STK11 (LKB1) and AXIN1 Tumor Suppressors

STK11 (Serine/Threonine Kinase 11, commonly designated Liver Kinase B1 [LKB1], encoded on human subtelomere Chr19p13.3 across 10 exons; 433 amino acids, ~50 kDa) and AXIN1 (Axis Inhibition Protein 1, encoded on subtelomeric Chr16p13.3 across 11 exons; 862 amino acids in isoform a, ~96 kDa) are evolutionarily conserved, high-GC subtelomeric tumor suppressor guardians that couple cellular bioenergetics, apico-basal epithelial polarity, and developmental Wnt/β-catenin morphogen signaling. Whereas STK11 (LKB1) functions as the master upstream serine/threonine kinase that phosphorylates Thr172 of the catalytic α-subunit of AMP-activated protein kinase (AMPK) and 12 additional AMPK-related kinases (SIK1–SIK3, MARK1–MARK4, NUAK1/2, BRSK1/2, and SNRK), AXIN1 functions as the concentration-limiting central scaffold of the cytoplasmic β-catenin destruction complex alongside Adenomatous Polyposis Coli (APC), Glycogen Synthase Kinase 3β (GSK3β), and Casein Kinase 1α (CK1α).

Both STK11 (NC_000019.10:1,205,798–1,228,434, situated ~1.2 Mb from the 19p telomere) and AXIN1 (NC_000016.10:287,440–352,723, situated ~287 kb from the 16p telomere) reside within recombination-active, CpG-dense (~62–68% GC) human subtelomeric compartments enriched in G-quadruplex and Z-DNA-forming repeats. Consequently, both loci are vulnerable to terminal 19p or 16p chromosomal deletions (loss of heterozygosity, LOH) paired with somatic truncating or missense mutations: biallelic STK11 inactivation occurs in 15% to 30% of non-small-cell lung adenocarcinomas (LUAD) and ~20% of cervical carcinomas (and in the germline causes Peutz–Jeghers syndrome), whereas biallelic AXIN1 loss occurs in 8% to 12% of hepatocellular carcinomas (HCC), medulloblastomas, and colorectal and gastric adenocarcinomas.

Allosteric Activation by the Pseudokinase STRADα and Scaffold MO25α

Unlike nearly all other eukaryotic protein kinases, which require phosphorylation of their activation-loop T-loop residue by an upstream kinase to adopt an active conformation, STK11 (LKB1) is allosterically activated without activation-loop phosphorylation by assembling into a 1:1:1 heterotrimeric complex with the STE20-related pseudokinase STRADα (STRADA) (or STRADβ) and the armadillo-repeat scaffolding protein MO25α (CAB39) (or MO25β).

Crystal structures of the heterotrimeric LKB1–STRADα–MO25α holoenzyme at 2.65 Å resolution reveal how a catalytically inactive pseudokinase activates an active kinase: STRADα lacks the conserved catalytic Asp and Mg²⁺-binding DFG residues required for phosphoryl transfer, yet still binds adenosine triphosphate (ATP) with high affinity. Binding of ATP and MO25α (via a C-terminal Trp-Glu-Phe motif on STRADα) locks STRADα in a "closed" active-like conformation. The pseudokinase cleft of STRADα and the concave surface of MO25α then clamp onto the N- and C-lobes and activation loop of LKB1, stabilizing the LKB1 αC-helix and activation loop in its catalytically competent geometry and exporting LKB1 from the nucleus into the cytoplasm (where C-terminal farnesylation at Cys430 tethers a subset of LKB1 to endomembranes and lysosomal surfaces).

AMP/ADP Sensing at the Lysosomal v-ATPase–Ragulator Complex

When cellular oxidative phosphorylation or glucose availability declines, adenylate kinase (2 ADP ⇌ ATP + AMP) converts small drops in ATP into large relative increases in the intracellular [AMP]:[ATP] and [ADP]:[ATP] ratios. Binding of AMP or ADP to the cystathionine-β-synthase (CBS) repeats (γ1/γ2/γ3 subunit) of heterotrimeric AMPK (αβγ)—together with fructose-1,6-bisphosphate depletion sensed by aldolase at the lysosomal v-ATPase–Ragulator–AXIN1 complex—recruits LKB1 to phosphorylate Thr172 on the AMPKα (PRKAA1/PRKAA2) activation loop, increasing AMPK kinase activity more than 100-fold.

Activated AMPK immediately restores bioenergetic homeostasis by phosphorylating anabolic and catabolic switches:

1. Inhibition of mTORC1 Anabolism: AMPK phosphorylates and activates TSC2 (Ser1387) while phosphorylating the mTORC1 subunit Raptor (Ser722/Ser792), shutting down cap-dependent protein translation and ribosome biogenesis.
2. Inhibition of De Novo Lipogenesis and Activation of β-Oxidation: AMPK phosphorylates and inhibits acetyl-CoA carboxylase 1 and 2 (ACC1 Ser79 and ACC2 Ser212), lowering malonyl-CoA levels to relieve inhibition of carnitine palmitoyltransferase 1 (CPT1) and drive mitochondrial fatty acid β-oxidation while preserving intracellular NADPH against H₂O₂ oxidative stress.
3. Activation of Autophagy and Mitophagy: AMPK directly phosphorylates ULK1 (Ser317/Ser555/Ser777), initiating autophagosome biogenesis and clearance of damaged mitochondria.
4. Transcriptional and Polarity Regulation via SIK and MARK Kinases: Independently of AMPK, LKB1 constitutively phosphorylates the Salt-Inducible Kinases (SIK1–SIK3) (which phosphorylate and sequester CRTC [TORC] coactivators and class IIa histone deacetylases in the cytoplasm) and Microtubule Affinity-Regulating Kinases (MARK1–MARK4 / PAR-1) (which regulate epithelial tight junctions and cell polarity).

Concentration-Limiting Phase Separation of the Destruction Complex

In the absence of extracellular Wnt ligands (Wnt-OFF state), cytoplasmic levels of the transcriptional coactivator β-catenin (CTNNB1) are kept below nanomolar thresholds by continuous phosphorylation and proteasomal degradation executed by the β-catenin destruction complex. Within this macromolecular machine, AXIN1 serves as the central rate-limiting architectural scaffold: in Xenopus extracts and mammalian epithelial cells, intracellular AXIN1 concentration (~20–150 pM to low nanomolar) is 1,000-fold to 5,000-fold lower than that of APC, GSK3β, CK1α, or β-catenin—insulating the Wnt pathway from cross-talk with other GSK3β-regulated pathways (such as insulin/PI3K–AKT signaling).

Structurally, AXIN1 contains four modular interaction regions connected by intrinsically disordered linkers:

1. N-Terminal RGS (Regulator of G-Protein Signaling) Domain (residues 88–211): Binds the SAMP (Ser-Ala-Met-Pro) repeats of the tumor suppressor APC.
2. Central GSK3β-Binding Helix (residues 380–404) and β-Catenin-Binding Domain (residues 465–500): Brings the kinase GSK3β and its substrate β-catenin into immediate spatial proximity, increasing the effective molarity and phosphorylation rate of β-catenin by >10,000-fold.
3. Casein Kinase 1α (CK1α) and PP2A Binding Regions: Recruits CK1α to prime β-catenin via phosphorylation at Ser45, which licenses processive hierarchical phosphorylation by GSK3β at Thr41, Ser37, and Ser33.
4. C-Terminal DIX (Dishevelled and Axin) Domain (residues 779–862): Undergoes head-to-tail dynamic polymerization (Kd ≈ 5–20 μM), cross-linking AXIN1 and multivalent APC molecules into micron-scale biomolecular condensates ("destruction complex puncta") that present phosphorylated pSer33/pSer37 β-catenin to the SCFβ-TrCP E3 ubiquitin ligase for 26S proteasomal destruction.

Tankyrase-Mediated PARylation and Hepatocellular Carcinoma Subtypes

Cellular AXIN1 protein stability is tightly governed by Tankyrase 1 and Tankyrase 2 (TNKS / TNKS2, PARP5a/b). Tankyrases recognize a conserved N-terminal hexapeptide motif (RPPVPG, residues 22–27) on AXIN1 and AXIN2, catalyzing NAD⁺-dependent poly(ADP-ribosyl)ation (PARylation) that recruits the RING E3 ubiquitin ligase RNF146 (Iduna) to degrade AXIN1/2. Small-molecule tankyrase inhibitors (XAV939, IWR-1, G007-LK, OM-153) block AXIN PARylation, stabilizing AXIN1 and AXIN2 protein levels to suppress oncogenic β-catenin signaling.

Notably, in human hepatocellular carcinoma (HCC), somatic truncating mutations in AXIN1 (~10% of HCCs) are mutually exclusive with gain-of-function exon 3 mutations in CTNNB1 (~30% of HCCs). Transcriptomic and epigenetic profiling shows that whereas CTNNB1-mutant HCCs drive high-level canonical Wnt target expression (GLUL, LGR5, AXIN2) in well-differentiated "immune-excluded" tumors, AXIN1-mutant HCCs induce a weaker canonical Wnt signature (partially buffered by residual AXIN2) while simultaneously disrupting the lysosomal AXIN1–v-ATPase–Ragulator–LKB1–AMPK and YAP/TAZ (Hippo) complexes—converging with the STK11 metabolic axis to drive a proliferative, YAP/Notch-active hepatocarcinogenic program.

Peutz–Jeghers Syndrome (PJS)

Heterozygous germline loss-of-function mutations in STK11 cause Peutz–Jeghers syndrome (PJS, OMIM #175200), an autosomal dominant disorder (incidence ~1 in 50,000 to 200,000 births) characterized by mucocutaneous melanocytic macules on the lips, buccal mucosa, and digits, multiple gastrointestinal hamartomatous polyps with a pathognomonic arborizing smooth-muscle core, and an 85% to 90% cumulative lifetime cancer risk across gastrointestinal, pancreatic, breast, gynecological (specifically minimal deviation adenocarcinoma of the cervix, or adenoma malignum, and sex cord tumors with annular tubules [SCTAT]), and testicular Sertoli cell sites. Intriguingly, mouse genetic models revealed that early PJS hamartomatous polyps arise from Stk11 haploinsufficiency in stromal myofibroblasts and T cells (via AMPK/SIK loss and TGF-β/IL-6 dysregulation) before epithelial LOH drives progression to invasive adenocarcinoma.

Mechanism of PD-1/PD-L1 Resistance in KRAS–STK11 Co-Mutant Lung Adenocarcinoma

In non-small-cell lung adenocarcinoma (LUAD), STK11 loss co-occurs frequently with activating KRAS mutations (KRASG12C, G12V, G12D; termed the KL subtype, often accompanied by KEAP1 mutation). Clinical landmark cohorts (SU2C, CheckMate, KEYNOTE) established that KRAS + STK11-co-mutant LUAD is a primary driver of intrinsic resistance to anti-PD-1/PD-L1 immune checkpoint blockade, exhibiting low PD-L1 (CD274) expression, near-complete exclusion of CD8⁺ cytotoxic T lymphocytes ("immune-cold" desert phenotype), and dense infiltration of immunosuppressive SiglecFhigh/CXCR2+ neutrophils despite a moderate-to-high tumor mutational burden (TMB).

Mechanistically, loss of STK11 in KRAS-mutant lung tumor cells silences innate antitumor immunity through two convergent pathways:

1. Epigenetic Silencing of STING (TMEM173): Loss of LKB1–AMPK signaling elevates intracellular SAM synthesis and activates DNMT1 and EZH2 (PRC2), depositing repressive 5mC DNA methylation and H3K27me3 marks across the STING1 promoter. Consequently, cytosolic mitochondrial DNA (mtDNA) released by damaged mitochondria fails to activate the cGAS–STING–TBK1–IRF3 Type I interferon (IFN-β / CXCL10) cascade.
2. Inactivation of SIK1–SIK3 and CRTC2-Driven IL-6/CXCL Secretion: Loss of LKB1-mediated phosphorylation of SIK1/SIK2/SIK3 permits nuclear translocation of unphosphorylated CRTC2, which co-activates CREB at the IL6, CXCL1, CXCL2, and CSF3 (G-CSF) promoters—flooding the tumor microenvironment with myeloid-derived suppressor cells (MDSCs) that exclude T cells.

How does AXIN1 physically link the Wnt pathway to the STK11 (LKB1)–AMPK energy-sensing pathway?

In 2013–2017, Sheng-Cai Lin and colleagues demonstrated that AXIN1 is the obligate molecular scaffold for glucose-starvation-induced AMPK activation. When intracellular glucose and fructose-1,6-bisphosphate (FBP) levels fall, FBP-unoccupied aldolase bound to the lysosomal vacuolar H⁺-ATPase (v-ATPase) induces a conformational change in the Ragulator (LAMTOR1–5) complex. Cytoplasmic AXIN1 carrying constitutively bound STK11 (LKB1) docks onto the v-ATPase–Ragulator surface alongside myristoylated AMPK, positioning LKB1 to phosphorylate AMPKα at Thr172 even before global cellular [AMP]:[ATP] ratios change—proving that AXIN1 directly coordinates both Wnt morphogen destruction and LKB1–AMPK metabolic sensing.

Why are STK11 and AXIN1 both located in GC-rich human subtelomeres (Chr19p13.3 and Chr16p13.3)?

Both Chr19p13.3 (STK11) and Chr16p13.3 (AXIN1, adjacent to the α-globin cluster) belong to the highest-GC, highest-gene-density isochores (H3 isochore, >60% GC) in the human genome, maintained over mammalian evolution by GC-biased gene conversion (gBGC) during meiotic double-strand break repair near chromosome ends. While this open euchromatic subtelomere architecture supports constitutive housekeeping transcription across epithelial tissues, the high density of G4/Z-DNA secondary structures and subtelomeric mitotic recombination hotspots increases the rate of terminal 19p and 16p loss of heterozygosity (LOH), accelerating Knudson's "second hit" in pre-neoplastic clones.

What therapeutic strategies restore sensitivity in STK11-mutant lung adenocarcinoma?

Because STK11-null tumors lack AMPK-mediated metabolic checkpoints and exhibit elevated mitochondrial reactive oxygen species and replication stress, preclinical and clinical trials target three vulnerabilities: (1) Metabolic biguanides (phenformin/metformin) or GLS1 glutaminase inhibitors (especially in STK11/KEAP1 co-mutant tumors dependent on glutamine anaplerosis), which induce bioenergetic crisis because STK11-null cells cannot activate AMPK to conserve ATP; (2) Epigenetic EZH2/DNMT1 inhibitors or direct STING agonists (ADU-S100, TAK-676) to re-express STING and restore CXCL10/T-cell recruitment for combination with KRASG12C inhibitors (sotorasib, adagrasib) and anti-PD-1; and (3) ATR or CHK1/WEE1 cell-cycle checkpoint inhibitors (ceralasertib), exploiting dNTP depletion in LKB1-deficient cells.

1. Hawley, S. A., Boudeau, J., Reid, J. L., et al. (2003). "Complexes between the LKB1 tumor suppressor, STRADα/β and MO25α/β are upstream kinases in the AMP-activated protein kinase cascade." Journal of Biology, 2(4), 28. https://doi.org/10.1186/1475-4924-2-28
2. Zeqiraj, E., Filippi, B. M., Deak, M., Alessi, D. R., & van Aalten, D. M. F. (2009). "Structure of the LKB1-STRAD-MO25 complex reveals an allosteric mechanism of kinase activation." Science, 326(5960), 1707–1711. https://doi.org/10.1126/science.1178377
3. Skoulidis, F., Goldberg, M. E., Greenawalt, D. M., et al. (2018). "STK11/LKB1 mutations and PD-1 inhibitor resistance in KRAS-mutant lung adenocarcinoma." Cancer Discovery, 8(7), 822–835. https://doi.org/10.1158/2159-8290.CD-18-0099
4. Zhang, C.-S., Jiang, B., Li, M., et al. (2014). "The lysosomal v-ATPase-Ragulator complex is a common activator for AMPK and mTORC1, acting as a switch between catabolism and anabolism." Cell Metabolism, 20(3), 526–540. https://doi.org/10.1016/j.cmet.2014.06.014