Some researchers have used an anti-idiotypic method for finding antibodies with beta-lactamase activity. They reason that the binding region of the antibody is a mirror image of the antigen structure it binds to. So, the first step is to immunize with beta-lactamase and isolate the antibodies to beta-lactamase. This gives you the mirror image of the beta-lactamase. They then immunize with the anti-lactamase antibody. The antibodies raised against the anti-lactamase antibody should have the positive image of the beta-lactamase, and possibly the functional fold itself. It’s analogous to the process of casting negative molds and making sculptures, if you are familiar with that.
It is worth noting that they are simply amplifying pre-existing B-cells. The immune system is not creating completely new antibodies in response to new antigens. V(D)J recombination occurs during embryonic development so that each individual is born with all of the antibodies it will ever have. However, there is somatic hypermutation in B-cell lineages that show initial specificity for a given antigen so there can be an increase in avidity and specificity during B-cell proliferation through positive selection.
This is why the other experiments using naive populations of B-cells is of importance. These populations had never seen beta-lactamase or antibodies raised against beta-lactamase, and yet they were still able to find antibodies that demonstrated beta-lactamase activity.