D.L. Nanney. Original paper presented at Symposium on "The Epigenetics of Cell Transformation and Tumor Development" American Association for Cancer Research, Eightieth Annual Meeting San Francisco, California, May 26, 1989. The work of the author was supported by a grant (R01 GM-07779) from the U.S. Public Health Service.

The author revisits the 1950s to reconsider the impact of the Double Helix on thinking about developmental biology and cytoplasmic inheritance. He interprets that interaction as a near disruption of an incipient merging of cybernetics with regulatory biology. The metaphor of "the genetic library", and the successful program for its elucidation, may have hindered the exploration of the systemic components of living systems, which are not just creatures reified from the "blueprints", but essential complementary components of life that reciprocally regulate the nucleic system.

I. DEDICATION

This essay is dedicated to the memory of the late George Wells Beadle (19xx-1989). My awareness of the essential connection between this discussion of "epigenetics" and Beadle's contributions came belatedly - after the penultimate draft of the essay had been stored away for a few months of 'maturation'. The linkage came about through a chance encounter in a hallway.

Rick Horwitz, a colleague concerned with the quality of undergraduate education, stopped me to "pick a bone with a geneticist". His complaint was that geneticists have quit teaching the understanding achieved during the era of "biochemical genetics". He said that advanced students in cellular and developmental biology are now well briefed about "molecular genetics" - about replication, transcription and translation, about the regulation of gene action and protein synthesis. They simply have no comprehension of how gene products participate in cellular and organismic functions. They think of genes with developmental effects as sending their agents directly to participate in the critical events. The association of genes with enzymes, and enzymes with biochemical pathways and metabolic networks, which we credit to Beadle and Tatum seems to have been passed by and forgotten. The corpus of "biochemical genetics" is dismissed, in effect, as a primitive and preliminary precursor of the subsequent rigorous and definitive molecular genetics.

During changing times, we have difficulty discriminating between what we were taught and what we teach, between what we know and what we think our students know. The messages received are often not the messages we thought we sent. As investigators in corners of a field, we have distorted perceptions of the events occurring around us. We need time to consolidate our understanding as well as to escape our myopic preoccupations and egoistic involvements. Evaluating the explosive advances in genetics during the last half century will require at least another half century of systematic and objective study. Even a century has not given us certain understanding and universal agreement about the contributions of Darwin and Mendel. This essay is an attempt to stimulate some reevaluation and integration across a span of a third of a century. I discover that it assumes acceptance of a more substantial component of the genetics of the 1940s than can be taken for granted.

Although I assume here the correctness of the Beadlian perspective, I must confess to a long-standing skepticism about Beadle's role in establishing that perspective. I can now perhaps explain and excuse that attitude as a consequence of my temporal position. As a graduate student in the late 1940s I complained to my thesis advisor - T.M. Sonneborn - about the intemperate praise being heaped upon the Neurospora workers. I claimed that the discoveries weren't news at all. I asserted that the concepts were so self-evident as to require little documentation. I took seriously Delbruck's (1946) criticisms of the tautology of the experiments. Sonneborn's response was, "You can't understand Beadle's accomplishments because you don't know how confused we were earlier". Although I have questioned the nature of Beadle's role as scientist and as publicist, and of the Neurospora work as "discovery" and as "demonstration" (Nanney, 1983), I have never doubted the essential concepts of "biochemical genetics".

Meanwhile, several additional major developments have been faced by several new generations of biologists. In the midst of the excitement I failed to notice that well-publicized and widely acclaimed understanding may have been slipping away. The achievement of Beadle and his collaborators may have been reduced to an empty slogan, and its deeper significance obscured in the flurry of even more remarkable new insights.

The essential insight of the 1940s was not, however, superseded by subsequent developments. That understanding of gene enzyme relationships established a permanent reference point in the dialectics of genotype and phenotype, and it is taken for granted in the discussion below. I will not develop that theme here except to note that historians of genetics have placed considerable emphasis on the semantic distinctions between the genetics of transmission and the genetics of expression (Allen, 1975, 1979). In order to achieve his results, Mendel must have conceived the distinction we now refer to as the genotype/phenotype dichotomy, even though he invented no terms to crystallize the concepts. Johannsen (1909) and Bateson (1908), major architects of of the discipline of genetics, made explicit the distinction and made obvious the challenge of a meaningful connection.

Perhaps the early geneticists expected the relationship between genotype and phenotype, of gene and phene, to be relatively simple and direct. In this they were of course quickly frustrated. A single assorting factor could affect several phenotypic characteristics; multiple assorting factors could influence the same phenotype. These mysteries were classified and camouflaged by the proliferation of terms - epistasis, pleiotropy, polygeny, penetrance.... No simple one-to-one correspondence could be sustained.

The semantic separation of transmission genetics from physiological genetics, combined with the intractability of genetic expression, provided the basis for an important disciplinary strategy. According to Sapp (1987), the definition of transmission genetics as a distinctive component of genetics allowed the reductionistic program of T.H. Morgan to be exploited without the burden of explaining the phenotype. That program, however, had other consequences - particularly the alienation of biologists who considered the separation of heredity and development to be inappropriate. It also alienated biologists commited to molecular mechanisms, such as the French heirs of Pasteur (Burian, 1988). What Beadle achieved was essentially the Pasteurization of genetics, which had been delayed by the strategy of dividing the genetic question.

This perspective on Beadle's scientific contributions, and on the state of genetic thought in the 1950s, is a significant part of the context within which the abortive junction of cybernetics and biochemistry is considered here. Most genes specify enzymes, which catalyze reactions in interlocking metabolic networks. Understanding the emergence of phenotypes - of complex forms and programs, must be achieved through an understanding of these compound/complex systems of biochemical reactions, which seem to have been neglected of late. We should not revert to the simplistic gene/phene correlations of naive Mendelians.

II. INTRODUCTION

I was recently invited to participate in a symposium, sponsored by the American Association for Cancer Research, on the "Epigenetics of Cell Transformation and Tumor Development". Though my abysmal ignorance of cancer research warned me against such a venture, I was flattered into accepting the invitation by the symposium organizer's recollection of a paper I had written long ago on "Epigenetic Control Systems" (Nanney, 1958). He observed that investigators of tumor transformations in culture (Rubin and Xu, 1989) and in vivo (Farber and Sarma, 1987) continue to struggle with the relationships between the adaptive responses of cells to environmental stress and their eventual genetic alterations. Rubin requested that the subject of my old paper be brought up-to-date, and I agreed to make the attempt.

Though intellectual tasks are sometimes difficult to begin, they may be equally hard to stop before they have run their course. I have been greatly stimulated by recent discussions with new-found colleagues, and by the effort to comprehend the significance of epigenesis in the contexts of both the 1950s and the 1980s. The result of that stimulation is this essay, which I am having some difficulty classifying. The essay is concerned with biological regulatory mechanisms, but it is also concerned with the social regulatory processes of science. It reflects personal memories of people and events, but it reaches for more general historical perspectives.

III. THE HISTORICAL CONTEXT

A. The Conference at Gif-sur-Yvette

I need to go back in time and recollect a now faded scientific environment. My most public involvement with epigenesis occurred at a conference on "Extrachromosomal Heredity" held in France, at Gif-sur- Yvette, nearly a third of a century ago.

This was a small conference to which Boris Ephrussi, the most prominent European spokesman for "cytoplasmic inheritance", had invited a few sympathetic leaders of European and American biology. It was basically a strategy conference concerning a disciplinary issue of deep concern to the participants. Much was at stake at the conference, because "extrachromosomal heredity" appeared to be at a critical juncture (Sapp, 1987). The double helix had been grasped but not assimilated. Long-standing questions seemed to be on the verge of answers, and the rhetoric of scientific discussion had shifted perceptibly. Previously stable disciplinary positions were being threatened.

I went to the conference as the guest of Tracy M. Sonneborn who had been my thesis advisor. Sonneborn was the leader of the American contingent - as Ephrussi was the acknowledged leader of the Europeans. I was not expected to participate very actively, but I quickly found myself in the center of an argument. My spontaneous public dispute with Ephrussi after his opening address surprised both of us. The argument was semantic, and dealt with how to phrase the scientific question. The discussion at Gif was the basis for my paper on epigenetic control systems.

I did not comprehend fully then what was going on at Gif, and I still do not. Periodically I have tried to evaluate what happened. I suspect that events at Gif may be in some way relevant to continuing issues in biology. Making the connection is not easy, however. Gif was three decades away, and three decades of modern science is a very long time - at least two informational doublings according to scientometrician D. J. de Solla Price (1983). Three fourths of our current biological knowledge has been acquired since 1957. Bridging that gap will not be easy.

B. The Metaphor of Merging Streams

To evaluate the Gif Conference we have to locate it in the context of its times, which means that we have to discuss what went before and what came after. To summarise the significant historical circumstances we have no choice but to employ metaphors, but we do have a choice of metaphors. Because that choice conditions the

evaluation, it not a trivial one. I will return later to a discussion of metaphors. For the moment I will use one which is perhaps not as loaded as others - the metaphor of converging streams.

A flowing stream has a trivial beginning in chance rivulets that gain strength and substance from the influx of other streamlets. It eventually acquires enormous power by the confluence of branches draining different intellectual basins.

This metaphorical river of 20th century biology needs to be distinguished from another famous stream, that of the early Greek philosopher Heraclitus of Ephesus. Heraclitus used the stream not as a symbol of progress, but of pointless dynamism. The stream of Heraclitus is an "eastern" stream, ever in flux, ever changing, but going nowhere. One cannot step twice in the same stream, but it doesn't matter.

To illustrate the modern "western" scientific stream, I diagram (Fig. 1) a portion of the history of ideas in biology over the last hundred years. The diagram visualizes the separate origins of many of our concepts and tools. One stream is identified as the bramch of biology that we call "cytology". This discipline was "technology-driven" as it emerged from the exploitation of the vastly improved light microscopes that were developed by the last quarter of the 19th century. The cytological stream made a belated junction (in 1902) with the concepts and observations flowing from the Benedictine monastery in Brno, now legitimized as the science of genetics.

The cytogenetic synthesis initially seemed far removed from that previously established major river of thought and observation that we associate with Charles Darwin and his statistically inclined cousin, Francis Galton. The Mendelian and Darwinian streams eventually turned together, however, and fused in what is proudly acclaimed as "The Modern Synthesis" of the 1930s.

Another confluence occurred only slightly later. It is not commonly recognized as a junction on this side of the Atlantic. The gene - enzyme linkage is popularly credited to Beadle and Tatum and considered a wholly indigenous development. Another view is that the linkage represents the confluence of the molecular reductionism of Pasteur with the genic reductionism of Mendel (Burian et al., 1988). The class of 1822 (Galton, Mendel and Pasteur, who were born in that notable year) finally had a reunion in the 1940s. In this interpretation of convergence, Boris Ephrussi transported the French perspective through his work with Beadle on the genetics of eye colors in Drosophila.

The next depicted junction is that we associate with the double-helix. From three decades away, "biochemical genetics" seems to merge smoothly into "molecular genetics", but the coalescence is misleading, and a distinction should be made. (See the Dedication, above). The discovery of the double helix of 1953 made possible a dramatic new confluence - between the small stream of nucleic chemistry started by Friedrich Miescher in the 1870s, and joined tentatively with the mainstream of biochemical genetics in 1943 (Avery et al., 1943). The differentiated linear nature of nucleic acids, and hence of proteins, and therefore of "biological information" generally, seemed to be a profoundly revolutionary perspective, and it was disseminated widely through masterful public relations.

This metaphoric history of biology holds important unspoken assumptions about the scientific process. The intellectual streams move forward through the cultural landscape, impelled by the forces of gravity toward great rivers and seas of understanding. Science in the western world still assumes a belief in forward motion, in "Progress", however skeptical we may be about an optimistic view of other kinds of history. Science is believed to improve our understanding of nature through an abrasive erosion of ignorance. The movement may be episodic, with periods of slow movement or stagnation punctuated by "Breakthroughs" and "Junctions", but the direction is always forward.

C. Cybernetics - The Neglected Junction

Keeping within this metaphoric landscape, we need to consider another junction of disciplinary streams that occurred very near the nucleic junction. This junction between Cybernetics and Molecular Genetics is not generally acknowledged. Nevertheless, it figured prominently in the Gif conference of 1957 and it has had a significant though subtle influence subsequently.

Cybernetics is a set of ideas about control processes in engineering that was of considerable interest to post-war biologists. The ideas and the language of Wiener (1948) and Shannon. the concepts of information, signals, control systems and feedback mechanisms were borrowed in the late 40s and freely adapted to biological phenomena.

Though the direct impact of cybernetics on biological thought is often discounted, the conceptual framework and the language of biology was significantly and permanently modified by the interaction. Delbruck's (1949) schematic interpretation of antigenic mutual exclusion in Paramecium was one of the first applications of cybernetic thinking in biology. The context of this diagram is now almost forgotten, but the issues were considered crucial. Delbruck's diagram was a very simple application of the concept of alternative stable states, maintained by negative interactions between pathways, but it was introduced in a charged atmosphere. The larger issue was the role of the cytoplasm in development and heredity. Because of the authority of Delbruck, and because of the aptness of the message, the example settled in principle for many of the disputants the "Conflict between the Nucleus and the Cytoplasm".

That "conflict" may be considered a consequence of a disciplinary strategy. The separation of the problems of genetic transmission from those of genetic expression (Allen, 1975; Sapp, 1988) at the beginning of the century released the new genetics from the immediate requirement to explain the biochemical nature of the gene and of gene function. It also separated the disciplinary practices and languages of embryology and genetics from their common roots in the thinking of 19th century biologists such as Darwin (1868), Weismann (1891), and E.B. Wilson (1896). An expression of their alienation was the "Developmental Paradox". How can cells possessing the same genetic components, reliably copied in every mitosis, come to have very different characteristics?

A widely discussed solution to the Developmental Paradox was the proposal of a dual biological informational system - a primitive cytoplasmic system localized in the cytoplasm and a derived nuclear system associated with the chromosomes (Sonneborn, 1950; Ephrussi, 1954). This dichotomy must not be confused with the much simpler and less divisive issue that was formulated after DNA was recognized as the primary genetic material: how much of the nucleic information is in the cytoplasm, and where (Willkie, 1954; Beale and Knowles, 1978)?

My 1958 paper (Nanney, 1958), based on the Gif discussions, elaborated on the lesson from the Paramecium antigens. I argued that differential gene action, masked by systemic properties of differentiated cells, could account in principle for many of the persistent stable states encountered in protists and in developing multicellular organisms. The discussion helped convince some of the most obstinate defenders of the cytoplasm, most notably the European leader of the cytoplasmic faction, Boris Ephrussi (1958).

The subsequent studies on "adaptive enzyme formation" carried out by Jacob and Monod, who participated in the conference at Gif, provided a concrete and conscious implementation of the cybernetic principles (Jacob and Monod, 1961; Monod and Jacob, 1961). The experimental work, as well as its extrapolations, provided convincing evidence of the ability of genic regulatory systems operating on basic cybernetic principles to explain a wide range of biological processes - clonal heredity, as well as sequential (cascading) events, and cyclical and calendric phenomena. And these extrapolations were not merely theoretical. They were based on real genes, and real metabolites, and real biochemical pathways.

This work became a centerpiece of molecular biology, but it was adumbrated by earlier studies on the same system (Novick and Weiner, 1956) which were discussed prominently at Gif.

Thus, a firm basis was established for a cybernetic/genetic junction by the early 1960s. Nevertheless, the generalization and expansion of cybernetics in theoretical biology failed to occur. Following the Gif conference, cybernetic applications in developmental biology and cell culture were often referred to as "epigenetic" (Harris, 1964). They were elaborated by a small number of theoreticians and became progressively sophisticated (Goodwin, 1963; Kacser, 1963; Kauffman, 1969, 1974). The confluence of cybernetics with other kinds of regulatory biology was widely ignored. A commonly held opinion among biologists was that cybernetics contributed nothing to theoretical biology.

D. Why did the Genetic/Cytogenetic Junction Fail?

The reasons for this judgement, and the reasons for the apparent rejection of cybernetics, deserve a more systematic analysis than I can provide here. I will simply note a few factors that might be considered.

One reason for the neglect of the junction might be the disciplinary immaturity of biological systems analysis at the time. Although primitive cybernetic lessons could be learned and simple model systems could be explored, the properties of complex systems could not be described until nonlinear mathematical equations were subjected to more effective analysis by steadily improving generations of computers. The kind of order that might emerge from sensitive nonlinear "chaotic" systems was accessible in the 1950s and 1960s only by intuitive leaps. The technology that can deal decisively with weather systems and embryonic programs has come only by exploiting steadily improving generations of computers (Gleick, 1987).

The lack of sophisticated epigenetic theory is by no means a complete explanation for the neglect of cybernetics in biology. The little that was available in the way of theory was often ignored. Another reason that needs to be added might be referred to as technological antagonism. Cybernetics is above all a perspective of ordering probabilistic events into deterministic patterns. Its logical affinities are with transmission genetics and statistics, not with biochemistry and molecular biology. The characteristic technology of cybernetics is that of the computer, not the lab bench. And molecular biologists are not comfortable with probability and artificial intelligence.

Although only anecdotal, several observations bear on the suggested technological antagonism. Many capable molecular geneticists claim incompetence when asked to teach formal genetics to undergraduates. Molecular geneticists often disapprove of the recruitment of population geneticists into their disciplinary units, or of converting wet lab space into computer rooms. Molecular geneticists prefer to consider their discipline a branch of biochemistry rather than a discipline associated with Mendel, Galton and Morgan. One remembers a classroom dictum of Salvador Luria, one of the founders of molecular biology: "If you have to use statistics to prove something, it isn't worth knowing." This resistance, even hostility, to formal analysis is a curious and pervasive attitude within the mainstream of molecular biology.

Another factor for the neglect of cybernetics is related to the technological bias, but it involves an even more sociological dimension. Cybernetics has always been of more interest to developmentalists than to geneticists, and the very term "epigenetics" was derived from embryology. Developmentalists through the first half of this century have been locked in a disciplinary ambivalence with genetics, recognizing common objectives while experiencing little but frustration in their interactions. Many developmentalists backed the proponents of cytoplasmic inheritance in the nuclear/cytoplasmic conflict because transmission geneticists, and eventually molecular geneticists, in their preoccupation with reducing mechanisms to their most elementary components, gave liitle attention to whole organisms.

In their search for a haven in a storm, developmentalists also earned the disrespect of many geneticists as they continued to be sympathetic with what geneticists refer to as "soft" mechanisms (Mayr, 1982). They entertained heretical thoughts about the possibility of directed adaptive responses to environmental cues and embryonic signals, even when the coupling of Lysenko to Lamarck had become a public scientific scandal.

Meanwhile geneticists exulted in their decisive demonstration of the randomness of genetic mutation to drug resistance. The Luria/Delbruck fluctuation test (1943) was one of the first recognized triumphs of molecular genetics (despite its initial probabilistic foundation, which was replaced as soon as possible by the more visual and intuitive demonstration of "indirect selection" (Lederberg and Lederberg, 1952), and it was emphatically opposed to the mechanisms that cytoplasmic geneticists and embryologists thought might be needed to explain their puzzling phenomena. The possibility of adaptive genetic change was considered not only experimentally nonfeasible, but intellectually unacceptable. Nevertheless recent students of bacterial mutation (Cairns et al., 1988) recognize that reasonable mechanisms for a limited range of adaptive responses can be proposed, and that categorical evidence against such mechanisms is not available.

I hope I have said enough to characterize somewhat the curious scientific turbulence that can be observed around developmental phenomena in the 1950s and 60s. Biologists have not been generally concerned very much with that turbulence because of our preoccupation with the far greater excitement associated with the description of the DNA double helix in 1953. That discovery seems different in kind from the relatively placid confluence of streams characteristic of "normal science". We are inclined to treat the double helix as a qualitatively different kind of thing - a "revolutionary" breakthrough.

E. The Power of an Image

The powerful image of the double-helix was perhaps in the long run more significant in obscuring the cybernetic connection than were the other technological and social factors. Whether the double helix was fundamentally revolutionary can be debated; without question it was a Surprise. The double-helix came as a real shock to theoretical biologists; nothing earlier in their experience had led them to anticipate anything like this. The enthusiastic acceptance of new biological perspectives transformed an unexpected metaphor into a dogma.

Before the double-helix was described we were sure that biological specificity was referable to the surface properties of three-dimensional macromolecules. This opinion was based on studies of the most specific biological agents that were then known - the enzymes. We thought biological specificity was the consequence of surface templating processes utterly different from the linear templates of nucleic acids. Together the alpha helix of proteins, and then the DNA double helix destroyed the older perspectives and demonstrated that essential biological information can be encoded in a linear order analogous to the written human language. The metaphors which encapsulated this new understanding were "The Genetic Library" and "The Genetic Blueprint".

These power of such images cannot, of course, be attributed soley to their appeal to the imagination. The image evoked a program of studies - a set of questions that had not been previously asked, but for which answers could be found, and for which technology was either at hand or soon to be provided. Probably no one anticipated the explosion that followed or its eclipse of most previous biological understanding.

Much of the subsequent history of biological thought has consisted of the elaboration of this image in a system of corollaries and supplementary mechanisms. The dogmatic synthesis concerning biological information systems is too familiar to be argued in detail. Its central claim is that the nucleic system is the "exclusive" reservoir of biological information. The synthesis does not deny that biological specificity is expressed by other systems and or even retained for considerable periods of time in the absence of nucleic differentials. It claims, however, that nucleic sequences are the ultimate and sufficient source of the information.

Moreover, the genetic information is believed to flow unidirectionally; the flow of information is essentially from sequences of nucleotides to sequences of amino acids; primary amino sequences determine secondary and tertiary protein conformation, the properties of component molecules govern their associations with other molecules to constitute organelles and cells and organisms through mechanisms of self-organization.

Another component of the new dogma is the old belief that the nucleic library is essentially unchanged in the course of embryonic development. Mitosis is an accurate means of distributing the library so that all cells of the body are informationally equivalent. The "developmental paradox" is explained by the differential expression of nucleic information in isolated cellular compartments, provoked by their exposure to programmed embryonic environments. Both the responses of the cells, and the cellular environments of the embryo are believed to be encoded in the library.

IV. SCIENTIFIC IMAGES

A. The Metaphor of Discontinuity

I have tried to evoke some of the tensions in biology during the 1950s by recalling some of the historical circumstances. The context was framed in the metaphor of converging streams. That metaphor, like any metaphor, emphasizes certain features of its analogue while ignoring others. The metaphor of junctions places emphasis upon the broadening or generalization of understanding by the union of previously discrete knowledge. That metaphor is not, however, the prevailing metaphor of our times. A metaphor much more common in the discourse of the new academic industry of "science watchers" is Revolution (Kuhn,1962). Most of us suspect that at least one major discontinuity in biological science occurred in the 1950s. Characterizing that discontinuity may be premature. For the time being we can label it loosely as the "Watson-Crick Revolution" or the "Breakthrough of the Double Helix".

References to scientific events as "breakthoughs" and

"discontinuities" are readily understood and accepted in the scientific culture as appropriate and accurate representations of scientific activities. To call the discovery of the double-helix revolutionary is to use a political metaphor. The metaphor of revolution emphasizes the changes the occur, and the conflicts that accompany them (Engelhardt and Caplan 1987). Both scientists and historians have long believed that the path of science is discontinuous. The concept of Discontinuity, particularly in its diminuitive form Discovery, is perhaps the most familiar metaphor commonly applied to the scientific process in modern times. It lies at the foundation of our system for recognizing scientific merit. Many "discoveries", however, are social artifacts that require later reevaluation.

Discontinuities and Discoveries, like all analogues and metaphors, fail to represent reality perfectly, but they are useful as social instruments and pedagogical devices so long as they do not mislead. Even if we accept as justified a general criticism of our metaphors, we must ask about particular cases. Was the Revolution of the Double-helix in fact only a Surprise. Future historians may find in Muller's (1926) "The Gene as the Basis of Life" the essence of the 20th century discontinuity, and the Double-Helix as its programmatic opportunity.

B. The Metaphor of Dialectics

Because of the hazards pf distortion when employing any metaphor, it may be instructive to consider an alternative to Revolutions. One that comes to mind is that of Dialectics. This metaphor is less reputable, perhaps because Karl Marx was fond of it. The metaphor is also less comfortable with the western world's concept of Progress. It suggests that our culture does not move primarily or solely through the confrontation and destruction of barriers - through Breakthroughs and Discoveries - but that we meander back and forth between fixed alternatives, or perhaps between drifting buoys in a Hegelian sea.

According to this perspective the new questions we confront can often be seen to be older questions, only slightly disguised. Examples of recurrent riddles abound once we learn the game. The Gradualism and Catastrophism of the 19th century reappear as Microevolutionary and Saltatory changes. Earlier debates about Heredity and Environment, about Nature and Nurture continue, but the new protagonists speak of Biological Determinism and Cultural Constraints. We continue to ask whether evolutionary directions are determined largely by random events or by selective forces, by Chance or by Necessity (Monod, 1971).

In most of these instances, we are aware that the argument is never decisively resolved. Neither point of view is firmly established as correct. Each generation renews the discussion, invents a new set of terms, and establishes a new consensus about the appropriate balance between essential perspectives. One point of view may be partially eclipsed for a time, but it is never far out of consideration.

Because deep questions concerning the natural order resist final solutions, the rhetoric of earlier debates often echoes in current discussions. The words may break loose from previous contexts and drift into inappropriate configurations. Our poorly developed historical sense also encourages us to redefine words and to link dialectics inappropriately. We tend to push the unresolved dialectics into a single polar perspective which we unconsciously label as Good and Evil; Right and Wrong; Ignorant and Enlightened. Because an earlier dialectic concerned Rationalism and Mysticism, any defender of an unpopular dialectic risks being labelled a mystic or a vitalist in a mechanistic culture (Crick, 1966).

We are primarily concerned here with the dialectics of development. In its oldest and most generalized form developmental interpretations are summarized by the terms "epigenesis" and "preformation" in the rhetoric of the Greek philosophers. The epigenetic and preformationist views are imposed on the broad backs of some of our earliest bearers of concepts - Aristotle and Plato. Aristotle noted the progressive emergence of order and structure out of the undifferentiated yolk and albumen of the fertilized hen egg. Aristotle trusted the testimony of his eyes. The appearance of specific organs was interpreted matter-of-factly as epigenetic. Structure appeared from featureless ground substance; pattern arose de novo.

Plato depended less on his observations, if indeed he ever looked. He was a theorist rather than an experimentalist. He might have considered the fact that different animals arose from the undifferentiated egg substance of different species. Perhaps he asked why limbs and feathers appeared in the developing eggs of chickens but not in the eggs of snakes. Though he couldn't see the pattern in the fertilized egg, he could be sure that the pattern, the "Idea", was there in some form. The invisible preformed pattern revealed itself through the processes of development, but the pattern was not re-created anew in each generation.

The development of the light microscope in the 17th century, and particularly the vast improvement of microscopy in the 19th century, allowed biologists to peer much more effectively into the realm of the very small. This careful search, like the unaided observation of the macroscopic egg, revealed no evidence of a preformed organism whose growth could explain the developmental process. Clearly embryogenesis is "epigenetic", even after the development of the electron microscope.

The double helix changed that. What could be a more definitive triumph for a dialectic alternative than the equation of the Platonic Idea of the animal body with the genetic library of the molecular biologist? Science, like politics, makes strange bedfellows. What could be more ironic than the rejection of the explanations of Aristotle, the founder of experimental mechanistic biology, by the most successful of the modern experimental disciplines? Is it possible that in awarding the prize to the idealistic Plato we are losing our historical perspective? Do Aristotle and epigenesis still have rounds yet to come?

V. SCIENTIFIC MECHANISMS

A. The Developmental Paradox

Our focal point is the 1950s. The developmental paradox of the 1950s emerged from experimental biology of the 20th century. The exploitation of Drosophila as a genetic instrument had made clear that most traits transmitted from generation to generation are referable to sites on the chromosomes. The chromosomal equivalence of somatic cells remained a matter of dogmatic belief, despite occasional "DNA puffs" in Drosophila, and chromatin elimination or heterochromatization in rare "aberrant" animals. The belief in nuclear equivalence was reinforced by observations on the developmental flexibility of cells and nuclei with different prospective fates which had been experimentally maneuvered into different embryonic locations early in their history.

At the same time, a substantial body of knowledge from "experimental embryology" and from cell cultures (Harris, 1964) indicated that distinctive and stable cell types are eventually established during development by a delimitation of developmental "potency". Cell culture studies were still primitive, but they had progressed far enough to allow biologists to anticipate the unequival demonstration of "cellular memory" which was to come in the 1960s. The cytoplasm seemed implicated by the correlation between regions of the egg cytoplasm and embryonic tissues. Occasional breeding studies going back to Carl Correns (1909), one of the rediscoverers of Mendel, demonstrated "maternal inheritance" which seemed to require a cytoplasmic or nonchromosomal genetic mechanism. Morover, by the 1940s breeding studies on eukaryotic protists seemed to be turning up exceptional numbers of cases of cytoplasmic inheritance. The petite mutants of yeast (Ephrussi, 1953), and the antigenic, mating type, and "killer" traits of Paramecium were widely recognized cases in point (Sonneborn, 1949).

The belief that "hereditary" cell variants arise regularly in the course of development, or in the course of cell culture in protists, without modifying the nuclear genetic apparatus, was the developmental paradox that led to the hypothesis of a dual genetic system. The nuclear hereditary system was assumed to be responsible for the transmission of traits between sexual generations while cellular heredity was regulated by a functionally different system in the cytoplasm (Sonneborn, 1949; Ephrussi, 1953).

This generally unsatisfactory resolution dissolved eventually in an interpretation referred to as "differential gene activity". This idea began to emerge as a possible resolution in the late 1940s, and was influenced importantly by microbial studies of "adaptive enzyme formation" (Spiegelman, 1948). Prior to this time (See Wright, 1945) gene action was usually considered to be constant. Cellular differentiation was considered a consequence of the interaction between a constant source of gene products with differential cytoplasmic and environmental conditions. The direct regulation of gene action, what we would now term transcriptional control, was not suggested.

Even transcriptional control, however, seemed unlikely to explain the persistence of the differentiated state, which was a central doctrine of developmental biology. As mentioned above, this gap in the explanation of embryonic change was repaired by the insertion of cybernetic systems. Also as mentioned above, the consensus of developmental biologists attributes only a limited role to the cybernetic apparatus.

B. Epigenetic Nucleic Changes

I need to add to this survey a subsequent complication of major significance. The conceptual distinction between genetic and epigenetic information, and between genetic and epigenetic cellular differences seemed reasonably clear in the late 1950s, though some of us doubted even then the dogma of nuclear identity during development. (Nanney, 1963). The assumption of nuclear equivalence - the somatic integrity of the genetic library - was based on two kinds of negative evidence. The first was the failure of development to be disturbed in amphibians when embryonic nuclei were displaced from their usual organismic fates. The work of Briggs and King (1952) with nuclei at later stages of development began to cast doubt on the previous interpretations, but failure of embryonic nuclei to sustain full development was often dismissed as evidence of technical damage.

Other techniques of nuclear analysis were capable of demonstrating chromatin elimination, heterochromatization, and quantitatively significant DNA puffs (Breuer and Pavan, 1953). These were all found in some developing systems. Systematic sequence variations associated with development could not be observed in any organisms, however, and were assumed dogmatically not to occur.

The idea of epigenetic informational alterations of somatic nuclei is a threat to the nuclear monopoly of biological information. The fact of epigenetic alterations of nuclear information is the foundation for a new synthesis of biological information systems in which the "genetic blueprints" are only one component of a complementary genetic/epigenetic system of information management.

The hallmarks of epigenetic systems were considered to be the relative ease, adaptability, and regularity of their changes. Epigenetic systems were considered to have fewer alternative states, but to be more capable of specific response. However, these characteristics were seen as quantitative and not qualitative, and the manipulations necessary to demonstrate the "ideal" characteristics required more information about the biological functions involved than is often available.

Lederberg (1958) suggested that the information systems be characterized in terms of their chemical foundations, as nucleic and epinucleic. These terms evoked more directly the distinctive mode of maintaining nucleic information, i.e., semi-conservative replication, as opposed to the various modes for self-reproducing patterns proposed for "systemic" hereditary systems. This terminology is more concrete in its references, but ascertaining whether sequence variations have occurred in appropriate nucleic regions of cells with distinctive hereditary traits was not then, and is not even yet, practical in most cases. We are left with a conceptual distinction without a clear generalizable operational application.

This operational difficulty is compounded by another, which is even more fundamental. Subsequent work has clearly shown that limited arrays of truly nucleic changes can occur reliably in development, or can be induced regularly under other defined environmental circumstances. Regular changes in DNA sequences, associated with changes in cellular phenotype, were already becoming known in the 1950s. One of the earliest examples came from phase variation of Salmonella serotypes (Lederberg and Iino, 1956). These antigenic alterations are now clearly demonstrated (Seifert and So, 1988) to be consequences of constrained movements of blocks of DNA sequences.

The mechanism underlying changes of mating type in yeast have also been shown to be associated with the controlled and adaptive movements of "cassettes" within the genome (Herskowitz, 1983; Nasmyth, 1983). The expression of antigenic types in Trypanosoma (Boothroyd, 1985) is controlled by other regular and reversible modifications of nucleic sequences. The classical example of ciliate serotypes (Doerder and Berkowitz, 1987; Epstein and Forney, 1984) has even been brought into the category of controlled nuclear change by the demonstration that the regulatory system of a conjugating cell can modify the maturation of the germinal (micro-) nuclear structure so as to exclude certain sequences. The system of mating type determination in the "karyonidal" species of Tetrahymena is also plausibly interpreted in terms of epigenetically modified nucleic sequences (Orias, 1981). Thus, limited changes in nucleic templates are now known to be capable of specific induction in a wide array of protists under circumstances that indicate that these are programmed adaptive changes under the control of epigenetic systems (Sonneborn, 1977).

All these microbial examples might be dismissed, or lumped with pathological insertions of viruses or other plasmids, were it not for the example from the vertebrate immune system (Tonegawa, 1983). The differentiation of immunocytes involves a regular pattern of rearrangement of specific chromatinic domains. And this systematic scrambling of genetic elements is responsible for the diverse capacities of cells to recognize and respond to various foreign compounds and to each other. Even though immunological maturation does not generate specific new cell types, it nevertheless produces a highly restricted array of recombinant cell types from which appropriate types can be selected.

These processes of genetic diversification and clonal selection are clearly NOT "epinucleic" processes. Immunocyte maturation involves "nucleic" modifications achieved by epigenetic mechanisms. All the examples concern normal, developmentally-regulated modifications of cellular information content. Even though the distinctive cell types are subsequently maintained by conventional nucleic replication, the generation of the diversity must be recognized as "epigenetic". To the conceptual problems raised by the immune system, must also be added the developmental phenomena associated with sequence amplification, and with reversible chromosomal inactivation, especially that associated with sex determination in mammals (Lyon, 1968). The process is clearly "epigenetic" in a general sense of a directed adaptive hereditary change. We do not yet know surely whether a persistent heterochromatized chromosome or an "imprinted" chromosome has undergone a "nucleic" alteration, in the original sense of Lederberg. Certainly the nucleic acids of the chromosome are changed in some way, but the changes have not been demonstrated to occur in their primary sequences. Are changes in the methylation patterns, or histone associations, which are perpetuated by some mechanism supplementary to semi-conservative replication, still to be considered "nucleic"? Are the chromosomal breaks and rearrangements commonly seen in cancer cells the causes, or the consequences of their metastatic state?

None of these complications would justify the resurrection of an abandoned and still problematical class of regulatory mechanism if the concept of "differential gene action" had been expanded to provide the mechanisms required for orderly development and embryonic regulation. Without doubt differential gene action is a central phenomenon in developmental processes, and the number of genes being regulated by these processes is increasing steadily. Nevertheless, the mechanisms for organizing differential gene action are still unknown. The serious attempts to generate ontogenesis from known processes (e.g. Edelman, 1988) fail to deal with some of the characteristic features of cellular differentiation in either multicellular organisms (Gurdon, 1988) or protists (Sonneborn, 1977; Frankel, 1989).

VI. CONCLUSIONS

At mid-century the disciplines of genetics and embryology were divided over "the developmental dilemma": How do cells with identical genetic composition acquire adaptive differences capable of being maintained in clonal heredity? One proposed solution was a dual system of cellular information management. The second system was assumed to be localized in the cytoplasm and to be responsible for developmental differentiation. The alternative view of "differential gene activity" became the more acceptable explanation once elementary cybernetic control system were understood to be capable of explaining in principle the properties of differentiated cells, including their "hereditary" stability.

The full capabilities of cybernetic (or "epigenetic") control systems were not examined fully, however, because of the explanatory power of the "genetic library". Under this metaphor the nucleic informational system seemed to require little assistance from auxiliary systems. In an almost unchallenged triumph of preformationist thinking, the "double helix" was assumed to encode all the emergent properties of living systems.

The monopolistic control of biological information is, however, now being reexamined, and the capabilities of epigenetic systems are being more fully explored. The reasons for the reevaluation are of several kinds. One is a perceived lack of progress in understanding how the features of organization can be effectively encoded in linear sequences and evoked in hierarchical systems. Another is the steady increase in our understanding of the capabilities of complex control systems. A third reason is the clear evidence from many biological systems that the nucleic library is coupled in complementary informational constraint with the systems whose components it specifies.

Sequential information is regulated within particular adaptive limits by epigenetic systems in a variety of organisms. Examples include microbial "shuffle" systems controlling antigens and mating types, cell recognition systems in vertebrate development, and other poorly understood developmental alterations such as chromosome inactivation and chromosome imprinting. One may no longer assume that a correlation between a genetic change and a cellular characteristic requires a final causal role for the genetic change.

The reactivation of interest in systemic properties of cells and organisms is a return to the epigenetic dilemmas that underlay the nucleocytoplasmic discussions in the 1950s. The fundamental connectedness of the genetic-cybernetic systems is, however, distinctly different from the dialogue of duality in the earlier discussions. Clearly the molecules of the system are specified by the genetic information system. Clearly also, however, the deployment of the genetic system, and in many cases its content, is regulated by its systemic creature. The complementarity of the linked information systems is a fundamental property of living systems in both ontogeny and in the longer phylogenetic time scales.

REFERENCES

Allen GE (1975): "Life Science in the Twentieth Century". Camb

Univ Press, Cambridge.

Allen GE (1979): Naturalists and experimentalists: the genotype and

the phenotype. In Coleman W and Limoges C, eds. Stud in the

Hist Biology 3:179-209.

Avery OT, MacLeod CM, McCarty, M (1944): Studies on the chemical

nature of the substance inducing transformation of pneumococcal types.

Induction of transformation by a deoxyribonucleic acid fraction isolated

from Pneumococcus type III. J Exp Med 79:137-158.

Bateson W (1908): "Mendel's Principles of Heredity". Canbridge Univ

Press, Canbridge.

Beale GH, Knowles JKC (1978): "Extranuclear Genetics". Edward Arnold,

London.

Boothroyd JC (1985): Antigenic variation in African trypanosomes.

Ann Rev Microbiol 39:475-502.

Brannigan A (1981): "The Social Basis of Scientific Discoveries".

Camb Univ Press. New York City.

Breuer ME, Pavan C (1953): Salivary chromosomes and differentiation.

Proc Ninth Intern Cong Genet Part II. Caryol 6 (suppl):778.

Briggs RW, King TJ (1952): Transplantation of living nuclei from

blastula cells into enucleated frog eggs. Proc Natl Acad Sci

38:455-463.

Burian RM, Gayon J, Zallen D (1988) The singular fate of

genetics in the history of French biology, 1900-1940. J Hist Biol

21:357-402.

Cairns J, Overbaugh J, Miller S (1988) The origin of mutants. Nature

335:142-145.

Correns C (1909): Vererbungsversuche mit blass (gelb) grunen und bunt-

blattrigen Sippen bei Mirabilis Jalapa, Urtica pilulifera und

Lunaria annua. Z indukt Abstamm u VererbLehre 1:291-329.

Crick F (1966): "Of Molecules and Men", Univ Wash Press, Seattle and

London.

Darwin C (1868): "The Variation of Plants and Animals under

Domestication". John Murray, London.

Delbruck M (1946). In Discussion, following Bonner D. Biochemical

mutations in Neurospora. Cold Spring Harb Symp Quant Biol

11:14-24

Delbruck M (1949): Discussion following Sonneborn TM, Beale GH,

Influence des genes, des plasmagenes et du milieu dans le

determinisme des caracteres antigeniques chez Paramecium aurelia

(variete 4). Unites biologiques douees de continuite genetique. Col

Intern CNRC 8:25-35.

Doerder FP, Berkowitz MS (1987): Nucleo-cytoplasmic interaction during

macronuclear differentiation in ciliate protists: genetic basis for

cytoplasmic control of SerH expression during macronuclear

development in Tetrahymena thermophila. Genetics 117:13-23.

Edelman GM (1988): "Topobiology; An Introduction to Molecular Embryology". Basic Books, New York. Engelhardt HT Jr, Caplan AL (1987): "Scientific Controversies: Case

Studies in the Resolution and Closure of Disputes in Science and

Technology", Camb Univ Press, Cambridge et al.

Ephrussi B (1953): "Nucleo-Cytoplasmic Relations in MicroOrganisms".

Oxford Univ Press, New York.

Ephrussi B (1958): The cytoplasm and somatic cell variation. J Cell Comp Physiol 52: Suppl 1, 35-54

Epstein LM, Forney JD (1984): Mendelian and non-Mendelian mutations

affecting surface antigen expression in Paramecium tetraurelia. Mol

Cell Biol 8:1583-1590.

Farber E, Sarma DSR (1987): Hepatocarcinogenesis: A dynamic cellular perspective. Lab Investig 56:4-22.

Frankel J (1989) "Pattern Formation: Ciliate Observations and Models". Oxford Univ Press, New York.

Gleick J (1987): "Chaos: Making a New Science". Penguin Books, New York.

Goodwin BC (1963): "Temporal Organization in Cells". Acad Press, London &

New York.

Gurdon JB (1988): A community effect in animal development. Nature

336:772-774.

Harris M (1964): "Cell Culture and Somatic Cell Variation". Holt,

Rinehart and Winston, New York.

Herskowitz I (1988): Life cycle of the budding yeast Saccharomyces

cerevisiae. Microbiol Rev 52:536-553.

Jacob F, Monod J (1961): Genetic regulatory mechanisms in the

synthesis of proteins. J Mol Biol 3; 318-356.

Johanssen, W. (1909):"Elemente der Exakten Erblichkeitslehre". G. Fischer, Jena.

Kacser, H. (1963): "The Kinetic Structure of Organisms". In "Biological Organization at the Cellular and Supercellular Level". Harris RJC, ed. Academic Press, New York.

Kaufmann SA (1969): Metabolic stability and epigenesis in randomly

constructed genetic nets. J Theor Biol 22:437-467.

Kaufmann SA (1974): The large-scale structure and dynamics of gene

control circuits: an ensemble approach.

Kuhn TS (1962): "The Structure of Scientific Revolutions". Univ

Chicago Press, Chicago and London.

Lederberg J (1958): Genetic approaches to somatic cell variation:

summary comment. J Cell Comp Physiol 52: suppl. 1, 383-401.

Lederberg J, Iino T (1956): Phase variation in Salmonella. Genetics

41:743-757.

Lederberg J, Lederberg EM (1952): Replica plating and indirect

selection of bacterial mutants. J Bact 63:399- .

Luria SE, Delbruck M (1943): Mutations of bacteria from virus

sensitivity to virus resistance. Genetics 28:491-511.

Lyon MF (1968): Chromosomal and subchromosomal inactivation. Ann Rev

Genetics 2:31-52.

Mayr E (1982): "The Growth of Biological Thought: Diversity, Evolution

and Inheritance." Harvard Univ Press, Cambridge.

Muller HJ (1926): The gene as the basis of life. Proc Intern Congress

Plant Science 1:897-921.

Monod J (1971): "Chance and Necessity: An Essay on the Natural Philosophy of

Biology". Alfred A. Knopf, New York.

Monod J, Jacob F (1961): General Conclusions: Teleonomic mechanisms

in cellular metabolism, growth and differentiation. Cold Spring Harb

Symp Quant Biol 26:389-400.

Nanney DL (1958): Epigenetic control systems. Proc Natl Acad Sci

44:712-717.

Nanney DL (1963): Aspects of mutual exclusion in Tetrahymena. In

"Biological Organization at the Cellular and Supercellular Level".

RJC Harris, ed. pp 91-110.

Nasmyth KA (1982): Molecular genetics of yeast mating type. Ann Rev Genetics 16:439-500.

Novick A, Weiner M (1957): Enzyme induction as an all-or-none

phenomenon. Proc Natl Acad Sci USA 43:553-566.

Orias E (1981): Probable somatic DNA rearrangements in mating type

determination in Tetrahymena thermophila: A review and a model.

Developmental Genet 2:185-202.

Price DJ deS (1983): "Little Science, Big Science and Beyond". Columbia Univ Press, New York.

Rubin H, Xu K (1989): Evidence for the progressive and adaptive

nature of spontaneous transformation in the NIH 3T3 cell line. Proc

Natl Acad Sci USA 86:1860-1864.

Sapp J (1987): "Beyond the Gene: Cytoplasmic Inheritance and the

Struggle for Authority in Genetics". New York, Oxford Univ Press.

Seifert HS, So M (1988): Genetic mechanisms of bacterial antigenic

variation. Microbiol Rev 52:327-336.

Sonneborn TM (1949): Beyond the gene. Amer Scientist 37:33-59.

Sonneborn TM (1977): Genetics of cellular differentiation: stable nuclear differentiation in eucaryotic unicells. Ann Rev Genet 11:349-367.

Spiegelman S (1948): BDifferentiation as the controlled production of

unique enzymatic patterns. Symp Soc Exp Biol 2:286-325.

Tonegawa S (1983): Somatic generation of antibody diversity. Nature 302:575-581.

Weismann A (1891): "Essays upon Heredity and Kindred Biological

Problems". Oxford Univ Press, London.

Wiener N (1948): "Cybernetics: or Control and Communication in the Animal

and the Machine". MIT Press, Boston.

Wilkie D (1964): "The Cytoplasm in Heredity". Methuen, London.

Wilson EB (1896); "The Cell in Development and Inheritance" .

Reprinted (1966), Johnson Reprint Corp, New York.

Wright S (1945): Genes as physiological agents: General considerations.

American Naturalist 79:289-303.