For enzymes in particular, subcellular protein localization plays a key role in the proper functioning of cells by enabling interaction with required substrates and preventing unwanted side reactions. Regulation of subcellular localization, studied for numerous nucleocytoplasmic-shuttling, signal-transducing proteins (Xu and Massague, 2004
), adds an additional layer of complexity, enabling changes in access to substrates as well as upstream activators and inhibitors depending on the needs of the cell.
As an example, the upstream AMPK activator LKB1, under different stimuli, can localize to either the nucleus or cytoplasm, greatly affecting its own activity. Some LKB1 mutations that cause human Peutz-Jeghers syndrome constitutively sequester LKB1 to the nucleus and despite being outside the kinase domain, are phenotypically indistinguishable from mutations that abolish enzymatic activity (Nezu et al., 1999
; Boudeau et al., 2003b
; Xie et al., 2009
). Although the mechanism of LKB1 localization is complex, it involves other proteins, including the STRADs (STE-related adapter) that form a complex with LKB1, promoting nuclear export and inhibiting nuclear import (Boudeau et al., 2003a
; Dorfman and Macara, 2008
). Although LKB1 is an upstream activator of AMPKα (Lizcano et al., 2004
) and both proteins are kinases, LKB1 can function without other subunits bound, whereas AMPK is generally thought of as an obligate trimer (Hardie, 2007
Previous studies have elucidated both nuclear and cytoplasmic targets of AMPK. In the cytoplasm, AMPK most notably phosphorylates and inhibits ACC, a rate-limiting enzyme required for fatty acid synthesis (Carling et al., 1989
). Conversely, there are several known nuclear targets of AMPK (Leff, 2003
; Bronner et al., 2004
; Jager et al., 2007
; Narkar et al., 2008
), including PGC1α and PPARα/γ/δ, which regulate transcription in the nucleus. Furthermore, AMPK accumulation in or dispersion from the nucleus can be regulated by exercise, cellular stress, and circadian rhythms. In one study, exercise increased induced nuclear translocation of AMPKα2 in skeletal muscle (McGee et al., 2003
), where AMPK is known to activate PGC1α and subsequent gene transcription (Jager et al., 2007
). Another more recent study has demonstrated that AMPKα1 in the nucleus fluctuates in a circadian manner, regulating the circadian clock by inducing degradation of cryptochrome 1 (Lamia et al., 2009
). Clearly, mechanisms that regulate AMPK subcellular localization are widely utilized to modulate its access to downstream substrates.
The nuclear pore complex (NPC) plays a key role as a molecular sieve to help compartmentalize proteins between nucleus and cytoplasm. Indeed, many nucleocytoplasmic shuttling proteins contain signals to direct them in and/or out through the NPC (Yasuhara et al., 2009
). An AMPK trimer would far exceed the generally accepted nuclear pore diffusional cutoff size of 40 kDa (Gorlich and Kutay, 1999
) and would thus also need such NPC shuttling signals. Despite distinct AMPK targets in both the nucleus and cytoplasm, the mechanisms for regulating its localization remain unclear, particularly in organisms that have only single α/β/γ subunits (e.g., Drosophila
), where localization models based on different genetically encoded isoforms are not applicable.
One previous model for nuclear AMPK localization proposes that AMPKα2, but not AMPKα1, contains an NLS in the kinase domain that becomes functional only upon addition of leptin (Suzuki et al., 2007
). Because not all organisms encode leptin and we did not observe any localization differences for AMPKα2 when mutating key residues in the proposed leptin-stimulated NLS in transfected HEK293 cells (data not shown), regulation of AMPK localization is likely cell type-dependent. This can also be seen in the differential localization between AMPKα1 and AMPKα2 isoforms in insulinoma cells (Salt et al., 1998
), in contrast to HEK cells, where α1 and α2 localize similarly (data not shown). These effects are also seen with the β subunit, because only AMPKβ1 enriches in the nucleus upon mutation of two phosphorylation sites in HEK cells (Warden et al., 2001
), whereas AMPK complexes containing β2 preferentially localize in the nucleus in C2C12 cells under leptin treatment (Suzuki et al., 2007
). Distinct SNF1 β subunits are also thought to promote differential subcellular localization in yeast (Vincent et al., 2001
). Altogether, these results suggest that cells differentially regulate AMPK localization, and thus activity, through multiple pathways, depending on their unique metabolic requirements and hormonal responses.
The findings herein identify amino acids at the carboxy-terminus of AMPKα that modulate its nuclear export ( and ) that are nearly universal, with these sequences found across phyla ( and ), closely matching the consensus sequence for the leucine-rich CRM1-dependent NESs (la Cour et al., 2004
). Further, we identify a stress treatment in vivo, heat shock, which causes nuclear translocation of AMPKα, which requires the NES for increased phopho-AMPKα under this stressor. Because phosphorylation of the activation loop is required for AMPK activity, this mechanism might be expected to be beneficial for surviving physiological stress.
As C-terminally truncated AMPKα localizes to the nucleus in vitro in HEK cells and in vivo in Drosophila under unstressed conditions, this suggests that AMPKα is basally imported to the nucleus and that regulation of AMPK localization in response to stress would predominantly be affected through modulation of the export pathway. Adding further complexity, localization of the AMPK complex and partitioning of specific subunits may also be both cell type– and context-dependent. For instance, in multicellular organisms certain tissues (e.g., fat) provide energy to other tissues/organs (e.g., muscle) at their own expense. In these cases, AMPK activation likely leads to different physiological outcomes between cell types, such as increased lipid mobilization in fat cells versus increased lipid uptake in muscle cells. In these situations, differential localization of AMPK in distinct cell types could be used to generate these different cellular responses.
A further avenue of inquiry in regulation of AMPK localization is in the possible effects of posttranslational modification of AMPK subunits on the accessibility of the carboxy-tail of AMPKα. As the AMPKα carboxy-tail folds into a pocket formed by the α and β subunits after a long flexible loop (Amodeo et al., 2007
; Xiao et al., 2007
), altering the strength of these interactions could change its accessibility to CRM1, thus activating or inhibiting nuclear export. Although there are conserved residues that could be posttranslationally modified in the AMPKα carboxy-tail adjacent to the putative NES, we have so far been unable to identify residues flanking the NES that change the subcellular localization of AMPKα in vitro or are required for genetic rescue in vivo, as described earlier. One tantalizing possibility is that the potential phosphoserine mutations in β1 increase nuclear localization of AMPK by enhancing β interactions with the AMPKα tail, thus blocking nuclear export.
Whatever mechanisms determine AMPK localization, they must take into account two general observations: 1) AMPKα1 and AMPKα2 are largely genetically and functionally redundant in the mouse and 2) many organisms encode only a single isoform for individual AMPK subunits. In many mouse strains AMPKα1 and AMPKα2 are genetically redundant as single α1 knockouts or α2 knockouts are viable, yet double knockouts are lethal (B. Viollet, personal communication), suggesting that different AMPKα isoforms are functionally redundant for activities required for life in vivo. Therefore the elucidation of mechanisms that regulate AMPKα subcellular localization beyond isoform distinctions, such as the ones identified in this study, is vitally important to the understanding AMPK regulation in vivo.