After antibody sequencing and binding confirmation, sequences of the mouse antibody are analyzed. My question is based on the sequence analysis, how to find a suitable human framework for CDR-grafting? If the identity of the heavy chain or light chain is not so high with other human antibody sequence in the database, how to do the antibody humanization? We have a mouse antibody with only 75% identity with all the known human antibody sequences, and we fail all the time for humanization. Is there a identity threshold for successful humanization?
Challenging for sure. Is it murine or other rat rabbit etc? IMGT has some interesting software. You can try a straight graft into the closest VH and VL match but will likely need tuning to have biological activity - assuming your parental mAb has bioactivity.
Dear Hao, Have a look at a tool called "TabHu" (in case you don't know it). Maybe it can help you with selecting the right width of the grafts.
http://circe.med.uniroma1.it/tabhu/index.php
In antibody humanisation, you try to optimise various properties: to keep the binding affinity, to keep and if possible improve the stability and folding efficiency and to reduce the immunogenicity.
While human germline family VH3 and VH1 have good biophysical properties, the other germline families often are suboptimal, which can become limiting especially for scFv. Sometimes residues outside the classical CDR definition will affect binding affinity, e.g. VL-VH interface residues in mouse lambda domains, residues in the outer loop or the N-terminal strand that pack against buried CDR-1 residues. Looking at a homology model of your graft, you want to keep all buried residues above (towards the paratope) of the disulphide bridge and the core Trp corresponding to the murine sequence, the same for residues fully buried in the VL/VH interface.
There is no specific identity threshold for successful grafts - for example, "Wörn, A., Auf der Maur, A., Escher, D., Honegger, A., Barberis, A., and Plückthun, A. (2000). Correlation between in vitro stability and in vivo performance of anti-GCN4 intrabodies as cytoplasmic inhibitors. J. Biol. Chem. 275, 2795-2803." (http://www.bioc.uzh.ch/plueckthun/index.php?pid=3-1-17800) represents a graft to a very different pair of frameworks to generate an scFv of exceptional stability, etc.
Are you directly working on full IgG, or with scFv or other constructs? Is the protein produced in decent amounts, or do you have a folding problem? Are you sure your murine sequence is correct? Did you compare the murine sequence to the closest murine germline sequences to identify unusual somatic mutations? Did you try chimeric constructs of murine VH and human VL and vice versa to identify which of the two domains causes the problem?
Annemarie makes good points too it is empirical. In the past when I was humanizing we always made chimerics as controls and usually more than one isotype to see if it would still function. Biovia (former Accelerys) has good software too -
Dear Michael,
The link below does not open any page. May be it stopped working. Do you know any other sites? Thanks.
http://circe.med.uniroma1.it/tabhu/index.php
Are you trying to reach one of Anna Tramontano's sites? They are no longer maintained since she sadly died.
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology
- Virology
- Virus