Single-cell and machine learning approaches reveal METTL14-mediated autophagy via PI3K/AKT signaling in invasive PitNET

OBJECTIVE: This study investigates the role of m6A regulators in invasive pituitary neuroendocrine tumors (PitNETs) by defining m6A-related molecular subtypes, constructing a nomogram, and elucidating the mechanistic role of METTL14 in PitNET progression. METHODS: GEO datasets were analyzed for m6A regulatory gene expression. Machine learning approaches were used to develop a nomogram. m6A molecular and gene subtypes were identified, and scRNA-seq from tumors characterized intratumoral subpopulations. Functional validation was performed using METTL14 overexpression and knockdown in PitNET cells, followed by proliferation, invasion, RT-qPCR, Western blotting, m6A-RIP, RIP, RNA stability, and luciferase reporter assays. RESULTS: Seven key m6A regulators were selected for the nomogram. Two m6A subtypes were identified, with Cluster B showing lower immune infiltration and higher m6A scores. Further classification revealed two gene-based subtypes with consistent patterns. scRNA-seq defined four PitNET clusters, including a proliferative TPC population with the highest invasion scores, strongly correlated with METTL14. Functional experiments confirmed that METTL14 promotes proliferation, invasion, and autophagy via PI3K/AKT activation. Mechanistically, IGF2 was identified as a novel downstream effector of METTL14, as METTL14 enhanced IGF2 expression through m6A modification, thereby activating PI3K/AKT signaling. CONCLUSION: This study provides a comprehensive characterization of m6A methylation in invasive PitNET, integrating bulk and single-cell transcriptomics. The nomogram offers clinical potential, while the discovery of TPCs and the identification of the METTL14–IGF2–PI3K/AKT regulatory axis highlight novel mechanistic insights and therapeutic targets for personalized invasive PitNET management.