CROTCHETS & QUIDDITIES

Dinner at Baby’s: Werewolves, Dinosaur Jaws,Hen’s Teeth, and Horse ToesKENNETH WEISS AND SAMUEL SHOLTIS

As we get older we have a tendencyto become nostalgic and think back onold times. We’ve recently seen a surgeof nostalgia for the 1950s; among theremarkable comebacks are the newold diners, like Baby’s here in StateCollege (Figure 1). These new-oldwonders proffer burgers and shakeslike they used to be in the good olddays. The decades of change in thecompetitive fast-food industry seemnot to matter at all. The old taste isback! Even Patsy Cline and Elvis arestill singing the same songs in thebackground.

There is a similar phenomenon inbiology. Nobody accepts Ernst Haeck-el’s famous recapitulation argumentthat, as embryos, we literally gothrough the adult stages of our ances-tors. Nonetheless, many seem to thinkthe evolutionary past can rise again.Can it?

“A CIRCUMSTANCE WELLWORTHY OF ATTENTION”

One of the key facts in Darwin’s“long argument” for evolution wereatavisms—“throwbacks”—to evolu-

tionary earlier states. In Descent ofMan, he argued that these reversionswere “well worthy of attention.” Suchtraits are easier to explain as reflectinghistorical connections than by cre-ationist arguments (for a populartreatment, see1). Many such traits oc-cur naturally, but recently some sur-prising examples have arisen out ofexperiments in developmental biol-ogy.

NATURAL “ATAVISMS”

The occasional presentation of ex-treme hairyness in humans is an ex-ample of a naturally occurring “ata-

vism.” Our primate ancestors werefully furred, and when a variant alleleor new mutation arises that causes aperson to be very hairy, he or she mayunderstandably be seen as ape-like.Such human atavisms, because wetend to think of ourselves as advancedand gentile, are often popularly por-trayed as haunts of a brutish past, asin the “abominably hairy” atavismRed Eye of Jack London’s book BeforeAdam who “was a monster in allways.” In our attention-hungry societythis has even been likened to were-wolves.2

Another common reversion dis-cussed by Darwin is supernumerarynipples (polythelia). The “milk line” iswell known in mammals. It runs alongthe thorax and abdomen on both sidesof the midline. Different species have

Ken Weiss is a biological anthropologist,and Sam Sholtis is a Weiss graduate fel-low (no relation to the first author), both atPenn State University.E-mail: kmw4@psu.edu

© 2003 Wiley-Liss, Inc.DOI 10.1002/evan.10125Published online in Wiley InterScience(www.interscience.wiley.com). Figure 1. Dinner at Baby’s (State College, PA).

Occasionally traits arise that appear to be atavistic throwbacks to the remotepast. How can this make evolutionary sense?

Evolutionary Anthropology 12:247–251 (2003)

differing numbers of nipples locatedalong this line, and this, as Darwinnoted, can vary naturally and be her-itable.

The evolution of digits has longbeen of interest to anthropologists.3

Julius Caesar’s horse is said to havehad toes (as have the horses of Napo-leon and Alexander the Great).Whether seeing these thundering toesin battle terrorized the barbarians wecan only speculate, but the fossilrecord shows that horses haven’t hadtoes since Merychippus during the up-per Miocene (and they were alreadyreduced and probably bore little or noweight).4

Another classic is the occasional ap-pearance of rudimentary hindlimbs inwhales.5 Whale legs and horse toes arerelated to Darwin’s notion of vestigialtraits, because vestiges of the horse’soriginal toes are present in the form ofsplint bones in the legs of all modernhorses, and many whales retain bonesor cartilage related to hindlimbs hid-den beneath their incredible mass.

EXPERIMENTAL ATAVISMS

Hen’s Teeth

Teeth develop through a process oftissue induction between two embry-onic tissues in the developing jaw(Figure 3B). The overlying dental lam-ina is an epithelial layer. Underneathit is a mesenchyme largely composedof neural crest cells that have physi-cally migrated to the jaws from the

early neural tube.6 The epithelium dif-ferentiates into ameloblasts that pro-duce enamel, while the mesenchymebecomes odontoblasts that producedentine. When these two tissues comein contact early in embryonic develop-ment, the information for initiatingdental patterning resides in the epi-thelial layer, which activates the mes-enchyme, and the two then interactduring tooth morphogenesis.

Kollar and Fisher7 wonderedwhether teeth could be induced inbirds. They grafted chick embryonicjaw epithelium to mouse dental mes-enchyme, and found the developmentof what appeared to be teeth (Figure4). This famous “hen’s tooth” experi-ment was questioned because itseemed implausible that chick epithe-lium could be re-awakened to maketeeth tens of millions of years afterchick ancestors had any choppers tochop with. The tooth seems to containenamel, but chicks appear not to havethe required gene (amelogenin), and itwas molariform, but chicks don’t havemolars in their family tree (We as-sume as an artifact that it looked morelike a human than a mouse molar.)

The standard explanation was thatthis was really a mouse tooth becausethe grafted mesenchyme was contami-nated with mouse epithelium. But theexperiments have been replicated8,9 inmany ways generating tooth-like struc-tures, but without enamel. Wang et al.9

reversed the combination (mouse epi-thelium and chick mesenchyme) and

showed molecularly that the mesen-chyme in fact expresses chick, notmouse, genes. There is still much to belearned about the crosstalk betweenthese two tissues required for tooth de-velopment as suggested by the seem-ingly contradictory results of two recentpapers. Chen et al.10 show that path-ways involved in tooth development areconserved and can be turned on inchick mesenchyme by the addition of asingle signaling factor (Bmp4) to thechick epithelium. However, the chickepithelium is perfectly capable of initi-ating tooth-like formation when chickneural crest is replaced by neural crestfrom a mouse.11

Dinosaur Jaws

It is exciting to perform an experi-ment for one reason, and discoversomething totally unexpected. Youscurry to the nearest textbook to findout what has occurred. If you wereworking on a detailed biomedicalproblem, but discovered something ofevolutionary interest—like an atavis-tic trait—it would make a quote-wor-thy thing to jazz up a paper.

This kind of surprise has resultedfrom transgenic mouse experiments,and a particularly interesting case in-volves mutations engineered to inacti-vate a gene believed to be involvedwith jaw and tooth development. Twocomponents made up the jaws ofmammalian ancestors, Meckel’s carti-lage in the lower jaw, and the palato-

Figure 2. Atavisms. (A) Werewolves, andother “atavisms.” Source: http://www.e-pub.org.br/cm/n09/fastfacts/atavis-mo_i.htm from “Evolution: Zufall oderSinn?” Bild der Wissenschaft, 4: 114-126,1979 (B) Horses’s toes (which are thehorse, which Caesar?). Source: Modi-fied from Bettman Archive.

248 Weiss and Sholtis CROTCHETS & QUIDDITIES

quadrate cartilage in the upper (Fig-ure 4A). The palatoquadrate has beenseverely reduced during mammalianevolution with an anterior survivingremnant, the alisphenoid, that formspart of the braincase, and a posteriorsurviving remnant that is the incusossicle of our middle-ear (Figure 4B).The Dlx-2 DNA regulatory gene is ex-pressed early in jaw development.When Dlx-2 was experimentally inac-tivated in a mouse, the abnormally de-veloping jaws possessed anomalouscartilage12 (Figure 4C). The investiga-tors suggested that this cartilage washomologous to the ancestral palato-quadrate, as if ancient dinosaur jawdevelopment had been recreated.Similar atavistic explanations havebeen offered for anomalous cartilagein experimental results inactivatingthe Hoxa2, MHox, Otx2, and retinoicacid receptor genes.

In fact, this probably reflects misun-derstandings of the path of evolutionand a simplistic notion of atavisms.13

The cartilage formed in these mutantsbears no real resemblance to the an-cestral condition, and thus fails a keycriterion for identifying atavisms.5

Comparing the results of the Dlx-2knockout with the appearance of theactual palatoquadrate cartilage showstoo vague a resemblance of these

structures to substantiate claims ofhomology, much less the atavistic rec-reation of an ancestral organized trait.

The assignment of homology oftenimplies similar developmental originof structures, but gene phylogeny andexpression by David Stock, working inour lab, showed that Dlx-2 existed andwas expressed in embryonic jaws longbefore reptiles came on the scene. Theancestral jaw developed with, notwithout that gene, so its experimentaldeletion could hardly duplicate ances-tral processes in dinosaur jaws. Smithand Schneider offer the more plausi-ble explanation that cartilage forma-tion relies on a threshold level of cellcondensation in development. Exper-imental disruption of early jaw devel-opment could lead to the piling up ofmigrating cranial neural crest cells inuncharacteristic locations, leading toanomalous local cartilage formation.Consistent with this is that the exper-imental animals had numerous othercraniofacial developmental anoma-lies.

GENETIC EXPLANATIONS

What Probably Isn’t

A general principle of evolution isexpressed by Dollo’s14 famous rulethat “an organism is unable to return,

even partially, to a previous stage al-ready realized in the ranks of its an-cestors.” Marshall et al.15 addressedthis from a genetic point of view,showing by plausibility calculationsthat accumulating mutations arelikely to destroy genes not maintainedby selection by around 10 millionyears. Subtle coordination and inter-action among genes are probably thefirst to go, but eventually unusedgenes simply mutate into oblivion.

Why can’t such a gene be restoredby reverse mutation? With few excep-tions (olfactory receptors being onepossible example), the probabilitythat reverse mutation could restorethe complex function of a gene issomewhere between impossible andunbelievable. A gene long silent (asbetween a chicken and it’s formerteeth) cannot be resuscitated. Avianamelogenin appears to be but a mem-ory.

Sometimes a given complex traitdoes recur independently in relatedlineages, as for example, complexeyes, or immature stages in variousamphibian or sea urchin larvae.16,17

Traits of similar general form may berather simple to initiate but these re-currences are probably not identicalat the gene level. And because theyoccur among evolutionary lineages,rather than within a lineage over time,we would not normally call them ata-vistic throwbacks to a former state.

What Probably Is

At any given time, traits can comeand go across generations. Recessivetraits are familiar examples. But blueeyes are not an interesting atavisticthrowback even if both of one’s par-ents had brown eyes but at least onegrandparent had blue ones. This isvariation still circulating in the popu-lation even if rare or masked in someindividuals. Similarly with polydacty-ly: it is only a kind of statistical typol-ogy to say that horses have hoovesrather than toes.

Sewall Wright18 and others madeformal breeding studies of existingdigit variation in guinea pigs early inthe 20th century. Wright showed evi-dence for genetic causation, as well asrandom effects among and within in-bred strains. Interestingly, becauseother South American Caviidae have

Figure 3. Hen’s tooth (A) original and (B) a confirmation showing the morphology andcorresponding different dental tissues. Mouse dental mesenchyme grown in culture incombination with chick “dental” epithelium. am, ameloblasts; en, enamel; de, dentine; od,odontoblasts. G and H are from the original panel notation. In G, e indicates area ofinvaginating mouse epithelium; arrows indicate chick mesenchyme. (A: Reprinted withpermission from Kollar and Fisher7) B. Wang et al.9 (Reprinted with permission, copyright 1998Developmental Dynamics, Wiley.)

CROTCHETS & QUIDDITIES Dinner at Baby’s 249

thumbs, big and little toes, Wright re-ferred to their occurrence in guineapigs as atavistic, despite the degree ofnatural variation he observed. Henoted that the genetic basis of theserecurrences was probably different indifferent strains, and attributed this topolygenic threshold causation. With

polygenic causation, the same traitarises in many different genotypes,implying that the same genetic basisof these toes in guinea pigs and otherCavids past or present is unlikely. It’ssimilarly unlikely that the genetic ba-sis of Caesar’s horse was the same asthe toes on ancestral horses.

Similar arguments apply to super-numerary nipples, variation in dentalformulas including transitory dentalrudiments in the diastema of develop-ing mouse jaws which have been de-scribed as representing “missing”teeth, and the occasional presence of“tails” in humans. Given their rathercommon natural occurrence, if theseare atavistic throwbacks, they are notthrowing very far back.

Then How Do Horses GetToes and Hens Get Teeth?

A lot has been written about thesesame traits, often in terms of develop-mental genetic “thresholds” and con-served developmental “programs,” butthis amounts to little more than handwaving. However, we can make ge-netic sense of it. Most examples ofatavisms are in traits produced by pe-riodic patterning processes, a pointfamously stressed by William Bate-son.19 These serially homologousstructures are produced by generic de-velopmental processes that involvesignaling-factor molecules that dif-fuse between cells in a particular de-veloping tissue. The relative local con-centration of these molecules affectsgene expression in the cells, resultingin the production of epithelial pla-codes, tissue-layer invagination,branching, and periodically spacedgrowth and inhibition zones wheremembers of a series (e.g., a tooth) de-velop.3,6

The same genes are used to patternmany structures, including the classic“atavism-prone” traits like limbs,mammary glands, teeth, vertebrae,and hair. The genes that make teeth inprimates are still around biting andkicking in hens. A tooth is develop-mentally also a hair, feather, scale, ornipple. For example, the close rela-tionship of teeth and hair is shown bythe interesting effect of mis-express-ing the regulatory gene Lef-1 in mice,which leads to tooth development inthe lip furrow or hair developmentamong developing teeth. (Is thatwoolly morning-after mouth a re-membrance of times past?) Natural aswell as experimental alteration ofthese developmental processes canproduce variation in the number, size,and other characteristics of the ele-

Figure 4. Dinosaur jaw “atavisms” that aren’t really. (A) Primitive tetropod jaw, (B) normalmouse and (C) Dlx-2 knockout “atavism.” (Black, upper first arch elements or palatoquad-rate cartilage; dark grey, lower first arch element, or Meckel’s cartilage; light grey, (ectopiccartilage.) (Source: Smith and Schneider13 with permission).

250 Weiss and Sholtis CROTCHETS & QUIDDITIES

ments in the structures they pattern—sometimes resulting in variation thatresembles traits from distant earlierages.20 The genetic pathway thatcould turn the splint bones in horsesinto actual toes still exists in thehorse’s remaining digit. Many genesare involved in these processes, butthey may actually be simpler than thepolygenic control envisioned byWright.

Hen’s teeth aren’t really teeth, be-cause some of the patterning instruc-tions and the enhancers needed to in-tegrate the pathway in one of itscontexts are gone. The chick epitheliallayer has lost these capabilities, andmouse mesenchyme can’t inducethem, but mouse epithelium has theStart instructions and can invoke theresponses in chick mesenchyme. It isinstructive that the missing ameloge-nin gene that is used only in the pro-duction of enamel was not protectedfrom obliteration by pleiotropy.

In most cases, only some aspects ofthe process have been anomalouslyexpressed, and atavistic changes areusually not completely normal. Thusthe hen’s tooth has no enamel, andsupernumerary nipples usually areonly imperfectly formed and do not orcannot function normally in humans.

BUT THERE AREN’T ANYCENTAURS

We can imagine all sorts of assem-blages of characters. One examplemight be the Centaurs of classical my-thology. These half-horse half-humancreatures have not actually been seenrecently, but we have evolutionaryreasons to assert that they, like uni-corns and mermaids, aren’t real. Itwould be impossible, based on every-thing we know, for the genetic anddevelopmental organization requiredto make a human torso also to make ahorse’s aft. The forelimb and hindlimbof mammals develop using at leastsome of the same genes, and there is agenetically based correlation in theform of the two limbs. The divergencebetween primate and horse genomeshappened so long ago that too much

independent evolution has occurredin each genome for one to be a part ofor function with the other. In thissense, unicorns might just be conceiv-able, but Centaurs and mermaids,sadly not. By contrast, at least in thepast, hens had teeth, and horses hadtoes, so hen-like and horse-like ge-nomes once were compatible withteeth and toes.

Still, from a gene-regulation, devel-opmental point of view, the verisimil-itude of atavism does indicate a con-nection to the past, though it’s not anawakening of long dormant develop-mental programs. Recent develop-mental genetics has been showinghow deeply conserved developmentalprocesses really are. Developmentally,hen’s teeth are truly tooth-like. Butthey are not truly teeth. Evolutiondoesn’t generally reverse itself, but thepast does seem to weave in and out ofcomplex traits, in that elements of themechanisms producing those traits doseem to be conserved in interestingways. The atavism story is subtle andwe’re now getting a genetic under-standing of why we see what we see.

Thus, from what we know Dollo willrule. Genetic engineers might producechickens with teeth, but they won’t bereal atavisms. It’s like that with retrodiners, too. They have a nostalgic ap-pearance and Patsy Cline may be sing-ing the same songs in the background,but most everything else in these din-ers has changed, and we’re “Sowrong,” if we think we can really be“Back in Baby’s arms.”

NOTES

We would welcome comments onthis column: kmw4@psu.edu. Crotch-etyComments are maintained on:http://www.anthro.psu.edu/weiss_lab/index.html. We thank Anne Buchanan,Kathleen Smith, and John Fleagle forcritically reading this manuscript.

REFERENCES

Many things discussed here can beprofitably explored by web searching.

1 Gould SJ. 1980. Hen’s teeth and horses toes.Natural History 89:24–28.

2 Figuera LE, Pandolfo M, Dunne PW, Cantu JM,Patel PI. 1995. Mapping of the congenital gener-alized hypertrichosis locus to chromosomeXq24-q27.1. Nat Genet 10:202–207.

3 Chiu C-H, Hamrick M. 2002. Evolution anddevelopment of the primate limb skeleton. EvolAnthropol 11:94–107.

4 Carroll RL. 1988. Vertebrate Paleontology andEvolution. New York: WH Freeman.

5 Hall BK. 1984. Developmental mechanisms un-derlying the formation of atavisms. BiologicalReviews 59:89–124.

6 Jernvall J, Jung HS. 2000. Genotype, pheno-type, and developmental biology of molar toothcharacters. Am J Phys Anthropol Suppl 31:171–190.

7 Kollar EJ, Fisher C. 1980. Tooth induction inchick epithelium: expression of quiescent genesfor enamel synthesis. Science 207:993–995.

8 Lemus D. 1995. Contributions of heterospecifictissue recombinations to odontogenesis. Int JDev Biol 39:291–297.

9 Wang YH, Upholt WB, Sharpe PT, Kollar EJ,Mina M. 1998. Odontogenic epithelium inducessimilar molecular responses in chick and mousemandibular mesenchyme. Dev Dyn 213:386–397.

10 Chen Y, Zhang Y, Jiang TX, Barlow AJ, StAmand TR, Hu Y, Heaney S, Francis-West P,Chuong CM, Maas R. 2000. Conservation of earlyodontogenic signaling pathways in Aves. ProcNatl Acad Sci U S A 97:10044–10049.

11 Mitsiadis TA, Cheraud Y, Sharpe P, Fontaine-Perus J. 2003. Development of teeth in chick em-bryos after mouse neural crest transplantations.Proc Natl Acad Sci U S A 100:6541–6545.

12 Qiu M, Bulfone A, Martinez S, Meneses JJ,Shimamura K, Pedersen RA, Rubenstein JL.1995. Null mutation of Dlx-2 results in abnormalmorphogenesis of proximal first and secondbranchial arch derivatives and abnormal differ-entiation in the forebrain. Genes Dev 9:2523–2538.

13 Smith KK, Schneider RA. 1998. Have geneknockouts caused evolutionary reversals in themammalian first arch? Bioessays 20:245–255.

14 Dollo L. 1893. Les lois de l’evolution. Bulletinde la Societe Belge de Geologie, de Paleontologieet d’Hydrologie 7:164–166.

15 Marshall CR, Raff EC, Raff RA. 1994. Dollo’slaw and the death and resurrection of genes. ProcNatl Acad Sci U S A 91:12283–12287.

16 Wray GA, Hahn MW, Abouheif E, Balhoff JP,Pizer M, Rockman MV, Romano L. 2003. TheEvolution of Transcriptional Regulation in Eu-karyotes. Mol Biol Evol. In press.

17 Weiss KM. 2002. How the eye got its brain.Evol Anthropol 11:215–219.

18 Wright S. 1968. Evolution and the genetics ofpopulations: a treatise. Chicago: University ofChicago Press.

19 Weiss KM. 2002. Good vibrations: the silentsymphony of life. Evol Anthropol 11:176–182.

20 Salazar-Ciudad I, Jernvall J. 2002. A gene net-work model accounting for development andevolution of mammalian teeth. Proc Natl AcadSci U S A 99:8116–8120.

CROTCHETS & QUIDDITIES Dinner at Baby’s 251