iCAFs secrete osteoprotegerin (OPG).

(A-D) Data are from female C57/BL6 mice injected with GFP-labelled EO771 cells. Tumors were collected 24 days after implantation. Stromal cells (GFP-CD45-) were sorted by FACS and prepared for 10X Genomics scRNA sequencing analysis. (A) Schematic diagram of the experiment and single-cell RNA sequencing. (B) Uniform manifold approximation and projection (UMAP) visualization of immune-depleted stromal cells from tumor-injected mice generated from 10X Genomics single-cell RNA sequencing (scRNA-seq) analysis. Distinct clusters of stromal cells are identified using the Leiden algorithm and are represented by different colors. UF: Universal fibroblasts, DC: Dividing cells, RF: Reticular fibroblasts, Postn+ CAFs: Cancer associated fibroblasts (CAFs) expressing high Postn, Acta2+ CAFs: CAFs expressing high Acta2.(C) Matrixplot showing high expression levels of literature-based CAF, iCAF, and myCAF biomarkers in clusters identified above as Postn+ and Acta2+ CAFs. Expression data were normalized with z-score transformation. Blue and red represent the low and high expression of a gene, respectively, relative to the median expression level. (D) UMAP plot highlighting the association between Postn+ CAFs and an elevated “SenMayo score,” calculated based on the average expression of a curated set of 118 genes known to identify senescent cells at the single-cell level. (E). Schematic diagram of the cell sorting experiment (F) Quantification of OPG level from the sorted cells in F (G) Quantification of OPG secreted from nontumor-derived fibroblast (NFs) and cancer associated fibroblast (CAFs). NFs and CAFs were isolated from one patient with a gastric tumor by FACS using CD45-CD90+CD326-CD31-. Data in G are representative of two independent experiments. Statistics were calculated using ordinary one-way analysis of variance test.

OPG inhibits cytotoxic function of T effector cells.

(A–B) Data are from single-cell RNA sequencing (scRNA-seq) of 12 human esophageal carcinoma tumor samples, derived from biopsies or resections, as reported by Strasser et al., 2023. (A) UMAP visualization of different clusters of stromal populations. Distinct clusters were identified using the Leiden algorithm and are shown in different colors. Five fibroblast subtypes were identified: universal stromal cells, PTGDS⁺ cancer-associated fibroblasts (CAFs), myofibroblasts, POSTN⁺ CAFs, myofibrotic CAFs (myCAFs), and immune cells. (B) UMAP feature plots showing gene expression of PDGFRA (top) and TNFRSF11B (bottom). (C–D) Data are from scRNA-seq of 26 primary breast tumors, including 11 ER⁺, 5 HER2⁺, and 10 triple-negative breast cancer (TNBC) samples, as described by Wu et al., 2021. (C) UMAP plot showing four major clusters: CAFs, perivascular cells, endothelial cells, and CE (Cycling Endothelial). (D) UMAP feature plots displaying expression of PDGFRA (top) and TNFRSF11B (bottom). (E–L) Data are from isolated CD8⁺ T cells derived from mouse splenocytes, stimulated with vehicle or 3 μg/mL CD3/CD28 for 24 hours. (E) Quantification of TRAIL expression in CD8⁺ T cells. (F) Expression levels of IFNγ in CD8⁺ T cells, normalized to total protein. (G) Expression of Ki67 in CD8⁺ T cells. (H) Expression of RANKL in CD8⁺ T cells. (I) Expression of CD69 in CD8⁺ T cells. (J) Cytotoxic activity of CD8⁺ T cells against L929 cells at a target-to-effector (T:E) ratio of 1:4, with increasing concentrations of recombinant OPG (rOPG).

Blocking OPG alters tumor microenvironment.

(A-C) Female p16luc/+ mice of 6 -8 weeks were used for EO771 injection. The mice were administered with 500 ng IgG or αOPG at 3, 10, 17, 28, and 35 days post-EO771 injection. (A) Schematic diagram of the experiment. (B) Tumor growth curve from IgG and αOPG-treated mice (n= 9 or 10 mice per group) (C) Tumor image (left)and weight (right) from IgG and αOPG-treated mice. (D) Tumor growth curve (left) and tumor weight (right) from age matched OPGKO and WT mice (N= 5 or 6 mice per group, The experiment has been repeated twice. (E) Irradiated C57BL6-CD45.1(CD45.1) recipient mice transplanted with bone marrow cells isolated from age- and sex-matched OPG KO or WT littermates. Tumor measurement was started at 12 days post tumor injection, and tumors were harvested at 26 days post tumor injection. (F-I) Female C57/BL6 mice were injected with GFP-labelled EO771 cells. Mice were treated with either αOPG or isotype control (IgG) at 3-, 10-, 17-, and 24-days post tumor implantation. Tumors were collected at 25 days post injection and processed for digestion and staining. Stromal cells (GFP-CD45-) and T cells (GFP-CD45+CD3+) were sorted by FACS and prepared for 10X Genomics scRNA sequencing analysis. (F) Schematic diagram of the experiment and single-cell RNA sequencing. (G) UMAP feature plots. UMAP plot of stromal cells from 10X Genomics scRNA-seq analysis (left). The stromal fraction was derived from mice injected with isotype control (IgG) or αOPG Colors indicate different clusters. UMAP plot showing high expression of Postn in cluster 3 (middle). Density plot showing the differences in density between IgG and αOPG treatment samples in Postn+ CAFs (cluster 3) (right). For each point in the UMAP space, calculated and visualized the likelihood of a UMAP region being associated with increased (positive log density ratio represented by red color) or decreased (negative log density ratio represented by blue color) proportions of IgG cells. (H) Heatmap showing relative expression levels of genes linked to inflamatory and Interferon (IFN) response pathway between IgG and αOPG treatment samples in Postn+ CAFs. Expression data are scaled between 0 and 1, in which 0 (dark blue) and 1 (dark red) represent the minimum and maximum expression, respectively. (I) UMAP plots showing expression of key iCAF and interferon response pathway genes in Postn+ CAFs. Data in B and C are representative of four independent experiments and D are representative of two independent experiments. Statistics were calculated using ordinary two-way analysis of variance test (B and D left panel) or two-tailed, unpaired Student’s t-test (C and D right panel).

Blocking OPG enhances immune cell tumor infiltration and T cell effector function in breast cancer model.

(A-E) Data are from C57/BL6 mice with 1×106 EO771 implantation and treated with IgG or αOPG at 3-, 10-, and 17-days post implantation. Tumors were harvested 23 days post-implantation for T cell function assessment. (A) Schematic diagram of the experiment (B) Quantification of the total number of tumor infiltration immune cells (CD45+, CD3+ T, CD4+ effector T, and CD8+ T cells) normalized by tumor weight (n= 10 mice per group). (C) Heatmap showing upregulation of genes associated with T cell activation in CD4+ T effector lymphocyte upon αOPG treatment. Gene enrichment analysis show αOPG treatment resulted in T cell activation. (D) Quantification of the total number of TNFα, IFNγ, Perforin A, and Granzyme B on CD4+Foxp3+ T cells normalized by tumor weight (n =10 mice per group). (E) Quantification of the total number of TNFα, IFNγ, Perforin A, and Granzyme B on CD8+ T cells normalized by tumor weight (n =10 mice per group). Data in B, D and E are representative of four independent experiments. Statistics were calculated using two-tailed, unpaired Student’s t-test

Blocking OPG enhances immune cell tumor infiltration and T cell effector function in pancreatic cancer model.

(A-C) Data are from C57/BL6 mice injected with 1×106 FC1199 and treated with IgG or αOPG at 7-, 14-, 21-, and 28-days post implantation. Tumors were harvested for T cell function assessment. (G) Schematic diagram of the experiment. (B) Tumor image (left) and tumor weight (right) on the day 34 post implantation (n=8 or 9 mice per group) (C) Quantification of the total number of CD3+ T cells, Cd4-CD8+ T cells, Granzyme B+, and Perforin A CD8+ T cells normalized by tumor weight (n= 8 or 9 mice per group) (D-F) Data are from C57/BL6 mice co-injected with 4×105 FC1199 and 4×105 adipose-derived stromal cells either from WT or OPG KO mice. Tumors were harvested at 25 days post co-implantation T. (D) Schematic diagram of the experiment. (E) Tumor weight on the day 25 days post implantation (n= 4 or 8 mice per group; combined two independent experiments). (F) Quantification of the frequency of IFNγ+ TNFα+ and Granzyme B+ Perforin A+ in CD8+ T cells (n= 4 or 8 mice per group; combined two independent experiments) Data in B, C, E and F are representative of two independent experiments. Statistics were calculated using two-tailed, unpaired Student’s t-test.

(A) UMAP visualization of different clusters in immune-depleted stromal cells derived from tumor-injected mice generated from 10x Genomics scRNA-seq analysis (first top left). UMAP feature plots showing expression of selected cluster-specific genes in mouse stromal cells. (B) Differential expression analysis across different clusters in mouse stromal cells derived from tumor-injected mice. Matrixplot showing expression of top 5 differentially expressed markers in each fibroblast subpopulation. Tnfrsf11b (green square) is one of the top differentially expressed markers for cluster 3 (identified as iCAFs). Expression data were normalized with z-score transformation, in which blue and red represent the high and low expression of a gene, respectively, relative to the median expression level. (C) Characterization of iCAFs (cluster 3) subpopulations: The cells derived from cluster 3 were selected and further analyzed by increasing the clustering resolution using the Leiden algorithm. Three subclusters were identified and visualized with UMAP plot. (D) The majority of iCAFs cells express inflammatory and senescence associated genes (subcluster 0) while there is a small subpopulation (subcluster 2) expressing cell cycle genes. Matrixplot showing the relative expression of senescence and cell cycle division markers across different subclusters in iCAFs. Expression data are scaled between 0 and 1, in which 0 (blue) and 1 (red) represent the minimum and maximum expression, respectively. UMAP feature plots showing the expression of selected senescence and cell cycle division markers in iCAFs.

(A) UMAP visualization of different clusters in stromal cells derived from human samples obtained from biopsies or resection. (B) Differential expression analysis across different clusters in human stromal cells from human samples obtained from biopsies or resection. Dotplot shows the expression of the top 3 differentially expressed markers in each fibroblast subpopulation. For each cluster, the fraction of cells expressing each gene is represented by the dot size. Expression data are scaled between 0 and 1, in which 0 (blue) and 1 (red) represent the minimum and maximum expression, respectively. (C) UMAP feature plots showing expression of selected cluster-specific genes in different human fibroblasts subpopulations. (D) Human PBMCs were isolated from human whole blood. CD8+ T cells were purified by human CD8+ T cell negative selection kit and activated with CD3/ CD28 beads for indicated time. TRAIL expression level was evaluated at 2-, 12-, 18-, and 30-hours post-stimulation by flow cytometry. (E) Human lung fibroblast cell line, Wi38, was used to generate senescent cells (Sen) by treating with etoposide for 48 hours. Wi38 cells without etoposide treatment was used as proliferative (Pro) control. Supernatant was collected for OPG measurement. OPG level was normalized to the protein content from the cells. (F) Senescent Wi38 cells (10,000) were seeded into 96-well plate. CD8+ T cells were activated by CD3/CD28 beads or vehicle for overnight and subsequently used for cocultivation with senescent cells with the T:E ratio as 1: 5. Human recombinant OPG was added to this co- cultivation system and incubated for 18 hours before cytotoxicity measurement.

(A) UMAP plot of stromal cells from 10x Genomics scRNA-seq analysis. (Left) The stromal fraction was derived from mice injected with IgG or αOPG. Colors indicate different clusters. (B) Differential expression analysis across different clusters in mouse stromal cells derived from tumor-injected mice. Matrixplot showing differentially expressed markers in cluster 3 (identified as iCAFs), and cluster 4 (identified as myCAFs). Expression data were normalized with z-score transformation, in which blue and red represent the high and low expression of a gene, respectively, relative to the median expression level. (C) UMAP feature plots showing high expression levels of selected genes in iCAFs and myCAFs (D) Violin plots showing expression of genes known to be the CAF markers, IFN responsive genes, and antigen-presenting molecules in different treatment groups (IgG and αOPG).

(A) Quantification of tumor isolation was shown as cells per mg. (B) UMAP visualization of different immune cell populations obtained from tumor-injected mice treated with IgG or αOPG. Distinct clusters of immune cells are identified using the Leiden algorithm and represented by different colors. Dotplot showing the expression of selected markers of various cell types. Dot size represents the fraction of cells expressing a specific marker. Expression data are scaled between 0 and 1, in which 0 (dark blue) and 1 (dark red) represent the minimum and maximum expression, respectively. (C) Heatmap showing the expression of selected genes associated with T cell activation in CD4+ T effector lymphocyte cluster (left) and CD8+ T lymphocyte cluster (right) between different treatment groups (IgG and αOPG). Expression data are scaled between 0 and 1, in which 0 (dark blue) and 1 (dark red) represent the minimum and maximum expression, respectively. (D) Gating Strategies for CD8+ T cells and CD4+Foxp3- T cells. (E) Tumor infiltration immune cells (CD3+ T, CD4+ effector T, and CD8+ T cells) from Fig 3D were quantified. (F) The activation of CD8+ T and CD4+ effector T cells from Fig 3D. was shown by intracellular cytokine, TNFα and cytotoxic marker, Perforin A. (G) Bone density of mice treated with IgG and αOPG were shown. Cortical analysis and Trabecular analysis of mice treated with IgG and αOPG were shown. (H) Serum bone alkaline phosphatase (BALP) was measured by ELISA.