SEC-MALS and SAXS should help to define the apparent MW of the active enzyme. MS on the native complex is also a possibility. Article Studying macromolecular complex stoichiometries by peptide-b...
In addition to the methods suggested by Dominique Liger, here are some older methods. The following pertains to the purified protein.
If the protein is a homo-oligomer (all identical subunits) then the monomer molecular weight is compared to the molecular weight of the whole protein to determine the stoichiometry. This is pretty straightforward as long as there aren't a very large number of subunits. You can use SDS-PAGE to measure the molecular weight of the subunit, or size exclusion chromatography under denaturing conditions. For the whole protein, you can use blue native gel electrophoresis, analytical ultracentrifugation, dynamic light scattering, or SEC-MALS under native conditions to measure the molecular weight. Divide the molecular weight of the whole protein by the molecular weight of the subunit to get the stoichiometry.
If the homo-oligomer is an enzyme, then you may be able to get the stoichiometry by using an active site titration with a potent inhibitor to measure the concentration of active sites, and compare that to the total concentration of protein.
It is more difficult to measure the subunit stoichiometry of proteins with multiple different subunits. One way is to use SDS-PAGE with Coomassie Blue staining and densitometry to measure the relative concentrations of each subunit, adjusting the amount of stain for the molecular weight. This is tricky, however, because they may not all stain with the same ratio of dye to amino acids due to differences in the amino acid compositions. But it is a simple way to start, as long as you can find a densitometer.
Another way is biosynthetically radiolabel the protein with 35S-methionine, purify the protein, run the gel, and quantify the amount of radioactivity in each band. Then use the known amount of methionine residues in each subunit to calculate the relative amounts of each.
A variation on this method would be to label all the lysine residues plus the N-terminus (if it isn't blocked) with an amine-reactive compound (radioactive or fluorescent), or all the Cys residues with a thiol-reactive compound, under denaturing conditions (e.g. 6 M guanidine-HCl), then quantify the amount of that compound in each subunit.
Still another method would be to do a quantitative N-terminal sequencing of the protein by Edman degradation. Using the known N-terminal sequences of each subunit, you would then be able to calculate the relative amount of each subunit.
To add to the other answers, you can capture the subunit by crosslinking with formaldehyde and compare the sizes by SDS-PAGE. This has the benefit over native-PAGE because you can compare the samples side by side on the same gel.
If you have access to an ultracentrifuge you can get the diameter of the complex and use that to figure out the number of subunits.