My reading of the literature, chiefly Ward et al, Photochem. Photobiol.
35: 803-808 (1982), suggests that GFP is much more prone to oligomerize than
suggested in the previous correspondence between Sam Henderson
(Shender at macc.wisc.edu) and Stu Kuhstoss (sk at lilly.com). Fig. 2 and Fig. 5
of the paper by Ward et al suggest that GFP is >50% oligomerized (probably
dimerized) at an ionic strength of 100 mM NaCl and a protein concentration
of only 0.16 mg/ml = 6 micromolar. The estimate of 5 mg/ml for the
threshold for dimerization might have come from Fig. 3 of that paper, but
that was obtained in 1 mM Tris-HCl as the only salt, which is fine for
protein biochemistry but rather different from the inside of most cells. As
we point out in our review on GFP, due to appear in the November issue of
Trends in Biochem. Sciences, this often-overlooked tendency to dimerize
would make the long-wavelength (470 nm) excitation peak of the wild-type
GFP even weaker. Also there is the possibility (or worry) of artificially
inducing the dimerization of whatever GFP might be fused to.