Steven A. Benner

The seventeenth-century Enlightenmentgenerated two very different approach-es to thinking about the natural world.

The first sought mathematical models toaccount for observations, beginning with the motion of the planets in the night sky.This approach generated ‘physical science’, aweb of mathematical and physical modelsdescribing and predicting the behaviour ofunits of matter smaller than the atom, largerthan the planets, and much in between.

The second approach emphasized col-lecting and classification. The archetypalpractitioner was the English gentleman, whonamed plants, rocks and fossils on his estate,then in his empire. From this enterprisecame ‘natural history’, a web of classificationsystems, non-reductive models and histori-cal statements about the planet and the life itcarries. The positions of continents came tobe understood as outcomes of historicalmovements of plates on the globe, obeyingphysical laws but too complicated to be pre-dicted by them. Life is also understood as theconsequence of historical events, occurringwithin the context of darwinian theory,again consistent with physical laws but notpredicted by them.

It is curious that these two very differentactivities receive the same name: ‘science’.This has generated much confusion, andsome rancour, with each group feeling itselfto be the more ‘relevant’. But the disjunctionbetween the two traditions is (I believe) aboutto end, and in a largely unheralded way.

Many chemical biologists and biophysi-cists view the future of biology as a metamor-phosis in which ‘biological’ phenomena willbe replaced by their underlying ‘physicalchemical’ components. This metamorpho-sis is ongoing, and very productive, but isunlikely to be the entire story. The surprisewill come when biophysicists and chemical

biologists discover that they need to researchthe history of biomolecules if they are truly tounderstand the physical behaviours that theyhave worked so hard to characterize.

An early sign of the need for natural histo-ry in the physical sciences was the struggle to predict how proteins fold. This seemed to be a good area for applying the physical-science paradigm to biology, so physicalchemists mounted a frontal assault, usinghuge computers to build physical models ofproteins in water and making guesses abouthow atoms interact. The assault failed. Theonly way to make the computation evenvaguely tractable, it turned out, required con-siderable abstraction of the physical modelfor the protein. The same physical theory thatinspired the computation suggested thatthese abstractions must compromise thecomputation’s value as a predictive tool.

Natural history offered an entirely differ-ent approach. Divergent evolution createsfamilies of proteins that have descendedfrom common ancestors. As proteins evolvefrom those ancestors, natural selectionrequires them to remain ‘fit’. The principalprerequisite for fitness in a protein is a fold.So proteins diverging from a common ances-tor generally conserve their folds. Thismeans that during the evolution of proteinsequences, mutations do not accumulate asthey would if proteins were formless, func-tionless organic molecules. Instead, aminoacids that are important to the fold suffersubstitution differently from those that arenot. A signal should lie in the pattern of pro-tein-sequence divergence — the differencebetween how proteins have divergentlyevolved in their past, and how they wouldhave evolved had they been formless, func-tionless molecules.

Natural history did not overcome thechallenges obstructing the frontal assault ofthe physical scientists; it went around them.Today, the secondary and tertiary structure ofproteins can reliably be predicted by exploit-ing the historical signal embedded in a set ofprotein sequences related by common ances-try. Since 1990, about 30 protein folds havebeen predicted using the history of proteinfamilies. In many cases, the prediction provided information about function as wellas form.

Genomics is driving the use of history inphysical science. Genomic-sequence data-bases contain historical information aboutgenes in an easy-to-use form. It can be used tobuild evolutionary trees and reconstruct thesequences of ancestral genes and proteins.Through recombinant-DNA technology,ancient proteins from extinct organisms canbe resurrected in the laboratory and studied

biochemically to test hypotheses about formand function. These techniques promise todeliver a comprehensive model for life, com-bining the physical and structural behaviourof biomolecules with two histories — the first told by palaeontological and geologicalrecords, the second by molecular sequences.

I believe that this historical model is asimportant for understanding the behaviourof biomolecules as the powerful instrumentsused for their physical characterization.Indeed, the human genome, now no morethan a set of chemical structures of theorganic molecules involved in inheritance,will need natural history to give it value.

Palaeogenomics models have already sug-gested how the global unification will lookwhen it is complete. For example, a historicalanalysis of resurrected proteases ancestral tothose regulating blood pressure indicatedwhich animal models were appropriate forstudying the pharmacology of pharmaceuti-cals targeted against these enzymes.

Combining natural history and physicalscience involves huge collaborations andfaces many obstacles, not least of which istraining. Compartmentalization of academ-ic departments and funding bodies ensuresthat natural history is only rarely incorporat-ed into biophysics curricula and that theconverse is true for natural historians. Butthe potential rewards are immense. The firstto reconcile the two traditions will be the firstto glimpse the ‘how’ and ‘why’ of life fromthe molecule to the planet. ■

Steven A. Benner is in the Biochemistry Division ofthe Chemistry Department, University of Florida,PO Box 117200, Gainesville, Florida 32611, USA.

FURTHER READINGChandrasekharan, U. M. et al. Science 271, 502–505(1996).Jermann, T. M. et al. Nature 374, 57–59 (1995).Behe, M. Darwin’s Black Box (Free Press, New York,1996). Benner, S. A. et al. Chem. Rev. 97, 2725–2843 (1997).Benner, S. A. et al. Adv. Enzyme Regul. 38, 155–180(1998).Goh et al. J. Mol. Biol. 299, 283–293 (2000).

WEBLINKS➧ http://www.stanford.edu/class/history133/Smoc/Unifying.html➧ http://www.embl-heidelberg.de/predictprotein/predictprotein.html

Natural progressionconcepts

NATURE | VOL 409 | 25 JANUARY 2001 | www.nature.com 459

Molecular history“The human genome, now no morethan a set of chemical structures ofthe organic molecules involved ininheritance, will need natural historyto give it value.”

Interweaving the histories of biomoleculesshould deliver a comprehensive model for life.

KE

N E

WA

RD

/SP

L

© 2001 Macmillan Magazines Ltd