Hoping to narrow this list down, scientists have tried to determine experimentally which glycoproteins CD22 binds to, but until recently their success was limited. It's nearly impossible to separate CD22 from the rest of the contents of the cell without also stripping away whatever CD22 is bound to.
Careful experiments in which CD22 was attached to an antibody and allowed to bind to different cellular glycoproteins in vitro yielded inconclusive results--the experiments identified not one but a number of glycoproteins to which CD22 can bind. This stymied scientists because they did not know which of all these possible targets CD22 bound to during the regulation of B cell activation. For instance, the B cell surface proteins CD45 and IgM were both identified as targets of CD22 in vitro. But there was no evidence that CD45 and IgM were the actual targets physiologically.
Taking a different approach, Paulson and his colleagues decided to settle the question by looking at the target of CD22 in a natural or "in situ" setting. As it turns out, the in vitro experiments that identified CD45 and IgM as candidates for recognition by CD22 were wrong.
Paulson and his colleagues designed a way to attach a common type of cross-linking agent known as a "photoaffinity label" to the sialic acid that would covalently attach to whatever protein it was close to when an ultraviolet light was shined on it. Then they fed the cells the modified form of the sialic acid so that all the glycoproteins on the outside of the cell incorporated this sugar.
This way, when they flipped on the UV light, the CD22 would become permanently attached to whatever glycoprotein to which it was in close proximity--essentially, to the protein to which it was bound. Then, by separating out the components of the cell, they would be able to tell what that molecule was.