western blot is not consistent with immunofluorescence - (Nov/28/2010 )

I used IF to dectect the localization the endogenouse protein of interest. Antibody specificity is confirmed by knock down in westernblot, and background staining can be excluded by knock down in IF.
IF showed predominant localization of this protein in nucleus (z-stack confirmed), however, when I performed nuclear fractionation and westernblot, my protein was barely detected in the nucleus but most in cytoplasmic fraction(and all fractionation controls were good; GAPDH only detected in the cytoplasm and sp1 only in the nucleus,loading control shows equal loading).
I don't understand how come with such a strong nuclear signal by IF, but I cannot detect it by western blot. Which result is more convincing? I am frustrated with those results and I will appreciate if any of you can give me an explaination.

The IF is looking at native protein, and I am assuming that your western was using denatured protein. Some antibodies work better on native than on denatured proteins.

I also encountered the same problems. The antibody i used is a commercial polycolonal antibody with high specificity in Western blot. I noted that my target is localized in nucleus no matter the cells were fixed with methanol/acetone or paraformaldehyde. The IF result is derived from a confocal microscope. In contrast, the target is only notable in cytoplasm fraction in Western blot no matter the fraction was processed by syringe or by a commercial kit. Can anyone give me a rational explanation?