Are ther specific inhibitors, natural or synthetic? Or are there different metal requirements or cofactors. It looked like b glucosidase produces single sugar molecules and the other cellulases make two carbon type molecules. So a HPLC assay may distinguish at least those activities.
One has to calculate first U/mL of the solution for the enzyme of interest then go for total protein content in the same solution (mg/mL). Afterwards divide former by latter u will get U/mg. This will be the concentration o f enzyme wrt to total protein or other enzymes.
I think the problem is in the assay. If Avicel is the substrate, and he cannot use HPLC, how does he distinguish Units of each protein without HPLC? would TLC work?
How precise does the answer need to be? If you can resolve them on an SDS-PAGE gel and you use an appropriate stain (like silver) then you can quantify the relative abundances with densitometry. Knowing the total concentration by some other method (like BCA) you will then be able to know the concentrations of the various enzymes in your assays. I wouldn't claim to be under 15% error with this approach, however, without using good statistical methodology.
The above would work if you don't use a substrate to measure the active amount of enzyme for each. But if you have no abs, and it is a mixture of proteins, how do you know which protein is yours? I like the idea perhaps of a native gel though, along the lines of Ryan's thinking, followed by making slices, and then incubating the slices with the substrate. You can even look for the right MWs, to pick certain slices first. or you can do this with a sizing column, but I think the gel would give better separation.