Spatial transcriptome and single-cell sequencing reveal the role of nucleotide metabolism in breast cancer progression and tumor microenvironment.

Background: The complexities of nucleotide metabolism in breast cancer (BC) cells are not yet fully understood. A deeper exploration of the various tumor subpopulations and their interactions with the tumor microenvironment (TME) could provide important insights into these clinically relevant signaling pathways. Methods: We integrated five distinct single-cell enrichment scoring methodologies to conduct a comprehensive enrichment analysis of BC cell populations. The analytical findings underwent subsequent validation using an independent single-cell cohort. Tumor cell clusters were categorized based on their average enrichment scores. Functional analyses were carried out using several tools, including CellChat, Monocle, CopyKAT, SCENIC, and CytoTRACE. The RCTD method was then employed to map the single-cell clusters onto spatial transcriptomics data, facilitating the evaluation of cellular dependencies and pathway activities to differentiate tumor cell subtypes. A prognostic framework was subsequently established using large-scale transcriptomic datasets, enabling prediction of immunotherapy responsiveness. Experimental validation further confirmed expression patterns of pivotal genes implicated in therapeutic outcomes. Results: Tumor cells exhibit significantly upregulated nucleotide metabolic activity, enabling their classification into two distinct subgroups: NUhighepi and NUlowepi. Cells within the NUhighepi subgroup demonstrate pronounced malignant phenotypes. Intercellular communication analysis performed with the stLearn platform revealed robust interactions between NUhighepi cells and fibroblasts. Supporting this finding, spatial transcriptomic analysis via MISTy revealed a distinct dependency of NUhighepi on fibroblasts. A robust prognostic model, developed using various machine learning algorithms, was able to predict survival outcomes and responses to immunotherapy. Furthermore, targeted drugs were identified for both the high and low scoring groups. Experimental investigations confirmed the expression of core genes in different breast cancer cells. Discussion: In conclusion, we developed a nucleotide metabolism-derived prognostic signature for BC, with DCTPP1 highlighted as a promising biomarker and therapeutic target. These findings provide a valuable framework for early clinical intervention and show promising potential for predicting responses to immunotherapy in BC patients.