While the epitope that I'm working on is a trisaccharide, I guess the same would apply if someone has experienced it working with peptide epitopes. I am wondering if binding specificity can change with the conversion from ScFv to full-IgG format.
Going from an scFv to a Fab or full Ig, you put more stringent constraints on the relative orientation of the VL and VH domain. For most antibodies, this is no problem, as the domain interaction is usually specific enough to by itself generate a relative orientation that is compatible with the constraints of larger constructs. However, antibodies recognizing small antigens with good affinity sacrifice a fair amount of VL-VH interaction interface to achieve a deep binding pocket allowing a larger antibody-antigen binding interface. Additional factors increasing the flexibility of the VL-VH interface are : lambda-like CDR-L1 (lacking the cis-Pro of the kappa CDR-L1 omega loop) and/or replacement of one of the conserved glutamine residues that ordinarily form a well buried pair of hydrogen bonds across the VL-VH interface (e.g. in antibodies based on the murine lambda 1 germline).