berberine); (2) directly, by binding at site 3 (e.g., ZMP, derived
from intracellular metabolism of the prodrug 5-aminoimidazole-4-
carboxamide riboside (AICAR)); or (3) directly, by binding at the
ADaM site (e.g., A769662) [3]. Compounds that activate AMPK
in intact cells generally fall into one of these categories. AMPK
activators that work via mechanism (1) increase the intracellular
concentration of AMP or ADP, while those that work via mecha-
nism (2) increase the intracellular concentration of an AMP ana-
logue. In either case, an intact AMP-binding site 3 on the AMPK-γ
subunit is required. There are a number of well-characterized
mutations within the CBS domains of theγsubunits that disrupt
AMP binding, and heterotrimeric complexes containing these
mutations are in general less sensitive to changes in AMP levels
within the cell [4]. We have used transient transfection of plasmids
encoding “RG” mutations affecting the binding of AMP at site 3 in
humanγ1 (R299G) or humanγ2 (R531G) to determine whether
AMPK activators require an intact site 3; if they do, this suggests
that they mediate their effects via mechanism (1) or (2) [3]. It is
sufficient to transiently transfect DNA encoding theγsubunit only;
the recombinantγsubunit will compete with the endogenousγ
subunit for binding to the availableαandβsubunits, thus acting as
a “quasi knock-in” or dominant negative mutant. Although the
transfected γ subunit does not usually completely replace the
endogenousγsubunit, this is not a problem if the transfectedγ
subunit is tagged (e.g., FLAG-tagged) and is immunoprecipitated
prior to assay using anti-epitope tag antibody (e.g., anti-FLAG).
The endogenousγsubunits will not interfere in such IP-kinase
assays, although they may still contribute to phosphorylation of
downstream targets, making any studies of the latter difficult to
interpret.
Using the Flip-In™system, we have also made cell lines stably
expressing WT or RG mutants, which has the advantages that
clones of cells that express the WT and RG mutant at similar levels
can be selected, and that the use of expensive transfection reagent is
minimized. While we have used these stable cell lines successfully
[3, 5], unfortunately the RG mutant cells appear to be rather
unstable during serial passage. We find that the expression of the
RG mutant declines, and/or expression of the endogenous WTγ
subunit is restored. We suspect that the cells are undergoing selec-
tion for promoter mutations that either reduce expression of the
RG mutant or increase expression of the endogenous WTγsub-
unit. It is possible that this problem could be overcome by starting
withγ1/γ2 knockout cells generated by CRISPR/Cas9, but in the
meantime, we describe below the transient transfection approach.
In crystal structures of humanα 2 β 1 γ1 complexes with ligands
(A769662 or 991) bound in the ADaM site, the ligands are bound
in a cleft between the carbohydrate-binding module (CBM) on the
β1 subunit and the N-lobe of the kinase domain on theα2 subunit240 Simon A. Hawley et al.