completely. I guess if there was a function to remove nipples altogether, any ‘leakage’ of that function to females would cost more in infant-deaths than the small cost of maintaining small nipples on men. Cant see the ‘danger’ to whales of eliminating hind limbs completely. Seems there is something to lern about the mechanisms of evolution, what is easy and what is hard to evolve & devolve.