Loss of the tumour suppressor LKB1/STK11 uncovers a leptin-mediated sensitivity mechanism to mitochondrial uncouplers for targeted cancer therapy

Abstract

Non-small cell lung cancer (NSCLC) constitutes one of the deadliest and most common malignancies. The LKB1/STK11 tumour suppressor is mutated in ∼ 30% of NSCLCs, typically lung adenocarcinomas (LUAD). We implemented zebrafish and human lung organoids as synergistic platforms to pre-clinically screen for metabolic compounds selectively targeting LKB1-deficient tumours. Interestingly, two kinase inhibitors, Piceatannol and Tyrphostin 23, appeared to exert synthetic lethality with LKB1 mutations. Although LKB1 loss alone accelerates energy expenditure, unexpectedly we find that it additionally alters regulation of the key energy homeostasis maintenance player leptin (LEP), further increasing the energetic burden and exposing a vulnerable point; acquired sensitivity to the identified compounds. We show that compound treatment stabilises Hypoxia-inducible factor 1-alpha (HIF1A) by antagonising Von Hippel-Lindau (VHL)-mediated HIF1A ubiquitination, driving LEP hyperactivation. Importantly, we demonstrate that sensitivity to piceatannol/tyrphostin 23 epistatically relies on a HIF1A-LEP-Uncoupling Protein 2 (UCP2) signaling axis lowering cellular energy beyond survival, in already challenged LKB1-deficient cells. Thus, we uncover a pivotal metabolic vulnerability of LKB1-deficient tumours, which may be therapeutically exploited using our identified compounds as mitochondrial uncouplers.

Keywords: Airway organoids; CRISPR/Cas9-mediated genome editing; Drug discovery; HIF1A-LEP-UCP2 axis; LKB1/STK11; Metabolic stress; Non-small cell lung cancer (NSCLC); Piceatannol; Tyrphostin 23; Zebrafish.

Fig. 1

Loss of LKB1/STK11 promotes HIF1A-mediated leptin activation accompanied by selective susceptibility to metabolic activators Piceatannol and Tyrphostin 23. (A) Schematic depicting the timeline of the synthetic lethality screen performed on wt and lkb1 zebrafish trunks. (B) Piceatannol and tyrphostin 23 were identified as synthetically lethal with loss of lkb1 in zebrafish trunks. (C-D) Survival analysis of wt and lkb1 larvae following treatment with 50 µM of piceatannol (pic), 25 µM of Tyrphostin (Ty23), or DMSO. (C) Pic-treated lkb1 larvae die from 24 h post-treatment (hpt) onwards (****), while wt larvae are only slightly affected (n.s.). (D) Ty23- treated lkb1 larvae die from 48 hpt onwards (****). Wt larvae are affected from 96 hpt onwards (***). ****P < 0.0001, ***P < 0.0005 compared to control treatments; Log-rank (Mantel-Cox) test. (E) Transcriptome analysis of total RNA isolated from wt and lkb1 larvae at 5, 7 dpf and wt trunks at 11 dpf. (F) q-PCR for lepb mRNA levels from total RNA extracted from wt and lkb1 trunks at 7 dpf. ****P < 0.005; two-tailed student’s t-test. (G) Top: heatmap showing reciprocal expression of LKB1 and LEP across LUAD tumour samples of the TCGA dataset (n = 483). Bottom: LKB1 and LEP expression density plots across LUAD (n = 483) and normal (n = 59) samples of the TCGA dataset. (H) Schematic depicting the experimental strategy implemented to acquire clonal TP53^KO or TP53^KO/LKB1^KO AOs via CRISPR/Cas9-mediated genome editing. (I) Western blotting for indicated markers in AO clones. (J) Representative images of normal (WT), TP53^KO and TP53^KO/LKB1^KO airway mutant organoids stained for haematoxylin and eosin (H&E) and Ki67. Scale: 10–30 μm. See also Fig. S3C. (K) Drug screening of indicated CRISPR/Cas9-engineered AO mutants, LUAD and WT AOs with various concentrations of piceatannol. For each concentration, stars represent statistical comparisons of TP53^KO/LKB1^KO AOs or LKB1^null LUAD AOs (as indicated by sample color) with WT AOs. See also Additional File 3. (L) Same as (K), for tyrphostin 23 (Ty23). See also Additional File 3. (M) Human LEP promoter map displaying binding positions of transcription factors in sense (+) or anti-sense (-) strands potentially regulating LEP. Identified motifs per transcription factor are displayed, along with their P values (P < 0.001). See also Additional File 4. (N) q-PCR for HIF1A and CEBPA mRNA levels in CRISPR/Cas9-engineered AOs. (O) HIF1A and CEBPA ChIP on the LEP promoter. (P) Immunoblotting for indicated markers in LKB1^null LUAD AOs treated with siRNA against HIF1A or non-targeting siRNA (siCTRL). ***P < 0.001 and **P < 0.01, of Student’s t-test. Error bars indicate s.e.m. N.s.; non-significant. Data shown are representative of at least 3 independent experiments

Fig. 2

Piceatannol and Tyrphostin 23 suppress LKB1-deficient tumourigenesis in vivo by increasing the energetic burden through an epistatic HIF1A-LEP-UCP2 axis. (A) Docking poses of piceatannol and tyrphostin 23 in drug-binding sites of the human VHL target protein (pdb id: 6GMR and 3ZRC). Illustration of predictions of the most stable binding geometries for piceatannol (burgundy carbons) and tyrphostin 23 (green carbons). Intermolecular hydrogen bonds are depicted as dashed red lines. (B) In vivo ubiquitination assay in HIF1A immunoprecipitates from indicated cell lysates. Immunoprecipitates and Input lysates are probed with displayed antibodies. (C) Immunoblotting of AO lysates with indicated markers. (D) Schematic of the experimental procedure of introducing LEP or UCP2 KOs in LKB1^null LUAD AOs. (E) Representative stainings and immunoblotting of two independent clones of LKB1^nullLEP^KO LUAD or LKB1^nullUCP2^KO LUAD organoids for displayed markers. See also Fig. S8 and S9A-B. Scale: 30 μm. (F) Drug screenings of indicated AOs with various concentrations of piceatannol/tyrphostin 23. For each concentration, stars represent statistical comparisons of LKB1^null LUAD AOs with the respective LEP or UCP2 KO mutants (as indicated by sample color). See Additional File 3 for relevant IC50s. (G) Survival and proliferation of AO lines (n ≥ 10) under treatment or not. See also Fig. S9C. (H) Schematic of drug administration in mice to determine potential toxic effects. See Additional File 5. (I) LKB1^wt or LKB1^null LUAD AOs were subcutaneously injected in immunodeficient mice and subsequently treated with piceatannol/tyrphostin 23 to determine potential effects on tumour growth. (J) Immunohistochemistry in formed tumours (n ≥ 6). Scale: 30–60 μm. (K-L) Quantification of Ki67 expression levels in (K) LKB1^wt and (L) LKB1^null organoid-derived tumours having received no treatment versus treated tumours. (M) Quantification of GLF16 immunofluorescence in displayed mouse xenografts. See also Fig. S10. (N) Tumour volume curves in mice. Stars represent statistical comparisons of LKB1^null LUAD + Pic or LKB1^null LUAD + Ty23 AOs with LKB1^null LUAD AOs receiving no treatment or LKB1^wt LUAD AO counterparts, as indicated by sample color. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05 of two-way ANOVA statistical analysis. Error bars indicate s.e.m. N.s.; non-significant. Data shown are representative of at least 3 independent experiments