Spatial epigenometabolomic niches of lactate‑H3K27ac co‑activation predict and drive resistance to PD‑1 blockade in glioblastoma

Hypothesis

Spatial co‑localization of elevated lactate and enhancer‑associated H3K27ac marks defines a immunosuppressive niche that drives resistance to PD‑1 blockade in glioblastoma.

Rationale

Multi‑omics integration outperforms genomics‑only approaches by uncovering functional resistance mechanisms invisible to DNA sequencing [1][2]. Spatial transcriptomics and metabolomics have already shown that locoregional metabolites such as L‑glutamine and DL‑dopamine regulate tumor growth in distinct cell states [4]. Epigenomic layers add further resolution: methylation profiles at distal regulatory elements capture heterogeneity and predict drug response [5][6], and histone demethylases KDM5A/B are implicated in resistance [1]. Yet no study has directly linked a specific metabolite‑epigenetic axis to the formation of resistance niches in the tumor microenvironment. We propose that lactate, a hallmark of glycolytic tumors, inhibits histone deacetylases (HDACs) locally, leading to increased H3K27ac at enhancers of immunosuppressive cytokines (e.g., CCL2, VEGFA) and checkpoint ligands. This epigenometabolomic reprogramming creates a niche that recruits T‑regs and macrophages, blunting PD‑1‑mediated T‑cell activation.

Testable Predictions

1. In glioblastoma samples from patients who progress on PD‑1 blockade, regions of high lactate signal (detected by imaging mass spectrometry) will spatially overlap with peaks of H3K27ac (detected by CUT&Tag‑seq) more frequently than in responders or untreated tumors.
2. Disrupting lactate transport (using MCT1 inhibitor AZD3965) or blocking HDAC activity (with a selective HDAC3 inhibitor) will reduce H3K27ac at immunosuppressive enhancers and decrease lactate‑driven immune suppression in ex vivo slice cultures.
3. Combining lactate transport inhibition with PD‑1 blockade will improve tumor infiltration by CD8⁺ T cells and prolong survival in orthotopic patient‑derived xenograft models compared with either monotherapy.

Experimental Design

Cohort: Collect fresh‑frozen glioblastoma resection specimens from three groups: (a) PD‑1‑treated progressors, (b) PD‑1‑treated responders, (c) treatment‑naïve controls (n ≥ 10 per group). Spatial metabolomics: Perform imaging mass spectrometry (IMS) for lactate and related metabolites (e.g., pyruvate) at 10 µm resolution. Spatial epigenomics: Apply CUT&Tag‑seq for H3K27ac on adjacent sections, followed by spatial barcoding to retain location information (similar to spatial‑CUT&Tag). Analysis: Register IMS and epigenomic maps, compute colocalization scores (Manders’ coefficient) for lactate/H3K27ac voxels. Compare scores across groups using Mann‑Whitney U test; predict higher scores in progressors (p < 0.01). Functional validation: Treat patient‑derived glioblastoma organoids with AZD3965 (1 µM) or HDAC3 inhibitor (RGFP966, 0.5 µM) for 24 h, then measure H3K27ac at CCL2/VEGFA enhancers by qPCR‑based CUT&Tag and lactate secretion by enzymatic assay. Assess immune suppression by co‑culturing with autologous microglia and measuring CD8⁺ T‑cell IFN‑γ production. In vivo test: Orthotopically implant luciferase‑labeled patient‑derived glioblastoma cells into immunocompetent mice (with humanized microglia). Randomize to four arms: control, anti‑PD‑1, AZD3965, anti‑PD‑1 + AZD3965 (n = 12 per arm). Monitor tumor growth by bioluminescence, survival, and flow cytometry of tumor‑infiltrating lymphocytes.

Potential Outcomes

If lactate‑H3K27ac colocalization predicts resistance and combined targeting restores response, the hypothesis is supported. Lack of spatial correlation or failure of the combination to improve outcomes would falsify the claim, indicating that lactate‑driven HDAC inhibition is not a dominant resistance mechanism in this context.

This framework directly extends the cited multi‑omics successes [1][2][3][4][5][6] by adding a mechanistic, spatially resolved epigenometabolomic link that can be therapeutically intersected.