AMPK is a serine threonine protein kinase

AMPK is a serine/threonine protein kinase composed of a catalytic α subunit with the activating phosphorylation site (Thr172) and two regulatory subunits, β and γ. The two AMPK variants, α1 and α2, show different cellular localization in mammalian cells. AMPKα2 is detected in nuclear and non-nuclear fractions whereas AMPKα1 is only found in the latter. An increase in AMP leads to the binding of AMP to the γ subunit, subsequent conformational changes of the AMPK heterotrimer and AMPK activation. Phosphorylation by the upstream kinases, LKB1 and Ca2+/calmodulin-dependent protein kinase kinase (CaMKK) increase AMPK activity [34]. Adiponectin stimulates translocation of LKB1 from the nucleus to the cytosol. APPL1 enhances cytosolic localization of LKB1 by interacting with this kinase via its BAR domain thereby facilitating phosphorylation of AMPK [35] (Fig. 2). Interestingly, the antidiabetic drug metformin also stimulates cytosolic localization of LKB1 and thereby may increase AMPK phosphorylation. CaMKK is another upstream kinase of AMPK that phosphorylates AMPK independent of AMP levels. Adiponectin induces Ca2+ release from the endoplasmic reticulum by activating phospholipase C thereby stimulating CaMKK mediated phosphorylation of AMPK. This pathway is independent of LKB1 and APPL1 and seems to play a minor role in adiponectin mediated activation of AMPK [35]. In myocytes adiponectin via binding to AdipoR1 also stimulates extracellular Ca2+ influx and subsequent activation of CAMKK and AMPK (Fig. 2) [36]. AMPK in concert with sirtuin 1 (SIRT1) promote phosphorylation and deacetylation of peroxisome proliferator-activated receptor y coactivator-1a (PGC-1α), thereby enhancing PGC-1α activity. Muscle-specific AdipoR1 knock-out mice display decreased PGC-1α, lower number of mitochondria, reduced type I fibres and lower exercise capacity [36]. However, whether APPL1 is involved herein has not been analysed so far. Full-length and globular adiponectin promote translocation of GLUT4 to the plasma membrane. Similar effects are observed in tak1 inhibitor where APPL1 is overexpressed whereas recombinant expression of APPL1 lacking the PTB domain impairs adiponectin-induced GLUT4 membrane translocation. The N-terminal BAR domain of APPL1 binds activated Rab5 and adiponectin induces the formation of this complex. Adiponectin stimulated GLUT4 membrane translocation is blocked by dominant-negative Rab5 and by APPL1 devoid of its BAR domain [35].
APPL2 down-modulates AdipoR1 signalling APPL2 is an APPL1 isoform with 54% identity in protein sequence and similar protein domain organization [37]. APPL2 binds to AdipoR1 and AdipoR2, and in addition forms heterodimers with APPL1 via an interaction of their BAR domains [38]. APPL2 negatively regulates AdipoR1-dependent signalling in C2C12 cells exploiting two distinct mechanisms: firstly APPL2 competes with APPL1 for the binding to AdipoR1, and secondly APPL2 anticipates APPL1 binding to AdipoR1 by forming a heterodimer with APPL1 (Fig. 3). The biguanide metformin as well as adiponectin promote dissociation of the APPL1/APPL2 complex and translocation of cytosolic APPL1 to the plasma membrane. Adiponectin furthermore stimulates disruption of the AdipoR1/APPL2 complex thus enabling binding of APPL1 [38] (Fig. 3).
Involvement of APPL1 in adiponectin-stimulated ERK1/2 phosphorylation Adiponectin activates the ERK1/2 MAP kinase pathway in vascular smooth muscle cells, endothelial cells, Hek293 cells, macrophages and hepatocytes [16], [39], [40]. Nevertheless, suppression of ERK1/2 by adiponectin has also been demonstrated [41], [42] and currently there is no simple explanation for these discrepant findings. While AdipoR1 seems to be important for ERK1/2 activation in hepatocytes [39], AdipoR2 is involved in ERK1/2 activation in the granulosa tumour cell line KGN [43] and both receptors stimulate ERK1/2 in Hek293 cells [16]. Some studies confirmed RNAi-mediated knock-down of AdipoR1 or AdipoR2 at the mRNA but not the protein level, and therefore, the different levels of protein reduction might partly explain the contradictory findings [16], [39], [43].