The surprising thing for me is that the naive precursors for these CD4 binding site antibodies bound well. This is in contrast to the V1-V2 binding human bNAbs found, say, in Doria-Rose et al 2014 and other papers. Different animal and different antibodies.
They inferred unmutated ancestors (UAs), which for some reason they call igLs, by taking a consensus of similar-looking sequences: “Based on HCDR3 length and homology, as well as by use of the same V and J germlines, we identified related sibling clones.” They could not find anything close to the UA in the pre-vaccination repertoire.
Rather than just doing ELISAs, they also did Biolayer Light Interferometry with Octet, which IIUC directly measures binding through time to get a clearer picture of on and off rates.
As a reminder, here’s a map of the various binding sites from a nice review.