Moving towards isoform-specific AMPK activation

The AMP-activated protein kinase (AMPK) αβγ heterotrimer is a highly conserved serine/threonine protein kinase that acts as a metabolic fuel sensor and is crucial for maintaining cellular energy homeostasis. Mammalian AMPK forms complexes in a 1:1:1 ratio made of unique subunit isoform variations (α1, α2, β1, β2, γ1, γ2, γ3) that allow for 12 distinct AMPK complexes to form, where each complex is subject to a range of modifications such as phosphorylation. Each isoform exhibits distinct tissue-expression signatures, α2 and β2 are expressed in a range of tissues but display high expression in skeletal muscle and γ3 shows the highest selectivity being exclusively expressed in skeletal muscle with small amounts recently found in brown adipose tissue.

The majority of direct allosteric AMPK activators bind at a hydrophobic binding pocket formed between the α and β subunits, termed the allosteric drug and metabolite site (ADaM site). AMPK is capable of stimulating glucose uptake independently of insulin signalling, where activating AMPK using ADaM site activators has been shown to improve key hallmarks of type 2 diabetes mellitus (T2DM). Despite this, recent studies show that chronically stimulating all 12 AMPK complexes (pan activation) is detrimental as it can lead to hypertrophic cardiomyopathy. This precludes pan activators from progressing to clinical trials. The current direction for the field is to develop isoform-specific activators that target AMPK expressed in select tissues. For the treatment of T2DM it is beneficial to target skeletal muscle as it is the primary site for glucose disposal, hence, α2, β2, and γ3 isoforms are the most favourable to target. This can be achieved by developing α2β2-specific ADaM site compounds or with novel drugs targeting γ3 directly.

Despite this, the current knowledge on isoform-specific AMPK regulation is limited, in particular the regulation by a range of phosphorylation sites across each subunit, the functional role of an NH2-terminal extension (NTE) unique to the γ3 subunit, and the structural mechanism for the isoform specificity of the currently limited range of direct allosteric activators. Therefore, the goal of this thesis is to tackle each of these questions to ultimately aid the field in the development of isoform-specific AMPK activators.

Our lab recently characterised the ADaM site activator SC4 as not only α2-selective, but also as a potent β2 activator. In Chapter 2 I perform structure/function analysis of SC4 by substituting the 2-hydroxyphenyl group with polar-substituted cyclohexene-based probes. This resulted in the formation of two compounds, MSG010 and MSG011, that do not display α2-selectivity and are hence classified as pan activators. A crystal structure of MSG011 complexed to AMPK α2β1γ1 revealed a similar binding mode to SC4. Interestingly, it highlighted the absence of an interaction that we saw in the SC4/α2β1γ1 crystal structure between the SC4 2-hydroxyphenyl group and α2K31, which may be important for directing α2-selectivity. These findings will guide future design of α2β2-selective AMPK activators. In addition, MSG010 and MSG011 will serve as important tool compounds in AMPK research as they are most potent pan activators available to date.

In Chapter 3 I use a targeted mass spectrometry approach to generated precise phosphorylation stoichiometry profiles of 18 phosphorylation sites across all 12 AMPK complexes. This uncovered important isoform-specific differences, particularly in the basal level of βS108 phosphorylation which is located in the ADaM site and dictates the potency of most activators. Mechanistic target of rapamycin complex 1 (mTORC1) is a nutrient-sensitive protein kinases that governs cell growth and proliferation. It has been known for some time that AMPK inhibits mTORC1 activity by phosphorylation, and our lab recently discovered AMPK is directly phosphorylated by mTORC1 on α2S345 to suppress activity, forming a fundamental negative feedback loop. I found seven phosphorylation sites on AMPK were sensitive to pharmacological mTORC1 inhibition, including four in the unique γ2-NTE and α2S377 which is located in the nucleotide-sensing motif. In particular, β1S182 and β2S184 were found to be mTORC1 substrates in vitro and near-maximally phosphorylated under cellular growth conditions. Lastly, I identify two phosphorylation sites on the γ3-NTE.

Despite γ3 being a promising therapeutic target, it is yet to undergo rigorous biochemical characterisation leaving it without a solved crystal or cryoEM structure and the function of its NTE remains unknown. In Chapter 4 I discover that removal of the γ3-NTE results in elevated AMPK activity, suggesting it contains an autoinhibitory region. Furthermore, I detected a direct interaction between the γ3-NTE and the α2 kinase domain which may partly explain the mechanism for AMPK autoinhibition. I characterise one of the γ3-NTE phosphorylation sites I discovered in Chapter 3, γ3S14, identifying it as an autophosphorylation site in vitro and in cellulo with the potential for alternative upstream kinases in cellulo.