We need more information. Is it a surface antigen? Does it accumulate intracellularly? Please specify the kind of antigen, the protocol, the experimental conditions, the positive and negative controls...
Assuming that there is no technical problem occurring during your staining procedure, the easiest explanation is as follows: The epitope recognized by the antibody is cryptic meaning hidden inside of the protein. Heat denaturing under reducing conditions in the presence of SDS leads to an unfolding of the protein which makes the otherwise hidden epitope available while under more native staining conditions as for FCM it is not accessible. You can check the ab by co-immunoprecipitation from total extracts whether or not the epitope is accessible on your native protein. Covalent post translational modifications such as glycosylation are unlikely as they should not be destroyed during heating/denaturing for SDS-PAGE. (I like to mention, although quite unlikely, sometimes phosphoepitopes can be lost including after/during SDS-PAGE/immunoblotting or immunostainings if alkaline phosphatases are in your buffer(s) or equipment e.g. from using dish soaps for cleaning the equipment or glassware without extensive rinsing afterwards)