Mapping metabolic reprogramming dynamics across pancreatic neuroendocrine tumor cell differentiation at single-cell transcriptomic resolution.

The metabolic switch between oxidative phosphorylation (OXPHOS) and aerobic glycolysis is a fundamental feature of tumor biology. Its dynamic regulation during pancreatic neuroendocrine tumor (pNET) cell differentiation remains poorly characterized, especially at single-cell resolution. We analyzed publicly available single-cell RNA sequencing (scRNA-seq) data from the Gene Expression Omnibus database to profile pNET cells at single-cell resolution. Sequencing data were processed through a standard analytical workflow including pseudotime ordering, low-dimensional visualization, co-expression network construction, and metabolic state estimation. To validate key transcriptional patterns, we measured the mRNA and protein levels of selected genes in BON-1 and QGP-1 cells using quantitative real-time PCR (qRT-PCR) and enzyme-linked immunosorbent assay (ELISA). Through single-cell analysis, we resolved the cell population into ten transcriptionally distinct clusters, annotated into major cell types including PT-like tumor cells, TAL-like cells, DCT-like cells, CNT/CD-like cells, endothelial cells, fibroblasts, immune cells, and podocyte-like cells, with PT-like tumor cells constituting the dominant fraction (67%). Canonical marker genes SLC34A1, SLC5A2, LRP2, CUBN, ALDOB, and GATM confirmed the identity of the PT-like tumor cell population on UMAP embedding. Metabolic state annotation identified three states-OXPHOS-high, Mixed, and Glycolysis-high-distributed differentially across PT sub-clusters. Differential expression analysis between glycolysis-high and OXPHOS-high states identified 247 significant genes: OXPHOS genes including ATP5F1B, ATP5F1A, COX4I1, NDUFA1, and NDUFS1 were strongly upregulated in OXPHOS-high cells, while glycolytic enzymes including HK1, HK2, ENO1, PKM, and ALDOA were enriched in glycolysis-high cells. A gradient boosting classifier distinguished metabolic states with ROC AUC = 0.673 and PR AUC = 0.689, with MALAT1 emerging as the most discriminative and conserved feature. Intercellular communication analysis identified prominent TGFB1→TGFBR1, SPP1→CD44, VEGFA→KDR, CXCL12→CXCR4, EGF→EGFR, and FGF2→FGFR1 signaling axes. qRT-PCR confirmed coordinated metabolic gene changes in BON-1 and QGP-1 cells (r = 0.91, P < 0.01). ELISA confirmed corresponding protein-level alterations. The metabolic program of pNET cells shifts from OXPHOS-dominant to glycolysis-dominant states along the differentiation trajectory. MALAT1, ATP5F1B, PKM, and NDUFS1 are positioned as key regulatory nodes. These findings refine current understanding of metabolic reprogramming during pNET differentiation and suggest targeting the OXPHOS-to-glycolysis transition as a potential therapeutic strategy in pancreatic neuroendocrine tumors.