Co-IP problems with band sizes - Different band sizes in input and pulled down (May/30/2009 )

Hey
I would like to get some opinions on my problem. I have performed standard co-immunoprecipitation with the antibody against protein A. I want to confirm previously published protein protein interactions. Between the above mentioned protein A and "my protein B'. However the input sample with antibody anti-B on WB gives band size 90 kda which is correct, but using the same antibody after Co-IP I see the band about 75-80 kDa. Is it possible that in complexes the protein got smaller? or is it more likely unspecific band. Thank you for time and efforts.

Photo?
Also, the salt concentration and detergents in your samples can affect running mobility.
Try to ensure that buffers in input and IP are consistent.