TL;DR
This study identifies a gene signature that distinguishes invasive from non-invasive early-stage lung adenocarcinoma (esLUAD). This signature, translated into an "Invasiveness Score" (IVS), predicts patient survival and invasiveness in cell lines. Crucially, the research pinpoints Aurora kinases as key regulators of this invasive phenotype, demonstrating that inhibiting them reduces tumor invasion and progression in both cell models and a mouse model, suggesting a promising therapeutic avenue.
What Problem Does This Paper Address?
Lung adenocarcinoma (LUAD) is the most common type of lung cancer, and while early detection has improved outcomes, a significant portion of early-stage patients still develop metastasis. This heterogeneity in invasiveness and clinical outcomes within LUAD presents a challenge for effective treatment. Existing classification methods often rely on histology, which can be subjective and may not fully capture the underlying molecular mechanisms driving invasion and progression. The authors aim to address this by identifying a molecular signature of invasiveness in early-stage LUAD and uncovering actionable therapeutic targets that can intercept tumor progression.
Methods
The researchers employed a multi-pronged approach:
- Gene Expression Profiling and Signature Development: They performed RNA sequencing on 53 early-stage LUAD samples to identify differentially expressed genes (DEGs) between invasive and non-invasive tumors. This led to the development of a "pro-invasive" and an "indolence" gene signature.
- Invasiveness Score (IVS) Calculation: Using the gene signatures, they developed a classifier (elastic net) to generate an IVS for individual tumors, quantifying their invasiveness.
- Network Analysis: They constructed a molecular causal network for esLUAD by integrating gene expression, copy number variation (CNV), and methylation data from TCGA. This network analysis helped identify key drivers and subnetworks associated with invasiveness.
- Drug Perturbation Analysis: The Connectivity Map (CMAP) database was used to identify small molecules that could perturb the identified invasive subnetworks, pointing towards potential therapeutic targets.
- In Vitro and In Vivo Models: They used lung cancer cell lines for migration and invasion assays, and a genetically engineered KrasG12D/Tgfbr2-/- mouse model of invasive LUAD for in vivo experiments.
- Immunohistochemistry (IHC): Protein expression of Aurora kinases was assessed in human tumor samples.
Datasets
- Primary Tumor Datasets:
- A cohort of 53 early-stage LUAD samples (used for initial signature development).
- Seven independent human LUAD cohorts (Shedden, TCGA, Okayama, Tang, Der, Rousseaux, Wilkerson) for validation of the invasiveness signature and IVS.
- A human tissue microarray (TMA) of 768 LUAD cases for IHC analysis of Aurora kinase protein expression.
- Cell Line Datasets:
- Gene expression data from 70 LUAD cell lines (CCLE dataset) for IVS calculation.
- Specific LUAD cell lines (H1373, SK-LU-1, H1792, A549, H2009, H1755, H1650, H1975, HCC78, H3255, Calu3, HCC1833, HCC2279) were used for in vitro experiments (migration, invasion, wound healing, proliferation, western blotting).
- Mouse Model Data:
- Gene expression data from a KrasTgfbr2-/- mouse model of invasive LUAD (GSE27717) for comparison with human signatures.
- A Kras(G12D)/Tgfbr2-/- genetically engineered mouse model was used for in vivo drug treatment experiments.
Key Insights and Ideas
- Invasiveness as a Molecularly Defined Phenotype: The study demonstrates that invasiveness in early-stage LUAD is not just a histological descriptor but a distinct molecular phenotype that can be captured by a gene signature. This signature is robust and generalizes across multiple independent patient cohorts.
- Aurora Kinases as Master Regulators: Network analysis strongly implicates Aurora kinases (specifically Aurora-A and Aurora-B) as central regulators of the invasive gene signature. This is a significant finding, as their role in early LUAD invasion was previously unclear.
- Redundant but Essential Roles: Genetic deletion experiments revealed that Aurora-A and Aurora-B have redundant functions; inhibiting either alone didn't significantly impact invasion, but simultaneous inhibition was required to achieve a substantial reduction. This highlights the importance of targeting the entire Aurora kinase family.
- Therapeutic Vulnerability: The identification of Aurora kinases as key drivers, coupled with drug perturbation analysis using CMAP, pointed towards Aurora kinase inhibitors as promising therapeutic agents. This provided a direct link from molecular discovery to potential clinical intervention.
- Beyond Proliferation: The invasiveness signature captures biological features beyond mere proliferation, as demonstrated by experiments where proliferation rates did not correlate with invasiveness in cell lines.
Key Results
- A gene signature was identified that accurately distinguishes invasive from non-invasive esLUAD tumors.
- The derived Invasiveness Score (IVS) significantly associated with survival across seven independent LUAD patient cohorts, with "invasive" tumors showing poorer outcomes.
- The IVS also correlated with invasiveness phenotypes in lung cancer cell lines, with high IVS cell lines exhibiting increased migration and invasion.
- Network analysis identified TPX2 and AURKB (Aurora kinase B) as key regulators within a pro-invasive subnetwork.
- Inhibition of Aurora kinases (using CRISPR/Cas9 or small molecule inhibitors like AMG900 and PF-03814735) significantly reduced migration and invasion in LUAD cell lines.
- Aurora kinase inhibition suppressed key signaling pathways (AKT/mTOR) and EMT markers associated with invasiveness.
- In a Kras(G12D)/Tgfbr2-/- mouse model, AMG900 treatment significantly reduced tumor burden, invasive tumor growth, and improved survival, while also decreasing tumor-infiltrating neovessels and collagen deposition.
Limitations and Future Work
- Tumor Microenvironment Complexity: While the study highlights tumor-intrinsic features regulated by Aurora kinases, the authors acknowledge that tumor invasion is complex and involves interactions with the tumor microenvironment. Further single-cell genomic studies are suggested to elucidate these complex interactions.
- KRAS Mutation Independence: While the study showed Aurora kinase inhibition's effect on invasion was independent of KRAS mutation status in cell lines, the interplay between driver mutations (like KRAS, LKB1) and Aurora kinase activity warrants further investigation.
- Clinical Translation: The authors propose that the invasiveness signature and AMG900 could be tested in unresected subsolid lung nodules to identify lesions at risk of progression and potentially intercept tumor progression and metastasis. This suggests a clear path for future clinical trials.
- Combination Therapies: The study notes that the Akt/mTOR pathway is a downstream target of Aurora kinase inhibition, suggesting opportunities for combination strategies to address cells that might bypass Aurora kinase inhibition.
Summary updated