1Mr Administrator,
My dear colleagues,
My dear friends,
Ladies and gentlemen,

2What is life? Though life appears self-evident, it actually remains quite elusive. Is it a substance, a structure, a process? For Henri Bergson, “life is evolution”; “the organism changes at every moment. That which is real is the continuous evolution of form: form is only an instantaneous snapshot of a transition”.1

3These words highlight a central property of living systems, namely an internal power of transformation, movement, and evolution; a dynamic peculiar to it. They also raise two formidable questions: what is the origin of this dynamic? How can we comprehend the order and permanence of living forms, and ours in particular, considering the dynamics that characterize their development, their incessantly renewed organization, and their history? For example, biological species, long conceived of according to a creationist belief, transform and evolve over millions of years. The majority of non-neuronal cells in humans are renewed in less than ten years, and in certain organs such as the intestine, in five days. Most of the molecules composing a cell are renewed in a few hours, while the components of every organelle are renewed in just a few minutes or seconds. In What is Life?2, Erwin Schrödinger contrasts the stability of forms, such as the prognathism of the Habsburg lineage and the stochastic fluctuations at the atomic and molecular scales modelled by late nineteenth-century statistical physics, which governs the chemical information underlying heredity. How do an organism and all biological structures retain a constant physiological state, that is, homeostasis? Where does the stable geometry of the living come from, considering that “a current of materials […] continuously traverses the body and renews it in its substance”3, as Claude Bernard so neatly put it? How does the biological order emerge from the chaos present on the molecular level? Revolutionary innovations in microscopy over the past twenty years have shed light on the unfathomed extent of these dynamics, allowing us, as we will see, to renew our understanding of the particularity of living matter and its complex self-organization.

4We have difficulty comprehending the multiple dynamics of the living, given the extent to which the fixedness of the representations constructed by our brains is a powerful tool for grasping the diversity of the world as it is perceived by our senses. It is even more difficult to understand the paradoxical union between a fixed organization and a ceaseless dynamic.

5By deciding to create the Chair of Dynamics of Living Systems the Faculty has recognized the importance of this paradox, and has understood that it is a path for exploring the quid proprium of living matter and the organization of its dynamic features. This inaugural lecture intends to illustrate and to explore this idea in more depth.

6I am very grateful to Edith Heard for having put her trust in me and for having presented and supported the creation of this Chair and my candidacy with patience, thoughtfulness, and enthusiasm; Alain Prochiantz, for his acute intelligence, humour, and attentiveness; and all of you, dear colleagues, for your warm welcome and ardent curiosity. I feel very honoured and truly delighted to find myself among you.

7I give particular regard to all of you to whom I owe my presence here today. From my mentors Claude Desplan, Stephen Cohen, and Eric Wieschaus, I retain not only the liveliness and creativity of their scientific minds, but also their generosity, their passion, and their kindness. My students, postdoctoral researchers, and engineer colleagues and co-workers, particularly the physicist Pierre-François Lenne, have given me the thirst to understand things together, shown me the necessity of doubt and how to avoid false paths, and also illustrated the joy of discovery by opening new vistas onto the great unknown. Lastly, I would like to thank my wife Marie, my constant support, who through her dynamism and confidence has imbued me with the spirit of adventure and the courage to keep going further; and as well as our four children, who have been our most relevant lesson in the dynamics of living systems!

8Now, I am faced with a colossal programme, the vast scope of which I am well aware. I am following in the steps of my predecessors, in particular Nicole Le Douarin, professor of Cellular and Molecular Embryology. The Collège de France has placed a great deal of emphasis on the study of the formation of organisms. What was once called “the theory of generation” became, during the nineteenth century, experimental or comparative embryology. Starting in the 1980s, driven by molecular genetics, we spoke of cellular and molecular embryology, followed by developmental biology and genetics, a discipline taught within these walls by Spyros Artavanis-Tsakonas.

9The title “Dynamics of Living Systems” marks a break and a thematic and methodological expansion. In this lecture, we shall see how the study of the development of organisms initially drew from other domains in biological sciences, in particular cellular biology, and then more decisively from experimental and theoretical physics. This multidisciplinary context has profoundly changed our understanding of living systems. The anchoring of biology in chemistry, the major accomplishment of the past century, has permitted the growth of approaches and models on the molecular level; for instance, the discovery of the structure of DNA, the molecular basis of heredity, and the central dogma of molecular biology formulated by Francis Crick. Yet a living system is a self-organized dynamic assembly whose understanding requires theoretical tools that are unknown to the majority of biologists. These tools have mainly been developed by statistical physics to study the collective properties of matter. Pierre-Gilles de Gennes, professor of Condensed Matter Physics, offers an eloquent example of this. He inspired and trained a generation of physicists who shed light on certain remarkable properties of living matter. Biologists owe him a great deal. Moreover, they recently gained access to an almost exhaustive description of the chemical components of living matter, of its multi-level organization, and its internal dynamics, without which physicists would be powerless. I will evoke the fruitfulness of recent links between biologists and physicists.

From the geometry to the dynamics of the living

10Following a century-long biochemical revolution, there are grounds to think that the twenty-first century could be that of the physics of living systems. This statement might provoke scepticism among biologists, who, despite the fact that they do not doubt the physical nature of living matter, nonetheless reject the possibility of accounting for its unique functioning by means of the quantitative and predictive models of physics. However, in the seventeenth century, physics, “the science with the goal of studying natural things” according to the Dictionnaire de l’Académie française in 1694, encompassed the study of inert bodies and living entities. Biology as an independent science did not yet exist and would emerge in the early nineteenth century only, with the coining of the term biologie by Jean-Baptiste de Lamarck in France and Gottfried Reinhold Treviranus in Germany. The natural philosophy of Galileo, followed by Gottfried Wilhelm Leibniz and Isaac Newton, applauded the possibility of stating the physical laws of nature in mathematical language.

11Let us go over, in very large steps, the evolution of physics models of living matter. The challenge was to analyze the autonomy of life sciences and, beyond scientific methods, the ontological specificity of a living system compared to an inert body. Does material unity exist in the natural world, between inert matter and living matter? Are the laws of physics of the world unified, or is it necessary to invoke forces peculiar to living systems? What is the particularity of the living?

12The materialist philosophy of the Ancient atomists – Leucippus, Democritus, and Epicurus – describes a universe composed of atoms and void, the combinations of which produce the variety of structures. These arrangements are subject to the randomness of interactions driven by eternal whirling movements. The effect of brute forces is therefore at work in all inert and living nature. While Aristotle shares the idea of a material continuity in the world, he nonetheless invokes form as the higher principle of the determination of matter. The soul constitutes the form and the internal principle of movement of the living body. The body is organized; in other words, it is literally the instrument of the soul. This matter/form duality had a lasting influence up until the seventeenth century. With Descartes and more generally in the Galileo-Cartesian mechanist context, the extended substance, thus also living matter, obeys a geometric representation. It is a passive substance, the organization and movement of which are exclusively the result of the mechanical laws governing the collisions between physical elements, following the principle of inertia. Life is therefore not an instance other than matter; it is simply a property “of the other more particular bodies that are on the earth”4, of a purely physical nature. In this context, the dynamics of living systems does not exist as such, because it is reduced to a purely deterministic mathematizable kinematics.

13Leibniz and Stahl oppose Descartes’ strict mechanism, albeit each very differently. Stahlian animism conceives of life as a force external to matter, which opposes it and resists death. Stahl, a German physician and contemporary of Leibniz, along with the entire vitalist movement that followed him, saw the organizing principle of life as a transcendent, intangible, and non-mechanical force. On the other hand, Leibniz, who first coined and used the term dynamics, led the way to a rationalist philosophy that moves beyond mechanism, without however refuting it. In this conception, the primary elements of reality are considered to have a specific activity, and not to be characterized only by their geometric extension. Leibniz argued that the world was intelligible according to mathematics, and in this sense, professed unity and continuity within the natural world. Hence, there is no true opposition between inert matter and living matter; a difference in their degree of organization is the only thing that sets them apart. Inert matter is infinitesimally dynamic. Life is a property of the substance that is force, whereas with Descartes, it is, in certain cases, a property of the substance, the main attribute of which is its extension. This primitive active power, non-extended and indivisible, is an organizing activity principle of matter. Leibniz’s dynamics, of a metaphysical essence, thus accounts for the spontaneously active nature of living systems.

14The inventors of differential calculus, Newton and Leibniz, had an additional influence on materialist models of natural sciences during the Enlightenment. While acknowledging Descartes’ mechanism, Georges-Louis Leclerc Buffon sought to broaden the repertoire of forces and invented a physics of “penetrating forces” which were internal, based on the Newtonian model of the gravitational force.5 Leibniz had an influence on the transition from pure mechanism to organism. The “organic molecules” invented by Buffon were not simple atoms representable in geometrical terms. Buffon argued that they were informed via an “internal mould”, the conception of which he based on a dynamic organization.6 In Buffon’s footsteps, Pierre-Louis Moreau de Maupertuis went even further. He materialized Leibniz because, according to him, the primary elements of reality are material and have specific dynamics. They have various properties of sensation, memory, and thought. Their dynamic interactions determined by their affinities underpin the formation of living bodies. In The Earthly Venus7 and even more clearly in his Essay on the Formation of Organized Bodies8, Maupertuis explains – easily he claims – the resemblance to parents, the appearance of freaks, or the existence of mixed races.

15Denis Diderot, who was inspired by reading this work, added a nuance to it. In On the Interpretation of Nature9, he attributes incorporeal properties such as thought or memory not to the particles of matter as such, but rather to their arrangement. This notion would be reused and developed several decades later in 1809 by Lamarck, who, in his Zoological Philosophy10, conceptualized the idea that the properties of living matter originate from its internal organization under the action of universal laws of physics and of time. Lamarck’s transformism encompassed both the formation of embryos and that of species. This theory stemmed from the key importance that he granted to the laws of physics, their effects on organized matter, and the time needed for these laws to form structures that are initially simple but subsequently and gradually become more complex. With Lamarck, time is not considered simply as an external dimension. It is a part of the organized structure of the living being and of its history. Note that the theory of spontaneous generation defended by Lamarck and by many of his predecessors proceeded from this unitary way of conceiving of living and inert matter, and from transformism. On the other hand, refuting this principle fed the suspicion of vitalist thinking. Lamarck’s oft-forgotten stroke of genius was to consider, within a context of universal physics, two essential properties of living systems: their organization and their historicity. This exonerates him from his mistakes regarding the mechanisms of evolution. By putting forward the theory of evolution by variation and natural selection, Charles Darwin marked a decisive step forward, because he opened the path to a non-finalistic scientific conception of living systems.

The difficulty of reconciling mechanism and epigenesis

16We will now see the impact of these conceptions of living systems on theories of biological development. Two main traditions going back to Greek Antiquity come forward. Aristotle defended the theory of epigenesis, according to which the formation of an embryo is a gradual process of elaboration and organization of living matter. He contrasted it with a conception closer to what would later be called “preformation” – advocated by Democritus and Hippocrates – which postulated the envelopment of the embryo in a microscopic state, already fully formed in the egg. Hence, the formation of the embryo simply consisted in its development by growth.

17Aristotelian epigenesis had a lasting influence. In the seventeenth century, William Harvey defended this theory in his treatise on the formation of animals: “an animal which is created by epigenesis attracts, prepares, elaborates, and makes use of the material, all at the same time; the processes of formation and growth are simultaneous”.11 Epigenesis was then associated with a clearly finalist conception making use of a formal cause (what Harvey called “plastic forces”) and conferring a vitalist connotation on epigenesis models.

18It is noteworthy that, at the time, the growth of mechanist thought progressively side-lined epigenesis, championing preformation instead. Throughout his life, Descartes proposed unsuccessfully a mechanical model of epigenesis, inspired by his “animal-machine” theory. He was the only one though, and unfortunately, he failed. Marcello Malpighi, a renowned representative of the Italian mechanist school of thought following in Galileo’s footsteps, and for whom tissues and organs contained small complex machines, supported preformation, nonetheless specifying that the form does not pre-exist fertilization. The embryo develops via growth under the effect of encased mechanical devices fed by heat. However, microscopes, recently developed at the time, allowed him to make unprecedented observations on the development of chicken, which should have led him to defend epigenesis.12

19The principle of order at the origin of form could therefore not be studied within a mechanist framework, as Malebranche pointed out in The Search after Truth.13 He defended the theory of emboîtement and of the pre-existence of germs, an extreme version of the theory of preformation, pushing the formation of all living beings back to creation.

20Consequently, at the beginning of the Enlightenment, there appears the refutation of the idea that efficient causes explained embryonic formation. The hegemony of preformation models would however only last for so long. In The Earthly Venus (1745), Maupertuis clearly explains why the observation of racial intermixing in humans is incompatible with preformation.14 Although very distant from Harvey’s Aristotelian hypotheses, Maupertuis comments on the observations of the latter in The Earthly Venus15 and presents, in his Essay on the Formation of Organized Bodies (1754), an epigenetic model via the self-organization of the elements of living matter.16 Epigenesis experienced a new fame in the eighteenth century, although inklings of vitalism still weighed down on this theory. The “internal mould” invented by Buffon to account for resemblances between generations17, or the “morphological forces” later posited by Claude Bernard illustrate the difficulty of articulating a mechanist model of development by epigenesis.18 Bernard distinguished between the composition of living matter, the protoplasm, and the organization of the living being. He adopted the physico-chemical composition of living matter as a topic of study, as it was more suitable for an experimental approach. According to him, life was peculiar to matter according to physical and chemical laws. However, he came up against the impossibility of explaining the organization of living beings. To do so, he invoked “morphological forces”, which he distinguished from physical and chemical forces, thus flirting with neo-vitalist ideas.

21Claude Bernard unfortunately lacked the chromosome theory of heredity, which under the impetus of Theodor Boveri and Walter Sutton, was developed shortly after him in 1902.

22The great founding figures of embryology in the nineteenth and early twentieth century, namely the Germans Karl Ernst von Baer, Ernst Haeckel, and his disciples Wilhelm Roux and Hans Driesch, were naturally strong supporters of epigenesis. This was largely due to the remarkable observations offered by microscopy. Let us consider for a moment the tensions between Roux and Driesch, because they tell us a great deal about the opposition between mechanism and vitalism. Roux based his research on the Entwicklungsmechanik, the mechanics of development. He observed that the removal of half of the two or four cells of a frog embryo led to the development of half-embryos. Therefore, he concluded that cells were determined already at this stage. This mosaic theory of epigenesis was refuted shortly afterwards by Driesch, on the basis of experiments on sea urchins. Driesch found that a single blastomere at the stage of two or four cells develops into a complete embryo, thus revealing its totipotent properties. Driesch’s disciple Hans Spemann confirmed his work and broadened it. However, Driesch, a vitalist, referred to Aristotle’s entelechy to explain his observations. Spemann himself, who in 1924, along with Hilde Mangold, proposed the concept of an “organizer”19, reusing an original idea of Ethel Browne20, was also a neo-vitalist. He found that transplanting the dorsal lip of the blastopore of a frog embryo resulted in the formation of Siamese twins, that is, the duplication of the anterior-posterior and dorsal-ventral axis. This induction phenomenon shows that a group of cells organizes the development of other cells remotely. However, due to his neo-vitalism, he did not seek out the chemical nature of the information produced by this organizer. Others would do so in his place.

23In the early twentieth century, cell theory and the chromosome theory of heredity had been established, and a mechanist model of developmental epigenesist appeared. Boveri demonstrated that anomalies in chromosome separation in doubly-fertilized sea urchin eggs resulted in developmental perturbations.21 From this, it was thus deduced that chromosomes contain morphogenetic information. Hence, the principle of order of living organisms is not an immaterial, external instance but rather an information inherent to living matter, of a chemical nature, transmitted from generation to generation by the germ cells and chromosomes that they contain.

24Embryonic development, just like origami, which is the result of a precise sequence of mechanical folding operations, is based on four notions: space, time, information, and mechanics.

25Let us now consider the complex nature of this information, before moving on to study the physical mechanics that it controls.

The dynamics of developmental information

Developmental genetics and positional information

26At the beginning of the twentieth century, a former student of Driesch, at the heart of the dispute with Roux, left the field of sea urchin embryology and the regeneration of annelids to focus on heredity, the laws of which had been discovered by Gregor Mendel in 1886. This student was Thomas Hunt Morgan. The rediscovery in 1900 of the forgotten laws of Mendelian heredity marked a major turning point. A new model organism also appeared, namely the fruit fly Drosophila, the main advantages of which were quick development and the existence of only four chromosomes. At Columbia University, Morgan induced the mutagenesis of fruit flies and discovered, step-by-step, the chromosome laws of heredity, which won him the Nobel Prize in 1933: genes are located along chromosomes like the beads of a necklace and combine with one another. Genetic information had thus been found to be visible and palpable, and the mechanical basis of heredity could consequently be understood. The following decades revealed the chemical nature of this chromosomal information, namely DNA, its replication, and its transmission during each cell division. Genetics then turned its focus to adult features. With the very noteworthy exception of Waddington, fruit fly geneticists were not interested in development.

27And yet, during this time, fruit flies progressively became a powerful model of embryology and regeneration, especially under the impetus of Ernst Hadorn. Subsequently, in the 1970s, developmental genetics became a new discipline that experienced a remarkable upsurge. It became possible to induce the homozygosity of mutations through X-ray irradiation and to analyse their effects in the clonal descendance of the mutated cell. These genetic studies, referred to as “mosaic”, were focused on fruit fly limbs: wings, legs, and antenna. Antonio Garcia-Bellido, amongst others, discovered that the genetic information of morphogenesis is expressed at the cellular level and that it can be inherited in cellular lineages.22 The embryo is therefore modelled as a sort of genetic automaton: during each cell division, genetic information is as if selected by each cell. Garcia-Bellido thus suggested differentiating so-called “selector” genes, which control developmental decisions, from the “cytodifferentiation” genes that execute them. The mutation of the homeotic Antennapedia gene offers a remarkable example of this selector gene: legs develop instead of and in the place of antennae.23

28DNA is therefore the medium of all of these developmental genes. However, it is much more than a structure that stores genetic information like a hard drive. DNA also constitutes the operating system allowing for its controlled use in each cell during development. In 1969, following François Jacob and Jacques Monod’s research on the lactose operon, Eric Davidson proposed a theory of the genetic regulation of development.24 DNA allows genetic information to be expressed, executed, and regulated in space and time.

29Let us now return to the Antennapedia mutation in order to introduce the central concept of “positional information”. The induction of mutant cell clones for the Antennapedia gene produces mosaic antenna and leg tissues, so that the position of each cell can be associated with its differentiated characteristics.25 Howard Schneiderman, the author of this research, discovered that the cells that preside over a given position on the antenna correspond, notwithstanding distinct differentiated traits, with the cells of the same position of the leg. There thus exists identical positional information on different limbs, which define the Cartesian coordinates of each cell. This concept of “positional information” had been proposed by Lewis Wolpert two years earlier, in 1969, in a seminal theoretical article that first earned him a great deal of misunderstanding.26 The prevalent viewpoint on development in the 1960s was strongly influenced by the study of molecular differentiation, following Jacob and Monod’s research. It posited that there was information specific to each differentiated cell and to each organism. Wolpert, by contrast, argued for the universality of the mechanisms that define the position of cells within a tissue, and which operate upstream of differentiation processes. This positional information gives spatial coordinates to each cell with respect to the boundaries of developing tissues.

30Positional information has a genetic origin. It is translated into biochemical activity produced and interpreted by cells. What internal logic gives this information the universal nature attributed to it by Wolpert? What are its organizing principles?

31Two radically different mechanisms for producing positional information can be distinguished. Genetic information can be directed by a few “conductor” genes that define, in space and time, the order of activation and suppression of all genes in each cell. Like project managers, these genes execute the construction plan and the recipe for the creation of forms. Inversely, forms can emerge spontaneously from local, stochastic interactions between all cells, although without the cells being orchestrated or directed by a few cells or singular genes. It is according to this principle that social insect hives, such as those of termites, organize themselves without a plan or an orchestrator. Morphogenesis simply emerges from local interaction rules, whether in the case of animals – like in this example –, cells, or molecules. Developmental information borrows from orchestration and self-organization strategies, like Gaudi’s cathedral and termite cathedrals (Fig. 1).

Figure 1

Orchestration of morphogenetic processes

32Let us first consider some of the principles of orchestration. The genetic studies carried out in the 1970s and 1980s on fruit flies revealed a very limited number of positional information genes, as well as the logic of their collective activity. These discoveries earned their authors the Nobel Prize in 1995. Christiane Nüsslein-Volhard first identified a category of maternal genes defining the main axes of the embryo, namely the anterior-posterior axis and the dorsal-ventral axis.27 Along with Eric Wieschaus, she identified around fifteen genes defining the segmented patterns of larvae resulting from embryonic development.28 Strikingly, these genes were grouped into four discrete functional categories: genes of a maternal origin that control the development of a very large portion of the embryo; other genes called gap genes that define more limited territories covering multiple segments; pair-rule genes that organize every second segment; and the last category of genes that polarizes the segments (Fig. 2). While the mechanisms of segmentation sometimes prove to be different in other organisms, we can retain the following general principles: positional information is deployed dynamically and according to a hierarchical scheme; these genes code for transcription factors that regulate other genes; and their cascading activity results in the progressive subdivision of the morphogenetic field that they organize, to the point of granting each cell its anterior-posterior coordinates.

Figure 2

33In parallel, Ed Lewis identified the bi-thorax gene complex, which groups together eight homeotic genes defining the identity of thoracic and abdominal segments, based on their positional information.29 These genes are transcription regulators that orchestrate all differentiation processes downstream. They are located on chromosomes according to an axis that prefigures their sequential expression along the anterior-posterior axis, following a principle called “collinearity”. The homeotic Antennapedia gene that we just mentioned is also part of a gene complex that operates on the basis of this principle. Homeotic Hox genes are present in all segmented organisms, including humans30, and translate positional information into differentiation information.

34Reading an article on the work of Ed Lewis around the age of twenty was my first incursion into the world of developmental molecular genetics, which was booming at the time. It was a revelation for me. A year later, when I was a student at the École normale supérieure, I had the good fortune to be able to join Claude Desplan’s team at his laboratory in New York. He had discovered the molecular bases of the specificity of homeodomain proteins and their role in defining the anterior-posterior axis in Drosophila. In particular, his laboratory was studying the DNA binding properties of the Bicoid transcription factor, which orchestrates the development of the anterior regions of the embryo. This orchestration is due to its morphogenetic features. That was in 1993. For five years, my research focused on the concept of “morphogen”.

Morphogens

35The term morphogen was coined by the mathematician Alan Turing, in his seminal article “The chemical basis of morphogenesis” published in 1952.31 Turing defined the morphogen as a chemical substance that produces form, and did not attribute a more specific meaning to the concept. He predicted the formation of spatial structures from morphogenetic chemical substances according to a general reaction-diffusion model.

36This concept was soon defined in more detail in the context of Wolpert’s positional information model. Even though Wolpert did not use the term morphogen, he posited that a positional information gradient in a tissue, characterized by the local concentration of a substance, defines the spatial coordinates of a cell with respect to the boundaries of the morphogenetic field. Crick predicted that the diffusion of a molecule from a source could constitute a concentration gradient on a timescale compatible with morphogenetic processes.32 The concept of a “morphogen” then took on the meaning that it still has today. In 1972, Francis Crick and Peter Lawrence studied cell polarization in insects, and analysed their organization subsequent to a rotation of cuticle fragments.33 They distinguished the scalar value and the vectoral value of the morphogen along the activity gradient. The concentration of the morphogen gradient at any point provides positional information within a tissue and defines different cellular states. The local slope of the gradient polarizes each cell. Therefore, the morphogen gradient provides positional information and polarization potential. In 1988, Christiane Nüsslein-Volhard and Wolfgang Driever identified the first morphogen in Drosophila: Bicoid.34 This transcription factor activates different target genes in different regions of the embryo. By doing so, the morphogen gradient converts a continuously decreasing information into separate domains that regionalize the embryo.

37A major breakthrough in the 1990s was the discovery of morphogens in multicellular tissues. Concretely, Bicoid is a unique example of a morphogen acting within a large syncytium, similar to a large cell. My PhD research, completed under the supervision of Stephen Cohen at Heidelberg University, afforded me the opportunity to demonstrate the existence of morphogens during Drosophila leg and wing development. The organizing properties of multiple molecules had just been established.35 For example, the Dpp growth factor results in the duplication of wings when it is expressed ectopically. Under normal conditions, Dpp is expressed within limited sections at the centre of the wing.

38I first demonstrated that Dpp organizes the anterior-posterior axis of the wing during the development of the adult tissue precursor called the “imaginal disc” by directly acting remotely from its source. Next, I showed that Dpp induces the expression of two target genes according to the morphogen gradient model theorized by Wolpert36: they are expressed in nested domains, and the spatial extent of their domain of expression is set by the local concentration of the morphogen (Fig. 3). Dpp presents an activity gradient, the form of which is regulated by the expression of its receptor, which is suppressed by Dpp and forms an inverse gradient. By suppressing its receptor, Dpp extends its field of action, because the receptor limits its diffusion or its transport in the developing tissue.37 This is an example of the ligand-receptor system forming a morphogen.

Figure 3

39Analogous results have shown that the combined action of two morphogens (Dpp and Wg) during the development of legs defines the proximal-distal information of cells.38 Positional information is the result of the coordinated action of the two morphogens, which presents an axial symmetry with respect to the centre of the imaginal disc of the leg, whereas in the wing, the morphogen Dpp acts on its own following a bilateral symmetry.

40My research then led me to analyse in more detail and consider anew certain notions associated with the concept of “morphogens”, by taking into account the growth of the tissue within which the morphogen acts. The tissue was generally considered to be static, whereas it experiences growth in response even to the action of the morphogen Dpp. How is the morphogenetic information provided by this morphogen influenced by tissue growth, and in particular by the fact that cells move away from the source as the gradient acts? I demonstrated that the transcription of certain genes induced in the developing wing or leg is “memorized” and persists during growth.39 This “cellular memory” contributes to the expansion of the action of the morphogen. Moreover, the gradual establishment of developmental patterns under the action of the morphogen Dpp is correlated with tissue growth, and that affects the form of the activity gradient.40 This research thus presents the still poorly-understood coupling between morphogen signalling dynamics and growth.

41The study of morphogens remains very active today and revolves around questions of general interest. How is gradient information read? What is the accuracy of the information and of its measurement? What properties grant morphogen gradients scale invariance? How does a morphogen control the growth of the tissue that it organizes? This field of research has greatly benefited from the contributions of physics and mathematics, which propose dynamic models enriched with quantitative biological data.

Self-organization

42I would now like to discuss the self-organization models according to which positional information emerges from local, stochastic interactions. A remarkable illustration of this is offered in Turing’s reaction-diffusion model. In its most simple formulation, an autocatalytic activator also activates its own inhibitor. If the diffusion coefficient of the inhibitor is greater than that of the activator, the numerical solutions to the differential equations associated with it result in stationary patterns, such as stripes, or in waves. Hans Meinhardt, the reinventor of these dynamic systems, further developed the study of reaction-diffusion models. A striking example of Turing instabilities is found in the pigmentation of animals – fish scales in mammal coats –, as well as the patterns of shells. For the latter, waves of shady pigmentation propagate over the course of the growth of the shell.41 In these examples, we see how complex structures emerge from simple local rules for the interaction between two chemical components.

43The self-organization of dynamic structures within space is also at the heart of segmentation processes in vertebrates. The segmentation clock discovered by Olivier Pourquié42 following a theoretical article by Christopher Zeeman and Jonathan Cooke inspired by catastrophe theory43, produces somites, the precursors to vertebrae, one by one. This clock stems coherent oscillations in gene expression in the presumptive mesoderm that result in waves of transcription.

44The implementation of planar polarity in all multicellular organisms offers another eloquent example of self-organization. Discovered by geneticists in the 1980s, planar polarity is a vectoral property contained in each cell of a tissue, such as the orientation of differentiated structures on the wings of flies, the coat of vertebrates, or the cochlea, as well as the orientation of dynamic processes such as cell division and growth. This process is now better understood thanks to the identification of the molecules responsible for breaking symmetry in each cell. These molecules contain activating and inhibiting interactions at cellular interfaces, spontaneously producing a local asymmetry and its global propagation within the tissue. We note that in this case, the consistent alignment of self-organized asymmetries can make use of the activity of a morphogen gradient, the derivative of which provides a polarization slope at any point. Therefore, orchestration and self-organization often constitute joint developmental information mechanisms.

***

45In The Logic of Life, François Jacob concludes with the following: “today, the world is messages, codes and information”.44 These words are highly relevant at a time when the nature and logic of morphogenetic information are increasingly studied. Whether it operates according to a remote orchestration principle or according to local self-organization rules, this information fulfils three main functions: a regionalization function characterized by the definition of territories via progressive subdivision; a polarization function, through the orientation of dynamic cellular processes; and last of all, a temporalization function, through which the temporal order of processes is established (Fig. 4).

Figure 4

46If, as we have seen, developmental information is of a chemical nature, we must not forget that the laws of physics, and in particular those of mechanics, are at the heart of the local and long-range interactions between biological structures, from the molecular scale up to the tissue scale and the scale of organisms. We will now see how mechanical constraints and control during morphogenesis interrelate with one another.

The mechanics of morphogenesis

47My research on morphogens convinced me of the relevance of changing my approaches and placing greater emphasis on the dynamics of developmental processes. Positional information operates concretely in each cell, the size, shape, mechanical properties, and dynamics of which cannot be ignored. I joined Eric Wieschaus’ team at Princeton in 1998, with a view to studying the dynamics of the membrane and actin cytoskeleton in fruit fly embryos. Eric Wieschaus was already one of the pioneers of the genetic study of gastrulation, a complex morphogenetic process in young embryos. At the time, I developed fluorescent microscopy methods to follow membrane dynamics in real time, by monitoring the cellularization process through which the 5000 epithelial cells of the embryo are formed. While there, I developed an intuition regarding the dynamic behaviours of the cell in a living embryo. During that very productive period, I took advantage of new methods in confocal videomicroscopy and the possibility of fusing a protein of interest with fluorescent proteins such as GFP, in order to observe its dynamics in vivo. At the time, these innovations revealed the still largely unknown world of internal cellular dynamics, and paved the way to a renewed approach to morphogenesis. During the creation of my research team in Marseille in 2001, I set myself an ambitious research programme focused on the dynamics and mechanics of morphogenesis and its control by developmental information.

48Observing the embryo of a sea urchin or any other organism with a microscope reveals a constant dynamic: the cells divide, change shape, or migrate. All of these cellular deformations demonstrate the action of the mechanical stresses exerted by the cellular material on the cellular material. What are the mechanical properties of cells? What is the origin of the stresses responsible for cellular deformations and movements? How does the embryo control these processes in time and space? My team has been attempting to answer these three important questions for more than fifteen years.

Surface tension models: from adhesion to contractility

49Among the original purely physical models of morphogenesis, it is worth citing D’Arcy Thompson, and in particular his book On Growth and Form45, the centenary of the publication of which we are celebrating this year. D’Arcy Thompson was without doubt the first person to unite a purely physicalist explanation of morphogenesis with a mathematical description of forms, as the continuous variations in morphology by conformal transformation. In the name of platonician idealism, he refused to accept Darwin’s evolutionary theory and relegated hereditary causes to the secondary level. He used the physical concept of surface tension and interfacial free energy between cells and with the environment in order to explain the shape of cells and cellular aggregates. Cells adopt configurations that minimize the interfacial free energy as they approach thermodynamic steady state. D’Arcy Thompson also included in his models of morphogenesis the mechanical properties of cells and tissues. Embryologists of his time showed themselves to be relatively unaccepting of the purely physical explanation of morphogenesis that he proposed.

50However, in the 1960s, Malcolm Steinberg took over a thermodynamic model of morphogenesis46. He based himself on old cellular sorting experiments carried out on sponges, followed by amphibians, in which it had been found that cells dissociated from an organism reorganize into concentric spherical envelopes determined by their developmental origin. His predecessors, and Johannes Holtfreter in particular, used the term “tissue affinity” (Gewebeaffinität) to describe this selective association phenomena.47 This affinity, which is reminiscent of Maupertuis’ theory of the organization of the elements of living matter48, is a specific quality of each cell type. According to the affinity model, the organizing information of cells is qualitative. This affinity therefore prevents the quantitative generalization required by a physical theory. Yet, in his 1963 article, Steinberg proposed a physical model of cell sorting based on a single quantitative parameter: the adhesion energy between cells.49 He put forward the differential adhesion hypothesis, according to which cells organize themselves during cell sorting and during morphogenesis as a function of adhesion energy, regardless of the nature of the molecules and the particularities of their interactions. Steinberg wrote: “it would appear that a not inconsiderable amount of the “information” required to produce, through morphogenetic movement, the anatomy of a body part may be expressed in a code whose sole element is quantity: more versus less”.

51The molecular identification of cadherins in the 1980s allowed for the differential adhesion hypothesis to be experimentally tested, at least in vitro. In the 1990s, Malcolm Steinberg, Masatoshi Takeichi and others made use of this hypothesis to explain the segregation of embryonic germ layers or the immiscibility of developmental compartments, to cite only two examples. In 1988, Gerald Eldeman developed a theory of morphogenesis that emphasized the pre-eminent role of adhesion.

52We are now going to see how this viewpoint that grants a central place to adhesion in morphogenesis models has changed considerably over the past fifteen years, how our research has contributed to it, and what the impact of this has been on the mechanical models of tissue morphogenesis.

53In Marseille, our first research projects focused on the study of adhesion forces in the embryonic epithelium. This epithelium consists of a single layer of cells and establishes a polarized boundary between the external and internal face of the newly formed embryo. Epithelial cells maintain a single-layer organization while the embryo undergoes major deformations. Intercellular adhesion forces guarantee epithelial cohesion during these deformations.

54In an epithelium, intercellular adhesion forces are often concentrated at the apical-lateral pole of cells. They are based on the polarized distribution of adhesion molecules called “cadherins”, and in particular E-cadherin. The E-cadherin present on the surface of a cell interacts directly with the E-cadherin in a neighbouring cell via its extracellular domain. These interactions, called “homophilic”, are at the origin of adhesion forces and ensure the cohesion of the epithelium (Fig. 5). My team characterized the dynamic organization of adhesion complexes and their regulation. These studies were enriched by particularly stimulating collaboration with my colleagues Pierre-François Lenne, a physicist in Marseille, and Satyajit Mayor in Bangalore, where I spent a sabbatical year in 2011. Among other things, this collaboration allowed to test the quantitative organization of E-cadherin complexes at the nanometre scale, via intramolecular energy transfer, to a few dozen nanometres, via super high-resolution microscopy. The E-cadherin is not distributed homogenously, and it forms aggregates whose size, location, and mobility are regulated locally.50 We demonstrated that this regulation is multiple: the actin cytoskeleton stabilizes these complexes and immobilizes them; endocytosis reduces their size; the biochemical network of cellular polarization and mechanical forces have varying effects, the characterization of which is still incomplete. We confirmed the direct nature of the mechanical coupling between adhesion complexes and subcellular contractile tensile forces, at a time when this coupling was being called into question.

Figure 5

55Intercellular adhesion is based on a discrete organization enabling the quantitative regulation of three parameters: the size and dynamics of these nanometric aggregates, and their mechanical coupling to the actin cytoskeleton. This mechanistic framework offers a better understanding of the spatial regulation of cellular mechanics during morphogenesis. Concretely, not only do adhesion molecules stabilize cellular interfaces, but they also transmit contractile tensile forces that destabilize cellular interfaces and are at the origin of tissue dynamics, as we are going to see. The discrete nature of E-cadherin complexes therefore results in the localized action of adhesion forces and contractile tension forces, and therefore in the compartmentalization, so to speak, of cellular mechanics (Fig. 6). Measurements taken on individual molecules by multiple researchers indicate moreover that the binding affinity of extracellular monomer interactions is very low. The concentration of these molecules in aggregates of discrete sizes therefore allows for the creation of complexes in which the effective affinity for intercellular molecular interaction is locally several orders of magnitude greater than that of the monomer. We can therefore model the constant exchanges between molecules diffusing in the membrane and molecules aggregated in a complex as a phase transition allowing the formation, in a few minutes, of points of adhesion capable of stabilizing an interaction between two cells. As we can see, intracellular adhesion constitutes a “glue” with very complex dynamic properties that are the subject of active research.

Figure 6

56Adhesion forces therefore stabilize cellular interfaces and explain the polygonal geometry of their organization, like soap bubbles. The minimization of the energy of cellular interfaces can largely explain the shape of the cells.

57Epithelia are however extremely dynamic, and new microscopy methods have enabled us to see the extent to which that is so. We focused on the elongation of an epithelial tissue, a very general category of morphogenetic process. Rapid imaging of embryos revealed an overall movement of cells towards the posterior section, similar to the flow of a fluid. The close monitoring of the relative positions of cells during this process indicates that the elongation of the ectoderm along the anterior-posterior axis is caused by cell intercalation, a process that leads cells to change neighbours in a very organized way.

58The absence of individual migration in the cell population and the coordinated nature of the overall process drew our attention to a more local analysis focused not on the polygonal geometry of cells but rather on the topology of their interfaces (Fig. 7). Cells initially lose contact with their anterior and posterior neighbours, subsequently establishing new contact with their dorsal and ventral neighbours. We thus demonstrated irreversible topological rearrangements.51 Moreover, these rearrangements are polarized on the epithelium level: in this model, only contact points with neighbouring anterior and posterior cells are remodelled (Fig. 7). This phenomenon demonstrates a new form of planar polarity.

Figure 7

59Our research subsequently determined the origin of these polarized rearrangements. They are not due to constraints originating from outside the tissue, compressing or stretching it. Rather, they have a local origin, at the cellular scale, or more specifically at the level of cellular interfaces. We initially thought that the remodelling would be directed mainly by local inhibition of intercellular adhesion, in accordance with the prevailing model of differential adhesion. However, our study showed that the remodelling of junctions depends on a local increase in contractile tension (Fig. 8).52 The molecular motor myosin-II is enriched at the interfaces between cells and destabilizes these junctions through the traction forces exerted by the contractile actomyosin network on E-cadherin adhesion complexes. In this way, the localization of myosin-II is planar polarized in the epithelium and induces a planar remodelling of cell junctions by increasing the contractile tension at these junctions. Laser ablation experiments of the actomyosin cortex, in collaboration with Pierre-François Lenne, demonstrated that the planar polarity of myosin-II induces an anisotropy of tension at cell junctions.53

Figure 8

60Cells thus break the symmetry of their polygonal geometry. What is the origin of this remarkable process? We demonstrated that the local polarization of cellular mechanics is controlled by the overall polarization of the embryo. The organized expression of pair-rule genes which, as we have just seen, defines positional information, is therefore at the origin of the cellular interfacial information that polarizes cells in the plane of the tissue.54

61Let us now compare the elongation of the ectoderm with the invagination of embryonic tissues (Fig. 8). Elongation and invagination constitute two very general categories of morphogenetic processes. Epithelial invagination, a process extensively studied by Eric Wieschaus and Maria Leptin, requires the apical constriction of cells via the contractile activity of myosin-II recruited apically in these cells. In this way, subcellular regulation of the localization of myosin-II in epithelial cells induces the polarization of contractile forces and orients cellular deformations: apical isotropic forces induce geometric deformations, apical constriction, and tissue invagination. By contrast, anisotropic forces induce topological rearrangements, cell intercalation, and the global elongation of the tissue.

62This research aroused much interest for three main reasons. First, it substantially reduced overall the complexity of the elongation of a tissue, likening it to a simple topological transition in cellular interfaces. Second, it revealed a new type of dynamic, whereby a planar polarity information controls="true" the ordered remodelling of cellular contacts. It thus opened the way to a characterization of how positional information (a scalar quantity) controls="true" cellular polarity (a vectorial quantity). Lastly, this research established the central role of mechanical tension in the remodelling of cellular contact and tissue plasticity. Since then, tissue plasticity and organization have no longer been explained simply in terms of adhesion forces, but also in terms of contractile forces. In 2007 we proposed a general model of morphogenesis based on the physical concept of surface tension, integrating the combined roles of adhesion and mechanical tension at the cellular level.55 It allowed us to extend the differential adhesion model, which overlooked the role of cortical tension in intercellular and inter-tissue surface tension. Even though it is a thermodynamics model and therefore cannot account for the short-term dynamics of processes, it nonetheless correctly predicts, over long time scales, the successive transitions in cellular configuration.

63Many studies have since generalized the idea that the distribution of these contractile forces is predictive of geometric or topological deformations on the cellular and tissue level.

Dynamic models and active matter

64Thermodynamics models of morphogenesis have the major drawback of not accounting for the real dynamics of biological processes. Since 2010 our research has revealed the importance of this problem and has led us to renew our understanding of cellular and tissue dynamics. Unlike inert matter, a living system is not defined simply by its internal energy. It is an open, dissipative out-of-equilibrium system in which energy flows are organized and sustained at the molecular level, as opposed to dissipative systems in which these exchanges take place at the boundaries of the system, as in Rayleigh-Bénard convection. In this context, molecular motors such as myosin-II, whose importance in morphogenesis we have seen, use adenosine triphosphate (ATP) hydrolysis to convert chemical energy into mechanical energy. Thereby, the actomyosin cytoskeleton which is composed of dynamic actin filaments deformed by the motor activity of myosin-II is maintained in a state out-of-equilibrium. Over these past years, the physicists Jacques Prost and Jean-François Joanny, along with Sriram Ramaswamy, Karsten Kruse, Frank Jülicher, Cristina Marchetti, and Madan Rao, have undertaken to develop a theory of “active matter” that explains, among other things, the emergence of contractility within actomyosin networks and the hydrodynamics of these active systems.56 It constitutes the ideal theoretical framework for understanding the dynamics of morphogenetic processes, in particular those that I will now discuss.

65New microscopes offering the rapid acquisition of high-resolution spatial images have revealed an unsuspected contractile dynamic during cell intercalation, as well as in other morphogenetic processes. The apical surface of cells undergoes spectacular pulsatile deformations. These deformations are induced by the semi-periodic contraction of the apical actomyosin network.57 The detailed observation of these deformations reveals internal flows in the contractile network, and namely a local convergence of myosin-II motors and actin filaments during the formation of a contractile pulsation. Moreover, the actomyosin network can present broad-range anisotropic flows that polarize cellular deformations. These polarized pulsatile dynamics are self-organized in the sense that we have seen above. They are the result of local interactions of a biochemical as well as mechanical nature. Active matter theory explains these morphogenetic processes on the molecular scale. It is also suitable for explaining the active dynamics of schools of fish or flocks of birds.

66In addition to pulsations and flows, contractile systems also form waves that are progressively propagated within cells or epithelial tissues. Recently, we observed the activation of the actomyosin network in cells of the dorsal epithelium. A contractile wave propagates within the tissue and induces a cellular constriction wave that results in the invagination of the tissue. This contractile wave also requires information of a mechanical-chemical nature, and namely a biochemical activation network and the mechanical feedback that sustains and propagates it.

67Therefore, the cells of an organism present a large repertoire of contractile dynamics during morphogenetic processes. Comparison of the invagination and elongation of the embryonic epithelium reveals a common strategy based on sequential deformation and stabilization of cellular shapes. Eric Wieschaus and his colleagues demonstrated the pulsatile contraction of cells during apical constriction.58 We revealed a similar process during intercalation.59 In both cases, the cells present periodic deformations and a subsequent phase of the stabilization of shapes, similar to a mechanical ratchet that makes the macroscopic process irreversible.

68The study of morphogenesis entails taking into account, in addition to contractile forces, the mechanical properties of the cellular material. The rheology of living matter is complex: within short time scales of just a few minutes, cells and tissues are mostly elastic materials in which strains are proportional to stresses and are reversible. Over longer time frames, we observe a range of dissipative behaviours, which by definition are irreversible, reflecting the internal viscosity of the cellular and tissue material. For example, the contractile actomyosin network offers a viscous resistance to the rate of deformation imposed by myosin-II motors. The viscosity of actomyosin networks depends on the dynamics of the renewal of its components, and in particular actin filaments. The dynamics of living matter thus underlies its visco-elastic mechanical properties.

Conclusion

69As we reach the end of this lecture, let us try to put into words a few of the main characteristic features of living systems. The dynamics of these systems are conveyed through the notions of space, time, information, and mechanics. The study of developmental processes, from the fertilized egg to the new born organism, teaches us that the language used to describe and understand the dynamics of living forms is of a mechanical and chemical nature. The multidisciplinary environment of life sciences, at the interface of biology, physics, computer science, and mathematics, provides the most suitable framework for deciphering this language. With it, we are entering into a new area in which we will be able to build a more quantitative and, hopefully, more predictive understanding of living systems. I will dedicate my teachings at the Collège de France to searching for the general principles of the mechanical-chemical information at the heart of transformation, growth, and renewal dynamics in living matter, on several organizational scales. There is no doubt that deciphering this new language will also afford a renewed and more profound understanding of pathological processes, because all of them are the result of alterations in mechanical-chemical controls="true" and constraints.

70The particularities of living systems must be sought within their internal dynamics. These unfold through space and time on various organizational scales, from molecular assemblies to ensembles of cells forming our tissues and our organs. Mechanochemical information structures the irreversible deployment of these processes within space. This information is clearly based on the chemistry of molecular interactions. It also follows the principles of thermodynamics and statistical mechanics. The morphogenetic information is unique because it changes slowly from generation to generation, giving molecular machines, like molecular motors, their asymmetrical structures and their incredible energy efficiency. Evolution writes in the genome the fundamental rules of the dynamic deployment of morphogenetic information according to the laws of physics, within the physical space of cells and tissues.

71Life deploys its dynamic forms in an in-between: the stochastic world of molecular interactions without which life could not be, and mutations without which life would only be a dull and eternal repetition; but also the deterministic world of physical and genetic laws, without which there would be no regularity. Between the two, life proves to be fragile yet certain, creative and – should we not rejoice over this? – unpredictable too.