Okay....so what do you do if somebody once created an antibody that works against your disease, but tragedy struck and you can't make any more of it? The cell line is gone, the material is limited...and its real important? Like, so important that only 2 antibodies to this disease have EVER been isolated?
Well, you could take the little bit of it that you have and crystallize part of it for high resolution NMR and you could give some of it to the awesome team at the NYU Proteomics Resource Center and maybe just to pull it all together you might call on the team at Protein Metrics.
And this is what you have here: this sick paper from Walter Bogdanoff et al., called "De Novo Sequencing and Resurrection of a Human Astrovirus-Neutralizing Antibody." Even the title is cool!
I'm going to have to skim over the NMR stuff -- its a crystal and you can see whats on the inside and outside and stuff and some inference of what amino acids are close to other ones.
On the mass spec side, though? There are some pros running the equipment up there in NYC!
They break out the Fusion and get an intact mass of the antibody. They mention that they don't get full isotopic resolution (I don't think anyone has ever got a full antibody resolved to baseline, but they seem disappointed...) but they get their starting masses for the intact chain.Then they reduce the antibody down to the heavy and light chains and pull full isotopic resolution on both of them. Wait. Isotopic resolution on the heavy chain??
Told you, this team is seriously good! They get some really good masses and notice that they've got some variations to the heavy chain. (To do this, they had to use "protein mode" on the Fusion and they run through how they worked their gas pressures and differentials to make this magic happen!)
Next they digest the subunits with multiple enzymes and run the digests out on a Q Exactive, get high res MS1s and MS2s and the data processing comes to center stage. I'm a little fuzzy on the de novo process they used, but it sounds like the best hits are used to make a database that is used in the next round of processing and so on. Traditionally, de novo has a pretty high FDR -- we worry about the size of our search matrix when we have a database. Here the search matrix is basically infinitely large. With that many possibilities there are bound to be some bad matches. Run this data through any de novo algorithm and you'll come up with a sequence, but really...is it really, really right? This is important -- we can't be messing around with some okay sequence here. Antibodies aren't real forgiving. The sequence has to be right.
They take the sequence information
And they use it to create a new antibody
And it binds to the antigens expressed by the disease, successfully resurrecting an antibody.
What a ridiculously good piece of work! I'm assuming this search strategy is something