CD301b+ monocyte-derived dendritic cells mediate resistance to radiotherapy

Abstract

Monocytes infiltrating tumors acquire various states that distinctly impact cancer treatment. Here, we show that resistance of tumors to radiotherapy (RT) is controlled by the accumulation of monocyte-derived dendritic cells (moDCs). These moDCs are characterized by the expression of CD301b and have a superior capacity to generate regulatory T cells (Tregs). Accordingly, moDC depletion limits Treg generation and improves the therapeutic outcome of RT. Mechanistically, we demonstrate that granulocyte-macrophage colony-stimulating factor (GM-CSF) derived from radioresistant tumor cells following RT is necessary for the accumulation of moDCs. Our results unravel the immunosuppressive function of moDCs and identify GM-CSF as an immunotherapeutic target during RT.

Graphical abstract

Figure 1.

Treg accumulation mediates tumor resistance to RT. (A) B16 tumors were implanted subcutaneously in B6 mice, and BRaf/Pten tumors were induced by 4-hydroxytamoxifen. Tumors were treated with 20 Gy or three 8 Gy (3 × 8 Gy) doses of RT at the indicated tumor sizes. (B) Left: individual B16 tumor growth with a ratio of surviving mice in parentheses. Right: survival curves (n = 16/group, 3–5 experiments [exp.]). (C) Survival curves for BRaf/Pten tumors (n = 10–13/group, 2–4 exp.). (D and E) B16 tumors were treated or not with 20 Gy RT. (D) TIL numbers 7 days after RT, by flow cytometry (mean + SD; n = 14–19/group, 3–5 exp.). (E) TILs producing IFNγ, TNF, and GzmB 7 days after RT, after ex vivo restimulation, by flow cytometry (violin plots show data distribution; n = 8–11/3–4 exp.). (F–H) B16-bearing Foxp3^DTR mice were treated or not with 20 Gy RT and DT. (F) Left: frequency of blood Treg within CD4⁺ T cells (violin plots show data distribution; n = 5–6/group, 2 exp.). Right: individual tumor growth (with a ratio of surviving mice in parentheses) and survival curves (n = 7–9/group, 3 exp.). (G) TIL numbers 7 days after RT, by flow cytometry (mean + SD; n = 7–9/group, 4 exp.). (H) TILs producing IFNγ, TNF, and GzmB 7 days after RT, after ex vivo restimulation, by flow cytometry (violin plots show data distribution; n = 7–9/group, 4 exp.). (I) Left: tumor-infiltrating cDC1 numbers in Irf8^WT and Irf8^Δ32 mice 7 days after RT, by flow cytometry (violin plots show data distribution; n = 3/group, 2 exp.). Right: average tumor growth (mean + SEM) and survival of mice treated or not with RT and DT (n = 3–7/group, 3 exp.). Statistics: two-way ANOVA plus Tukey’s post hoc test for mean tumor growth (B, F, and I); Mantel–Cox’s test for survival curves (B, C, F, and I); two-tailed unpaired t test for the rest. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure S1.

T cell analysis after RT. (A) B16-bearing B6 mice were treated or not with 20 Gy RT, and tumors were analyzed 7 days later, after ex vivo restimulation, by flow cytometry. Left: representative flow cytometry plots. Right: frequency of CD4⁺ TILs (mean + SD; n = 7/group, 3–4 experiments [exp.]). (B) BRaf/Pten tumors were treated or not with 20 Gy RT. TIL numbers and TIL-to-Treg ratio 7 days after RT, by flow cytometry (mean + SD; n = 11–18/group, 3–5 exp.). (C) B16 tumors treated with 20 Gy RT followed by CD8⁺ T cell depletion. Left: frequency of blood CD8⁺ T cells (violin plots show data distribution; n = 8–9/group, 2 exp.). Right: individual tumor growth (with a ratio of surviving mice in parentheses) and survival curves (n = 9–15/group, 2–3 exp.). (D) Same as C but depleting CD4⁺ T cells. Left: frequency of blood CD4⁺ T cells (violin plots show data distribution; n = 7–9/group, 2 exp.). Right: individual tumor growth (with a ratio of surviving mice in parentheses) and survival curves (n = 7–9/group, 2 exp.). (E) B16-bearing Itgax-Cre × Irf8^f/f mice were treated with 20 Gy RT. Left: tumor-infiltrating cDC1 numbers at endpoint (violin plots show data distribution; n = 6/group, 2 exp.). Right: individual tumor growth (with a ratio of surviving mice in parentheses) and survival curves (n = 7–8/group, 2 exp.). Statistics: two-way ANOVA plus Tukey’s post hoc test for mean tumor growth (C–E); Mantel–Cox’s test for survival curves (C–E); two-tailed t test for the rest. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure 2.

RT induces accumulation of moDCs in the TME. (A and B) B16-bearing B6 mice on days 0 (−RT) and 5 after RT (+RT), analyzed by CyTOF. Mononuclear phagocytes were gated as live/singlets/CD45⁺/CD3⁻/CD19⁻/CD335⁻/CD8⁻, and eosinophils/neutrophils were excluded (n = 2/group, 2 experiments [exp.]). (A) Left: UMAP generated using FlowSOM. Right: UMAPs of all cells and treatments (contour plot) were overlaid with cell populations on days 0 (−RT) and 5 after RT (+RT). (B) Heatmap of the Z-scored expression of markers in each cluster identified by FlowSOM. (C and D) B16 tumors analyzed by flow cytometry. (C) Representative gating strategy of mononuclear phagocytes gated on live/singlets/CD45⁺/Ly6G⁻/lineage⁻ (lineage staining includes CD3, CD19, and NK1.1). (D) As in C, but the expression of F4/80, CD11c, and CD26 on cells within gate E is shown. (E) Representative expression of canonical markers, by flow cytometry. Numbers indicate gMFI ×10². Myeloid DCs are shown on days 0 (−RT) and 5 after RT (+RT). All the rest of the populations are shown on day 5 after RT (1 of 3 exp.). (F) Cell numbers per mg of tumor on days 1, 3, and 5 after RT, by flow cytometry (mean + SD; n = 4–20/group, 4–5 exp.). Statistics: one-way ANOVA plus Tukey’s post hoc test. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. (G) Mononuclear phagocytes were analyzed on days 0 and 5 after RT, by scRNAseq. 5,000 mononuclear phagocytes (excluding granulocytes and pDCs) were plotted. (H) Heatmap displays the top five DEGs for each cluster. Gene expression levels are normalized as log-transformed counts per cell for each cluster, as indicated in Table S2. (I) Normalized gene expression levels for the indicated genes, by scRNAseq. (J) moDC signatures overlaid onto the UMAP of scRNAseq data. Expression values represent log-normalized counts for genes listed in Table S1. gMFI, geometric mean fluorescence intensity.

Figure S2.

Myeloid cell analysis after RT. (A and B) B16 tumors analyzed by CyTOF. (A) Representative gating strategy of live/singlets/CD45⁺/Ly6G⁻/lineage⁻ (lineage staining includes CD3, CD19, and NK1.1). (B) Left: UMAP showing clusters generated using FlowSOM (Fig. 2 A). Right: cell populations identified via CyTOF gating were overlaid onto the depicted UMAP. (C) Representative expression of canonical markers, by flow cytometry. Numbers indicate gMFI ×10². Myeloid DCs are shown on days 0 (−RT) and 5 (+RT) after RT, while all the rest of the populations are shown at 5 days after RT (n = 1 of 3 experiments [exp.]). (D) BRaf/Pten tumors analyzed on days 1, 3, and 5 after RT, by flow cytometry (mean + SD; n = 5–19/group, 3–5 exp.). Statistics: one-way ANOVA plus Tukey’s post hoc test (D). *P ≤ 0.05, **P ≤ 0.01. gMFI, geometric mean fluorescence intensity.

Figure S3.

Gating strategy and gene signature assignment. (A) Cells were purified from B16 tumors on days 0 and 5 after RT, by FACS. Gating strategy of presorted cells (top), postsorted TAMs (middle), and postsorted DC2s/moDCs (bottom) (1 of 3 experiments [exp.]). (B) As in A, but myeloid DCs were sorted based on the expression of CD301b (1 of 2 exp.). (C) Gene signatures were overlaid onto the scRNAseq UMAP from Fig. 2 G; expression values represent log-normalized counts (genes listed in Table S1). The gene signatures of monocytes and mono^ACT were obtained from NanoString data as previously described (Tadepalli et al., 2023), while TAM and myeloid cell signatures were obtained from NanoString data of sorted cells, as described in A.

Figure 3.

CD301b is preferentially expressed on moDCs after RT. (A and B) Mononuclear phagocytes in Ms4a3^Cre × Rosa^LSL-TdTomato B16–bearing mice on days 0 and 5 after RT, analyzed by flow cytometry. (A) Left: representative histograms of TdTomato expression. Numbers represent the frequency of cells expressing TdTomato. Right: frequency of TdTomato⁺ and TdTomato⁻ cells in Ms4a3^Cre × Rosa^LSL-TdTomato tumor–bearing mice (mean; n = 4/group, 2 experiments [exp.]). (B) Total number of TdTomato⁺ and TdTomato⁻ myeloid DCs in Ms4a3^Cre xRosa^LSL-TdTomato tumor–bearing mice (mean + SD; n = 4/group, 2 exp). (C) Bone marrow CD45.2 Ccr2^RFP+Cx3cr1^EGFP+ monocytes were adoptively transferred into B16-bearing CD45.1 mice 2 days after RT. Left: experimental design. Right: numbers and identity of transferred cells 5 days after RT, by flow cytometry (mean + SD; n = 3–4/group, 2 exp.). (D) Expression of Ki-67 by mononuclear phagocytes on days 0 and 5 after RT, by flow cytometry. (E) Top: relative expression of Cd301b is overlaid onto the UMAP from Fig. 2 G, by scRNAseq; expression values represent log-normalized counts. Bottom: relative expression of CD301b is displayed onto the UMAP of all populations on days 0 and 5 after RT (Fig. 2 A), by CyTOF; expression values represent an arcsinh transformation of marker intensity. (F) Relative expression of CD301b in B16 tumors on days 0 and 5 after RT, by flow cytometry (n = 1 of 2 exp.). Numbers represent the frequency of positive cells. (G) Number of CD301b⁺ myeloid DCs per mg of tumors in B16 (left), and BRaf/PTEN (right) on days 0 and 5 after RT, by flow cytometry (mean + SD; n = 7–9/group, 2–3 exp.). (H) Left: expression of CD301b and Ms4a3^Cre × Rosa^LSL-TdTomato on days 0 and 5 after RT, by flow cytometry. Right: number of TdTomato⁺ CD301b⁺ myeloid DCs per mg of tumor on days 0 and 5 after RT, by flow cytometry (mean + SD; n = 4/group, 2 exp.). Statistics: two-tailed t test (A–C, G, and H). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure S4.

Functional and transcriptional features of moDCs. (A–C) B16-bearing B6 mice after RT, and αCcr2 or control Ab inoculation. (A) Cells per mg of tumor 5 days after RT, by flow cytometry (mean + SD; n = 5–6/group, 3–4 experiments [exp.]). (B) Left: individual tumor growth with ratio of surviving mice in parentheses. Right: tumor volume 10 days after RT (mean + SD; n = 15/group, 3 exp.). (C) TIL numbers per mg of tumor and TIL/Treg ratio 7 days after RT, by flow cytometry (mean + SD; n = 5–8/group, 3–4 exp.). (D) B16-bearing Itgax-Cre × Irf4^f/Δ mice after RT. Left: individual tumor growth and survival curves with a ratio of surviving mice in parentheses (mean + SD; n = 9–10/group, 2 exp). Right: number of moDCs per mg of tumor 5 days after RT analyzed, by flow cytometry (mean + SD; n = 4–5/group, two to three exp). (E) B16-OVA–bearing Cd301b^DTR+ after RT, with or without DT inoculation. Cells per mg of tumor 5 days after RT, by flow cytometry (mean + SD; n = 10–13/group, 4–7 exp.). (F) B16-bearing B6 mice after RT, and αCD115 or control Ab inoculation. Individual tumor growth and survival curves with a ratio of surviving mice in parentheses (n = 10/group, 2 exp.). (G) B16-bearing MM^DTR mice after RT and DT inoculation. Individual tumor growth and survival curves with a ratio of surviving mice in parentheses (n = 9–10/group, 2 exp.). (H) DEGs between DC2s (day 0; −RT) and moDCs (day 5 after RT; +RT) are shown using a log₂ fold change cutoff of 0.5, by scRNAseq. (I) NanoString transcriptomic analysis of sorted DC2s/moDCs versus TAMs on days 0 (−RT) and 5 after RT (+RT). PCA of normalized sample counts is shown. Statistics: two-way ANOVA plus Tukey’s post hoc test for mean tumor growth (B, D, F, and G); Mantel–Cox’s test for survival curves (D, F, and G); two-tailed t test (A–E). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure 4.

CD301b ⁺ moDCs mediate Treg generation after RT. (A–C) B16-bearing Cd301b^DTR+ and Cd301b^DTR− mice after RT and DT inoculation. (A) Cells per mg of tumor 5 days after RT, by flow cytometry (mean + SD; n = 4–6/group, 2 experiments [exp.]). (B) Individual tumor growth and survival curves with a ratio of surviving mice in parentheses (n = 10–12/group, 2 exp.). (C) TIL numbers and TIL-to-Treg ratio on days 0 and 7 after RT, by flow cytometry (mean + SD; n = 7–9/group, 3 exp.). (D–F) B16-OVA–bearing Cd301b^DTR+ after RT, with or without DT inoculation. (D) Individual tumor growth and survival curves with a ratio of surviving mice in parentheses (n = 8–10/group, 2 exp.). (E) TIL numbers and TIL-to-Treg ratio on days 0 and 7 after RT, by flow cytometry (mean + SD; n = 13–17/group, 4–6 exp.). (F) TILs producing IFNγ, TNF, and GzmB 7 days after RT, after ex vivo restimulation, by flow cytometry (violin plots show data distribution; n = 13–17/group, 4–6 exp.). (G) Tumor-infiltrating moDCs and TAMs were sorted (5 days after RT; Fig. S3, A and B, for purification strategies) and co-cultured with naïve CD4⁺ T cells purified from LNs of B6 mice. Analysis was done 4 days later. Left: experimental schematic. Middle: frequency of Treg within CD4⁺ T cells (mean + SD; n = 5–8/group, five to seven exp). Right: representative flow cytometry plot showing Treg generation by sorted CD301b⁺ moDCs versus CD301b⁻ myeloid DCs and TAMs (n = 1 of 2 exp.). Statistics: two-way ANOVA plus Tukey’s post hoc test for mean tumor growth (B and D); Mantel–Cox’s test for survival curves (B and D); one-way ANOVA plus Tukey’s post hoc test (G); two-tailed t test (A, C, E, and F). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure 5.

moDCs are transcriptionally similar to DC2s and differ from TAMs. (A) UMAP of DC2s and moDCs on days 0 and 5 after RT, respectively, by scRNAseq. (B) DEGs between DC2s (day 0) and moDCs (day 5 after RT) are shown using a log₂ fold change threshold of 1, by scRNAseq. (C) Expression of the indicated genes in DC2s (day 0) and moDCs (day 5 after RT), by scRNAseq. (D–F) NanoString transcriptomic analysis of sorted DC2s/moDCs and TAMs before (0 days) and after (5 days) RT (mean + SD; n = 3/group in 3–4 experiments [exp.]). (D) Pathway enrichment analysis was performed on normalized counts. The size indicates the significance of the score. (E) Heatmap of normalized counts comparing DC2s/moDCs with TAMs 0 versus 5 days after RT, by NanoString (see Table S4). (F) Left: DEGs significantly upregulated in DC2s/moDCs relative to TAMs. Right: normalized expression counts for the indicated genes, stratified by cell type and treatment. Statistics: one-way ANOVA plus Tukey’s post hoc test. ***P ≤ 0.001. (G) Pearson’s correlation matrix of indicated genes in patients undergoing RT, derived from The Cancer Genome Atlas for radioresistant tumors (melanoma, glioblastoma, and head and neck cancer).

Figure S5.

Role of GM-CSF on CD301b ⁺ moDCs. (A) B16 tumors analyzed by CyTOF. Left: UMAP showing clusters generated using FlowSOM (Fig. 2 A). Right: relative expression of CD116 on days 0 (−RT) and 5 after RT (+RT). Expression values represent an arcsinh transformation. (B) Generation of the B16^∆Csf2 cell line. The line above indicates the gRNA position. ICE software analysis of Sanger sequencing data confirmed a deletion in the B16^∆Csf2 cell clone (below). (C)Csf2^−/− mice transplanted with B16 and B16^∆Csf2 tumor cells. Left: cells per mg of tumor at 5 days after RT, by flow cytometry (mean + SD; n = 8–9/group, 4–5 experiments [exp.]). Right: TIL numbers, TIL/Treg ratio, and CD8⁺ TILs producing IFNγ, TNF, and GzmB at 7 days after RT, after ex vivo restimulation, by flow cytometry (mean + SD and violin plots of data distribution; n = 8–11/3–5 exp.). Statistics: two-tailed t test. *P ≤ 0.05.

Figure 6.

Tumor-derived GM-CSF drives the accumulation of CD301b ⁺ moDCs. (A) Relative expression of selected genes was overlaid onto the UMAP from Fig. 2 G; expression values represent log-normalized counts. (B) B16-bearing mice on days 0 and 5 after RT. Left: tumor Csf2 transcript levels, by qPCR (mean + SD; n = 5–6/group, 2–4 experiments [exp.]). Right: GM-CSF protein levels in tumor lysate, by ELISA (mean + SD; n = 6–9/group, 4 exp.). (C) Expression of CD301b by DC2s/moDCs, by flow cytometry. Left: frequency of CD301b on DC2s (day 0) and moDCs (day 5 after RT) (1 of 3 exp.). Right: frequency of CD301b⁺ expressing cells (mean + SD; n = 3–11/group, 2–5 exp.). (D) B16 and MC38 tumor cells were irradiated with 20 Gy RT and cultured for 4 days. Left: tumor cell Csf2 transcript levels, by qPCR (mean + SD; n = 7–9/group, 3–4 exp.). Right: GM-CSF protein levels in the culture supernatant, by ELISA (mean + SD; n = 4–9/group, 3–4 exp.). (E) B16 and B16^∆Csf2 tumor cell lines were irradiated with 20 Gy and cultured for 4 days. GM-CSF protein levels in the culture supernatant (mean + SD; n = 3/group, 3 exp.). (F–J) B16 and B16^∆Csf2 tumor–bearing mice treated with 20 Gy RT. (F) GM-CSF protein level in tumor lysate 5 days after RT, by ELISA (mean + SD; n = 7–9/group, four exp). (G) Cells per mg of tumor 5 days after RT, by flow cytometry (mean + SD; n = 6–22/group, 4–5 exp). (H) TIL numbers and TIL/Treg ratio 7 days after RT, by flow cytometry (mean + SD; n = 9–11/group, 4 exp). (I) TILs producing IFNγ, TNF, and GzmB 7 days after RT, after ex vivo restimulation, by flow cytometry (violin plots show data distribution; n = 9–11/group, 4 exp.). (J) Left: average tumor growth with a ratio of surviving mice in parentheses (mean + SEM). Right: tumor volume day 10 after RT (mean + SD; n = 14/group, 3 exp.). (K–N) As in F–J, but Csf2^−/− mice were transplanted with B16 and B16^∆Csf2 tumor cells. (K) Left: GM-CSF protein level in tumor lysate 5 days after RT, by ELISA (mean + SD; n = 8–16/group, 4–5 exp.). Right: moDC numbers 5 days after RT (+RT) normalized to day 0 (−RT), by flow cytometry (mean + SD; n = 8–9/group, 4–5 exp.). (L) Cells per mg of tumor and CD8⁺ TIL-to-Treg ratio 7 days after RT, normalized to day 0 (−RT) (mean + SD; n = 8–11/group, 4–5 exp.). (M) TILs producing IFNγ, TNF, and GzmB 7 days after RT, after ex vivo restimulation, normalized to day 0 (−RT) (violin plots show data distribution, n = 8–9/group, three to five exp). (N) Left: average tumor growth with a ratio of surviving mice in parentheses (mean + SEM). Right: tumor volume 10 days after RT (mean + SD; n = 6–14/group, 2–3 exp). Statistics: two-way ANOVA plus Tukey’s post hoc test for mean tumor growth (J and N); one-way ANOVA plus Tukey’s post hoc test (C–E); two-tailed t test for the rest. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.