Adenine nucleotides are thought to serve as second messengers in the c
ontrol of beta-cell function by glucose, e.g. by regulating the activi
ty of ATP-dependent K+ channels. However, their localization in differ
ent intracellular pools may mask the biologically relevant changes and
complicate the interpretation of measurements in whole cells. The pla
sma membrane of mouse islet cells was selectively permeabilized by the
alpha-toxin from Staphylococcus aureus to allow diffusion of cytoplas
mic nucleotides. After permeabilization of cells from freshly isolated
islets, approximately 68% of ATP, 45% of ADP, and 52% of AMP rapidly
diffused out of the cells, whereas the insulin content hardly varied.
The nondiffusible pool of nucleotides was stable for at least 90 min a
t 4 C, which suggests that it is contained in cellular organelles. The
size of this nondiffusible pool decreased proportionally to insulin s
tores when these were lowered by stimulating secretion to different de
grees during culture before permeabilization. From these results, it c
an be calculated that nondiffusible nucleotides are mainly contained i
n insulin secretory granules, with a small proportion in another, prob
ably mitochondrial, compartment. Approximately 80% GTP and 30% GDP wer
e present in the diffusible pool, and their relative proportions in th
e granular pool were only about 20% that of adenine nucleotides. Incub
ation of the cells in 20 instead of 2 mM glucose before permeabilizati
on did not affect the nondiffusible pool, which indicates that the inc
rease in the ATP/ADP ratio measured in intact cells occurred in the di
ffusible pool. Cytoplasmic nucleotide levels could be evaluated by sub
tracting the nondiffusible pool from the measurements in intact cells.
It emerges that glucose induces large changes in the ATP/ADP ratio in
the cytoplasmic pool, and that these changes are largely due to a fal
l in ADP.