Introduction

STK11 (Liver kinase 1, LKB1) was first identified as a tumor suppressor gene through its association with the Peutz-Jeghers Syndrome (PJS)1. STK11 appears to be inactivated or mutated in sporadic cancers whose spectrum of tumor types suggests cooperation with exposure to environmental carcinogens. Thus, alterations in LKB1 have been found in non-small-cell lung cancer (NSCLC), malignant melanoma, and cervical cancer among others2,3,4.

The serine/threonine kinase LKB1 belongs to the calcium calmodulin family, which is ubiquitously expressed in several tissues and highly conserved among eukaryotes. Over the past 15 years, LKB1 has been implicated in a number of essential biological processes such as: cell cycle control5,6, cellular energy metabolism7,8, angiogenesis9,10, cell polarity11, and DNA damage response12. The sub-cellular localization and activity of LKB1 is controlled through its interaction with STRAD and the armadillo repeat-containing mouse protein 25 (Mo25)13,14. LKB1 regulates the activity of at least 14 downstream kinases related to the AMPK family15 and phosphorylates other substrates including STRAD16, PTEN17, and p21CDKN1A12. LKB1 is phosphorylated on at least eight residues, and evidence suggests that LKB1 auto-phosphorylates itself on at least four of these, whereas the other four are phosphorylated by upstream kinases8,16. While these post-translational modifications seem not to modify its kinase activity, they are involved in the different biological responses associated with LKB1, and likely in its interactions with other partners.

Up to date, more than 400 unique mutations have been described for the STK11 gene, where ~70% of these mutations promote the truncation of the protein and the other 30% represent missense mutations (COSMIC and TCGA-Bioportal). As a tumor suppressor, a number of studies have shown the contributions of the genetic loss of LKB1 to tumorigenesis. It has been demonstrated that LKB1 controls cell cycle through the transcriptional regulation of Cyclin D1 and p21CDKN1A56, where re-expression of LKB1 leads to G1 cell cycle arrest. The role of LKB1 in controlling cell metabolism through AMPK signaling has been widely documented. We know that the LKB1-AMPK axis controls lipid and glucose metabolism, and acts as a negative regulator of the Warburg effect suppressing tumor growth8,18. LKB1 is also important in the regulation of catabolic pathways leading to the increase of glucose uptake and modulation of glycolysis19 or the mobilization of lipid stores by stimulating lipases, such as adipose triglyceride lipase, to release fatty acids from triglyceride stores20. LKB1-AMPK-stimulated pathways also include increased turnover of macromolecules by autophagy, allowing the turnover of old and damaged molecules, or the replenishment of nutrient stores under starvation20. Additionally, several investigations have suggested the role of LKB1 in regulation of physiological21 and pathological angiogenesis22 through the regulation of VEGF, MMP-2, MMP-9, bFGF, and NOX1 expression, and its participation in neurophilin-1 degradation23,34; in fact LKB1D194A is considered a dead kinase. However, LKB1D194Y still has ~20% of the in vitro LKB1WT kinase activity, possibly because the deprotonated OH group of the Tyr residue still has a negative charge that might help to conserve some amount of activity. Residue Gly135 is also located in the ATP binding cleft; thus, it is tempting to speculate that substitution of a Gly for a charged amino acid might affect the ATP binding to LKB1 and, consequently its catalytic activity. Importantly, the described kinase activity of the different isoforms correlated with the metabolic stress response showed by the different cell lines expressing the different isoforms, according to the amounts of the direct LKB1 target AMPK.

The subcellular localization and activity of LKB1 is controlled through its interaction with STRADα and the armadillo repeat-containing mouse protein 25 (Mo25)13,14. All mutants except LKB1Y49D showed nuclear-cytoplasmatic localization. LKB1Y49D localized in the nucleus and had impaired activity, which is in agreement with its diminished binding capability to STRADα. Interestingly, recent studies support that LKB1Y49D mutation promotes variations in the binding energy pertaining to spatial conformation and flexibility, impairing the binding to STRADα and MO2535. Thus, the tumor suppressor activity linked to LKB1 kinase activity could be acquired through STRADα-dependent (LKB1G135R and LKB1D194Y) or STRADα-independent mechanisms (LKB1Y49D); the later also affect the subcellular localization and most likely other processes, such as transcriptional regulation6. In addition, LKB1Y49D, LKB1R87K, and LKB1D194Y showed a significantly shorter half-life than LKB1WT, and LKB1R87K was the isoform showing the shortest half-life. In this case, the substitution of the Arg residue by a Lys residue could promote post-translational modifications that might affect the protein stability.

Beyond its tumor suppressor activity, it is known that LKB1 serves as a FAK repressor to stabilize focal adhesion sites, contributing to cell adhesion and motility31. Both LKB1R87K and LKB1D194Y failed to repress FAK phosphorylation compared to the wild-type isoform. Interestingly, the LKB1R87K mutant promoted cell motility together with cytoskeletal reorganization. This effect on cell motility was also induced by LKB1D194Y, although it was less significant than the effects seen in LKB1R87K cells. In relation to this, it is known that LKB1 induces the degradation of β-catenin, a molecule involved in cell–cell contact in epithelial cells36. Both LKB1R87K and LKB1D194Y promoted more β-catenin degradation than the wild-type counterpart and induced the downregulation of TGFB1I1 and CDCP1, two molecules also involved in adhesion and motility37,38.

It is clear that reconstitution of LKB1 implies changes in the transcriptional profile of cells that were supported by previous studies33. Some genes involved in vesicle trafficking, regulation of autophagy, and inflammation (RUSC2 and AP1S3) failed to be upregulated by the mutant isoforms, which might have some effects in autophagosome maturation and distribution, as well as proinflammatory consequences39,40,41,42. The latter, could be more prominent in LKB1Y49D expressing cells that showed a stronger dysregulation of AP1S341,42. Another particularly interesting set of regulated genes, given to their clinical relevance, is cytokines (i.e., IL8 and CXCL1) which become downregulated upon STK11 expression. Our data suggest that loss of LKB1 and/or somatic mutation of this protein in tumor cells might lead to the upregulation of these cytokines contributing to the deregulation of the immune response and tumor angiogenesis. In particular, both LKB1Y49D and LKB1D194Y not only failed also repress IL8 and CXCL1 transcription but induced their transcriptional regulation. Vascular abnormalities in part due to deregulation of VEGF have been described in LKB1-deficient mice. Our observations suggest that loss of STK11 and/or LKB1Y49D and LKB1D194Y somatic mutations in tumor cells could also contribute to angiogenesis and immune response through upregulation of cytokines. This piece of data could be especially relevant in relation to the clinical responses to immunotherapy observed in LKB1-deficient KRAS-mutated lung tumors43. TDO2 an enzyme catalyzing the production of kynurenine, which promotes immune-tolerant dendritic cells (DCs) and regulatory T cells and thus contributes to an immunosuppressive environment, is also transcriptionally upregulated through the loss of STK11 or mutated isoforms. The contributions of these molecules to immunotherapy responses in a STK11-deficient or STK11-mutated context are currently under investigation.

Comparative secretome analysis of samples not only confirmed the role of LKB1 in regulating processes involved in motility and cell adhesion such as extracellular matrix organization and morphogenesis, but also revealed its participation in processes such as vesicular transportation and cytokine production. In fact, these results support both the possible role of LKB1Y49D in regulating IL8 and CXCL1, 2, 3, and 5, and promoting inflammation and angiogenesis, and the contribution of LKB1R87K to cell motility and adhesion through the regulation of extracellular matrix-remodeling molecules such as MMP2 and protein networks related to this phenotype. LKB1G135R and LKB1D194Y mutants showed certain similarities: both mutants lost the tumor suppressor capability of LKB1 and both mutations affect the LKB1 ATP binding cleft. In addition, the secretome analysis also identified similar protein–protein networks for these two isoforms, including processes networks involved in antigen processing, which could be relevant for immunotherapy.

Notably, the tumor suppressor activity of LKB1WT and LKB1R87K was also observed in vivo. Unfortunately, we did not observe any increased metastasis promoted by LKB1R87K. Interestingly, all tumors expressing LKB1Y49D showed signs swelling supporting a role of LKB1Y49D in regulating cytokine production and inflammation-related processes. These data were braced by an increased expression of IL8 and GROα, β, and γ (CXCL1, 2, and 3) in these tumors compared to the expression of these in parental cells or LKB1WT−expressing tumors. LKB1G135R and LKB1D194Y not only did not function as tumor suppressors but also the promoted tumor growth. Since parental cells lack the tumor suppressor activity of LKB1, these results suggest that LKB1G135R and LKB1D194Y mutants promote in vivo tumor growth by a mechanism independent of their kinase activity, a result that is reflected in the proliferation index of the tumors (Ki67 staining). In agreement with this, it has been described that LKB1 mutants that are catalytically deficient enhance cyclin D1 expression6, which contributes to tumor growth. This finding also supports the ability of these mutants to differentially regulate the expression of specific genes (i.e., LKB1Y49D regulation of IL8 and CXCL1) that might contribute to tumor development and progression.

In summary, we show that beyond the role of the non-mutated protein as a tumor suppressor, missense LKB1 somatic mutations could contribute to tumor development and/or progression by modifying not only intrinsic cell capabilities such as proliferation, motility, or adhesion but also the tumor microenvironment, affecting inflammatory responses and likely the immune system. Interestingly, these effects can be both kinase dependent and kinase independent, unveiling possible roles for LKB1 independent of its enzymatic activity. These results will contribute to clarify the unknown significance of missense somatic LKB1 mutations in human cancer, assisting with the diagnosis of diseases to help guide optimal treatment.

Methods

Reagents

Doxycycline, cycloheximide, and Ponceau S solution were obtained from Sigma-Aldrich Quimica (Madrid, Spain). Horseradish peroxidase and secondary fluorescent antibodies were obtained from GE Healthcare (Little Calfont, UK) and Thermo Scientific (Fremont, CA, USA), respectively. FITC-phalloidin was purchased from Abcam (Cambridge, UK); antibodies against LKB1 (ley37D/G6), IL-8, and GROα,β,γ (CXCL1, CXCL2, and CXCL3) were obtained from Santa Cruz Biotechnology (Heidelberg, Germany); p-AMPKα, AMPKα, Anti-LKB1 (D60C5F10), and Anti-β-catenin antibodies were obtained from Cell Signaling (Leiden, The Netherlands); anti-Ki67 antibodies were purchased from Abcam (Cambridge, UK); and anti-vimentin and anti-ERG antibodies were obtained from ROCHE, Ventana (Basel, Switzerland). Anti-E-cadherin antibodies were purchased from R&D Systems (Minneapolis, MN, USA). Anti-GAPDH antibodies were purchased from Trevigen (Gaithersburg, MD, USA).

Construct generation

The pLenti-rtTA2-IRES-H2B-GFP doxycycline-inducible plasmid was obtained from S. Tenbaum, HG Palmer’s Lab (Vall d´Hebron Institute of Oncology, VHIO). The human LKB1 sequence was subcloned from pCMV5-Flag-LKB1WT to obtain pLenti-rtTA2-LKB1WT-IRES-GFP. The different mutant isoforms were generated via site-directed mutagenesis using the Quick-Change II Kit (Stratagene, Cedar Creek, TX, USA) and the following primers: LKB1Y49D: ATCGGCAAGGACCTGATGGGG, LKB1R87K: AAGTTGCGAAAGATCCCCAAC, LKB1G135R: GCGTGTGTCGCATGCAGGAAA and LKB1D194Y: AAAATCTCCTACCTGGGCGTG.

Cell culture

A549 lung cancer cells, G361 melanoma cells and HeLa cells, all of which were null for STK11 mutations, were obtained from ATCC. Cells were grown in Dulbecco’s modified Eagle medium (DMEM) (Biowest, Riverside, MO, USA) supplemented with 10% fetal bovine serum (FBS) (Biowest) and 100 µg/mL penicillin/streptomycin (Thermo Scientific, Waltham, MA, USA) and maintained at 37 °C and 5% CO2. Cells were infected with the doxycycline-inducible construct rtTA2-H2B-GFP containing the different isoforms of LKB1 (LKB1WT, LKB1Y49D, LKB1R87K, LKB1G135R, or LKB1D194Y). Cells were induced with 1 µg/mL doxycycline (Sigma) for 48 h, and green fluorescent cells were sorted via a FacsAria Digital Cell Sorter (BD Biosciences, San Jose, CA, USA). For glucose starvation experiments cells were culture in DMEM without glucose for 3 h before total protein was recollected.

Immunoblots

Cells were lysed in RIPA lysis buffer, equal amounts of protein were subjected to SDS-PAGE and transferred to a PVDF membrane. Immunoblots were performed as previously described44,45.

Quantitative-reverse transcriptase polymerase chain reaction (qRT-PCR)

Two hundred micrograms of RNA per sample was used to generate cDNA using the SuperScript III First-Strand Synthesis System for RT-PCR Kit (Invitrogen, Carlsbad, CA, USA). Quantitative PCR analysis was performed using the SYBR Green PCR Master Mix Kit (Applied Biosystems Inc., Foster City, CA, USA) and the ABI Prism 7900HT Fast Real-Time PCR System (Applied Biosystems Inc.). The primers used are shown in Table 1. The measurements were calculated by employing the ΔΔCt method using SDS 2.3 Software (Applied Biosystems, Inc.). We applied geNorm algorithms to select TATA-binding protein (TBP) and human peptidyl-prolyl cis-trans isomerase A (HPPIA). to select TATA-binding protein (TBP) and peptidylprolyl isomerase A (cyclophilin A, PPIA) as the most stable reference transcripts. The geometric means of the expression values for both housekee** genes were used to normalize the expression and to calculate the normalized SD of all transcripts analyzed. Relative expression levels were calculated after normalization. Data were represented as mean ± SD of triplicates from three independent experiments (biological replicates).

Table 1 List of primers used for qPCR.
Full size table

Gene expression analysis

For microarray analyses, we used a genome-wide Human Gene 1.0 ST Array (Affymetrix, Santa Clara, CA, USA). Genes were considered differentially expressed in A549-LKB1-WT cells if the fold change was >1.2 and the p was <0.05 (noninduced cells versus cells treated with doxycycline for 48 h) using a two-tailed one-way ANOVA test.

Proliferation and colony-formation assays

We seeded 0.16 × 106 cells per well in 6-well plates. The data were collected in triplicate at 24, 48, 72, and 96 h after the initial seeding. For every time point, viable cells were counted (using a Neubauer chamber). Data are presented as the fold change with respect to the first measure (0 h.). For colony-formation assays, 300 cells per well were seeded in triplicate for every condition tested. Cells were maintained at 37 °C in 5% CO2 for ~20 days. Then, the cells were washed twice with PBS and fixed for 10 min with 4% paraformaldehyde in PBS at RT. Cells were dyed for 10 min with a crystal violet staining solution. After distaining, colonies were photographed and counted. Colony quantification was performed manually and by using ImageJ software. At least two biological replicates with three technical replicates each were performed for every cell line.

Cell cycle analysis

Cells were grown in complete media and treated for 48 h with doxycycline 1 µg/mL to determine LKB1-isoform expression. Time point treatments were performed in triplicate. Then, the medium and cells were collected, and after centrifugation, the cells were fixed and stained with the Cell Cycle Analysis Guava reagent (Guava Technologies, Hayward, CA, USA). Samples were analyzed with the Guava PCA cytometer (Guava Technologies Hayward, CA, USA).

Metabolic profiling

Mitochondrial function and glycolytic function were assessed using Seahorse technology (Seahorse XF Cell Mito stress kit, Agilent Technologies, Wilmington, DE, USA). Briefly A459 cells harboring the different STK11 isoform constructs were cultured on Seahorse XF-24 plates at a density of 75,000 cells per well. Cells were grown in the presence of doxycycline for 4 days before experiment. On the day of metabolic flux analysis, cells were changed to unbuffered DMEM (DMEM base medium supplemented with 10 mM glucose, 1 mM sodium pyruvate, 2 mM Glutamine, pH 7.4) and incubated at 37 °C in a non-CO2 incubator for 1 h. All medium and injection reagents were adjusted to pH 7.4 on the day of assay. Four baseline measurements of OCR and ECAR were taken before sequential injection of mitochondrial inhibitors. Three readings were taken after each addition of mitochondrial inhibitor before injection of the subsequent inhibitors. The mitochondrial inhibitors used were oligomycin (1 µM), FCCP (0.5 µM), and rotenone (0.5 µM). OCR and ECAR were automatically calculated and recorded by the Seahorse XF-24 software. After the assays, plates were saved and protein readings were measured for each well to confirm equal cell numbers per well. The percentage of change compared with the basal rates was calculated as the value of change divided by the average value of baseline readings.

Motility assay

Cells were seeded at low confluence in 24-well plates in duplicate. After 24 h, cells were induced with doxycycline and placed in the IncuCyte Live Cell Analysis Platform (Essen, Ann Arbor, MI, USA). Pictures were captured at 30 min intervals from five separate 950 × 760 µm2 regions per well using a ×20 objective for 5 days. Motility rates were measured for 30 to 50 GFP-positive cells individually with ImageJ software and graphed.

Kinase assay

A specific LKB1 kinase assay was performed as previously described in Lizcano et al.15. Briefly, different combinations of His-tagged LKB1 isoforms and FLAG-tagged STRADα were expressed in 293 cells and the complexes purified on cobalt binding resin. Protein complexes were washed twice with 1 ml of lysis buffer (50 mM Tris/HCl pH 7.5, 1 mM EGTA, 1 mM EDTA, 1% (w/v), Triton-X 100, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose, 0.1% (v/v) 2-mercaptoethanol and ‘complete’ proteinase inhibitor cocktail (one tablet/50 ml) containing 0.5 M NaCl, and twice with 1 ml of Buffer A (50 mM Tris/HCl pH 7.5, 0.1 mM EGTA, and 0.1% (v/v) 2-mercaptoethanol) and eluted from column with elution buffer containing 50 mM sodium phosphate, 300 mM NaCl and 150 mM imidazole. Phosphotransferase activity towards the NUAKtide peptide (SNLYHQGKFLQTFCGSPLYRRR residues 241–260 of human NUAK2 with three additional Arg residues added to the C-terminal to enable binding to P81 paper), was then measured in a total assay volume of 50 μl consisting of 50 mM Tris/HCl pH 7.5, 0.1 mM

EGTA, 0.1% (by vol) 2-mercaptoethanol, 10 mM magnesium acetate, 0.1 mM [γ32P]ATP (~200 cpm/pmol), and 200 μM NUAKtide peptide. The assays were carried out at 30 °C with continuous shaking, to keep the immunoprecipitates in suspension, and were terminated after 10 min by applying 40 μl of the reaction mixture onto p81 membranes. The p81 membranes were washed in phosphoric acid, and the incorporated radioactivity was measured by scintillation counting.

Secretome proteomics and statistical analysis

The secretomes were prepared as previously described46. In brief, 4 × 106 cells were seeded in 150 cc tissue culture plates and allowed to grow for 48 in the presence or absence of doxycycline 1 µg/ml. After that, media was aspirated and cells were washed twice with PBS and then three times with serum-free media. Then, cells at 60–70% confluency were maintained in serum-free media for 24 h before the collection of the conditioned media (secretome). Secretomes were spun down, and filtered through a 0.22-μm pore filter. Then, secretomes were concentrated using a 10,000 MWCO Millipore Amicon Ultra filter (Millipore) until a final volume of 50 μL was reached. The protein concentration was determined using the Pierce BCA Protein Assay kit (Thermo Scientific). All samples were digested with trypsin in-solution prior to analysis by liquid chromatography−mass spectrometry (LC − MS) as previously described (1). Tryptic digests were analyzed by shotgun proteomics using an LTQ Velos-Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). The RAW files of each MS run were processed using Proteome Discoverer (Thermo Fisher Scientific), and MS/MS spectra were searched against the human database of Swiss-Prot using the MASCOT (Matrix Science, London, U.K) algorithm. The results files generated from MASCOT (.DAT files) were then loaded into Scaffold (Proteome Software, Portland, OR), resulting in a nonredundant list of identified proteins per sample achieving a protein false discovery rate (FDR) under 1.0%, as estimated by a search against a decoy database.

Relative spectral counting-based protein quantification analysis was performed on the different samples analyzed using Scaffold. Files containing all spectral counts for each sample and its replicates were generated and then exported to R software for normalization and statistical analysis47. All statistical computations were done using the open-source statistical package R. The data were assembled in a matrix of spectral counts, where columns represent the biological conditions and rows represent the identified proteins. An unsupervised exploratory data analysis (EDA) by means of principal components analysis and hierarchical clustering of the samples on the SpC matrix was first performed. Then, the GLM model based on the Poisson distribution was used as a significance test47. Finally, the Benjamini-Hochberg multiple test correction was used to adjust the p-values with control on the false discovery rate (FDR).

Animal study

Animal experiments were conducted and designed according to protocols approved by the Institutional Animal Care and Use Committee of Vall d´Hebron Institute of Research. Nude mice (athymic nu/nu, female 4-6weeks old, Harlan Laboratories) were used for xenograft studies.

In vivo tumor growth experiments

For xenograft animal models, 5 × 105 luciferase-expressing A549-LKB1 cells were subcutaneously implanted into 8-week-old female athymic nude-Foxn1nu mice (n = 5 per group) (Envigo, Indianapolis, IN, USA). Tumors were measured with a digital Vernier caliper, and the mice were weighed twice a week. Tumor volume was calculated as D × d2/2, where D was the major diameter and d was the minor diameter. Bioluminescence imaging (IVIS Spectrum In Vivo Imaging System), PerkinElmer Life Science, Waltham, MA, USA) was performed at the end of the experiment for metastasis screening, both in vivo and ex vivo (viscera). Once the treatment was started, the animal weights and tumor sizes were monitored every 2 days. Representative tumor pictures were taken and tumor samples were obtained for further analysis.

Immunohistochemistry and immunofluorescence

Formalin-fixed paraffin-embedded tumor samples were subjected to immunocytochemistry according to the manufacturer’s antibody protocol. Samples were developed either by using either secondary antibodies linked to horseradish peroxidase or secondary antibodies linked to a fluorophore. Immunostaining was performed on 4 µm sections from formalin-fixed paraffin-embedded tissues. Staining was performed either manually or on the automated immunostainer Beckmarck XT (Ventana Medical Systems, Roche, Tucson, AZ, USA). Antibodies were visualized by the UltraViewTM Universal DAB Detection Kit (Ventana Medical Systems). For samples processed manualy antigen retrieval was performed using target retrieval solution pH 6.0 (Dako,Agilent, Santa Clara CA, USA)) Samples were scanned (panoramic slide digital scanner) and evaluated by two independent pathologists (using 3DHistech software).

Statistics and reproducibility

Statistical tests used are reported in the figure legends. In summary, statistical analyses were performed in GraphPad Prism 6 (GraphPad Software Inc.) using a two-tailed Student’s t test to compare differences between two groups or one-way ANOVA with multiple-comparisons tests to compare 3 or more groups. Statistical tests used for the analyses of transcriptomes (microarrays and gene set enrichment analysis) were performed in Partek Genomic Suite software (Partek Inc.). Number of biological replicates was ≥3 and is indicated in the figure legends.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.