I haven't worked specifically with the SUMOylation of FXR, but I have experience with other SUMOylated proteins.
If you have performed the classical IP, then of course, you will not see any modification on the desired protein unless you use stringent conditions to block de-SUMOylation. SUMOylation is a reversible modification, and during IP, the SUMO-modification of the desired protein can be lost. Therefore, it is necessary to use ~20 mM of NEM. Alternatively, you can also try IP with TCA and SDS (this worked very well for me). You can adopt the protocol from the following source: https://pubmed.ncbi.nlm.nih.gov/6290490/.