We did this but had to make sure gels were embedded on edge, not an
easy thing to do, and keep it flat.  Our best success was to fix, process,
and embed in GMA, then use the skinny embedding molds to make sure the
gel was on edge.  One could also embed, then turn the block on edge,
and embed again with the gel on edge  This actually
proved to be the easier technic, to embed gel in GMA, then do the
double embedding.      It was important to fix the gel while it was on a
plate, or it fell apart if one tried to lift it out unfixed.  The sample
could be cut out with scalpel after fixation.  Still dicey!

The embedded blocks were carefully examined with a hand lens before
cutting to make sure the gel with cells was in the
proper orientation, adjust block and section. We were able to detect single
cells growing into the gel fairly easily, using NBF fixation,
H&E staining.  We experienced better resolution of cells in GMA than
with paraffin.  No IHC was done, however.

Gayle Callis