How to make a machine that turns bits into molecules.
A few notes:
Light-based control of enzymes is a popular fount of...speculation, but too unreliable for the requisite level of precision. You might get a fast on-rate for, say, azobenzene switching, but the relaxation/off-rate will be too slow and stochastic. Do you have a method in mind for detecting when the ribosome has arrived at the desired site, or has traversed up to 19 "wrong" sites? A single protein molecule product is useless, so this needs to be synchronized between (at least) millions of ribosomes, some of which may proceed faster or slower over the template than others. This would also make detection of e.g. fluorescent excitation or quenching events unreliable as a method of determining ribosome location.
You mention the ribosome "gliding" around on the template, but not relying on the hypothetical motor protein for this movement, as the motor protein is used to melt through the blocking hairpin(s)? EF-G doesn't really act as a motor to continually translocate the ribosome forward; it only acts to push the A-site tRNA into the P-site after elongation. Since you're not continually adding amino acids as the ribosome glides, I don't think you can rely on EF-G to move the ribosome forward repeatedly. I wouldn't recommend using some kind of mutated EF-G for this purpose since it'd interfere with the actual elongation process. If we're going to be speculative, perhaps you could have some kind of nanotechnological "turbocharger" on the motor protein, so that it can provide both the forward gliding and the hairpin-overcoming.
The A-site "door" needs to be open long enough to ensure that the correct tRNA is added, and incorporated into the polypeptide, which varies heavily as the chance of the correct tRNA being inserted into the A site is somewhat randomly determined. It might be the first tRNA to enter the pocket, or the fiftieth. Leave it open too long, though, and you risk multiple incorporations, as you mentioned. Or, risk the ribosome sliding around on the template. You'd have to ensure that the ribosome is still able to perform some sliding, which allows the shuffling of the A-site tRNA into the P site during elongation, so you can't simply clamp it in place for a while.
I wouldn't say the the ribosome only moves codon-by-codon. Sliding usually ends at a site on the mRNA that matches the P-site tRNA's anticodon, which may include frameshifts. This P-site codon-anticodon interaction is fundamental to stabilizing the ribosome reaction, though I suppose not an insurmountable challenge. Similarly, after you've shoved the ribosome(s) to the correct site, the P-site tRNA is whichever was previously added to the protein, so unless there are 400 sites (20 x 20, one for each combination of P-site and A-site codons), the P-site tRNA won't match whatever codon it's sitting on.
Circular mRNAs are certainly useful, and seemingly essential for this approach, but I wouldn't highlight stability as a primary benefit. Since your reaction is running in vitro, you don't need to worry about RNA-degrading enzymes - unless you're having contamination issues.
It would seem that a better method of DNA synthesis, in terms of speed, cost, and length of contiguously synthesized DNA, would mostly solve the issues this proposal is trying to overcome. I agree that most new DNA synthesis companies aren't trying to compete on price, only length or turnaround time. Until we get past the crude hack of assembling hundreds of oligos together, that won't change, though we are working on that.
Sounds cool, thanks for sharing this!
I was wondering how you may re-set the device, to bring all ribosomes back to AUG before starting a new cycle. Perhaps you could use a third wavelength to trigger motion from the start codon (and maybe a fourth for the stop codon..?), so that when you’re done synthesising your protein, you can just slide everything forward e.g. 100 times with the standard wavelength for sliding; in this way all the ribosomes should slide until they reach the start codon, which they can’t leave without the special wavelength. Now your device should be ready to start over with a new protein :)
(Please let me know if it’s not clear!)
Amazingly creative! The fifth industrial revolution is accelerating very fast. Best luck to you speculative geniuses.
A high res TV screen, lying face up in the lab. The edges covered in waterproof sealant. The surface covered in a thin layer of bacteria engineered to produce these protein printers. Each pixel is flashing out its own protein at a rate of 6 amino acids / second. (60hz, 20 amino acids so on average 10 moves before each write)
Thought-provoking essay and creative ideas! But why so complicated? Wouldn't a "simple", non-templated (blocked-)amino-acid ligase suffice to recapitulate the idea of terminal transferase-enabled DNA synthesis also with proteins?
My startup Enthereal is working on the first part, enzyme design. We selected 3 kinds of reactions for pharma that intend to replace many of the chemorganic reactions currently in use.
We love the idea of protein printers, but there could be a few other completely different approaches. Think about nucleic acid synthesis, how is that done? Anything from nanopore tech to help out? Template-free tech? One more: how about simply immobilizing minimal ribosomal systems? I like the creativity. We're hiring in AI and enzyme design.