How can i keep sugars fixed in protein during the MD simulation?

Hello all.
I want to perform an MD simulation of a protein-ligand complex on a laccase containing 9 NAG( carbohydrate) in a mixture of water and a ionic liquide.I followed the Justin A. Lemkul tutorial for the protein-ligand complex, but istead of cgenff server, I used swissparam for generating parameters.
The tutorial went smoothly, but when i used vmd for visualizing the md.gro, I realizes, 2 of the NAGs (a monomer and an dimer) which were complexed to proteine, changed their place and were far away from protein, at the end of simulation time(100ns). Is this simulation reliable or not? how can i keep NAGs fixed in protein during the MD simulation?

to restrain the distance between two molecules (or part of them) you can use define virtual site on COM or COG of the molecules (in your case the sugar and closeby residues) using virtual sites_n and then apply bond type 6 between the two virtual sites
For syntax see File formats — GROMACS 2021.2 documentation

Best regards