Integrated spatial omics of metabolic reprogramming and the tumor microenvironment in pancreatic cancer

Abstract

Metabolic reprogramming is a defining feature of pancreatic cancer, influencing tumor progression and the tumor microenvironment. By integrating single-cell transcriptomics, spatial transcriptomics, and spatial metabolomics, this study visualized the spatial co-localization of metabolites and gene expression within tumor samples, uncovering metabolic heterogeneity and intercellular interactions. Spatial transcriptomics identified distinct pathological regions, which were further characterized using single-cell transcriptomic data and pathologist annotations. Pseudotime trajectory analysis revealed metabolic shifts along the malignant progression, while single-cell Metabolism (scMetabolism) delineated metabolic differences between pathological regions, classifying them as hypermetabolic or hypometabolic. Notably, aberrant cell communication between cancer cells, macrophages, and fibroblasts was observed, with key receptor-ligand pairs significantly co-expressed in malignant regions and correlated with poor prognosis. Spatial metabolomics imaging identified signature metabolites, highlighting metabolic alterations in amino acid metabolism, polyamine metabolism, fatty acid synthesis, and phospholipid metabolism. This integrated analysis provides critical insights into pancreatic cancer metabolism, offering potential avenues for targeted therapeutic interventions.

Keywords: Cancer; Metabolomics; Microenvironment; Transcriptomics.

Graphical abstract

Figure 1

Flow chart and spatial transcriptomic landscape of pancreatic cancer (A) Schematic diagram exhibiting the detection and analysis of ST and SM in pancreatic cancer. (B) Unsupervised clustering analysis UMAP divided all spots from six samples into 16 clusters. (C) H&E staining images (left), UMAP plots (middle), and ST feature plots (right) of 16 clusters in six samples. (D) The tissue of six samples was divided into four regions based on the histopathological features, including malignant, normal, immune, and stroma regions. H&E staining images (left), ST images of spots with tissue regions annotated by different colors (middle), and the proportion of spots and corresponding clusters in each tissue region (right) were presented, respectively. (E) ST feature plots showing the expression of representative marker genes in each spot of different tissue regions.

Figure 2

Pseudotime analysis, scMetabolism, and cell communication reveal metabolic reprogramming and interactions (A) Trajectory reconstruction of PDAC consisted of three branches: pre-branch (before bifurcation), T1 branch (bottom), and T2 branch (top). Each point corresponds to a spot. (B) BEAM heatmap plot displaying the expression patterns of pseudotime-specific genes and the corresponding GO pathway terms (hierarchically clustered into three profiles) in the malignant trajectory. (C) Heatmap shows the metabolic score representing glycolysis, pentose phosphate pathway, oxidative phosphorylation, and glutathione metabolism for each region. All spots were categorized into two clusters based on their metabolic activity: hypermetabolic, and hypometabolic. (D) Deconvolution results indicate the proportion of representative cell type of each spot in different types of metabolic regions. (E) Representative spatial transcriptomic tissue slides show the spatially projected metabolism clusters. (F) Heatmap showing the counts and strength of cell-cell communication between different cell types in the two metabolic regions.

Figure 3

Interactions between cell types in different metabolic regions and potential clinical significance (A) Dot plots showing ligand–receptor pairs that are significantly expressed in hypermetabolic regions. (B) ST feature plots exhibiting the expression level and spatial distribution of representative ligand–receptor pairs. The correlation between transcriptomic level and prognosis were also presented.

Figure 4

Spatial metabolomic atlas of pancreatic cancer (A and B) MSI images showing the abundance and distribution of representative metabolites and m/z information in each histopathological region in negative pattern (A) and positive pattern (B). (C) MSI images exhibiting the abundance of representative differential metabolites in different regions.

Figure 5

Visualization of metabolic reprogramming of amino acid metabolism in pancreatic cancer (A) The interconnected pathways of amino acids (AAs) metabolism. Glutamine and glutamate have a central role and can each be used for the synthesis of other AAs. Glutamate can be utilized to generate alanine, aspartate, serine, proline and also histidine. Aspartate is further utilized to generate asparagine. Serine makes glycine and donates methyl groups for one-carbon metabolism. Serine can also generate cysteine via the trans-sulfuration pathway. Violin plot show expression levels of key genes in amino acid metabolism. (B) Schematic maps of polyamine metabolism, including urea cycle (blue arrows), polyamine synthetic metabolism (yellow arrows), ornithine salvage synthesis (red arrows), polyamine catabolic metabolism (green arrows), and methionine salvage metabolism (brown arrows). (C–F) The spatial distribution feature of metabolic products and enzymes in the urea cycle (C), polyamine synthetic metabolism (D), polyamine catabolic metabolism (E) and ornithine salvage synthesis (F). Symbols: ns. denotes non-statistically significant, ∗ indicates a p-value <0.05, ∗∗ represents a p-value <0.01, and ∗∗∗ signifies a p-value <0.001. Data are represented as mean ± SEM.

Figure 6

Visualization of metabolic reprogramming of lipid metabolism in pancreatic cancer (A) Schematic maps of lipid metabolism, including fatty acid de novo synthesis (red arrows), Kennedy pathway (blue arrows), CDP-DG pathway (yellow arrows), and Lands cycle (green arrows). (B) MS images of representative lipids in pancreatic cancer tissues (intensity in color scale is relative value). (C) Expression levels of representative lipids in different regions of pancreatic cancer tissue. (D) Spatial expression images of key genes in lipid metabolism (intensity in color scale is log2 transformed). Symbols: ns. denotes non-statistically significant, ∗ indicates a p-value <0.05, ∗∗ represents a p-value <0.01, and ∗∗∗ signifies a p-value <0.001. Data are represented as mean ± SEM.