Researchers who reanalyzed two publicly available single-cell RNA-sequencing datasets have mapped the cellular landscape of breast cancer that has spread to bone, singling out a macrophage population defined by expression of the gene SPP1 and predicting, through computational modeling, that these cells exchange signals with specific T-cell subsets. The work, published in Cancer Reports, is a purely computational characterization rather than a laboratory experiment, and its authors are careful to frame the predicted macrophage-T-cell crosstalk as a hypothesis that now requires experimental validation.
After strict quality control and removal of suspected doublets, the integrated analysis retained 141,056 cells spanning three tissue compartments: normal breast, primary breast cancer, and the bone-metastatic niche. Unsupervised clustering resolved eleven major cell populations, including T cells, fibroblasts, epithelial cells, myeloid cells, perivascular-like cells, endothelial cells, mast cells, B cells, plasma cells, cycling cells, and a cluster interpreted as a heat-shock-induced stress state marked by HSPA6.
Because the metastatic and primary or normal specimens came from different datasets and were not patient-matched, the authors stress that the comparison is cross-sectional rather than longitudinal, and that observed differences reflect a combination of disease stage, tissue environment, and inter-individual variation. Even so, the magnitude of some differences was striking. The median proportion of myeloid cells rose from 0.58 percent in normal breast tissue to 29.6 percent in bone metastases, while the median T-cell proportion rose from 2.1 percent to 17.6 percent. The authors report effect sizes with confidence intervals for these comparisons and explicitly caution that dataset and tissue-environment effects may contribute alongside any biology of metastasis.
Within the myeloid compartment, clustering identified nine subpopulations, including two monocyte-like clusters, conventional dendritic cells, a LAMP3-positive dendritic cell state, cycling myeloid cells, and three macrophage subsets named Mac-SPP1, Mac-CXCL3, and Mac-APOE. The Mac-SPP1 population stood out because 2,132 of its 2,135 cells, or 99.9 percent, came from the bone-metastatic dataset. By contrast, Mac-CXCL3 and most Mac-APOE cells derived from the primary and normal tissue dataset. The authors interpret Mac-SPP1 as a dataset-associated cluster rather than a proven metastasis-specific cell state, since biological condition is fully confounded with source dataset in this design.
Functional scoring gave the Mac-SPP1 cells an M2-polarized signature, a gene-expression pattern conventionally associated with immunosuppressive, pro-tumor macrophage activity, and the overall macrophage population in metastases also skewed toward the M2 state. Relative to other macrophage subsets, Mac-SPP1 cells showed higher expression of TGFB1, an immunosuppressive factor; VEGFA, a pro-angiogenic factor; and the markers SPP1, MSR1, and TREM2. Expression of IL10, MMP9, and MMP12 was not elevated. Pathway enrichment analysis of differentially expressed genes in macrophages from primary and metastatic groups pointed to osteoclast differentiation among the enriched pathways, a finding consistent with the bone environment in which the metastatic cells reside.
To probe regulatory mechanisms, the team used pySCENIC to infer transcription factor activity and found high activity of HOXA9, MAF, FOS, and JUND in Mac-SPP1 cells, factors linked in prior literature to M2 macrophage polarization. Metabolic gene-set enrichment scoring, which measures the expression of metabolic pathway genes rather than actual metabolic flux, showed Mac-SPP1 cells with the highest enrichment among macrophage subtypes, and the metastatic group with higher enrichment than the normal and primary groups, again with the dataset confound acknowledged. Notably, when the authors tested whether a five-gene Mac-SPP1 signature predicted overall survival in the TCGA breast cancer cohort of 1,082 primary tumors, the answer was no: the association was not significant as a continuous predictor, after adjustment for age, stage, and molecular subtype, or in a median-split Kaplan-Meier analysis, so any prognostic value of this signature remains unestablished.
On the T-cell side, clustering resolved nine subpopulations, including CD8 subsets marked by IL7R, GZMK, GNLY, and AFF3, and CD4 subsets marked by LEF1, IKZF2, and S100A9. The CD4-S100A9 population, present almost exclusively in the bone-metastatic dataset, was confirmed as genuine T cells through robust co-expression of the canonical markers CD2, CD3D, and CD3E, since S100A9 is conventionally a myeloid marker whose expression can be induced in T cells under stress. These cells displayed high cytotoxicity and exhaustion scores simultaneously, which the authors interpret as an activated yet exhausted, stress-responsive CD4 state within the metastatic niche. Pseudotime analysis positioned CD4-S100A9 cells at one end of an inferred trajectory and CD4-LEF1 cells at the other, with CD4-IKZF2 cells in an intermediate position, though the authors emphasize that pseudotime infers transcriptional ordering rather than demonstrated lineage ancestry.
Among CD8 T cells, the CD8-AFF3, CD8-GNLY, and CD8-GZMK subsets showed high cytotoxicity and exhaustion scores, and cytotoxic and exhausted CD8 cells tended to be more abundant in the metastatic group. The inferred CD8 trajectory placed central-memory-like CD8-IL7R cells at its root, with CD8-GZMK cells in an intermediate position retaining an effector program while beginning to acquire exhaustion features, and the trajectory resolving into two terminal states: a cytotoxic CD8-GNLY population and an exhausted CD8-AFF3 population. Transcription factor analysis highlighted ETV1 activity in CD8-GZMK cells, and pathway enrichment of differentially expressed genes between metastatic and primary groups implicated ribosome, T-cell receptor signaling, and PD-1 checkpoint pathways.
Osteoclasts and osteoblasts, which could not be resolved as separate clusters in the integrated data, were excluded from the communication analysis.
The authors situate their findings against prior reports that SPP1-positive macrophages interact with exhausted CD8 T cells in liver and gastric cancers and promote progression in colorectal and head and neck tumors, while noting that macrophage-T-cell dynamics in breast cancer bone metastasis had not previously been reported. They also lay out the study’s limitations candidly: reliance on public datasets with few bone-metastasis samples, complete confounding of biological condition with source dataset, contrast of different tissue environments without spatial information, incompletely reported patient characteristics such as age, menopausal status, subtype, and treatment history, and the associative rather than causal nature of computational reanalysis. Confirming whether the predicted Mac-SPP1 to CD4-S100A9 and CD8-GZMK interactions represent a therapeutically tractable axis will require in vitro coculture and in vivo models, along with larger, matched cohorts and spatial profiling to corroborate and extend these observations.
Subject of Research: Single-cell characterization of SPP1-expressing macrophages and predicted T-cell interactions in breast cancer bone metastasis
Article Title: Single‐Cell Characterization of SPP1‐Expressing Macrophages and Their Predicted Interactions With T Cells in the Breast Cancer Bone Metastatic Microenvironment
Article References: Xuan, Z., Wu, Z., Zhong, Y., Jiang, J., Zhang, Y., Cheng, L., Zhang, J., & Wang, K. (2026). Single‐Cell Characterization of SPP1 ‐Expressing Macrophages and Their Predicted Interactions With T Cells in the Breast Cancer Bone Metastatic Microenvironment. Cancer Reports, 9(11), Article e70732. https://doi.org/10.1002/cnr2.70732
Image Credits: AI Generated
DOI: 10.1002/cnr2.70732
Keywords: breast cancer, bone metastasis, macrophages, SPP1, T cells, single-cell RNA sequencing, tumor microenvironment, CellPhoneDB, M2 polarization, computational prediction, Single, Cell
News Source: Nathaniel Bowman. (November 6, 2026). SPP1-Expressing Macrophages Predicted to Talk to T Cells in Breast Cancer Bone Metastases. Scienmag.
