Single-cell and spatial transcriptomic profiling of primary versus recurrent glioblastoma identifies PLOD2 as a core orchestrator of hypoxic niche remodeling and mesenchymal transition

Technical Summary: PLOD2 as a Core Orchestrator of Hypoxic Niche Remodeling and Mesenchymal Transition in Recurrent Glioblastoma

Problem Statement
Glioblastoma (GBM) recurrence is characterized by acquired treatment resistance and extensive remodeling of the tumor microenvironment (TME). Despite advances in bulk and single-cell RNA sequencing, identifying conserved drivers of recurrence remains elusive due to significant inter-cohort heterogeneity, limited sample sizes in individual studies, and technical discrepancies across analytical platforms. Current knowledge lacks robust, reproducible consensus on the core mechanisms that universally orchestrate tumor relapse and microenvironmental reprogramming, particularly regarding the transition to a mesenchymal-like (MES-like) phenotype and the associated extracellular matrix (ECM) remodeling.

Methodology
The study employed a rigorous, multi-scale omics screening strategy to decode recurrence mechanisms:

• Data Integration: The authors integrated two high-quality single-cell RNA-seq (scRNA-seq) and two single-nucleus RNA-seq (snRNA-seq) datasets, mapped to the GRIT-atlas reference, to identify recurrence-associated subpopulations.
• Cross-Cohort Validation: A robust transcriptional signature of malignant cells was defined using three large-scale single-cell cohorts (two public, one in-house) and cross-referenced with an 890-sample bulk transcriptomic cohort. Weighted Gene Co-expression Network Analysis (WGCNA) was used to identify modules associated with recurrence and drug resistance.
• Spatial Transcriptomics: High-resolution 6000-plex CosMx Spatial Molecular Imager (SMI) profiling was performed on primary GBM samples to map the spatial distribution of target genes within specific niches (e.g., perivascular, hypoxic peri-necrotic).
• Functional Assays: In vitro experiments included stable PLOD2 knockdown in U251 MG cells using lentiviral shRNA, followed by Western blotting and high-throughput ECM-receptor interaction antibody arrays (RayBio) to profile secretome changes.
• Pan-Cancer Analysis: Transcriptomic profiles from ten solid tumor cohorts were analyzed to evaluate hypoxia scores, tumor-stromal communication (via CellChat), and the correlation between hypoxia and PLOD2 expression.
• Bibliometrics: A comprehensive literature search and keyword co-occurrence analysis were conducted to contextualize PLOD2 research trends.

Key Contributions and Results

1. Identification of Recurrence-Associated Subpopulations: The study identified 11 distinct cellular subpopulations significantly enriched in recurrent GBM compared to primary tumors. These included proliferating cells, neuronal-like states, specific vascular endothelial populations, matrix-remodeling cancer-associated fibroblasts (myCAFs), and specific immune subsets (e.g., NKG2Ahi NK cells, T follicular helper cells, APOE+ tumor-associated macrophages).
2. Discovery of PLOD2 as a Hub Gene: Through intersection analysis of recurrence-associated modules and high-risk prognostic markers, Procollagen-Lysine, 2-Oxoglutarate 5-Dioxygenase 2 (PLOD2) was pinpointed as the paramount hub gene.
   • Clinical Correlation: Elevated PLOD2 expression correlated with higher histological grade and increased mortality risk (Hazard Ratio = 1.25).
   • Spatial Localization: PLOD2 was predominantly expressed by malignant cells and showed pronounced enrichment within the Pseudopalisading Necrosis (PAN) zone, a specialized hypoxic niche. Its expression correlated positively with tumor stemness and CD8+ T-cell exhaustion signatures.
3. Mechanism of Action (PLOD2/LAMA4 Axis):
   • ECM Remodeling: PLOD2 knockdown significantly reduced the secretion of LAMA4 (Laminin subunit alpha-4), a key component of the vascular basement membrane.
   • Spatial Co-localization: Spatial profiling confirmed the single-cell in situ co-localization of PLOD2 and LAMA4 transcripts within vascular-rich (S3) and myeloid-enriched (S1) niches.
   • Hypoxia-Driven Rewiring: Pan-cancer analyses revealed that hypoxic cancer cells exhibit significantly upregulated PLOD2 and enhanced signaling communication with cancer-associated fibroblasts (CAFs). This suggests hypoxia remodels tumor-stromal interactions via PLOD2, reinforcing mesenchymal transition and physical barriers.
4. Pathological Loop: The authors propose a "vicious pathological feed-forward loop": Mesenchymal-like glioma cells utilize PLOD2 to upregulate LAMA4, accelerating aberrant microvascular proliferation (MVP). This dysfunctional vasculature leads to perfusion deficits and blood-tumor barrier failure, exacerbating ischemia and accelerating the transition of MVP zones into necrotic PAN zones, which further enriches the MES-like niche and immunosuppressive environment.

Significance and Claims
The paper claims to identify PLOD2 as a pivotal driver of GBM recurrence that modulates the recurrent TME via a potential PLOD2/LAMA4 axis. The study establishes a spatial-functional link between PLOD2 expression in the hypoxic PAN region and GBM relapse, an aspect previously underexplored.

The authors suggest that PLOD2 orchestrates ECM cross-linking and niche stiffness, creating a physical barrier that fosters immune exclusion and therapy resistance. By delineating the PLOD2/LAMA4 axis, the study offers a potential translational strategy to disrupt microenvironmental barriers and alleviate therapy resistance. The findings position PLOD2 not only as a prognostic biomarker for recurrence but also as a potential therapeutic target to dismantle the microenvironmental barriers that sustain GBM relapse.

The authors acknowledge limitations, noting that their findings rely on preclinical and computational models and that future work requires prospective, harmonized multi-center cohorts and pharmacological validation to confirm the clinical utility of targeting this pathway.