The Molecular Architects of Body Design Putting a human gene into a fly may sound like the basis for a science fiction film, but it demonstrates that nearly identical molecular mechanisms define body shapes in all animals

by William McGinnis and Michael Kuziora

ll animals develop from a single tems that mold our form, we humans Antennapedia adults are rare excep- fertilized egg cell that goes may be much more similar to our far tions, because most mutations in ho- A through many rounds of divi- distant worm and insect relatives than meotic genes cause fatal birth defects sion, often yielding millions of embry- we might like to think. So similar, in in Drosophila. Nevertheless, even those onic cells. In a dazzling and still myste- fact, thatÑas our work has shownÑcu- dying embryos can be quite instructive. rious feat of self-organization, these rious experimenters can use some hu- For instance, Ernesto Sanchez-Herrero cells arrange themselves into a com- man and mouse Hox genes to guide the and Gines Morata of the Independent plete organism, in which bone, muscle, development of fruit-ßy embryos. University of Madrid found that elimi- brain and skin integrate into a harmo- The story of these universal molecu- nation of three genes in the bithorax nious whole. The fundamental process lar architects actually begins with the complexÑUltrabithorax, abdominal-A is constant, but the results are not: hu- pioneering genetic studies of Edward and Abdominal-BÑis lethal. Yet such mans, mice, ßies and worms represent B. Lewis of the California Institute of mutant embryos survive long enough a wide range of body designs. Technology. Lewis has spent much of to develop specialized structures that Noting that variation, biologists have the past 40 years studying the bithorax indicate all eight abdominal segments often supposed that the molecular ar- complex, a small cluster of homeotic are replaced by thoracic segments. chitects of body formÑthe genetic pro- genes in the fruit ßy Drosophila mela- Most people would be unnerved by cesses that control embryonic develop- nogaster. The Greek word homeo means analogous birth defects in mammals, ment in diÝerent speciesÑwould also Òalike,Ó and the ßy homeotic genes are but these grotesque defects in ßies can be quite diverse. There is compelling so named because of their ability, when be observed with equanimity. evidence, however, that an interrelated mutated, to transform one body seg- group of genes, called HOM genes in ment of the fruit ßy into the likeness of rom his original genetic studies invertebrates and Hox genes in verte- another. Mutations in bithorax complex of the bithorax complex genes, brates, governs similar aspects of body genes usually cause such developmen- F Lewis derived two key insights. design in all animal embryos. tal defects in the posterior half of the The Þrst was that the normal function In at least some of the molecular sys- ßy body plan. Thomas C. Kaufman of of these homeotic genes is to assign Indiana University and his colleagues distinct spatial (or positional) identities have discovered and studied a second to cells in diÝerent regions along the cluster of ßy homeotic genes, the An- ßyÕs anterior-posterior axis. That is, C WILLIAM M GINNIS and MICHAEL tennapedia complex (named for the they ÒtellÓ cells that they are part of KUZIORA investigate the eÝects of the homeobox supergene family on animal founder gene of the complex, Antenna- the ßyÕs head or thorax or abdomen. development. McGinnis is professor of pedia). Mutations in these genes usual- These identities are to some extent ab- molecular biophysics and biochemistry, ly cause homeotic defects in the anteri- stract, in that the positional coordinates with an appointment in biology, at Yale or half of the ßy body plan. assigned by homeotic genes are inter- University. After earning his bachelorÕs It is often the case in biology that preted in dissimilar ways in diÝerent degree in biology from San Jose State bizarre defects in odd organisms con- developmental settings. Antennapedia University in 1978, he continued his tain the clues to solving important assigns thoracic identity during both studies in molecular biology at the Uni- versity of California, Berkeley. While problems, and few biological phenome- the embryonic and pupal stages of the working at the University of Basel dur- na are more bizarre than the disrup- ßyÕs life cycle, even though the struc- ing the mid-1980s, McGinnis and Mi- tions in body design caused by home- tures (sense organs, legs, wings and so chael Levine discovered the homeobox otic mutations. For example, some mu- on) that develop along the thorax diÝer sequence motif in genes of Drosophila tations in the Antennapedia gene can in larvae and adults. fruit flies. Kuziora has been assistant cause the antennae on the head of the LewisÕs second important insight was professor of biology at the University of fruit ßy to be transformed into an ex- that the linear order of the bithorax Pittsburgh since 1991. He received his tra pair of thoracic legs. Surprisingly, complex genes on the fruit ßyÕs chro- Ph.D. from the Baylor College of Medi- cine and worked with McGinnis for five some of the animals that develop the mosome exactly paralleled the order of years at Yale as a postdoctoral fellow. extra legs survive, feed and even mate the body regions they speciÞed along with normal ßies. the embryoÕs anterior-posterior axis.

58 SCIENTIFIC AMERICAN February 1994 Copyright 1994 Scientific American, Inc. ANTERIOR HEAD EMBRYOS of vertebrate animals as diverse as Þsh, sala- manders, birds, rabbits and humans show great similar- labial ities early in their development. Drosophila fruit ßies and other invertebrates develop along a very diÝerent path, yet at the earliest stages they and the vertebrates proboscipedia share a common pattern of expression of the so-called homeobox genes. That discovery reveals that despite the diÝerences in the Þnal appearance of the animals, Deformed they use closely related genes to specify parts of the body along the anterior-posterior (or head-tail) axis.

Antennapedia

Abdominal-B

POSTERIOR TAIL

DROSOPHILA FISH SALAMANDER CHICKEN RABBIT HUMAN

SCIENTIFIC AMERICAN February 1994 59 Copyright 1994 Scientific American, Inc. dia mutations in adult ßies are caused by activity of Antennapedia in the head, where that gene is normally turned oÝ. In summary, the genetic evidence indi- cates that each HOM complex gene is needed to specify the developmental fate of cells in a certain position on the anterior-posterior axis: the posterior head, anterior thorax and so on. More important (and more instructive about their biological function), the activity of HOM complex genes is apparently suÛcient to determine the fate of at least some cells, even when those cells would not normally fall under a given geneÕs inßuence.

he genes of the HOM complex are virtually the only ones in T Drosophila that have those prop- erties. They also share an interesting resemblance at the structural level be- cause all of them are members of the homeobox gene family. Homeoboxes are DNA sequences that carry the de- scriptions for making a related group of protein regions, all about 60Ðamino acid residues in size, called homeodo- mains. The homeo- preÞx in the name of these domains stems from their ini- tial discovery in Drosophila HOM pro- teins. Since then, however, homeodo- mains have been found in many other NORMAL FLY Antennapedia MUTANT FLY proteins with varying degrees of simi- larity. The homeodomains of the Dro- HOMEOTIC TRANSFORMATIONS, in which body parts develop in the wrong posi- sophila HOM proteins are especially tions, occur in fruit ßies that have mutations in their homeobox genes. Mutations similar to one another, which suggests of the Antennapedia gene, for example, can cause belts of thoracic denticles (spikes) to appear on the heads of larvae (top right). Another developmental con- they are closely related. For that rea- sequence of the mutation is that the mutant adults have legs growing in place of son, they are often referred to as An- antennae (bottom right). A normal larva and adult are shown at the left. tennapedia-class homeodomains. What do these HOM proteins do at the biochemical level? Only a super- The same relation also holds for the tivation overlap, but each HOM com- Þcial answer can be given at present. genes of the Antennapedia complex. plex gene has a unique anterior bound- They belong to a large group of pro- United by these shared characteristics, ary of activation in the body plan. teins whose function is to bind to DNA the genes in the bithorax and Antenna- If deletion of a gene or some similar in the regulatory elements of genes. pedia groupings are collectively re- incident interferes with the expression The right combination of these bound ferred to as the HOM complex. of a HOM protein, then embryonic cells proteins on a DNA regulatory element that normally contain high levels of will signal the activation or repression partial understanding of how the that protein often undergo a homeotic of a geneÑthat is, to start or stop mak- HOM complex genes determine transformation. That transformation ing that geneÕs encoded protein. Inves- A axial positions in the fruit-ßy occurs because of a backup HOM gene tigators have shown that the homeo- body plan can come from looking at that is already active in the same cells domain region of the HOM proteins is where those genes are active in em- and that can substitute its own posi- the part that directly interacts with the bryos. The HOM genes are present in tional information. For instance, if the DNA binding sites. the DNA of all of a ßyÕs cells but are ac- function of the Ultrabithorax gene is We are fascinated by the contrast be- tive only in some of them. When acti- eliminated from cells within a ßyÕs an- tween the structural similarity of the vated, the HOM complex genes are cop- terior abdominal region, Antennapedia HOM proteins and their varied, speciÞc ied as molecules of messenger RNA, will take over the development of that eÝects. Here is a family of proteins that which serve as templates for the syn- region. As a result, structures normally all bind to DNA and are presumably thesis of HOM proteins. During early associated only with the thorax (which derived from a single ancestral Anten- developmental stages, before regions Antennapedia helps to specify) also ap- napedia-class protein. Yet their roles in of the embryo show any signs of their pear more posteriorly. development are remarkably diverse: eventual fates, the diÝerent HOM com- Homeotic transformations can also one protein assigns cells to become plex genes are activated in successive result from mutations that cause a parts of the head, another assigns cells stripes of cells along the anterior-pos- homeotic gene to become active in an to become thorax and so on. It seems terior axis. Some of these stripes of ac- inappropriate position. The Antennape- likely that HOM proteins designate var-

60 SCIENTIFIC AMERICAN February 1994 Copyright 1994 Scientific American, Inc. ious positions along the anterior-poste- body, and those genes would produce protein and put an Ultrabithorax ho- rior axis by regulating the expression chimeric proteins at any stage of devel- meodomain in its place, the chimeric of what may be large groups of subor- opment if we simply raised the temper- protein lost the ability to regulate De- dinate genes. The functional speciÞcity ature of the ßiesÕ growth chamber to formed gene expression in embryos. In- of the HOM proteins can therefore be 37 degrees Celsius for a brief period. stead the new protein acted on the ex- deÞned by the diÝerences between (Drosophila prefer to live at 25 degrees pression of the Antennapedia geneÑ them that allow them selectively to reg- C but can tolerate 37 degrees C for an much as a normal Ultrabithorax pro- ulate certain genes in embryos. hour or two with no ill eÝects.) Using tein would. By transferring the Ultra- To learn more about this speciÞcity, these animals, we could assay the abili- bithorax homeodomain to Deformed, we decided in 1986 to construct chime- ty of the chimeric proteins to act on we had apparently also transferred its ric HOM proteins that had components the regulatory elements of target genes selective regulatory abilities. Another derived from diÝerent sources. (The in their normal chromosomal positions homeodomain swap experiment gave chimera, a monster of Greek mytholo- and in their natural embryonic environ- us similar results. A Deformed protein gy, was part lion, part goat and part mentÑa demanding test that closely carrying an Abdominal-B homeodo- snake.) By testing the function of these mimics the usual conditions under main instead of its own mimicked the chimeric proteins, we thought it would which these proteins operate. regulatory speciÞcity of an Abdominal- be possible to deÞne which subregions B protein. of the HOM proteins determined their ecause HOM proteins have high- The chimeric proteins did not behave selective regulatory abilities. ly similar homeodomains, they exactly like the protein from which For the subjects of our Þrst experi- B bind to nearly identical DNA sites their homeodomain was derived. Both ments, we chose the HOM proteins De- when tested in the laboratory. It there- the Deformed/Ultrabithorax chimera formed, Ultrabithorax and Abdominal- fore initially seemed likely that the fea- and the Deformed/Abdominal-B chi- B. These proteins have structurally sim- tures giving each protein its functional mera activated expression of their tar- ilar homeodomains: that of the De- speciÞcity would be found outside the get genes, whereas the normal Ultra- formed protein is identical to that of homeodomainÑin the parts of the pro- bithorax and Abdominal-B proteins re- Ultrabithorax protein at 44 of its 66 teins that were most individual. Yet as pressed expression of the same genes. amino acidsÑbut they share no exten- often happens when simple deductive Presumably, regions of the Deformed sive resemblance in other regions. Each reasoning is applied to biological prob- protein outside the homeodomain re- of these proteins also exerts an inßu- lems, that expectation was wrong. gion supply a strong activation func- ence on other genes in the HOM family. We found that if we removed the na- tion that can work with any of these Thus, the Deformed protein selectively tive homeodomain from a Deformed HOM homeodomains. Consistent with activates the expression of its own gene; Ultrabithorax protein represses DROSOPHILA EMBRYO the expression of the Antennapedia HEAD THORAX ABDOMEN gene; and Abdominal-B protein regu- lates its own gene and those of others in the HOM complex, including Anten- napedia, Ultrabithorax and abdominal- A. We knew we could use these auto- and cross-regulatory relationships in tests of the speciÞc functions of chi- meric HOM proteins. The Þrst challenge was to create genes that would make the chimeric homeotic proteins we desired. Recom- binant DNA techniques make that feat possible through the splicing of bits and pieces of genes at the DNA level. If lab pb Dfd Scr Antp Ubx abd-A Abd-B the gene engineering is done with care, protein domains can be very precisely DROSOPHILA HOM GENES moved from one protein to another while retaining their functional charac- B1 B2 B3 B4 B5 B6 B7 B8 B9 teristics. We then had to make sure that the chimeric genes would be active MOUSE HoxB GENES in all embryonic tissues. We therefore used a method worked out a few years A1 A2 A3 A4 A5 A6 A7 A9 previously by Gary Struhl, now at Co- lumbia University, that involves attach- MOUSE HoxA GENES ing the gene to regulatory DNA sequenc- D1 D3 D4 D8 D9 es that can be activated by a mild heat shock. Finally, we inserted our heat-in- HUMAN HOXD GENES ducible HOM gene chimeras into Dro- sophila chromosomes by a technique HOMEOBOX GENE COMPLEXES have been identiÞed in both invertebrates and ver- called P-element transformation. tebrates. Drosophila have HOM genes, which occupy the same order on the ßy chro- The Drosophila ßies that we trans- mosome as the anterior-to-posterior order of body regions whose development formed in this way thereafter carried they control. Mice and humans have Hox genes, which are closely related to mem- the chimeric genes in every cell of their bers of the HOM complex and show the same spatial and functional arrangement.

SCIENTIFIC AMERICAN February 1994 61 Copyright 1994 Scientific American, Inc. HOMEOBOX PROTEIN H2N COOH HOMEODOMAINS are the highly similar 60Ðamino acid regions of the proteins VARIABLE REGION HOMEODOMAIN made by all homeobox genes. Each let- ter in the consensus string represents an amino acid; deviations from that consensus are shown for several close- CONSENSUS ly related HOM and Hox proteins. RKRGRTTYT RYQTL EL EKEFHFNRYL TRRRR I E I A HAL CL TERQI KIWFQNRRMKWKKEN Labial NNS Ð Ð ÐNFÐ NKÐ L TÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð AÐ Ð Ð Ð Ð Ð NTÐQÐ NÐ T Ð VÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð QÐ Ð RV PGGL Ð ÐNFÐ TRÐ L TÐ Ð Ð Ð Ð Ð Ð Ð Ð KÐ Ð SÐ AÐ Ð VÐ Ð Ð ATÐGÐ NÐ T Ð VÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð QÐ Ð RE studied in frogs, mice and humans, HoxB1 where they are called Hox (short for ÒhomeoboxÓ) genes. In both mice and Deformed humans, Hox genes cluster into four PÐ Ð QÐ ÐA ÐÐ ÐHÐ I Ð Ð Ð Ð Ð Ð Ð Ð YÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð TÐ VÐ SÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð DÐ large complexes that reside on diÝer- PÐ Ð SÐ ÐA ÐÐ ÐQÐ VÐ Ð Ð Ð Ð Ð Ð Ð YÐ Ð Ð Ð Ð Ð Ð Ð Ð VÐ Ð Ð Ð Ð Ð Ð Ð SÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð DH ent chromosomes. In their organiza- HoxB4 tion and patterns of embryonic expres- Antennapedia sion, the genes of the Hox complexes Ð Ð Ð Ð ÐQÐ ÐÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð share intriguing likenesses to the genes Ð Ð Ð Ð ÐQÐ ÐÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð YÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð TÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð of the ßy HOM complex. For example, one can identify Hox genes that struc- HoxB7 turally resemble the HOM genes labial, Abdominal-B proboscipedia, Deformed, Antennape- VRKKÐ KPÐ SKFÐ Ð Ð Ð Ð Ð Ð Ð Ð L Ð ÐAÐ VSKQKÐWÐ L Ð RNÐQÐ Ð ÐÐ ÐVÐ Ð Ð Ð Ð Ð Ð Ð Ð ÐNÐ ÐNS dia and Abdominal-B. The equivalent SRKKÐ CPÐ Ð KÐ Ð Ð Ð Ð Ð Ð Ð Ð Ð L Ð ÐMÐ Ð Ð Ð DÐ ÐHÐ VÐ RL ÐNÐ SÐÐ ÐVÐ Ð Ð Ð Ð Ð Ð Ð Ð ÐMÐ ÐL Ð Hox and HOM genes are arranged in HoxB9 the same linear order within their re- spective complexes. this notion, Deformed does have a few too, support the idea that much (though A further parallel has been observed regions of protein sequence that are not all) of the functional speciÞcity of both by Denis Duboule of the Universi- rich in the types of amino acids charac- the HOM proteins resides in the small ty of Geneva and Pascal DollŽ at the teristic of Òactivation domainsÓ in oth- diÝerences within or immediately adja- CNRS Laboratory of Eukaryotic Molecu- er gene regulatory proteins. cent to the homeodomain regions. lar Genetics in Strasbourg and by Robb Similar experiments on the function- To us, all those Þndings also suggest- Krumlauf and his colleagues at the Na- al speciÞcity of HOM proteins have also ed that certain long-shot experiments tional Institute for Medical Research in been carried out by Richard Mann and already ongoing in our laboratory, for London. They have assembled convinc- David S. Hogness of Stanford University which we had only faint hopes of suc- ing evidence that the patterns of ex- and by Greg Gibson and Walter J. Gehr- cess, actually had a chance of yielding pression for the two types of genes are ing and their colleagues at the Univer- interpretable results. Those experiments alike. That is, the Hox genes are activat- sity of Basel. Their experiments were involved functional assays of mouse ed along the head-tail axis of the early based on evaluations of the homeotic and human homeodomain proteins in mouse embryo in the same relative transformations that mutant and chi- Drosophila embryos. To convey the sig- order that the HOM genes are activat- meric HOM proteins induced in devel- niÞcance of those tests, we need to re- ed on the anterior-posterior axis of oping ßies. Because they were looking view what is known about the mam- Drosophila. at the developmental eÝects of the HOM malian Hox genes. Structural similarities between the proteins rather than just at their ef- During the past nine years, genes that mouse and ßy proteins are mainly lim- fects on gene expression, those investi- contain Antennapedia-class homeobox- ited to the homeodomain regions. Fly gators were using a more demanding es have been found in the chromosomes Antennapedia and mouse HoxB6 are measure of HOM protein function than of many animal species besides Dro- nearly identical in the amino acid se- the one we applied. Yet their results, sophila. Such genes have been carefully quence of their respective homeodo-

EXPRESSION of the Deformed gene in ßy embryos, as re- however, will produce the Deformed protein in every cell of vealed by a brown dye, is normally conÞned to a band of cells their body after a brief exposure to heat (right). The devel- that become posterior head structures (left). Genetically en- opmental abnormalities found in such embryos can be used gineered embryos that carry heat-inducible Deformed genes, to infer HOM gene function.

64 SCIENTIFIC AMERICAN February 1994 Copyright 1994 Scientific American, Inc. mains (they diÝer at only four of 61 positions), which means that these two CHIMERIC Deformed/Ultrabithorax PROTEIN proteins resemble each other more than Antennapedia does any other ßy ACTIVATE HOM protein. In an evolutionary sense, this information argues that HoxB6 and Antennapedia are structural homo- loguesÑthat is, they descended from a common ancestral gene diÝerent from Deformed GENE Antennapedia GENE the one that gave rise to, say, Abdomi- nal-B or Deformed. ACTIVATE REPRESS By the same reasoning, the similarity between the entire HOM complex and the Hox complexes argues that the most Deformed PROTEIN Ultrabithorax PROTEIN recent common ancestor of Drosophila, mice and humansÑa wormlike creature GENETIC TARGETS of homeodomain proteins are largely determined by the ho- that lived about 700 million years ago, meodomain regions of those proteins. For example, a chimeric Deformed protein give or take a few hundred million carrying an Ultrabithorax homeodomain acts on the same genes as does Ultra- yearsÑhad a protocomplex of Anten- bithorax. Yet the regulatory eÝect of the chimeraÑactivationÑis more like that of napedia-class homeobox genes. The ex- Deformed because of protein regions outside the homeodomain. act type and arrangement of genes in that complex remain a mystery. Never- theless, we can be conÞdent, using the created strains of mice whose embryos whether Hox proteins can take the modern HOM and Hox complexes as produce HoxA7 protein in the head place of HOM proteins in developing guides, that the ancient protocomplex and anterior cervical region. Normally, Drosophila embryos. Ideally, one would contained structural homologues of HoxA7 protein (which is similar to the accomplish this swap by completely re- labial, proboscipedia, Deformed, Anten- Antennapedia and Ultrabithorax pro- placing the HOM gene of a ßy with its napedia and Abdominal-B. This overall teins of the HOM complex) is most Hox homologue; the Hox gene might view of HOM and Hox gene evolution is abundant in the posterior cervical and then be expressed only when and strongly supported by research on bee- anterior thoracic regions and is exclud- where the HOM gene would be ordinar- tle homeotic genes by Richard W. Bee- ed from more anterior parts. Some mice ily. Unfortunately, such an experiment man of the U.S. Department of Agricul- in which HoxA7 is expressed inappro- is not yet feasible, because the genes in ture and Rob E. Denell of Kansas State priately develop deformities of the ear their entirety are too big to be manipu- University and by recent reports from and palate and occasionally have ho- lated by current technology. Still, we many laboratories that the primitive meotic transformations of the cervical could do the next-best possible thing: roundworm Caenorhabditis elegans also vertebrae. by using Hox DNA sequences linked to has a HOM complex distantly but rec- The diÛcult converse experimentÑ heat-inducible regulatory elements, we ognizably related to the Drosophila knocking out Hox gene functionÑhas could make all the cells of a developing HOM and vertebrate Hox complexes. been accomplished for HoxA3 by Os- ßy express a Hox protein. amu Chisaka and Mario R. Capecchi of ll this structural evidence, though the University of Utah and for HoxA1 he Þrst protein that we and our suggestive, still does not directly by Thomas Lufkin and Pierre Chambon laboratory colleague Nadine Mc- A tell us whether HOM and Hox and their collaborators at CNRS in T Ginnis tested in this way was the proteins do serve the same develop- Strasbourg. Their work has shown that human HOXD4 protein, the equivalent mental function in embryos. After all, some structures in the anterior regions of a mouse HoxD4. (When referring the mouse and ßy gene complexes have of mouse embryos do depend on those speciÞcally to human genes, the HOX been in diÝerent evolutionary lineages genes. Mutation of the HoxA3 gene re- label is capitalized to conform with for hundreds of millions of years, with sults in mice that die just after birth standard genetic nomenclature.) The plenty of time to evolve new or diver- with a complicated set of head and gene for this human protein, which has gent abilities. So the similarities in struc- neck deformities, including abnormally a homeodomain like that of the ßy De- ture and expression might be histori- shaped bones in the inner ear and face formed protein, was isolated and char- cal quirks and not trustworthy indica- and the absence of a thymus. acterized in 1986 by Fulvio Mavilio and tors of functional resemblance between Such deformities are reminiscent of Edoardo Boncinelli and their colleagues present-day HOM and Hox proteins. a human congenital disorder called Di- at the Institute for Genetics and Bio- One approach to the problem is to GeorgeÕs syndrome, raising hopes that physics in Naples. explore the biological eÝects of Hox the study of HOM and Hox genes will In Drosophila, when the Deformed genes in vertebrate embryos and to be of practical beneÞt in explaining gene is expressed outside its normal compare them with what is known some human birth defects. Much more anterior-posterior limits, the adult ßies about the eÝects of HOM genes in in- research needs to be done before biolo- suÝer a variety of head abnormalities, vertebrates. For instance, does the in- gists have a good understanding of how such as the absence of a ventral eye. appropriate activation or speciÞc inhi- Hox genes participate in the develop- We were amazed to Þnd that the hu- bition of Hox gene function during mental design of mice and humans, but man HOXD4 protein, when expressed mouse development cause homeotic these and other initial experiments cer- in developing ßy cells, caused the same transformations? In one eÝort to an- tainly suggest that the Hox and HOM deformities. We could not attribute swer this question, Peter Gruss and his genes serve comparable purposes. these changes entirely to the human colleagues at the Max Planck Institute In our own work, we have tried to protein, however : our experiments in- for Biophysical Chemistry in Gšttingen make a direct comparison by testing dicated that the human protein was

SCIENTIFIC AMERICAN February 1994 65 Copyright 1994 Scientific American, Inc. hark back to the classical observations of Karl Ernst von Baer, who in the 1820s concluded that if one examined early embryonic morphologies, all ver- tebrate forms seemed to converge to- ward a common design. The story, which sounds too good to be true, is that von Baer came to this epiphany af- ter the labels fell oÝ some of his bot- tled specimens of early embryos, and he realized with some chagrin that he could not be sure whether the embryos were lizards, birds or mammals. The structure and function of the HOM and Hox gene systems suggest that this de- velopmental convergence embraces the early development of a great many ani- mal species. But only at the level of molecular pattern can the developmen- tal convergence of such diÝerent em- bryos be Òseen.Ó Sometime between 600 million and a billion years ago, the HOM/Hox system evolved; it has proved so useful that many animals have since relied on its NORMAL FLY Deformed MUTANT FLY fundamental abilities to determine axi- DEFORMED MUTANT FLIES have a variety of head abnormalities, including the ab- al position during development. Is it sence of the lower half of the compound eyes. The same abnormalities can be in- the only developmental genetic system duced by making the immature ßies express the human HOXD4 protein. This hu- that has been so conserved? That seems man protein resembles the Deformed protein and seems to have a similar function. unlikely. Researchers have found hints that some other regulatory genes in ßies and mice are highly similar in promoting the expression of the ßyÕs adult homeotic transformations were structure and are activated in the same Deformed gene as well. (Remember much like those caused by the inappro- or homologous tissues. Exploring the that one normal eÝect of the Deformed priate expression of Antennapedia pro- functions of those novel genes, and protein is that it activates its own gene, tein throughout the body. how they interact with the HOM/Hox in a cycle of positive feedback.) The hu- system, promises to reveal many more man HOXD4 protein was therefore hat can one make of these evo- fascinating insights into the evolution mimicking the eÝects of inappropriate lutionary swap experiments? and mechanism of the ancient genetic Deformed expression becauseÑat least W First of all, they reinforced our systems that serve as the molecular ar- in partÑit was causing inappropriate conclusions that the homeodomains chitects of animal body plans. Deformed expression. Nevertheless, we themselves determined much of the could see that HOXD4 did act like a regulatory speciÞcity of the proteins: weak but speciÞc replica of its Drosoph- the homologous ßy and vertebrate pro- ila homologue. teins have little in common outside the FURTHER READING Encouraged by this result, Jarema homeodomain region. In addition, the HOMEOBOX GENES AND THE VERTEBRATE Malicki, a graduate student in our lab- experiments suggested that from a BODY PLAN. Eddy M. De Robertis, oratory, tested the function of the functional standpoint, the homologous Guillermo Oliver and Christopher V. E. mouse HoxB6 protein in developing proteins are at least somewhat inter- Wright in Scientific American, Vol. 263, No. 1, pages 46Ð52; July 1990. ßies. HoxB6, which Klaus Schughart and changeable and have similar Òmean- MOUSE HOX -2.2 SPECIFIES THORACIC Frank H. Ruddle identiÞed and charac- ingsÓ for early embryos. The system for SEGMENTAL IDENTITY IN DROSOPHILA terized a few ßoors away from us at determining anterior-posterior axial po- EMBRYOS AND LARVAE. Jarema Malicki, Yale University, has a homeodomain sitions has evidently changed little in Klaus Schughart and William McGinnis that is highly similar to Antennapedia the past 700 million years. in Cell, Vol. 63, No. 5, pages 961Ð967; protein. The eÝects of HoxB6 protein If one were to imagine the complicat- November 30, 1990. expression in developing ßy cells was ed network of interactions between HOMEOBOX GENES AND AXIAL PATTERN- ING. William McGinnis and Robb Krum- spectacular and unmistakably homeot- gene regulatory proteins inside an or- lauf in Cell, Vol. 68, No. 2, pages 283Ð ic. In Drosophila larvae the HoxB6 pro- ganism as a jigsaw puzzle, then the ho- 302; January 24, 1992. tein caused much of the head region to mologous ßy and mammal proteins are THE MAKING OF A FLY: THE GENETICS OF develop as if it were thoracic: instead pieces that can Þt in the same places. ANIMAL DESIGN. Peter A. Lawrence. of a larval head skeleton, the trans- Looking at the HOM/Hox system in this Blackwell Scientific Publications, 1992. formed ßies produced denticle belts, way also highlights how much we still THE MOUSE HOX -1.3 GENE IS FUNCTION- rows of spikes that are usually ar- have to learn: the other puzzle pieces ALLY EQUIVALENT TO THE DROSOPHILA ranged on the bellies of Drosophila. In that enable the HOM and Hox proteins SEX COMBS REDUCED GENE. J. J. Zhao, R. A. Lazzarini and L. Pick in Genes and Drosophila adults, HoxB6 caused a ho- to regulate genes and to have a speciÞc Development, Vol. 7, No. 3, pages 343Ð meotic transformation of the antennae function have yet to be identiÞed. 354; March 1993. into thoracic legs. Both the larval and In a way, these experiments also

66 SCIENTIFIC AMERICAN February 1994 Copyright 1994 Scientific American, Inc.