NB. As I was writingÂ thisÂ post, a couple of comments came up on Twitter on whether the machine metaphor was a useful one for biological systems. The discussion did not change the view that is expressed here: a machine is a good metaphor and one that, to a large extent, remains untested.
I am no fan of science fiction but there is a novel which I have always liked â€“for the philosophical background more than the plot- and that has come to my mind when recently thinking about genes and cells: Carl Saganâ€™s â€˜Contactâ€™. The key element of the story is the reception of an extraterrestrial message by scientists which, after some toiling, is revealed to contain instructions for building a machine. The instructions do not contain any hint of what the machine is for and only vague ideas of how its detailed shape will look like; both emerge as it is built. The story has many twists and turns, deals with some of the preoccupations of Sagan and, in some ways, the recent â€˜Interstellarâ€™ touches in a more saccharine manner, with some of the same issues. But the bit that has grabbed my attention over the last few weeks has been the notion of having a set of instructions without a clear picture of what they are for (self assemble furniture instructions feel like this sometimes) that, when followed, generate a machine and only when the machine emerges, one can think of what it is for.
The reason I was thinking about â€˜Contactâ€™ has to do with a preoccupation to figure out what it is that we want to know. I am well aware of things that we need to know, but given that there is always limited time, one needs to decide what it is that one would like to know; understanding is ambitious and, probably, out of reach. Modern cell and molecular biology have placed in front of us a formidable technical arsenal which, in principle, allows us to explore any question we may have. As I have hinted at before, it is unfortunate that a collusion of editorial and career interest are giving precedence to classifications and listings, sometimes one at a time, over real questions butâ€¦â€¦.. it is also true that questions are difficult to find and, more importantly, to answer. And it is in trying to understand what are important tractable questions, that something caught my eye, something which I am sure some of you either know or have thought about in a different manner. The problem of Developmental Biology is how the information in the DNA is decoded and transformed into the tissues and organs that configure an organism. Most seminars and reviews on the subject start like this only to then proceed with genetic or molecular screens. Nonetheless, as a consequence of this work, classical and popular genetics have created the mirage that there are genes for this and that i.e. that the DNA harbours instructions for digits, eyes, hair colour or height and, in the worst twist often promoted by newspapers, that there are genes for diseases. This is the basis of our current understanding of how the information in the DNA is interpreted and turn into an organism butâ€¦â€¦. if one thinks about it and thinks from first principles, genes are ONLY instructions to build machines (M in the figure below): ribosomes, transcription and replication enabling machines, membranes, cytoskeletal devices, etcâ€¦.their remit does not go beyond this. These machines, once built (through the central dogma), become assembled, much like a 747 or a transatlantic liner, into a larger product, a device with the capability to process information, react to it, do work, sense the environment. This device is what we call A CELL. In multicellular animals, subtle variations in the composition and performance of the component machines, lead to different cells that can be further assembled into tissues and organs and these into an organism. Thus, donâ€™t forget that the instruction in the DNA do not code for much that is 3D, and certainly lack any information about the function of the machines or the devices they code for. Like the extraterrestrial message in Saganâ€™s novel, the instructions are for a machine whose purpose only becomes clear once it is built.
Fig1. Top, A general principle whereby functional interactions between machines built by instructions from a blueprint, configure a device that works according to the laws of Physics and Chemistry. Bottom, a translation of the principle to a biological system.
Something interesting happens once the machines and their suprastructures, the cells, are assembled: space arises (the information in the DNA does not contain, convey or encode space, maybe time, but not space). As a consequence of the generation of space (surfaces and volumes), mechanics makes an appearance and it does so at two levels: the molecular one (the machines) and the cellular one (the device). Â Thus, the emergence of space leads to the emergence of mechanics which feeds-back on the processes of decoding and assembly and, in more than one way, tissues and organs and ultimately organisms, are the outcome of these feedbacks and the interactions they create. The decider of what the output of these interactions should be is function, as dictated in a blind manner by Natural Selection.
Fig 2. Tissues and organs doÂ not arise from instructions but result from the feedbacks that the emergent properties of the performance of the machines assembled in cells. More explicitly from the feedbacks that generate new activities and behaviours in those machines and cells.
Much has been made of the selfish gene hypothesis (and I hasten to add that I am not an evolutionary biologist and know little about evolution) which suggests that the machines and the devices have the aim of reproducing the genes which encode them. I find this naÃ¯ve, deterministic and anthropocentric. There might me no purpose in the assembly of those machines, maybe all is a game of molecular flaunting and selection is not simply looking at the replication of the genes but at the performance of the machines (I realize that I shall need to expand this here but, like Fermat, for that I need more space than the one I can afford here). Perhaps it is a widespread unconscious focus on the genes that for a long time has placed our emphasis on the decoding, which misses many of the important points and problems the need to be addressed. In this regard, it is interesting that while cell biologists have focused on the machines and, sometimes, how they are connected, developmental biologist have focused on the decoding in a kind of naÃ¯ve manner: a gene for each season. I would surmise that the important problems in developmental biology require, at least for now, thinking about the machines, the devices they configure and their functioning rather than in their component parts. Only in this manner we shall be able to go beyond genes/instructions and start to look at tissues and organs from the same perspective as Nature does: the ability of the cell to integrate and process information. Work on the decoding of gradients in developmental systems (James Briscoe, Marcos Gonzalez Gaitan, Thomas Gregor, Johannes Jaegger, and Arthur Lander) or of oscillators during somitogenesis (Alexander Auhlehla and Andrew Oates) do just that and, to my mind, are some good examples of the way forward, of how to integrate quantitative cell biology, hypothesis testing and modelling. These efforts gauge the information processing ability of cells rather than in the micromanaged organization that most molecular biologist do. Perhaps phage and E. coli supervise closely every step of their biology but, while not impossible, this is not probably the way multicellular developmental systems operate. They are likely to use coarse graining, space and time averaging of molecular events, system level strategies that we can easily miss by focusing on the details of the molecular events.
Much of what we have done so far to understand higher order organization of biological systems (and there is work to do) deals with temporal aspects of the system: simple gene regulatory networks that try to capture sequences of gene expression linking them to specific cellular events. The challenge comes when we face the nature and details of the feedback that space creates on the decoding, the assembly and performance of the machines and their coming together into the devices that we call cells. It is the feedbacks involved in these processes that we need to understand and represent. Turing when thinking about the problem of development in his 1952 paper, was well aware of the challenge when he wrote:
â€œâ€¦â€¦one proceeds as with a physical theory and defines an entity called ‘the state of the system’. One then describes how that state is to be determined from the state at a moment very shortly before. With either model the description of the state consists of two parts, the mechanical and the chemical. The mechanical part of the state describes the positions, masses, velocities and elastic properties of the cells, and the forces between them.(â€¦.) The chemical part of the state is given (in the cell form of theory) as the chemical composition of each separate cell; the diffusibility of each substance between each two adjacent cells rnust also be givenâ€¦.(â€¦). The interdependence of the chemical and mechanical data adds enormously to the difficulty (of understanting the state of the system) and attention will therefore be confined, so far as is possible, to cases where these can be separatedâ€
This is a pre-molecular way of looking at the problem but remember that Turing was trying to think about Development and doing this in a very prescient manner. The interdepence he refers to is the essence of the feedbacks mentioned above which turn cells into tissues and organs. Along similar lines, a few years ago F. Julicher pointed out something to me which highlights the essence of the problem and, at the same time, the difference between Physics and Biology. It can be stated simply: whereas in a physical system, Chemistry generates mechanics and the reverse is not true, in biological systems both work (Physics generates Chemistry and Chemistry generates Physics) and it is in this feedback and interactions that, probably, lies the essence of biological systems.
The current focus and craze on screens and single cell transcriptomics (proteomics is coming) should not make us forget that these endeavours only address the parts. Moreover, as I have discussed before (https://amapress.gen.cam.ac.uk/?p=1094) the system we are trying to understand has evolved to respond to selection (and remember that this is what screens are about, a highly selective selection of the ability of cells to respond to a stimulus) and the response is based on dynamic heterogeneities that we still do not understand but which, one suspects, have something to do with the feedbacks I have outlined. The exciting thing (to me) in â€˜Contactâ€™ is the machine, what it does and how both its structure and purpose unfold as the system is built. The parts that configure the system could have been used to build anything but the instructions turn them into a space-time machine. In Biology, right now there is too much emphasis on the parts. This would not be a bad thing were it not because we endow parts with functions that correspond to the wholes they are part of. How do we avoid this? The real aim of Systems Biology is to avoid falling into this trap. It is therefore unfortunate that in most realms, Systems Biology has become a proxy for data analysis. Cell and Developmental biologists should embrace a proper version of Systems Biology because right now it is the only way to get out of the describe-the-parts loop and move towards understanding on the way to re-engineer cells and tissues.
Acknowledgement: I want to thank F. Julicher and S. Grill for very enlightening and inspiring discussions on the subject of this post. Also, note that there is a I in the title. There will be a II which aims to deal with â€˜questions in Biologyâ€™.
Epilogue: The Twitter feed I mentioned above led to many interesting suggestions. They all described different kinds of machines. Even the notion of City, much liked in the thread, can be construed as a machine. What this reflects is that machines are ways to describe (and then engineer) assemblies of parts with some functional aim. It is difficult not to see this in the fabric of a cell and, in many ways, it is a useful, working notion. If properly used it could be helpful as in its day was the vision of the central dogma, as information transfer and decoding.