Context:
While the scientists were unsuccessful so far in crystallizing PG16's sister molecule PG9, they were able to glean insight into its action from biochemical studies using both molecules. By
switching a small (seven-amino acid) segment of the CDR3 subdomain of PG9
for a similar segment from PG16, the team changed the subset of HIV isolates neutralized by the antibody. This confirmed the loop in question was the "business end" of the antibody and suggested that it might be possible to create other interesting variants of the antibody by manipulating this region.
http://www.sciencedaily.com/releases/2010/06/100601171846.htm