CROTCHETS & QUIDDITIES

Is the Medium the Message? Biological Traits andTheir RegulationKENNETH WEISS

We live in the information age. In the1960s when Marshall McLuhan saidthe medium is the message, he meantthat in our electronic times the me-dium itself was an important carrierof our total cultural message. The waywe acquire information may affect usmore than the information itself.What sometimes matters most is notthe product, but getting the product tomarket, and what is done to get itthere may bear little resemblance tothe product itself. The purpose of ad-vertising is to manipulate the individ-ual, McLuhan said, making us want tobecome what we behold. He used au-tomobile marketing tactics as an ex-ample, and I had my own ’60s experi-ence of that. Single, young, and frisky,I was lured into buying the latestmodel little red sports car, lured I’msure by beautiful women in the ads(Figure 1, model not shown). I wassure I’d get the girls!

Science, too, has become a com-modity to be sold via the media. Butthere is another sense in which biol-ogy and culture converge. Evidence isaccumulating that a comparably largefraction of biological activity itself isused simply to bring phenotypes to

the evolutionary competitive market.This fact may affect our perception ofwhat’s going on in evolution, and howhighly specific traits—such as those ofcentral interest to biological anthro-pology—are produced by surprisinglyarbitrary mechanisms.

Thousands of proteins are used incomplex ways in the 4-dimensions ofspace and time even within a singlecell, all specified by a universal, com-pact DNA coding system. The codemetaphor may be overstated (Kay,2000), but information is certainlystored in DNA sequences so that whilea cell or organism may die, its DNAmolecules are transmitted from gen-eration to generation. The programoutlives the computer. Indeed, we of-ten hear the genome described as theprogram for “computing” the organ-ism.

Like C�� or Basic, the genetic sys-tem is a true code in that there is nodirect functional connection betweencoding mechanism and protein struc-ture. Nothing about the nucleotidetriplets that code for the amino acidthreonine relates to the chemicalproperties of threonine. Nor does thetriplet correspond to the traits inwhich it’s used. The same nucleotidetriplets code for the same set of aminoacids, whether the protein makes hairor bone or leaves, binds oxygen, pro-duces color, responds to light, or givescell walls their rigidity. Nor has it any-thing to do with being human, oak, orslime mold. The coding system is log-ically necessary, because we need tomake (say) hemoglobin, but is func-tionally arbitrary, because it doesn’t

matter what codes for it. For example,four different triplet codons specifythreonine (ACT ACC, ACA, ACG), buta protein that needs a threonine,doesn’t care which codon is used tospecify the threonine.

The separation of coding from func-tion allows the coding system to workquietly in the background, so evolu-tionary biologists can concentrate onwhat has traditionally been viewed asthe real aspects of life—morphology,eating, escaping, breathing, flying,and what we do with people attractedby our sports cars—without having toworry directly about how the geneproducts related to those traits weremade. Natural selection may screenproteins for a threonine in some par-ticular position, but it doesn’t carehow it got there.

How genes code for proteins is wellunderstood, but what makes multicel-lular life possible is differentiation.This has to do with when a gene spec-ifies a protein and which genes areexpressed in a given context. The tim-ing and context of gene expression in-volve an additional coding infrastruc-ture, also embedded within the DNAsequence but working in a very differ-ent way. Gene regulation greatly in-creases the amount of functionally ar-bitrary biological activity that goes onin an organism and raises some inter-esting evolutionary problems.

DIFFERENTIATION ANDCOMPLEX ORGANISMS

It may not be appreciated how im-portant the context-specific use ofsubsets of genes is to life. Makingmuscle and making hair are donefrom the same genome contained inthe organism’s cells. An organism pro-

Ken Weiss is Evan Pugh Professor of An-thropology and Genetics at Penn StateUniversity.

Evolutionary Anthropology 11:88–93 (2002)DOI 10.1002/evan.10028Published online in Wiley InterScience(www.interscience.wiley.com).

Much about the nature of biological processes is logically necessary but func-tionally arbitrary. What does this mean about biology?

88 Evolutionary Anthropology

duces multiple tissues and organs viaa developmental tree of descent froma single initial cell (fertilized egg) dur-ing its lifetime. But how is selectivegene expression controlled?

There are nearly countless ways inwhich the cells, tissues, and organ-isms differentiate, but the logic of theprocess is shared. As illustrated sche-matically in Figure 2, gene regulationinvolves short sequences, called recog-nition sequences (referred to as “pro-moters” or “enhancers”), in the DNAthat flanks the protein-coding regionof a gene. A gene is expressed when itsenhancer sequences are physicallybound by proteins called transcriptionfactors (TFs). A TF is a protein codedfor by its own gene somewhere in thegenome, whose structure leads it tobind to a specific enhancer sequence,because the chemical “shape” of theTF molecule fits the “shape” of DNAwith that sequence. Like the “Drinkme!” tags on growth-potion bottles inAlice in Wonderland, enhancers are agene’s “Bind me!” tags for specificTFs. An enhancer is a sequence-basedcode in DNA, but one that carries in-formation related to gene expressionrather than protein coding.

A common way these elements areused to activate a “target” gene isshown in Figure 3. A cell producesreceptor molecules specific to its par-ticular function. The receptor residesin the cell membrane, with an extra-cellular part protruding out of the cell,that “looks” for a circulating signalingfactor molecule that fits the extracel-lular binding region. The signalingfactors are produced and secreted by

other cells, and move through the ex-tracellular space “looking for” a par-ticular receptor. When the signalingfactor finds a cell that expresses thereceptor of its desires, the two bindtogether in Holy Regulamony. Thisevent alters the receptor’s intracellu-lar part, which triggers a cascade ofinteraction among second messengermolecules that, like hand grenades,wait around inside the cell for theirpins to be pulled. Eventually this cas-cade activates TFs that themselves arealso primed and present in the cell,which then bind enhancers upstreamof target genes, causing the latter to beexpressed.

A dizzying hierarchy of such activ-ity is required before a cell can be-come a skin or nerve cell (e.g., Figure4). It is the combination of a numberof specific TFs that occurs in a given

cell that determines what it does. Agene must be flanked by the appropri-ate set of enhancer sequences for eachcontext in which it’s used. But ofcourse, that can only happen in cellsthat already “know” to produce theright set of TFs, secondary messen-gers, or receptors. These are all codedfor by other genes, which means thatearlier in the cell’s developmental an-cestry similar mechanisms have beenused to anticipate its later function.Differentiation is thus hierarchical indevelopmental time. Like the prover-bial fleas, a regulatory cascade has an-other cascade upon its back to expressit, and so ad infinitum—back to thefertilized egg.

FUNCTIONALLY ARBITRARYMECHANISMS ARE UBIQUITOUS

AND EXTENSIVE

The extent of functionally arbitraryregulatory mechanisms in any organ-ism is vast. There are many families ofTF, receptor, signaling factor, andmessage-transduction genes in our ge-nome. Each family can have tens ofmembers—different genes descendedfrom a common ancestral gene by ahistory of gene duplication events.And there’s more. Many if not mostgenes are alternatively spliced, that is,use different combinations of theirprotein coding regions (exons), in dif-ferent situations; this requires specificsplicing mechanisms, coded for byother genes, for each such instance.And yet more: the genome providescode for a whole world of RNA mole-

Figure 1. The medium: Ad for a 1963 Triumph TR4. (Photo courtesy Brian Sanborn. (http://www.net1plus.com/users/sanborn/).

Figure 2. Schematic of information-holding regions of DNA in and near a gene. Shapessymbolize different functional sequence elements along the DNA. The intervening se-quence is shown as a line.

CROTCHETS & QUIDDITIES Evolutionary Anthropology 89

cules that have biological functions oftheir own, based on the way they foldup and interact with other things inthe cell. These include the familiartransfer RNA and ribosomal RNA. Butmore: part of this RNA activity has todo with other sequences scattered inmany places in the genome, called im-printing centers, that regulate the ex-pression of nearby genes by modify-ing their enhancer regions.

There’s another strange new worldof RNA molecules, also coded for inthe genome, whose function appearsto be to bind to specific mRNA mole-cules to regulate their translation intoprotein, or to degrade them rapidly soas to control delicate developmentaltiming switches. Such antisense genesare common in the genome, and thereare also mechanisms to activate anti-sense-RNAs at context-specific times.Also, the 3� end of mRNAs can includesequence signals that help localizetheir coded protein within a cell, us-ing microtubule machinery as aguide, to control its translation intoprotein, or for other functions.

And did I mention that the packagingof DNA by histone proteins also relatesto DNA sequence signals and requireshistone producing and processinggenes? But that’s still not all! A protein

that is finally coded by a gene typicallyhas to be molded, combined, protected,transported, cleaved, folded, spindled,or mutilated (are you old enough to re-member that McLuhan-age phrase?)before it becomes functional. Each ofthese processing functions is achievedby gene products, and there are many

others whose job is to clean up the messof misformed proteins that error-pronemechanisms leave behind. These genes,of course, require their own regulatorycascades.

The basic mechanisms probablywere already present in the earliestcells. Many DNA binding regions for,or interaction pathways among spe-cific regulatory factors seem to bedeeply conserved phylogenetically.These “circuits” comprise a tool kitused in so many ways that it’s not tobe tinkered with. You may vary whereyou use a screw, but you use the samescrewdriver. There is phylogeneticconservation in regulatory pathwaysassociated with many basic functions,like polarity in an early embryo, pho-toreception, or the differentiation ofneural, muscular, heart and other ba-sic cell types. This can have importanteffects on how we view evolution.

RECONSTRUCTION: THEANCIENT WORLD

Since the 1800s, biologists have re-constructed ancestral body plans forgroups of animals (e.g., the first chor-date) by building an hypothetical Coo-tie with all the shared elements, likefour limbs, two eyes, kidneys, mus-cles, heart, etc. But this may bewrong. Rather than an assembly of

Figure 4. A hierarchy of circuits contributing to activate a set of functional genes.

Figure 3. Schematic illustration of the cascade of receptor-mediated induction of geneexpression.

90 Evolutionary Anthropology CROTCHETS & QUIDDITIES

ready-made proto-organs, such ances-tors may be better reconstructed ashaving a set of primitive regulatorycircuits for the basic functions repre-sented in those organs (see Davidson,2001). Fly, octopus, and human eyedevelopment share the use of a ho-mologous pax6 regulatory gene circuitfor photoreceptors, but their actualeyes are so different that for centuriesthey were used as exemplars of anal-ogy—independent convergent evolu-tion. But the regulatory circuits revealan underlying element of what may betrue homology. We have viewed theworld in terms of macroscopic pheno-types, when it may be more useful tothink of the regulatory types thatmake them possible. Who knows whatsuch an UrCootie may have actuallylooked like!

These notions have reassuring im-plications for the evolution of com-plex organisms. Once evolved, the reg-ulatory toolkit remained available foruse. Because enhancer sequences areonly a few base pairs long (e.g., TTT-TATGG), they can arise even in ran-dom sequences flanking a gene by justa few mutational changes. This would“attract” TFs to that gene, perhaps toactivate it in some new context. Sincea gene is only expressed when a par-ticular combination of TFs binds itsnearby DNA, there may be little costto littering the genome with enhancersequences. It may be easier to evolvenew function by changing expressionpatterns of existing genes than to re-quire an entirely new gene. A cost ofdoing business this way, however, isthe high fraction of an organism’s en-ergy, and the information in its ge-nome that is required. Yet it is thefinal traits rather than the way theyget delivered that have traditionallyattracted the attention of evolutionarybiologists, are used in comparativeand cladistic studies, are found in thefossil record, and appeared to show ushow organisms evolved adaptively.Obviously, the cost was worth it.

Where is the Reality? In thePhenotype?

Each function in an organism re-quires its own subset of regulatory cir-cuits, and since the toolkit is limitedand the factors diverse, this means

that evolution has produced an ex-quisitely complex buzz of regulatoryactivities within, and differencesamong, cells. These achieve a high de-gree of specificity (fingers rarely con-tain eyes or teeth!). But surprisingly,this logically necessary toolkit of reg-ulatory mechanisms is functionallyarbitrary in the same way that proteincoding is.

As an example, TTTTATGG is anenhancer sequence that is bound by aTF called cdx1. Experiments in mousehave shown that binding of the cdx1TF protein to its enhancer near thehoxc8 gene induces caudal (tail-end)development in the early embryo, bytriggering a cascade of caudal-struc-ture-related development (Fig. 4). Butthere is nothing about TTTTATGGthat is chemically related to the struc-ture of vertebrae, or tails—or evenmice when it comes to it, because thesame hoxc8 gene is used in many con-texts within the same animal and inmany different species, including youand me, and even flies. Nor is any-thing about the cdx1 protein, that in-duces hoxc8, physically related to ver-tebrae, tails, or mice. And nothingabout hoxc8 is physically related tovertebrae, or tails, or mice. The onlything about the cdx1 3 TTTTATGG3 hoxc8 regulatory cascade that is re-lated to anything caudal is that the cas-cade occurs in appropriately primedcells in the caudal end of the embryo.

The arbitrariness of these mecha-nisms can be seen in the fact that it’sbecoming possible to identify gene ex-pression patterns by screening DNAsequences on a computer to find clus-ters of known motifs (like TTT-TATGG). That reveals genes regulatedby the corresponding TF without re-quiring any knowledge whatsoever offunction (e.g., Michelson, 2002).

A mechanism that is functionallyarbitrary is replaceable, and there aremany examples of substitution of onesuch tool for another. Different spe-cies may use partly or entirely differ-ent regulatory mechanisms to achievea trait they have shared since theircommon ancestor. Even genes ex-pressed in the same cell at the sametime can use entirely different regula-tory pathways, even if those genes aredescendants of a common ancestralgene and their usage has been con-

served since that time. The joint ex-pression in red blood cells of the twogenes required to produce hemoglo-bin (alpha and beta globin) is an ex-ample.

It is not clear how often one regula-tory circuit has become replaced byanother in this way. But we do knowthat we can’t characterize all the genesexpressed in a given context by thesame set of enhancer elements. Inwhatever way it happened, over evo-lutionary time and the complex his-tory of cellular differentiation duringdevelopment, the expression (or re-pression) of the thousands of genesused in a given cell is controlled bymany different pathways. But onething these share is that they are arbi-trary relative to the functional charac-teristics of the cell.

In fact, there may be more interest-ing aspects of the evolution of generegulation. The hundreds of interact-ing genes and enhancer sequenceshave natural variation just as anystructural genes or traits do. This vari-ation can affect binding efficiency,speed or duration of gene action,strength of expression, and so on.Much of the quantitative variation wesee in phenotypes in natural popula-tions is probably related to that regu-latory variation. However, there aredistinctions between the evolution ofregulatory mechanisms and “final”phenotypes like limbs and hair.

The use of the same regulatory cir-cuit in many different ways within thesame organism (sometimes even inmultiple ways during the evolution ofa given organ) is a kind of pleiotropythat exposes the circuit to potentiallyheavy natural selection in many con-texts that may be vital to the organ-ism’s survival. The more numerousthese contexts the more opportunitythere is for problems due to mutationsin the circuit’s enhancers. That multi-ple jeopardy could be a brutal way tokeep an embryo honest, and may alsoexplain the high degree of evolution-ary conservation found in regulatorycircuits (TFs or binding factors canhave similar action when experimen-tally engineered into distant species,like between flies and mice). But theremay be countervailing selection.

If hoxc8 is used in the developmentof (say) 20 tissues, the organism could

CROTCHETS & QUIDDITIES Evolutionary Anthropology 91

fail due to mutations in any of therequired TTTTATGG’s. A form of pro-tection that may evolve could be in-creases in the number of copies of anenhancer near each regulated gene, ormore variation in the enhancers’ se-quence (that is, the TF evolves to re-quire less precision in the enhancersequences it binds to). If this is a com-mon evolutionary buffering response,we should find a correlation betweenthe number of contexts in which a TFis used and the number of copies of itsenhancers near the genes it regulatesand/or variation in the enhancer se-quence. A recent study has found justsuch evidence in a screen of a bacte-rial genome (Sengupta et al., 2002).Note that here the “trait” being se-lected is the joint set of variant TTT-TATGG’s across the genome.

Such redundancy may buffer the or-ganism against mutation, but that maytether strange sets of straits together inthe evolutionary struggle. Redundancyor slippage in enhancers may make itharder for these traits to evolve inde-pendently, because it may be more dif-ficult for evolution to remove or modifya context by mutating the enhancers.Traits may thus coevolve because theyare under joint constraints that havenothing to do with the function of traitsthemselves (flat teeth, pointy ears, longtail, . . . ). This could lead to otherwise

curious suites of apparent coadaptationthat biologists from Cuvier to cladistshave gone to great lengths to explainfunctionally. Ironically, such coevolu-tion may reflect highly specific selec-tion—but for a set of functionally arbi-trary enhancers. That would be abizarre twist in the nature of “adapta-tion” indeed!

SO WHAT IS THE MESSAGE?WHAT IS THE MEDIUM?

A lot has been going on behind themorphological scenes over the long his-

tory of evolution. We’ve been blithelyconcentrating on morphological phe-notypes all this time, as if they were thereal business end of life. But at the mo-lecular level, the induction of one activ-ity by another through protein-proteinor protein-nucleic acid binding musthave long preceded structural featuresthat arose only in the last fraction of thehistory of life since cells evolved.

If regulation is the primary activityof life, what then is form? To rephrasethe chicken and egg quip, maybe aperson is a regulatory circuit’s way ofmaking another regulatory circuit. Atleast, it is gene regulation that turns afertilized egg into an organism thatcan make another egg, and the turn-ing is probably more complex a jobthan maintaining a body once it’sbeen produced. Most of the informa-tion in the genome may be consumedin early stages of embryogenesis. Thisis ironic, because from the earlyLamarckian incarnation of biologicalevolution, through Darwin and his ac-olyte Ernst Haeckel and many others,the notion of terminal addition hashad a lot of importance in biology.

The idea of terminal addition is thatby one means or another, natural selec-tion adds modifications onto traitspresent in adults to render them moresuited to their environments. This wasthe basis of many variations of Haeck-el’s famous biogenetic law that ontog-eny recapitulates phylogeny, becausewith terminal addition embryos reallywould reflect adult stages of ancestralspecies. Key to this was the view that

Figure 5. Rough distribution of functional types of human genes.

Figure 6. The message: Fix me! Junk me!

92 Evolutionary Anthropology CROTCHETS & QUIDDITIES

selection doesn’t act on embryos. Thisview is implicitly still widespread inpractice, even if every biologist knowsbetter and would say so. Much of thepervasive regulatory activity that I’vebeen discussing occurs and is likely tobe selected at embryologic as well asadult stages of life. But keep in mindthat what is being selected is the action,not the nature, of functionally arbitrarygenes.

Recent analysis of the entire ge-nomes of humans and other speciesshow that perhaps fewer than 10% ofall genes are concerned with the struc-tural aspects of the phenotypes wehold so near and dear (Figure 5). Kingand Wilson (1975) long ago specu-lated that the apparently great pheno-typic differences between chimps andhumans, who differ only slightly atthe gene level, might be due mainly toslight differences in the regulation ofdevelopmental timing. This notionseems ever more likely to be truerather than just speculative.

At the time, McLuhan’s phrase thatthe medium is the message appealedto a lot of people, but to me he seemedjust to be playing cute word gameswith our commercialized culture. It’strue that selling a product often usesarbitrary tactics having little if any-

thing to do with the product itself.McLuhan’s notion may also be highlyrelevant to biology. Phenotypes andthe regulatory networks that makethem are connected, perhaps like adsand cars, but their arbitrary natureand substitutability allows traits andregulation to go their separate waysover evolutionary time. While biolo-gists have been paying most of theirattention to the end products, manip-ulating the organism to get its prod-ucts to market may be the main en-gine of evolution.

Speaking of engines, I certainly feltmanipulated by advertising (Fig. 6).My ‘60s experience was no triumph,because it was a real Triumph: nogirls, and after a few thousandmiles—no engine, either! If you everowned one, you’ll know what I mean.

NOTES

I would welcome comments on thiscolumn: kmw4@psu.edu

I thank Anne Buchanan for criticallyreading and purging this manuscript.

TO READ

Most things discussed here includ-ing basic descriptions of the DNA-

RNA-protein coding system can beeasily explored by web searching.Carroll SB, Grenier JK, Weatherbee SD. 2001.From DNA to diversity: molecular genetics andthe evolution of animal design. Malden, MA:Blackwell Science.

Davidson, Eric H. 2001. Genomic regulatory sys-tems development and evolution. San Diego, Ac-ademic Press.

Hall BK. 1999. Evolutionary developmental biol-ogy. Dordrecht, Kluwer Academic.

Kay L. 2000. Who wrote the book of life? A his-tory of the genetic code. Palo Alto, StanfordPress.

King M-C, Wilson A. 1975. Evolution at two lev-els in humans and chimpanzees. Science 188:107–16.

Michelson A. 2002. Deciphering genetic regula-tory codes: A challenge for functional genomics.Proc Nat Acad Sci USA 99:546–548.

Sengupta A, Djordjevic M, Shraiman B. 2002.Specificity and robustness in transcription con-trol networks. Proc Nat Acad Sci USA 99:2072–2077.

Shashikant C, Bieberich C, Belting H-G, Wang J,Borbely M, Ruddle F. 1995. Regulation of Hoxc-8during mouse embryonic development: identifi-cation and characterization of critical elementsin early neural tube expression. Development121:4339–4347.

Sownward J. 2001. The ins and outs of signaling.Nature 411:759–762.

Weiss K, Fullerton SM. 2000. Phenogenetic driftand the evolution of genotype-phenotype rela-tionships. Theor Pop Biol 57:187–195.

Wolpert L, Beddinton R, Brockes J, Jessell T,Lawrence P, Meyerowitz E. 1998. Principles ofdevelopment. Oxford, Current Biology.

© 2002 Wiley-Liss, Inc.

Books Received

• Peacock, J.L. (2002) The Anthro-pological Lens: Harsh Light, SoftFocus. Xvii � 156pp. New York:Cambridge University Press.ISBN 0-521-80838-3 (cloth)$55.00.

• Marks, K. (2002) What It Meansto Be 98% Chimpanzee: Apes,People, and Their Genes. 325pp.Berkeley: University of Califor-nia Press. ISBN 0-520-22615-1(cloth) $27.50.

• Schwartz, J.H. and Tattersall, I.(2002) The Human Fossil Record:Volume One—Terminology andCraniodental Morphology of Ge-nus Homo (Europe). Xi � 388pp.New York: Wiley-Liss. ISBN0-471-31927-9 (cloth) $125.00.

• Salzano, F.M. and Bortolino,M.C. (2002) The Evolution andGenetics of Latin American Pop-ulations. xvi � 512pp. NewYork: Cambridge UniversityPress. ISBN 0-521-65275-8(cloth) $90.00.

• Calvin, W.H. (2002) A Brain ForAll Seasons. Human Evolutionand Abrupt Climate Change.341pp. Chicago: University ofChicago Press. ISBN 0-226-09201-1 (cloth) $25.00.

• Larsen, C.S. (2000) Skeletons inOur Closet: Revealing Our PastThrough Bioarchaeology. xvii �248pp. New Jersey: PrincetonUniversity Press. ISBN 0-691-09284-2 (paper) $18.95.

• Clark, J.D. (2001) Kalambo FallsPrehistoric Site. Volume III. xix� 701pp. New York: CambridgeUniversity Press. ISBN 0-521-20071-7 (cloth) $375.00.

• Leonard, W.R. and Crawford,M.H. (eds.) (2002) Human Biol-ogy of Pastoral Populations. xi �314pp. New York: CambridgeUniversity Press. ISBN 0-521-78016-0 (cloth) $80.00.

• Beckerman, S. and Valentine, P.(eds.) (2002) Cultures of MultipleFathers: The Theory and Practiceof Partible Paternity in LowlandSouth America. vii � 291pp.Gainesville: University of Flor-ida Press. ISBN 0-8130-2456-0(cloth) $59.95.

CROTCHETS & QUIDDITIES Evolutionary Anthropology 93