Peptide synthesis is in one of those topics that changes character depending on the direction you’re seeing it from. From one perspective, it’s basically a solved problem. These compounds have been made by conventional solution-phase organic synthesis for many decades, and taking off from Bruce Merrifield’s Nobel-winning work on solid-phase synthesis, the first automated solid-phase-support peptide synthesizer was marketed in the late 1960s (more on this in another couple of paragraphs). You can order all sorts of custom peptides to be made for you or use one of those machines to make ‘em yourself, and untold numbers of them are constantly being produced, from the microgram scale all the way up to kilos. So what needs to be fixed, eh?
Well, the process works, but that doesn’t mean it’s a great solution. All of the major synthesis methods work via a similar series of steps. They have to, because of the very nature of a peptide chain: you have amino acid building blocks with an amine end and a carboxylic acid end, and you string things together in a series of amide couplings, which is simple enough - until you realize (quickly) that unprotected amino acids will couple with themselves and/or in the wrong order with other unprotected ones. So you have to find some way to have one end (the acid or the amine) taped over in any specific building block in order to have things work smoothly. In living cells, this problem is solved by the use of tRNA molecules, which have specific amino acids attached to them via their carboxylic acid ends (aminoacyl tRNAs). That gives you both a temporary protecting group (leaving the amino end free to react) and a recognition element as the transfer RNAs line up in the ribosomal machinery according to their pairing with the messenger RNA strand that’s being translated into a protein chain. The enzymatic machinery involved is extremely impressive in its speed, fidelity, and error-detection capabilities.
We humans have to work in a more brute-force manner, and we generally do it with amino acids that have protecting groups on their amino groups (Boc or Fmoc, for the most part, which can be subsequently removed by acidic or basic conditions, respectively). In the solid-phase method, the leading C-terminal amino acid has its carboxylic acid blocked by being coupled to a solid-phase resin linker. Then you bring in the next amino acid in a form that has its COOH free and its amino end blocked by one of those protecting group, and couple them. Then you take that amino protecting group off of the new resin-linked dipeptide so you can bring in the next N-protected building block and couple those, and so on and so on. At the end of the synthesis you liberate the new peptide from the solid support and wash it off to collect it.
There have been numerous refinements over the years, but that cycle of having to use protected building blocks whose protecting groups are then removed is a constant feature. You may already be picturing some of the limitations. For one, peptide synthesis in this manner is generally limited to a few dozen amino acids in the chain. Impurities start to build up, and in some cases the chain starts folding around on itself to where the exposed end isn’t so exposed any more. You have to use excess reagents and excess building blocks at each step in order to keep the yields high: if every cycle goes in 90% yield, then by the time you’re out to 40 amino acids long you have a 1.5% overall conversion, which is pretty hard to take. Not to mention the expense of washing all that stuff down the waste line, which is an unappealing process even if it were free. This paper addresses the issue, and links to estimates that every kilo of synthetic peptide generates several tons of associated waste (five to ten times the waste stream generated for small molecules).
So this new paper could be of interest. As opposed to the C-to-N direction of the solid-phase route, it describes an N-to-C (“inverse”) synthesis direction. That’s been tried in many ways as well, but it tends to suffer from epimerization at the chiral center, which is a bit more labile under those conditions, especially as the peptide chain grows. They start with the N-terminal amino acid with its amine protected, either by a conventional group or by attachment to a resin. Then they treat that with an ynamide reagent (N-methyl-N-tosyl aminoacetylene) which reacts with the carboxylic acid to make a reactive vinyl ester. Adding in the next amino acid, which has its own carboxylic acid transiently protected by addition of bis (trimethylsilyl)acetamide, forms the peptide bond, and the silyl group of the new dipeptide falls off in the workup, leaving it ready for another round of ynamide and the next transiently protected amino acid. A good deal of experimentation established the optimum conditions (solvents, reagent loads, etc.) for multi-amino-acid couplings with high yield and suppressed epimerization.
This gives you a one-pot no-washing-needed operation for each peptide bond, which is considerably simpler. The ynamide reagent can be remade from the N-tosyl-N-methylacetamide produced in each coupling step, which makes this route more appealing on scale, and the paper demonstrates a gram-scale synthesis of a known hexapeptide in solution, and the synthesis of a 21-mer peptide on solid support (and while using unprotected amino acids as starting materials rather than the pre-protected ones that you’d be using under normal solid-phase conditions).
I’ll be interested to see if this takes off. Getting rid of so many operational steps would seem to be a big advantage and then you have the use of cheaper starting materials as well. You’d want some other experienced peptide synthesis labs to kick the tires here, looking at their own experiences with reproducibility, scalability, solvent/reagent costs, and waste stream, but I’m sure that this will be investigated!