Progress in Biophysics and Molecular Biology (2013)

March 18-21, 2012, Oxford Workshop on “Conceptual Foundations of Systems Biology


Rethinking the (Im)Possible in Evolution


James A. Shapiro

Gordon Center for Integrative Science W123B

University of Chicago,


Introduction: The philosophical background


            Science inevitably operates in ignorance of future developments. Results and concepts that seem inconceivable in one period become conventional wisdom in later decades and centuries. The history of science is replete with examples (Kuhn 1962). Moreover, it is often the case that we cannot perceive the blinders we impose on ourselves out of philosophical commitments rather than empirical necessities.


            Evolutionary thinking began in the 18th Century, at the same time as other fields in biology were transforming into more professional and rigorous disciplines (Stott 2012). In the second half of the 19th Century, the Darwinian ideas of gradual change and natural selection as a creative force engaged in a fierce battle with religious ideas of divine creation over the explanation of biological diversity. In order to combat the teleological arguments of William Paley for a divine watchmaker (Paley 1802 (republished 2006)), the evolutionists rigorously excluded all notions of goal-oriented activity from their theories. In keeping with 19th Century mathematical thermodynamics, they insisted upon randomness at the microscopic level as the basis for macroscopic effects.


As evolutionary thinking integrated Mendelian genetics into the neo-Darwinian Modern Synthesis (Huxley 1942), it adopted the mechanistic thinking that prevailed following the intense Mechanism-Vitalism debate of the early 20th Century. The vitalists, like Hans Driesch, argued that there must be something special about living organisms that informed their activities (Driesch 1908). The mechanicists, led by Driesch’s fellow student, Wilhelm Roux, insisted that only demonstrable physical or chemical entities could be invoked to account for biological phenomena (Roux 1895). Since the vitalists could not explain the nature of their hypothetical special life force, the mechanists prevailed for the rest of the 20th Century.


The issues in the Mechanism-Vitalism debate survive to the present day. In the 1950s, molecular biology and the identification of DNA as “the secret of life” were seen as the final triumph of the mechanists’ physico-chemical view of living organisms. It became possible to describe the cell and multicellular organisms in precise molecular terms. However, the rest of the 20th Century and the beginning of the 21st Century provided a finely ironic turn to the philosophical debate.


As molecular biology advanced, it began to uncover ever more complex and sophisticated multi-molecular networks that carry out sensory, communication, regulatory and decision-making activities within and between cells (Gerhart and Kirschner 1997; Alberts, Johnson et al. 2002). At the same time, the 20th Century development of cybernetics, computers and electronic information-processing systems began to provide real-world examples for capacities the vitalists saw at work in living organisms. The information revolution had come to biology.


This contribution to the workshop attempts to outline how the biological information revolution and its underlying molecular observations impact our thinking about evolution. The intentionally ambivalent title is there for the following reason. Showing how previously excluded (i.e., impossible) notions have been supported by empirical observations inevitably allows us to consider previously excluded concepts as feasible (i.e., possible) hypotheses.


Lessons we have learned about the sources of hereditary variation since 1953


Transmissible hereditary changes provide the raw material for evolution. The elucidation of the structure of DNA made it possible to examine how change arises at the molecular level. The results have been revelatory (Shapiro 2011)[1]:


1.     Genome change is not the result of stochastic errors but of biochemical (i.e., cellular) action.


2.     Genome modifications do not arise solely within vertical lineages. Genome components from different lineages can be combined. The single most important evolutionary change, the origin of nucleated eukaryotic cells, involved the fusion of at least two different kinds of prokaryotic cells lacking defined nuclei.


3.     DNA change is a non-random process in the sense that it results from well-defined biochemical operations, each leaving a characteristic signature in DNA structure. Collectively, these are called “natural genetic engineering” operators. Cells synthesize, recombine, cut and splice, and otherwise modify their genomes in well-defined reactions.


4.     DNA change is non-random in the sense that it is subject to life-history regulation. The natural genetic engineering operators are subject to inhibition and activation by cellular regulatory regimes, often epigenetic in nature. These regulatory regimes respond to various sensory inputs in ways that activate natural genetic engineering when cell or organismal reproduction is challenged (McClintock 1984).


5.     DNA change is non-random in the sense that natural genetic engineering events can be targeted within the genome. Targeting occurs by a variety of molecular mechanisms that have distinct specificities: at certain DNA sequences, at certain DNA structures, or as a result of specific processes, such as replication or transcription.


6.     Evolutionary DNA change occurs rapidly at all genomic levels of complexity, from altering a single nucleotide to doubling the entire genome. These rapid changes are often combinatorial in nature and generate novel functionalities, either within a single protein molecule, by creating previously non-existent submolecular domains, or by modifying the structure of multimolecular networks.


7.     Cells execute purposeful DNA restructuring events during normal life-cycles in a non-random but also non-deterministic fashion. These goal-oriented natural genetic engineering processes occur in many organisms, including ourselves. The most integrated and highly regulated natural genetic engineering processes known occur in the vertebrate immune system, as it evolves and refines antibody receptors to recognize unpredictable invaders.


Seven widely held evolutionary opinions invalidated by modern observations


            A convenient way to see how much we need to revise our basic assumptions about evolutionary change is to look at seven statements that are widely accepted and generally go unchallenged in public discourse about evolution. The fact that there are well-documented counterfactuals to each indicates how useful a new set of basic evolution concepts could prove to be.


1. All heredity transmission occurs from parent to progeny. The counterfactuals to this opinion include:


(a) The susceptibility to all groups of organisms to virus infection, incorporation of viral sequences into the genome, and acquisition of unrelated genome segments by means of viral vectors;


(b) the ability of cells from all groups of organisms to incorporate DNA from the environment into their genomes;


(c) the prevalence of horizontal DNA transfer among prokaryotes, including endosymbiotic bacteria, and (as we are increasingly discovering) among eukaryotes as well as between prokaryotes and eukaryotes;


(d) the frequent occurrence in life history of symbiogenetic cell fusions creating organisms with merged genomes, generally separated into two or more specific subcellular compartments (Margulis and Sagan 2002).


2. Mutations are the result of inevitable replication errors. The counterfactuals to this opinion come in two categories – (i) cellular capacities to remove errors and (ii) mutational events that involve the action of dedicated natural genetic engineering functions:


(a) Exonuclease proofreading during DNA replication;


(b) postreplication mismatch repair to remove misincorporations;


(c) introduction of localized “point” mutations by Y class “mutator” polymerases


(d) incorporation of reverse-transcribed RNA sequences into the genome;


(e) insertion of mobile genetic elements or modules (transposons and retrotransposons) into new genomic locations (often the most common source of “spontaneous” mutation);


(f) genome rearrangements, including duplications (Ohno 1970) and further amplifications.


3.Mutations occur at constant low probabilities over time (= there are "mutation rates"). The counterfactuals to this opinion come from all studies of mutagenesis, which invariably uncover stimuli that make the observed mutation frequency a function of life history events:


            (a) Treatment with inorganic mutagenic agents, chemicals and radiation;


            (b) viral or bacterial infection;


            (c) nutritional or environmental stress;


(d) matings between different populations (“hybrid dysgenesis”) or interspecific hybridization.


4. Virus infection cannot induce DNA changes giving heritable resistance. This opinion has long cited the famous 1943 Luria-Delbrück experiment (Luria and Delbrück 1943) as evidence that viral infection cannot induce heritable resistance determinants in the infected cells. However, it turns out that such a sweeping conclusion is not valid. Prokaryotes have the CRISPR acquired immunity defense system, and animals can generate so-called “piRNA” molecules after viral or mobile element infection. Both prokaryotes and eukaryotes have the capacity to incorporate fragments of genome sequence from invading nucleic acids and use these fragments to generate small RNA molecules that block invader reproduction. As recently as the end of the 20th Century, such adaptive genomic defense strategies would have been dismissed as inconceivable. Yet their existence and functions have been incontrovertibly documented in this century.


5. Mutations cannot be targeted within the genome. This opinion is an integral part of the idea that genome change must be a random process. Conventional evolutionists have had to recognize mobile genetic elements and other natural genetic engineering systems, but they have held on to the idea of randomness by asserting, without empirical support, that change cannot be targeted. The truth is quite different, and we know that multiple well-documented molecular mechanisms of targeting DNA changes are at work in real time as counterfactuals to this unfounded opinion:


(a) Many nucleases or transposases target specific DNA sequences for double-streak breakage to initiate various genome changes, including homologous recombination in meiosis, mating-type switching in yeast, VDJ recombination in lymphocytes, and insertion of inteins, group I introns, and certain classes of retrotransposons;


(b) many viruses insert their genomes into special bacterial chromosome attachment sites by site-specific recombination, a process that is also used for constructing multiple antibiotic resistance determinants (integrons), constructing long multi-coding sequence virulence and other arrays in bacterial genomes (superintegrons), rearranging DNA to turn genome expression on and off (phase variation), removing “intervening DNA” from protein coding sequences during terminal differentiation of specialized bacterial cells, and resolving structures containing duplications that form in chromosome replication and replicative transposition;


(c) special accessory transposition proteins target the bacterial Tn7 transposon either to a dedicated “attTn7” site in the bacterial chromosome (Craig 1991) or to replicating plasmid DNA during intercellular transfer (Wolkow, DeBoy et al. 1996);


(d) yeast retrotransposons are targeted by protein-protein interactions with transcription factors or chromatin proteins to insert upstream of transcription start sites in euchromatin (Ty1-Ty4) or in silent heterochromatin (Ty5);


(e) Drosophila P transposons are “homed” to preferred chromosome regions when they contain cis-acting regulatory sites active in those regions;


(f) group II “retrohoming” introns recognize their specific insertion sites by pairing between nucleotides in the spliced RNA and target DNA;


(g) both V region-specific somatic hypermutation and isotype switching of heavy chain exons in activated B lymphocytes are limited to certain transcribed regions and targeted by signals controlling where transcription occurs.


6. Spontaneous hereditary changes are localized and limited to those of small effect. This opinion originated with Darwin’s adherence to his geology professor Charles Lyell’s Uniformitarian philosophy that long accumulation of gradual changes would result in the major transformation currently observable (Lyell 1830). Darwin stated this view in his famous quote from Chapter 6 of Origin of Species: “If it could be demonstrated that any complex organ existed, which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down. But I can find out no such case.” (Darwin 1859) Although conventional evolutionists acknowledge larger genome changes, they typically treat them as random events of no particular significance for the main lines of evolutionary theory. The counterfactuals to this opinion include:

(a) Protein evolution frequently occurs not by changes in single amino acids but by “domain shuffling,” the exchange of longer stretches of DNA encoding semi-independent functional “domains” – a process that involves cutting and splicing DNA rather than simple replication errors;

(b) “homeotic” mutations in higher-level genome patterning components, such as the animal Hox complexes, have dramatic and extensive effects on the structures of multicellular organisms; such mutations were studied by proponents of saltational changes in evolution such as Bateson (Bateson 1894) and Goldschmidt (Goldschmidt 1933);

(c) episodes where natural genetic engineering has been activated by unusual mating or stress frequently results in the occurrence of multiple coincident changes to the genome;

(d) the many examples of whole genome doubling following interspecific hybridization in real time or documented to have occurred at critical transition points in the DNA sequence record.

7. Cells cannot integrate DNA change operations with adaptive needs. Part of the determined “materialism” of conventional evolutionary theory is the denial that natural genetic engineering operations can be integrated with the adaptive needs of the organism experiencing genome change. Once more, this is a philosophical position without empirical support. The counterfactuals include:


(a) Germline or other tissue specificity of action by many mobile genetic elements in animals;


(b) regulation of nuclease action to initiate adaptive processes like meiotic recombination, mating type switching, and provirus expression;


(c) coordination of the multiple cleavage, splicing and telomere addition events in ciliate protist macronucleus development after sexual exchange;


(d) the various recombination and DNA rearrangement events underlying microbial phase variation and protein engineering for immune evasion and modulation of intercellular contacts;


(e) the tightly regulated and highly coordinated natural genetic engineering processes in lymphocytes that result in antigen receptor formation as well as antibody diversification, affinity maturation, and functional redeployment during the primary and secondary immune responses.


How do we change our thinking? Genomes as Read-Write (RW) memory organelles


            The overarching take-home lesson from all the preceding counterfactuals is that we need to rethink how the genome functions as a cellular memory device (Wilson 1928). Conventional evolutionary theory treats the genome as the source code for cell and organism characters – essentially as a read-only memory (ROM) with no active input and subject to change through copying errors. The 21st Century alternative view is to treat the genome as a read-write (RW) memory system, more like an iPod than a blueprint or even a DVD[2].


            The active processes of inscribing information onto the RW genome are multiple and operate at roughly three different time scales:


-       Within cell cycles, there are transient forms of information stored in the form of meta-stable nucleoprotein complexes that influence the operation of processes such as replication, transcription, repair, and cell division.


-       Over multiple cell cycles, the predominant form of inscription is epigenetic modification of chromatin states. Chromatin formatting is inheritable at cell division but can also be modified in response to stimuli or during cell differentiation. Many chromatin states are “imprinted” during gamete formation; they are reset at each sexual generation and expression often depends on the gender of the contributing parent. In microbes, reversible DNA modifications also contribute to inscriptions heritable for a limited number of cell generations, and there is emerging evidence that DNA modifications may play a role in diversifying terminally differentiated cells in somatic tissues (Kano, Godoy et al. 2009; Astolfi, Salamini et al. 2010; Singer, McConnell et al. 2010; Baillie, Barnett et al. 2011; Eickbush and Eickbush 2011; Kazazian 2011).


-       Over evolutionary time, the chief form of inscription is a change in the structure of the genome. As mentioned, these inscriptions range from small changes to massive duplications, amplifications and rearrangement of cellular DNA molecules. Genome restructuring at any level is rare during well-adapted growth conditions but can increase by orders of magnitude when reproduction is threatened or after an interspecific hybridization. Since suboptimal conditions and exceptional mating events are most prevalent following ecological collapse, it is likely that significant evolutionary episodes are linked to disruptions of the biosphere. This would explain “punctuated equilibrium” patterns (Eldredge and Gould 1972) and the linkage between mass extinctions followed by bursts of origination in the fossil record (Erwin 2001; Jablonski 2001; Pave, Herve et al. 2002).


Functional innovation, not selected optimization of fitness, is the key problem in evolutionary change


            Given all the changes introduced into our knowledge of hereditary variation and evolution by molecular genetics and genome sequencing, it is time to re-examine our most basic assumptions about evolution. Ever since Darwin, the mainstream focus has been on optimizing reproductive success (fitness) by natural selection of random variants.


I have argued in my book (Shapiro 2011) and online that it may be a fundamental misapprehension to think of natural selection as the creative force in evolutionary diversification. By definition, selection can only operate after change has taken place. As long as changes were “numerous, successive, slight” (Darwin 1859), it could be argued that the gradual accretion of many changes over long periods of time guided the formation of evolutionary novelties, like new cell types, adaptive functions, and features of multicellular organisms, like limbs and organs.


Decades of study on fitness optimization notwithstanding, we have now learned about processes that generate rapid genome-wide change, like symbiogenesis and hybridizations, and we can see traces of these events in the genome sequence record. Thus, it has become necessary to question what biological processes truly constitute the sources of adaptive novelty.


Selection works by testing changed organisms against their unchanged progenitors. Hereditary changes are thus the ultimate source of novelty. We know that natural genetic engineering is non-random, sensitive to external inputs, and provides all the molecular tools necessary for controlling the genome restructuring process. Could it be that cell regulation of natural genetic engineering is the true source of complex evolutionary innovations that have adaptive utility? Today we are in a position to ask this question empirically at the cellular and molecular levels. My guess is that doing so in the decades to come will prove as eye-opening as the last 60 years of molecular biology.







Alberts, B., A. Johnson, et al. (2002). Molecular Biology of the Cell New York and London, Garland Science.

Astolfi, P. A., F. Salamini, et al. (2010). "Are we Genomic Mosaics? Variations of the Genome of Somatic Cells can Contribute to Diversify our Phenotypes." Curr Genomics 11(6): 379-386.

Baillie, J. K., M. W. Barnett, et al. (2011). "Somatic retrotransposition alters the genetic landscape of the human brain." Nature.

Bateson, W. (1894). Materials for the Study of Variation Treated With Especial Regard to Discontinuity in the Origin of Species. London, Macmillan.

Craig, N. L. (1991). "Tn7: a target site-specific transposon." Mol Microbiol 5(11): 2569-2573.

Darwin, C. (1859). Origin of Species London, John Russel.

Driesch, H. (1908). The science and philosophy of the organism : Gifford lectures delivered at Aberdeen university, 1907-1908 by Hans Driesch. Aberdeen, Printed for the University.

Eickbush, M. T. and T. H. Eickbush (2011). "Retrotransposition of R2 elements in somatic nuclei during the early development of Drosophila." Mob DNA 2(1): 11.

Eldredge, N. and S. J. Gould (1972). Punctuated equilibria: an alternative to phyletic gradualism. Models in Paleobiology. T. J. M. Schopf. San Francisco, Freeman, Cooper and Company: 82-115.

Erwin, D. H. (2001). "Lessons from the past: biotic recoveries from mass extinctions." Proc Natl Acad Sci U S A 98(10): 5399-5403.

Gerhart, J. and M. Kirschner (1997). Cells, Embryos, and Evolution. Malden, MA, Blackwell Science.

Goldschmidt, R. (1933). "Some aspects of evolution." Science 78(2033): 539-547.

Huxley, J. (1942). Evolution: the modern synthesis. London, Allen & Unwin.

Jablonski, D. (2001). "Lessons from the past: evolutionary impacts of mass extinctions." Proc Natl Acad Sci U S A 98(10): 5393-5398.

Kano, H., I. Godoy, et al. (2009). "L1 retrotransposition occurs mainly in embryogenesis and creates somatic mosaicism." Genes Dev 23(11): 1303-1312.

Kazazian, H. H., Jr. (2011). "Mobile DNA transposition in somatic cells." BMC Biol 9: 62.

Kuhn, T. S. (1962). The Structure of Scientific Revolutions Chicago, Univ. of Chicago Press.

Luria, S. E. and M. Delbrück (1943). "Mutations of Bacteria from Virus Sensitivity to Virus Resistance." Genetics 28(6): 491–451.

Lyell, C. (1830). Principles of Geology. Edinburgh, John Murray.

Margulis, L. and D. Sagan (2002). Acquiring Genomes: A Theory of the Origins of Species. Amherst, MA, Perseus Books Group.

McClintock, B. (1984). "The significance of responses of the genome to challenge." Science 226(4676): 792-801.

Ohno, S. (1970). Evolution by Gene Duplication London, George Allen and Unwin.

Paley, W. (1802 (republished 2006)). Natural Theology, or Evidences of the Existence and Attributes of the Deity, Oxford University Press. .

Pave, A., J. C. Herve, et al. (2002). "Mass extinctions, biodiversity explosions and ecological niches." C R Biol 325(7): 755-765.

Roux, W. (1895). Gesammelte Abhandlungen & Ueber Entwickelungsmechanik der Organismen.

Shapiro, J. A. (2011). Evolution: A View from the 21st Century. Upper Saddle River, NJ, FT Press Science.

Singer, T., M. J. McConnell, et al. (2010). "LINE-1 retrotransposons: mediators of somatic variation in neuronal genomes?" Trends Neurosci 33(8): 345-354.

Stott, R. (2012). Darwin's Ghosts: In Search of the First Evolutionists. London, Bloomsbury.

Wilson, E. B. (1928). The Cell in Development and Inheritance, 3rd ed., revised and enlarged. New York, MacMillan.

Wolkow, C. A., R. T. DeBoy, et al. (1996). "Conjugating plasmids are preferred targets for Tn7." Genes Dev 10(17): 2145-2157.



[1] Almost all the references for the statements that follow are among those posted online and updated periodically at the website for my book ( and will be given principally in the form of links to individual pages accessible there. Further discussions of many individual points can be found at my blog

[2] Thanks to Perry Marshall for this analogy.