# Artificial ribosome So much of the promising work on molecular machines has the flavor of building a crucial piece of a future system. This is incredibly important work! But at some point we need to start concretely talking about (and more importantly working on!) that system. Otherwise, the field will end up looking like a pile of legos without an instruction book and a picture on the box that we have no clue how to actually start building. <[[Academic-driven fields feel like a whole bunch of mixed legos]]>. This system-building work is beyond the capability of any single lab and as a result, the hunch is that it will require an ARPA-like program to coordinate and fund different projects towards the same goal. The motivating question then becomes *how do we frame concrete goals for the future programs around systems of artificial molecular machines*? It’s a truly massive design space! One strategy when approaching a massive design space is to look to those who have succeeded in it before. Nature is the undisputed champion of building systems of molecular machines. Of course, we’ve had the sense that designed nano-scale systems should take ‘inspiration’ from biology for a long time, but biology does *many* things at that scale, even if we’re just talking about factory-like-systems. Adam and I want to argue that the most important thing to be inspired by is the ribosome, and specifically its system architecture. Currently there is one programmable system that builds[^1] atomically precise structures: the ribosome.[^2] Through that lens, one system-level goal could be framed as creating an artificial ribosome: a reprogrammable tool that turns stochastic building blocks into a deterministic structure by choosing between forming *otherwise chemically equivalent* covalent bonds. You could leave an artificial ribosome as a poetic vision to strive for, but taking the idea of building an artificial ribosome seriously leads to some startling corollaries. We can use the ribosome as a starting point for breaking the big intractable goal down into modular subsystems. The ribosome framing highlights the obstacles that nature has overcome and provides a straightforward challenge for any system to “do something that a ribosome can’t.” Of course, the actual system may end up looking as much like a ribosome as an airplane looks like an artificial bird. However, birds and planes still have the same subsystems: lifting bodies, thrust generation, stabilization, altitude and directional control. To set the stage I’ll run through a disgustingly oversimplified version of how a ribosome works. The systems input is a strand of mRNA that stochastically encounters the ribosome (of course, like almost everything in the cell, there are proteins that increase that probability). The ribosome feeds the mRNA through itself to a binding site. Here, the ribosome exposes one mRNA codon at a time. At the same time there are tRNA-bound amino acids floating around. The ribosome captures and adds the amino acid that corresponds to the exposed codon to the amino acid chain (or starts the chain if it is a start codon) and advances the mRNA to the next codon. The ribosome releases the amino chain and mRNA when it hits a stop codon. We can break this system down into roughly seven subsystems. 1. Sequence information — this is the subsystem that tells the larger system which blocks go where. The ribosome uses mRNA as something similar to a punched paper tape, but you could imagine sequence information being transmitted by light or pH. 2. Start/stop — this is the subsystem that tells the larger system when to start adding blocks and when to stop. In the ribosome, start and stop codons on mRNA play this role. 3. Physical block selection — this is the subsystem that physically selects blocks from the environment or some block reservoir and shifts them from being outside of the system to inside the system. 4. Block positioning — this is the subsystem that positions blocks correctly so they can become part of the product. 5. Block binding — this is the subsystem that creates the covalent bond between the new block and the work product. 6. Error checking — this is the subsystem that counteracts the stochastic nature of the nanoscale world. In a ribosome it is built into the fact that each amino acid is encoded for by three bases instead of one. This mechanism is ok, but ribosomes actually have a fairly high error rate. 7. Product release — this is the subsystem that decouples the final product from the system.  Note that in the case of the ribosomes, there is overlap between some of the subsystems — the error correction mechanism is simply built into the way the sequence is encoded. However, this is not absolutely necessary. Of course, it’s also important to note that ribosomes don’t do their work floating randomly in a sea of amino acids and mRNA. Cells act as systems of systems: most ribosomes are embedded in the rough endoplasmic reticulum which allows the cell to concentrate mRNA and amino acids near them and to transport the products to where they’re needed. There are also a number of proteins that midwife the whole process. The proteins produced by the ribosomes are often themselves not end products, but instead are catalysts for reactions that many steps later yield a useful output. The cell isn’t a homogenous soup either, but is divided into compartments that can encourage a certain set of processes by concentrating inputs. The upshot is that there is a lot of work to be done in order to make an artificial ribosome useful outside of the ribosome itself. Another tricky issue that nature bypasses is the question of “how do we as 1-meter-scale people program the 10^-8-scale artificial ribosome?” Currently we program natural ribosomes through an incredibly indirect process where we write down a sequence of bases, synthesize DNA with that sequence, inject that DNA into an organism that then produces mRNA templated on that DNA and then feeds that mRNA into a ribosome. This is like needing to get an external firm to manufacture punch cards that you aren’t even allowed to put in the computer yourself. In addition to the system architecture, we can draw inspiration from what biology *can’t* do. The technology for designing and creating proteins is getting so good that if you can do it with a protein, you probably should. Proteins can do a truly astounding number of things, but it’s easy to forget that there is a massive space of useful functions they can’t perform. One way to uncover those functions is to look at the constraints of proteins: * 2D and 3D structures in proteins can only be created by hydrogen bonds and van-der-waals forces. As a result, proteins denature in adverse environments. * * Each amino acid in a protein is joined to its neighbors with a single covalent bond, which makes it hard to go from a desired function to a structure to a sequence of building blocks. Recent advances like AlphaFold might make this much easier. #### Pushbacks A pushback to the framing of an artificial ribosome is that it is positing a specific solution to a broader goal of creating large molecules with specific functions and desirable properties. If, for example, we had a straightforward system to go from a desired function to a completely self-assembling set of standard ingredients, it might achieve the same goal. ### Outline * Nature has created the only systems that turn stochastic building blocks into deterministic structures without depending purely on the properties of those building blocks and emergent behavior. * Ribosomes are the only reprogrammable instances of these systems * Programmability is important when it isn’t worth it to rebuild a factory every time you want a new product. * Non-programmability = high volume or high value only * Analog computers vs. digital computers * In broad strokes ribosomes lay out the system architecture for any reprogrammable positional chemistry system * At first glance there are six major sub-systems * Sequence Information * On/off initiator * Physical Block selection * Block binding * Block connecting * Error checking * If you squint, a scale-agnostic positional chemistry architecture is the closest thing we have to the Feynman path * At the large scale, you could do construction with tens of nanometers-wide proteins as building blocks as long as you are specifying covalent bonds between them. * At the smallest scale, the building blocks could be a few carbon atoms to build diamondoid structures. * The ribosome framing also suggests ways to do things that nature cannot * Annoying things about proteins * Single bonds * 2 and 3D structure dependent on non-co * Folded nature means that you need a lot of mass and structure for small sites * Stronger bonds than the single bonds in proteins * 2D or 3D bonded structures * Ribosomes might suggest the important super-systems that an artificial ‘cell’ would eventually need. * Places to concentrate products * Ways to artificially concentrate * A reprogrammable ‘machine’ that can create specified covalent bonds between nanoscale building blocks could enable everything from new catalysts to unthinkable materials. * [[Atomically Precise Manufacturing is more a matter of systems engineering than discovering new physics]] * People * [[Grigory Tikhomirov]] ### Related * [[Stochastic to deterministic lie on a continuum]] * [[What are the ways that nature uses to reduce stochasticity?]] ### References * From [[Andrew Turberfield]]’s programming chemical synthesis * [[Sequence-specific synthesis of macromolecules using DNA-templated chemistry]] * [[Multistep DNA-Templated Reactions for the Synthesis of Functional Sequence Controlled Oligomers]] * [[An autonomous molecular assembler for programmable chemical synthesis]] * [[Directing Otherwise Incompatible Reactions in a Single Solution by Using DNA-Templated Organic Synthesis]] * Grant for [[Andrew Turberfield]] [An Artificial Ribosome](https://gow.epsrc.ukri.org/NGBOViewGrant.aspx?GrantRef=EP/T000562/1) with [[Rachel O’Reilly]] * [[The Evolution of DNA-Templated Synthesis as a Tool for Materials Discovery]] * [[Programmable one-pot multistep organic synthesis using DNA junctions]] * [[Anisotropic polymer nanoparticles with controlled dimensions from the morphological transformation of isotropic seeds]] * [[Jason Chin]] * [[A network of orthogonal ribosome·mRNA pairs]] * [[Reprogramming the genetic code - chin and de la torre]] * [[Controlling orthogonal ribosome subunit interactions enables evolution of new function]] * [[Michael Jewett]] * [[Protein synthesis by ribosomes with tethered subunits]] * [[Anna Schepartz]] * [[Translation of Diverse Aramid- and 1,3-Dicarbonyl-peptides by Wild Type Ribosomes in Vitro]] * [[Farren Isaacs]] * [[Evolution of translation machinery in recoded bacteria enables multi-site incorporation of nonstandard amino acids]] * [[Exploring Strategies To Bias Sequence in Natural and Synthetic Oligomers and Polymers]] * [[Building tomorrow’s nanofactory]] * [[Phillipa Hemmings]] ran the program * https://www.linkedin.com/in/philippa-hemmings-8801441/ * [[Ned Seeman]] * [[Harris Makatsoris]] * [[Lee Cronin]] * [[Phillip Moriarty]] * [[Rasmita Raval]] * [[An artificial molecular machine that builds an asymmetric catalyst]] [^1]: As opposed to leaning purely on the intrinsic properties of the components to arrange in an eventually-energetically-favorable state. “Mixing and shaking.” [^2]: Arguably [[Polyketide synthase]] does this as well, creating a `highly diverse polyketide chain` [Web URL for this note](http://notes.benjaminreinhardt.com/Artificial+ribosome) [Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Artificial+ribosome)