If you knew exactly how an important class of antibodies interacted with its target, could you design a (relatively) small molecule to act the same way. These people did, and the work has very exciting implications for infectious disease [ Science vol. 358 pp. 450 – 451, 496 – 502 ’17 ].
The influenza virus is a very slippery target. Its genome is made of RNA, and copying it is quite error prone, so that mutants are formed all the time. That’s why the vaccines of yesteryear are useless today. However there are things called broadly neutralizing antibodies which work against many strains of the virus. It attacks a vulnerable site on the hemagglutinin protein (HA) of the virus. It is in the stem of the virus, and binding of the antibody here prevents the conformational change required for the virus to escape the endosome, a fact interesting in itself in that it implies that it only works after the virus enters the cell, although the authors do not explicitly state this.
Study of one broadly neutralizing antibody showed that binding to the site was mediated by a single hypervariable loop. So the authors worked with a cyclic peptide mimicking the loop. This has several advantages, in particular the fact that the entropic work of forcing a floppy protein chain into the binding conformation is already done before the peptide meets its target.
The final cyclic peptide contained 11 amino acids, of which 5 weren’t natural. It neutralized pandemic H1 and avaian H5 influenza A strains at nanoMolar concentration.
It’s important that crystal structures of the broadly neutralizing antibody binding to HA were available — this requires atomic level resolution. I’m not sure cryoEM is there yet.
Chemiotics II 
Comments
Interesting article. I wonder what proportion of antibody-antigen interactions could be mimicked by a cyclic peptide? How rare are those small, linear binding sites? I guess cyclic peptides like this are also only useful clinically if the binding itself is sufficient for an effect, like blocking a conformational change as in this case.
BTW, I do have one small suggestion for your blog posts. When you reference a paper, would it be possible to include either a link to it or something that could be quickly Google searched to find it? The title, DOI, or PubMed ID (PMID) would all probably work.
The format you use for your references obviously gets the job done, but it does add a step or two to tracking down papers. For example, the Science archives webpage (http://science.sciencemag.org/content/by/year) only lists volume and issue numbers, not page numbers, so it took a few tries for me to find the right issue. Not a big deal, but it would be nice to be able to get straight to the article from your blog.
Anyway, I do appreciate your posts, so keep them coming!
I’ll see what can do about the links to the papers. I read them on the web so the links should be available. However, I have a huge database (100 K) of text notes on the papers I’ve found interesting over the years. When I want to find my notes on a given paper, I search for them using [ Science vol. 358 pp. 450 – 451, 496 – 502 ’17 ]. The notes are always much shorter than the paper itself.
Okay, thanks for considering it. I can totally appreciate sticking with a system that works for you for the sake of consistency and being able to search your own database of notes, but a direct link or a title as well would be a nice addition. Anyway, thanks again.
CryoEM is capable of atomic resolution imaging and has been used to solve the structures of enzymes bound to small molecule inhibitors, e.g. see http://science.sciencemag.org/content/348/6239/1147
Bryan, thanks — I stand corrected, but most cryoEM that I see has resolutions over 3 Angstroms.