Evolution of the jaw


What do you know…just last week, I posted an article dismissing a creationist’s misconceptions about pharyngeal organization and development, in which he asks about the evidence for similarities between agnathan and gnathostome jaws, and what comes along but a new paper on the molecular evidence for the origin of the jaw, which describes gene expression in the lamprey pharynx. How timely! And as a plus, it contains several very clear summary diagrams to show how all the bits and pieces and molecules relate to one another.

The short summary is that there is a suite of genes (the Hox and Dlx genes, which define a cartesian coordinate system for the branchial arch elements, Fgf8/Dlx1 genes that establish proximal jaw elements, and Bmp4/Msx1 genes that demarcate more distal elements) that are found in both lampreys and vertebrates in similar patterns and roles, and that vertebrate upper and lower jaws are homologous to the upper and lower “lips” of the lamprey oral supporting apparatus.

Start here. This is a simplified diagram of the pharyngeal structures of a typical embryonic chordate— there is a series of repeated, similar bars of mesenchyme and developing cartilaginous frameworks and other tissues, from a mandibular arch (MA) in front, to a hyoid arch (HA), to a chain of branchial arches (Ba). These are tissues that contribute to a cage of supporting elements that will form various features of the face and neck as development proceeds, with each getting modified in characteristic ways. MA, for instance, will go on to form the core of the upper and lower jaws, while HA contributes to the hyoid bone in our neck, and the Ba structures make gill supports in fishes and diversify into other features in us gill-less tetrapods.

As the diagram illustrates, the information that specifies the identity of each pharyngeal arch is defined by a Cartesian grid made up of the overlapping expression patterns of a set of well known genes. The anterior-posterior identity is set by the array of Hox genes; no Hox genes are turned on in the mandibular arch, making that a kind of default state, while more and more Hox genes are activated farther and farther back.The dorsoventral identity is encoded in the pattern of Dlx gene expression.

Developmental mechanism of the gnathostome jaw as an element of pharyngeal arches. The regional specification of the pharyngeal ectomesenchyme is performed by the Cartesian grid consisting of the Hox code that specifies the anteroposterior identity, and the Dlx code for a dorsoventral basic pattern of the arch skeletons. Note that the MAis devoid of Hox gene expression (Hox code-default state). Abbreviations: Ba1-5, branchial arches 1-5; HA, hyoid arch; mo, mouth; ps1-5, pharyngeal slits 1-5.

Getting away from the simplified ideal, here are the structures in a sturgeon and lamprey. This paper is focused on the anterior specializations that differentiate us gnathostomes (“jawed mouth”) chordates from the Agnatha (“jawless”) fishes like lampreys and hagfish, so the specific structures of interest are color-coded: the MA derivatives are reddish purple, while a yet more anterior area, the premandibular region (PM) is colored green. The question the authors are addressing is how the homologies between gnathostome and agnathan visceral arch structures should be drawn, and the colors here reflect their hypothesized answer.

B: Actual shape of the visceral skeleton in young adult sturgeon redrawn from refs. The MA skeleton (shaded red) consists of upper (UJ) and lower (LJ) jaws. The rostral part of the neurocranium (shaded green) has been shown to be derived from the PM ectomesenchyme and is called the prechordal cranium (1-4 trabecula at early stages). C: Schematic illustration of an ammocoete larva. The larva is cut saggittaly and seen from the medial view to show the oral apparatus that consists of the upper lip (UL), lower lip (LL), and the velum (vel). According to the hypothesis in this review, MA-derivatives are shaded red, and the PM-derivatives light green. Colored orange indicates the position of the lamprey trabecula (ltr).

So how do they come to the conclusion that the upper lip of a lamprey is homologous to the premandibular skeleton of the vertebrate, and the lower lip/velum is homologous to the mandible? They compare tissues and molecules. It turns out that the distribution of neural crest cells, a plastic population of migratory cells that contribute to many features of the head and face, is nearly identical in agnathans and gnathostomes. That is one level of similarity.

Another is the distribution of molecules that specify these structures. I mentioned above that the spatial coordinates of the pharyngeal arches are defined by the Hox and Dlx genes; there’s another set of genes that are turned on earlier and are upstream of Hox/Dlx, the Fgf8 and Bmp2&4 genes. Fgf8 and Bmp2&4 are also expressed in similar patterns in lampreys and vertebrates, but there are also differences. Differences are good to see; since lampreys develop rather differently from vertebrates and lack all jaws, we expect that there will also be earlier differences in gene expression that can be correlated with the morphological differences.

What we see here is that there is an inner zone of Fgf8 expression, bracketed by an inner, posterior ring of Bmp2&4, and an outer edge of Bmp2 expression. The difference is that these zones extend farther anteriorly in the lamprey—one of the evolutionary differences between us and lampreys is that we gnathostomes seem to have acquired an additional repression of the extent of Bmp2&4/Fgf8 expression. Those genes seem to be more tightly focused in their activities.

Trigeminal crest cells and the concept of pharyngeal rings in the ectoderm. A: The early deployment of the crest cells is basically identical between agnathans and gnathostomes. The trigeminal crest cell population can be subdivided into supraoptic (SOc), postoptic (POc), and MA (MAc) crest cells based on the position of the eye (e) and mesodermal components, the PMm and MAm. The distribution of growth factors (red bar, BMP2/4; blue bar, FGF8/17) that specify the oral ectomesenchyme differs between gnathostomes and lampreys. B: A schematic diagram of the chicken head at late neurula stage. The caudal portion of trigeminal crest cells is specified as the MA region under the influence of ectodermally derived FGF8. Green broken line indicates the PMM boundary. This Fgf8 domain is a part of Fgf8 ring that lies between the inner Bmp2&4 and outer Bmp4 rings. Abbreviations: pp1, the first pharyngeal pouch; st, stomodeum.

We can play games with these regions of gene expression. Here is a face view of an embryonic chick; on the left is the normal pattern of gene expression, with a region labeled MA (mandibular arch) that is destined to form the upper and lower jaws (UJ and LJ). Next to it is the effect of adding a bead impregnated with Fgf8 to the embryo—the MA region expands anteriorly. Adding a Bmp4 bead inhibits the MA rgion, shrinking it into nonexistence.

In part B is a diagram of a lamprey. It’s remarkable similar to the face of the chick with a Fgf8 bead implanted in it.

Comparison of the growth factor distributions. A: Diagrams of pharyngeal rings in early chicken embryos. In the normal development (left), the MA or oral domain is specified by the rostral part of the Fgf8 ring, which is limited anteroposteriorly by Bmp2/4 expression domains. Green line indicates the PMM boundary. In an FGF8-implanted embryo (middle), the oral region is expanded, with the outer Bmp4 ring shifted rostrally. Green broken line indicates the original PMM boundary, and the solid line the apparent rostral shift of the oral identity. Exogenous BMP4 (right) results in a reduction of the oral domain. B: A schematic diagram of the lamprey pharyngeal rings. Both the PM and MA regions participate in the oral specification as seen in the distribution of LjFgf8/17- and LjBmp2/4a rings. Green solid line indicates the oral-preoral boundary. The PMM boundary (green broken line) crosses in the middle of the lamprey oral region, reminiscent of the FGF8-bead implanted chickenembryo (A;middle). Abbreviations: Ba1, branchial arch 1; HA, hyoid arch; mo, mouth; nhp, nasohypophisial plate; ph, pharynx; pp1, the first pharyngeal pouch; tr, trabecula.

Those face views of the embryos are a little confusing, but these side views help clarify what the two sets of genes are doing. Fgf8 (and its lamprey ortholog, LjFgf8/17) are expressed centrally and proximally, back towards the joint of the upper and lower jaws. Bmp4 and LjBmp2/4a, on the other hand, are expressed more distally, in the region of the upper and lower lips. Look at the similarities in their expression; it’s hard to argue that these tissues are not homologous!

The dotted line in these diagrams does show one difference. The anterior Bmp2/4 expression in the lamprey is not confined to the mandibular arch, but extends forward into the premandibular (PM) area. This is the gist of their evolutionary hypothesis, that what happened in gnathostome evolution was a shift that split the anterior mesenchyme into well-defined mandibular and premandibular components.

Comparison of PD patterning genes expression. Fgf8/17 cognates are expressed in the epidermis of the greater-proximal parts of the upper (UJ) and lower (LJ) jaws in the chicken embryo, and in the epidermis ranging between the upper (UL) and lower (LL) lips in the lamprey. Their downstream genes Dlx1 cognates are expressed in the underlying mesenchyme. Bmp2/4 cognates are expressed in the distal epidermis, respectively, of the jaws in the chicken and of the lips in the lamprey. Their target genes, Msx cognates, are expressed in the distal ectomesenchyme in each embryo. Note the position of the PM mesoderm (PMm) does not coincide with the gene expression patterns between these animals. The thick broken line indicates the PMM boundary. Abbreviations: Ba1, branchial arch 1; HA, hyoid arch; mo, mouth; nhp, nasohypophisial plate; ph, pharynx; pp1, the first pharyngeal pouch; tr, trabecula.

And here is there summary diagram of their model. The authors postulate that there was a rearward shift in the expression of Fgf8 and Bmp2&4 that set up a new structure, the trabecula (tr, in yellow), and defined a new oral region.

Hypothetical scenario of the jaw evolution—developmental background. In each evolutionary stage, only three rostralmost pharyngeal arches are shown. A: Hypothetical agnathan stage. Head ectomesenchyme is divided into oral region (PMand MA) and visceral arches (HA, Ba1 and posterior). The basic Hox code has already been established to specify the hyoid arch against the more posterior arches. The oral ectomesenchyme is defined by the Hox default state. From this ectomesenchymal material, both the dorsal and ventral oral protrusions are assumably differentiated through the molecular cascades involving growth factors FGF8/17 (blue lines) and BMP2/4 (red lines) derived from the epidermis. Morphological patterning of each visceral arch may be at a low level, as seen in the lamprey skeleton that lacks a clear DV specification. Inserted on the left is two tandemly arranged visceral cartilages of an adult lamprey, which is almost symmetrical dorsoventrally. The axis of symmetry is indicated by a red broken line, along which pharyngeal slits (phs) are positioned. B: Hypothetical pregnathostome stage. More Hox genes have acquired colinear expression patterns in the pharyngeal arches. Due to the heterotopic caudal shift of the growth factors, the anterior-most ectomesenchyme is divided into the PM and MA components. The PM ectomesenchyme is now free from an oral apparatus shaping program, to be incorporated into the prechordal cranium, trabecula (tr; yellow). C: Gnathostomes stage. The gnathostome-type basic structure of the visceral skeleton has been established. The significance is the DV subdivision that is presumably related to Dlx genes, which may also have been utilized for jaw articulation in the MA. D: Visceral arch skeletal system of the sturgeon as an example of species-specific shaping. The trabecula is shown as the anteriormost part of the prechordal cranium (yellow). Abbreviations: Ba1-3, branchial arches 1-3; HA, hyoid arch; mo, mouth opening; n, notochord; na, neural arches; ps1-5, pharyngeal slits1-5; sp, spiracle in gnathostomes.

The trabecula would then be an example of a novel structure in evolution, produced by a relatively simple positional shift in the expression domains of a few genes. It doesn’t seem like much—an extra chunk of ectomesenchyme in an embryo—but the key feature is that it creates a new relationship between closely apposed structures, and generates opportunities for unique epigenetic interactions. The shift would have been a minor feature in the pre-gnathostome ancestor, but it provided a cue from which future complexity would cascade.

Shigetani Y, Sugahara F, Kuratani S (2005) A new evolutionary scenario for the vertebrate jaw. BioEssays 27:331-338.


  1. Charlie says

    I think the creationist response to this will be:

    deception by technobabble!!!


  2. says

    Charlie, you don’t know how right you are.

    Tangled up in Blue entry

    If science is too hard to understand, Intelligent Design provides parsimony by not having to worry about details.

    One of the great things about Intelligent Design is that it simplifies things. Unlike the theory of evolution it passes the test of Occam’s razor the universally true scentific axiom that given a choice between a complex theory and a simple one, the simple one has the greatest probability of being right. It’s another way of saying that tall-tales and just-so stories are usually false. ID is so simple that even people who are un-trained in biological sciences can make great and astounding progress. What could be a greater indicator of it’s truth than that?(sic)

    This is another post by P-Zed that I need to print and save. I like this science-y stuff.

  3. amph says

    Charlie, note that this is a repost of a two-year old story, so your prediction is by now verifiable: in the meantime the creationists have had ample time to comment, but AFAIK they haven’t. Too complicated for them to read anyway, I suppose, although it is so easy for them to say, ‘Well this is just how G’d designed i’t.
    BTW: great story, fully worth the re-post.

  4. Wes says

    Okay, I’m a philosophy student, not a scientist, so I have to go through these kinds of things pretty slowly in order to understand them. Let me know if I’m getting this right (I’m gonna have to dumb this down into language I can understand):

    There are a set of arches in the embryos of jawless and jawed animals which are similar, and eventually develop into various similarly situated parts of the face and neck of each. However, the missing piece is the pointy part at the far end of the upper jaw. It turns out this whole apparatus is controlled by a specific set of genes which turn on in a specific pattern, each arch governed by a gene and the overall pattern governed by a sequence of Hox genes. The parts these genes govern and how they are expressed can give us hints of how the transition took place. There must have been a transitional period during which a primitive bit of stuff in the head-end of certain animals, somewhere in the geological column between jawless lampry-like creatures with no jaws and later creatures with fully formed upper and lower jaws. If fossils resembling such creatures are ever found, it would confirm the hypothesis in the paper (not that that’s the only way it could be confirmed or disconfirmed).

    Voila! A testable hypothesis! That is, if I’m actually following this correctly. It’s really frustrating to look words up on http://www.dictionary.com only to find that you don’t understand the words in the definition of the word you don’t understand. :-/

  5. says

    When I was in Biology last year, I thought that evolution of the jaw from pharyngeal gill arches from “jawless fish” to “bony fish” was one of the coolest things. Probably second in line after the Endosymbiotic Theory.

    Thanks for the article.

  6. says

    Err… Just-so stories are characterised by a lack of supporting data. I see plenty of it here. And keep in mind this is just a summary of a full paper, so it’s bound to be lacking some of the details.

    But I guess it’s easier to just dismiss this sort of thing out of hand, isn’t it?

  7. says

    Ah, the troll has already been disemvowelled. Huzzah! It’s much less fun when it happens to such a short post, though.

  8. Kseniya says

    That is, I can’t believe how obstinantly and tenaciously those people (#2, the author of the quote in #3) cling to (and promote!) their ignorance while projecting its attributes, and the mechanisms used to perpetuate, it back this way. It astonishes me, really. It’s like mass psychosis.

  9. K says


    Haste makes waste (and misplaced commas!)

    …attributes, and the mechanisms used to perpetuate it, back this way.