2.3 DNA as a Developmental Resource

The organization of a living thing is not stored as a complete plan anywhere. That claim is easy enough to state and rather harder to accept, especially when we turn to DNA. If form is not kept in the starting materials, what exactly does DNA contribute to development?

When thinking about the role of genes in development, especially in the early development of the embryo, it is tempting to say that information in DNA drives the process. In this picture, genes are in control: they cause things to happen and direct the formation of the organism. The appeal is obvious, because it gives us a simple answer to the question of why development is so reliable. But it also gives DNA a kind of agency it doesn’t have. DNA doesn’t do anything on its own; its effects depend on the organized cellular context in which it is used.

In his book Understanding Metaphors in the Life Sciences, Andrew Reynolds (2022) describes the language of genetics as a thicket of metaphors, and it is a tangled one. Genes act, DNA contains a code, and the genome is a blueprint, a program, or a book waiting to be read. These different metaphors are not all the same, and each one draws our attention to some features of molecular biology while overlooking others.

Code, Blueprint, Program

The genetic code is a good example. At the molecular level, the language of a “code” has a specific and useful meaning: sequences of nucleotides in DNA are transcribed into RNA, and groups of three nucleotides correspond to amino acids in a polypeptide chain. That is a real and precise relationship. But it doesn’t follow that DNA codes for an organism in the same sense. Even making a single protein depends on an already organized cell, with its RNA molecules, ribosomes, enzymes, membranes, and sources of energy.

A blueprint makes a different kind of claim: it represents the structure that is to be built. The metaphor of a genetic program adds something more, not just stored information but a sequence of operations that gets executed and controls what happens.

These metaphors encourage us to picture development as the realization of something already present in DNA. But once we stretch them from proteins to whole organisms, the analogy starts to fall apart. What reads the blueprint, or runs the program? DNA can’t do either; it is acted on by the very cellular system it is supposed to be directing.

None of this makes genes unimportant. DNA sequences are indispensable developmental resources, and changes in those sequences can have profound effects. The problem is the leap from “genes participate in producing proteins” to “the genome contains a plan for the organism.” The molecular processes are real enough; it’s the bigger interpretation we’ve built on top of them that needs rethinking.

One response is to replace the blueprint with another metaphor. The developmental biologist Alfonso Martinez Arias (2023), for example, describes cells as “master builders” that draw on a genetic toolkit. This is useful because it shifts attention from DNA to the activity of cells, and it helps explain how cells with the same DNA can become very different kinds of cells. But it is still a metaphor, and we should not ask it to do too much. Cells do not stand outside development, consulting DNA and deciding what to build. They are themselves products and participants in the developmental processes we are trying to understand.

The program metaphor has a more sophisticated form in contemporary developmental biology. Research on developmental gene regulatory networks has shown how regulatory DNA and interacting transcription factors help organize remarkably precise sequences of gene expression. Peter and Davidson (2015), for example, describe these networks as genomic programs that causally specify many aspects of cell fate and body-plan development. This is a more specific mechanistic account than the popular image of a blueprint for the whole organism.

Even so, these networks do not operate independently, since their activity depends on maternally supplied factors, chromatin state, signaling between cells, geometry, and the mechanical conditions of developing tissues. Genomic regulation plainly matters. The question is whether it can stand as the source of developmental organization, or whether it is one part of an organized developmental system.

Starting From an Organized Cell

If cells are the builders, we still have to explain where their organization comes from. Following that question takes us back to the cell from which development begins.

Development begins from a single cell, but not from an unorganized one. The zygote contains not just DNA but proteins, RNAs, mitochondria, membranes, and cytoskeletal architecture inherited largely through the egg. During the oocyte-to-embryo transition, maternal RNAs and proteins, organelles, and spatial organization sustain early development and help establish the conditions under which the zygotic genome becomes active (Conti & Franciosi, 2018). Localized molecules, gradients, and cytoplasmic movements contribute to the spatial and temporal patterning of early gene expression. DNA always begins and continues within an already organized cellular context. There is no stage at which DNA issues instructions to otherwise unorganized matter.

You cannot create an organism from DNA alone, because DNA has to function within an appropriately organized cell and developmental system. This is easy to lose sight of in media accounts of attempts to bring the woolly mammoth and other extinct animals “back to life.” In the case of the mammoth, the proposed process begins with living elephant cells, which are edited to introduce selected mammoth-associated variants, and an edited nucleus is then placed in an elephant egg (Colossal Biosciences, n.d.).

The result, if it can be produced, would therefore be an engineered elephant with some mammoth-like traits. That is not a trivial achievement, but it is not the same as reconstructing a mammoth from preserved DNA, because the developmental system doing the work is a contemporary elephant one. Whatever such an animal is called, DNA by itself is not enough to bring it into being.

Where the Instructions Actually Are

If DNA is not an instruction manual, it is fair to ask what is doing the organizing instead. One part of the answer that does not get enough attention is that a good deal of organization comes from the physical properties of the materials themselves. Collinet and Lecuit (2021) examine how genetically programmed processes and self-organization work together through biochemical signaling, mechanics, and geometry during morphogenesis.

The physicist Raghuveer Parthasarathy (2022) discusses self-assembly, in which organization arises through the physical properties of the materials themselves. Molecules interact in the ways they do because of their configuration and chemical structure. The constraints on what can happen next are therefore inherent in the materials rather than imposed on them from somewhere else. What can look like decision making at the cellular level is realized through the activities of molecules whose possibilities are built into their structures and modes of interaction.

Consider cell migration in the embryo. Migrating cells respond to mechanical as well as chemical features of the tissues through which they move, and changes in tissue mechanics can alter the forces and signaling processes through which migration occurs (Blackley et al., 2021). Nothing in that sequence needs to be read off from a central set of instructions. Mechanical properties of the surrounding tissue become part of the conditions to which cells respond, and those responses alter what happens next. The outcome can be reliable without being specified in advance.

This scales upward without becoming a one-way hierarchy. The structure of biological molecules constrains how they interact; those interactions constrain what cells can do together; and the organization of cells and tissues in turn alters the conditions under which molecular interactions occur. These relationships change what the organism can do next, connecting molecular activity to the space of possibilities discussed in The Riddle of Form and Matter.

The developmental biologist Stuart Newman has developed this point through the idea of dynamical patterning modules, in which gene products work together with physical processes such as adhesion, diffusion, viscoelasticity, and oscillation (Newman & Bhat, 2008). Layers, cavities, folds, segments, and appendages can arise because cohesive and chemically active masses of cells have particular physical possibilities, not because each form is separately specified in DNA.

Genes remain consequential because their products help give cells the properties through which these processes occur, but the organizing work is distributed across molecular products, physical forces, cellular activity, and the conditions in which they operate. This also helps explain why disrupting a gene can have reliable effects without making that gene the original source of organization.

What DNA Is For

If DNA is not a plan, what is it doing there at all, and why is it so central to living things?

Life cycles, and life itself, are processes in which cellular activities, molecular patterns, and physical forces maintain and reconstruct each other. Exactly how DNA arose within the earliest forms of life remains uncertain. Different accounts emphasize RNA-based replication, early metabolic networks, membrane-bound compartments, or some combination of these, and laboratory work has shown that parts of such systems can interact in ways that begin to resemble heredity and reproduction (Joyce & Szostak, 2018).

None of this gives us a settled history of what happened. For the argument here, the important point is that these proposals concern organized chemical processes rather than DNA acting alone. They investigate how replication, catalysis, and compartment formation could become coupled. The precise transition to DNA-based heredity remains an open historical question. Describing DNA as a developmental resource does not depend on settling that question: in living cells today, its replication, repair, and use depend on the wider cellular organization.

This distinction is important for understanding what DNA’s stability means. Cells copy DNA with high accuracy and repair damage to it. This helps maintain sequences across generations and contributes to developmental reliability. But it should not lead us back to seeing genes as instructions. Storage implies the preservation of patterns, whereas instructions imply an agent following directions. DNA is a particularly steady resource within a broader system of processes that maintain and reconstruct themselves through time, which is a different thing from being a plan.

This returns us to the life cycle discussed in Life Cycles and the Direction of Development. DNA is one of the resources used in its reconstruction. Whether it is a resource like any other, or whether its stability makes it special, is a question taken up in Developmental Systems Theory.

Are Genes in Control?

If genes do not specify phenotypes in any simple way, we might still want to ask whether they are at least in control of development. The idea that genes control biological processes is pervasive in how biology is taught and communicated. But as McKenna and colleagues have pointed out, biologists who use the language of genetic control seldom define what they actually mean by it. The word “control” is used as if its meaning were obvious, but when you press on it, the concept falls apart.

Consider what happens when we ask what controls gene expression. A gene is turned on by a transcription factor. But what activated the transcription factor? A signaling cascade. What initiated the signaling cascade? A signal from a neighboring cell. What determined that the neighboring cell would send that signal? Its own developmental history, including its own patterns of gene expression, which were themselves shaped by earlier signals. And so on. You can trace the chain of causes backward indefinitely without ever arriving at a starting point that is “in control.” The search for a master controller leads to an infinite regress, because development is a system of mutually dependent processes rather than a chain of command.

McKenna and colleagues make this point with a simple analogy. Think about trying to identify the master controller of an automobile. The engine seems essential, but the engine depends on the electrical system, which depends on the battery, which is charged by the alternator, which is powered by the engine. Each component is necessary, but none is in control of the vehicle’s overall function. Something similar can be said of organisms: genes and their products play important roles, but they can only function within appropriate cellular environments, and those environments are themselves products of developmental processes that involve far more than gene expression. They call the search for a primitive causal controller in an automobile a “fool’s errand” and extend the argument to organisms (McKenna et al., 2022, p. 91).

The infinite regress is only a problem if you are looking for a controller. Search the organism for the component that is in charge and you won’t find it. That isn’t because you haven’t looked hard enough; it’s because it isn’t there.

The importance genes have is different from what the language of control suggests. Genes provide molecular sequences that cells use: they are resources, not controllers. The metaphors deserve attention here, because when we describe genes as controlling development, or the genome as a blueprint or program, we are implying a level of understanding and causal clarity that does not actually exist. As McKenna and colleagues note, these metaphors allow us to talk efficiently about processes we do not fully understand, but to the outside world they suggest a level of knowledge that is not really there.

The late Evelyn Fox Keller (2014) argued that these issues have consequences that go beyond genetics. The genome is largely reactive to its cellular environment, since what genes do depends on signals from transcription factors, neighboring cells, and the molecular context in which they operate.

If that is the case, the genome is itself responding to its environment. That calls into question the entire distinction between nature, usually taken as genes, and nurture, usually taken as the environment. If much of what the genome does is respond to its surroundings, then dividing developmental influences into “genetic” and “environmental” categories does not make much sense, because the genome is not on one side of a divide but is part of a system in which the two are entangled from the start.

Mutations in DNA can, of course, be associated with serious diseases and developmental disorders. If genes are developmental resources, it makes sense that the path of development may be altered when there is a problem with those resources. At the same time, the same pathogenic variant can produce a wide range of outcomes in different people, from no apparent clinical effect to severe disease. Geneticists refer to these patterns as incomplete penetrance and variable expressivity (Kingdom & Wright, 2022). The variation can reflect genetic background, regulatory differences, mosaicism, developmental timing, stochastic cellular processes, environmental conditions, and compensatory mechanisms. A mutation can therefore be highly consequential without functioning as an independent instruction for the phenotype.

The Ontogeny of Information

We have been talking about information in a fairly loose way. What does the term actually mean in the context of development?

In 1985, the scientist and philosopher of biology Susan Oyama published a book called The Ontogeny of Information that challenged the way biologists and psychologists talked about genes and development (Oyama, 2000). Oyama argued that the entire way we think about genetic “information” is confused, and she wanted to take apart the assumption that information exists in the genes, waiting to be read out. Her argument was that development is not the unfolding of preexisting information, and that information itself is constructed through developmental processes.

What does this mean? Consider that the same DNA sequence can have very different effects depending on when and where it is active during development. In a liver cell, a particular gene might produce a specific protein, whereas in a neuron the same gene might be silenced entirely, or might produce a different protein through alternative splicing. The sequence itself has not changed, but its role in development, and what that sequence means for the system, is completely different (Taylor et al., 2023).

So information is not a fixed property of DNA waiting to be decoded. It emerges through the encounter between a DNA sequence and a particular cellular context at a particular moment. The relationship also runs both ways. What the cell does depends on which DNA sequences are available and active, so neither side can be fully specified on its own. The DNA’s significance emerges only in relation to the cell, and the cell’s activity emerges only in relation to its DNA.

Yes, this is a version of asking which came first, the chicken or the egg, but that’s where this ends up. There is no independent starting point at which DNA first acquires its developmental significance before the cellular context exists, or the cellular context exists independently of DNA. But that’s OK, and at least we understand what the problem is.

This is what Oyama means by the ontogeny of information. The informational relationships that matter for development are not present at the start as content compressed within the zygote’s DNA. They emerge, and are reconstructed, as development proceeds. Information, in this sense, has a developmental history.

Humberto Maturana made a closely related point when he treated information as belonging to an observer’s description rather than as something that explains the operation of a living system (Maturana & Varela, 1980). It can describe relationships that have already been established, but it does not explain how those relationships came into being.

What is inherited, then, is not genetic information on one side and an empty context on the other. The zygote is already an organized developmental system in which DNA sequences have important roles, but what those sequences do, and what they come to mean for development, depends on the cellular and developmental contexts in which they participate.

Oyama also argues against a “ghost in the machine,” meaning some central, fixed interpreter of information that drives development by reading off what is in the genes. But if the information content of a gene is not constant, there can be no fixed interpreter or interpretation. A related point she raises is that the genome cannot be the “unmoved mover” from which development proceeds, because the genome itself is constructed, maintained, and transmitted by the very developmental system it is supposed to direct.

Where This Leaves the Blueprint

Preformationism, in its broadest sense, is the idea that something present at the beginning of development specifies the outcome. The little person in the sperm was one version of this, and the idea that DNA contains a blueprint for the organism is another. The language changed, but the logic did not: the outcome is already specified at the start and development is just the unfolding of what was already there.

What we have seen in this reading makes the simple molecular version of that position very difficult to sustain. The genomic sequence does not by itself specify the phenotype, and the relationship between genotype and phenotype is mediated by regulatory networks, cellular organization, mechanical and biochemical interactions, and developmental history. Gene regulatory networks are unquestionably causal, but that is different from saying that the organization of the phenotype is already contained in DNA and then read out during development. In this sense we have moved back towards epigenesis, the view that development is a process in which new organization arises rather than one in which a preexisting plan is simply unfolded.

But as we saw in Made or Given?, epigenesis has always faced a challenge of its own. The original preformationists charged it with vitalism, meaning that it relied on some mysterious and ill-defined life force to explain how new structures and forms could arise during development. That charge is not answered by showing that genes do not determine development, because pointing out what fails to explain something is not the same as explaining it.

Part of the answer was developed in The Riddle of Form and Matter. While genetic and cellular processes are not deterministic in any simple sense, the developmental system as a whole is robust, reliably producing organized outcomes through constraints that have been stabilized over evolutionary time and reconstructed in each generation.

Marc Kirschner and colleagues anticipated a shift toward this view in a 2000 article entitled “Molecular Vitalism?”, written before the completion of the Human Genome Project. They pointed out that since gene products function in multiple pathways, and the pathways themselves are interconnected in networks, there are many more possible outcomes than there are genes. For Kirschner and colleagues, the genotype cannot be predictive of the actual phenotype, and can only provide knowledge of what they called the “universe of possible phenotypes” (Kirschner et al., 2000, p. 87), which you may recognize as another way of describing a space of possibilities.

Living systems have evolved to restrict these phenotypes, and in self-organizing systems the phenotype depends as much on external conditions and contingent events as on the molecular components encoded in the genome. Yet out of this potentially nondeterministic world, organisms have fashioned remarkably stable physiology and embryology. It is this robustness, Kirschner and colleagues argue, that originally suggested vital forces, and it is this robustness that we need to understand.

That seems to me a better place to be than where we started. The blueprint offered an explanation of developmental reliability that turned out not to be an explanation at all, since it simply relocated the organization that it was supposed to account for. We are left with the research problem of how that reliability is produced.

Bigger and More Interesting Questions

When we give up the idea that genes direct development from outside the process, or that DNA is a blueprint or an instruction manual, bigger and more interesting questions appear. In The Master Builder, Alfonso Martinez Arias (2023) develops the same shift by treating DNA as a resource for making RNAs and proteins while placing the construction of bodies in the organized activity of cells. The question then becomes how cells build bodies without a plan to follow, and where the organization comes from if not from something stored somewhere waiting to be read out.

The Riddle of Form and Matter drew on Aristotle to argue that what makes a living thing the kind of thing it is cannot be found by looking harder at what it is made of. This reading has reached the same conclusion about DNA. Looking more closely at DNA, or sequencing it more completely, does not reveal a complete plan for development, because no such plan is there to find. The organization of a living thing is not stored as a complete description in any one place. It is realized through developmental processes that recruit many resources, DNA among them.

Sources

Blackley, D. G., Cooper, J. H., Pokorska, P., & Ratheesh, A. (2021). Mechanics of developmental migration. Seminars in Cell & Developmental Biology, 120, 66–74. https://doi.org/10.1016/j.semcdb.2021.07.002

Collinet, C., & Lecuit, T. (2021). Programmed and self-organized flow of information during morphogenesis. Nature Reviews Molecular Cell Biology, 22, 245–265. https://doi.org/10.1038/s41580-020-00318-6

Colossal Biosciences. (n.d.). Woolly mammoth de-extinction project & process. Retrieved October 16, 2026, from https://colossal.com/mammoth/

Conti, M., & Franciosi, F. (2018). Acquisition of oocyte competence to develop as an embryo: Integrated nuclear and cytoplasmic events. Human Reproduction Update, 24(3), 245–266. https://doi.org/10.1093/humupd/dmx040

Joyce, G. F., & Szostak, J. W. (2018). Protocells and RNA self-replication. Cold Spring Harbor Perspectives in Biology, 10(9), a034801. https://doi.org/10.1101/cshperspect.a034801

Keller, E. F. (2014). From gene action to reactive genomes. The Journal of Physiology, 592(11), 2423–2429. https://doi.org/10.1113/jphysiol.2014.270991

Kingdom, R., & Wright, C. F. (2022). Incomplete penetrance and variable expressivity: From clinical studies to population cohorts. Frontiers in Genetics, 13, 920390. https://doi.org/10.3389/fgene.2022.920390

Kirschner, M., Gerhart, J., & Mitchison, T. (2000). Molecular “vitalism.” Cell, 100(1), 79–88. https://doi.org/10.1016/S0092-8674(00)81685-2

Martinez Arias, A. (2023). The master builder: How the new science of the cell is rewriting the story of life. Basic Books.

Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and cognition: The realization of the living. D. Reidel. https://doi.org/10.1007/978-94-009-8947-4

McKenna, K. Z., Gawne, R., & Nijhout, H. F. (2022). The genetic control paradigm in biology: What we say, and what we are entitled to mean. Progress in Biophysics and Molecular Biology, 169–170, 89–93. https://doi.org/10.1016/j.pbiomolbio.2022.02.003

Newman, S. A., & Bhat, R. (2008). Dynamical patterning modules: Physico-genetic determinants of morphological development and evolution. Physical Biology, 5(1), 015008. https://doi.org/10.1088/1478-3975/5/1/015008

Oyama, S. (2000). The ontogeny of information: Developmental systems and evolution (2nd ed.). Duke University Press. https://doi.org/10.1515/9780822380665

Parthasarathy, R. (2022). So simple a beginning: How four physical principles shape our living world. Princeton University Press. https://doi.org/10.1515/9780691231617

Peter, I. S., & Davidson, E. H. (2015). Genomic control process: Development and evolution. Academic Press.

Reynolds, A. S. (2022). Understanding metaphors in the life sciences. Cambridge University Press. https://doi.org/10.1017/9781108938778

Taylor, J., Weiss, S. M., & Marshall, P. J. (2023). Genes, genomes, and developmental process. Behavioral and Brain Sciences, 46, e204. https://doi.org/10.1017/S0140525X22002199

Explore Further

Pigliucci, M. (2010). Genotype–phenotype mapping and the end of the “genes as blueprint” metaphor. Philosophical Transactions of the Royal Society B: Biological Sciences, 365(1540), 557–566. https://doi.org/10.1098/rstb.2009.0241

A discussion of why the blueprint metaphor breaks down when genotype–phenotype relations are understood as developmental processes.