Depending on the site of labeling and the environmental conditions, the bond that results from labeling cysteine in proteins with maleimides can vary in stability. In the case you described, it may have been a particularly unstable labeling site, probably because of the identities of adjacent residues. You may have more luck with a different labeling chemistry.

Thank you. I can imagine that some bonds are instable and can be lost. But first change, from invisibility after run to visibility after first washing processes was strange. Does it mean that labeling occured but the fluorescence was quenched (what mechanism ever)?

The fluorescence intensity (quantum yield) may have increased due to the fixative. However, the acidity of the fixative may also have contributed to the loss of the dye. With a good labeling reaction, you should be able to see the fluorescent protein band in the gel without any fixing or washing.

Another way to tell if your labeling reaction has worked is to remove the excess labeling reagent by passing the protein through a desalting column (or two in sequence, for high label concentrations) equilibrated with a neutral pH buffer. Measure the average label stoichiometry of the recovered protein using the visible wavelength absorbance spectrum of the dye to measure the dye concentration, and measure the protein concentration using a colorimetric protein assay (e.g. Bradford).

You can also do intact protein mass spec to see the distribution of label molecules per protein molecule (0, 1, 2 , etc.).

Thank you, I think we will start further experiments...