Reprogramming tumour-associated macrophages to outcompete cancer cells

Abstract

In metazoan organisms, cell competition acts as a quality control mechanism to eliminate unfit cells in favour of their more robust neighbours^1,2. This mechanism has the potential to be maladapted, promoting the selection of aggressive cancer cells^3-6. Tumours are metabolically active and are populated by stroma cells^7,8, but how environmental factors affect cancer cell competition remains largely unknown. Here we show that tumour-associated macrophages (TAMs) can be dietarily or genetically reprogrammed to outcompete MYC-overexpressing cancer cells. In a mouse model of breast cancer, MYC overexpression resulted in an mTORC1-dependent 'winner' cancer cell state. A low-protein diet inhibited mTORC1 signalling in cancer cells and reduced tumour growth, owing unexpectedly to activation of the transcription factors TFEB and TFE3 and mTORC1 in TAMs. Diet-derived cytosolic amino acids are sensed by Rag GTPases through the GTPase-activating proteins GATOR1 and FLCN to control Rag GTPase effectors including TFEB and TFE3^9-14. Depletion of GATOR1 in TAMs suppressed the activation of TFEB, TFE3 and mTORC1 under the low-protein diet condition, causing accelerated tumour growth; conversely, depletion of FLCN or Rag GTPases in TAMs activated TFEB, TFE3 and mTORC1 under the normal protein diet condition, causing decelerated tumour growth. Furthermore, mTORC1 hyperactivation in TAMs and cancer cells and their competitive fitness were dependent on the endolysosomal engulfment regulator PIKfyve. Thus, noncanonical engulfment-mediated Rag GTPase-independent mTORC1 signalling in TAMs controls competition between TAMs and cancer cells, which defines a novel innate immune tumour suppression pathway that could be targeted for cancer therapy.

Extended data fig. 1. mTORC1 signaling supports the MYC-driven ‘winner’ status of cancer cells.

a, Frequency of MYC amplification across human malignancies with breast cancer being the most prevalent cancer type. b, Pam50 molecular subtypes in groups of human breast cancer samples with none, either or both of MYC amplification and PIK3CA driver mutation. c, Immunofluorescence images showing expression of cleaved PARP1 (C-PARP1), MYC, phosphorylated 4E-BP1 (p-4E-BP1), and EpCAM in 6 mm x 6 mm tumor tissue samples from mice of the indicated genotypes. Scale bar: 20 μm. Representative ‘loser’ cells marked by magenta dashed lines and ‘winner’ cells marked by yellow dashed lines. e, f, Quantification of percentage of C-PARP1-positive cancer cells (d), percentage of cells with high nuclear MYC expression or high p-4E-BP1 expression among C-PARP1-positive or −negative cancer cells from PyMT (n = 10), S100a8-cre-Rosa26^LSL-MYC/+PyMT (n = 10), S100a8-cre-Rptor^fl/flPyMT (n = 8) and S100a8-cre-Rosa26^LSL-MYC/+Rptor^fl/flPyMT (n = 6) tumors (e, f), and cancer cell size for tissue sections in Fig. 1d (g). h, Immunofluorescence images showing expression of cleaved caspase 3 (CC3), MYC, phosphorylated S6 ribosomal protein at serine 240/244 (p-S6), and EpCAM in 6 mm x 6 mm or 12 mm x 12 mm tumor tissue samples from S100a8-cre-Rosa26^LSL-MYC/+Rptor^fl/flPyMT mice. Scale bar: 20 μm. Representative ‘loser’ cells marked by magenta dashed lines and ‘winner’ cells marked by yellow dashed lines. i, j, Quantification of percentage of cells with high MYC expression or high p-S6 expression among CC3-positive or −negative EpCAM-expressing cancer cells from S100a8-cre-Rosa26^LSL-MYC/+Rptor^fl/flPyMT mice (n = 5). k-m, RT-qPCR analysis of Rptor, human MYC, and cre mRNA expression in cancer cells isolated from tumors of the indicated genotypes and tumor size (n = 4 mice for each group). All statistical data are shown as mean ± S.D. One-way ANOVA with the Tukey multiple comparison test correction in (d-g, k-m), and Two-sided Student t-test in (i, j).

Extended data fig. 2. MYC induces expression of amino acid metabolism genes in PyMT tumors.

A heatmap showing significantly differentially expressed genes (adjusted P < 0.05) from RNA-sequencing experiments between S100a8-cre-Rosa26^LSL-MYC/+PyMT (MYC-PyMT) and control-PyMT tumors, categorized in groups that encode for proteins involved in amino acid synthesis, amino acid transporters, and ribonucleoproteins.

Extended data fig. 3. A low protein diet triggers TAM reprogramming in MYC-PyMT mice.

a, Immunofluorescence images showing expression of F4/80, phosphorylated 4E-BP1 (p-4E-BP1), and EpCAM in 6 mm x 6 mm tumor tissue samples from mice of the indicated genotypes and dietary treatment conditions. Scale bar: 5 μm. b-c, Quantification of percentage of cells with high p-4E-BP1 expression among EpCAM-expressing cancer cells and F4/80-expressing macrophages (TAMs) in the tumor parenchyma. n = 12 mice for control-PyMT + NP or LP, n = 7 mice for MYC-PyMT + NP, and n = 10 mice for MYC-PyMT + LP. d, Immunofluorescence images showing expression of F4/80, MRC1, and VCAM1 in 6 mm x 6 mm tumor tissue samples from mice of the indicated genotypes and dietary treatment conditions. MRC1 is highly expressed in mammary tissue macrophages (MTMs) in the tumor interstitial region, while VCAM1 is highly expressed in tumor-associated macrophages (TAMs) residing in the tumor parenchyma. Scale bar: 50 μm. T: tumor parenchyma, S: interstitial region or stroma of the tumor tissue. Parenchyma-interstitial region borders marked by yellow dashed lines. e-g, Contour plots and histograms showing the gating strategy and side scatter SSC-A of monocytes, MTMs, and TAMs isolated from mice of the indicated genotypes and dietary treatment conditions. Siglec-F, B220, and Ly6G markers were used to exclude eosinophils, B cells, and neutrophils, respectively. (e). The percentage of monocytes, MTMs, and TAMs among CD45⁺ cells (f) and their median SSC-A (g) are quantified (n = 8 for each group). NP: control diet with 15% protein in weight (5CC7, TestDiet), LP: low protein diet with 2% protein in weight (5BT9, TestDiet). All statistical data are shown as mean ± S.D. One-way ANOVA with the Tukey multiple comparison test correction in (b, c, f, g).

Extended data fig. 4. Transcriptome profiling of TAMs and MTMs.

a, A schematic diagram showing the Rosa26^LSL-MYC allele that encodes human MYC (hMYC) and the internal ribosome entry site (IRES)-driven truncated human CD2 (hCD2) protein as a cell surface reporter. b, Reporter assay showing the specificity of S100a8-cre-mediated recombination of the Rosa26^LSL-MYC allele and high reconstitution efficiency of wild-type (WT) bone marrow cells. c, Experimental design of bulk RNA-sequencing experiments for tumor-associated macrophages (TAMs) and mammary tissue macrophages (MTMs) isolated from control-PyMT and MYC-PyMT mice reconstituted with WT bone marrow cells and treated with a low protein (LP) diet (2% protein in weight, 5BT9, TestDiet). d, Flow cytometry gating strategy for TAMs and MTMs. Siglec-F, B220, Ly6G, and Ly6C were used to exclude eosinophils, B cells, neutrophils, and monocytes, respectively. e. PCA analysis of TAMs and MTMs isolated from control-PyMT and MYC-PyMT mice reconstituted with WT bone marrow cells and treated with a LP diet, based on the log2 fold change of gene expression compared to the mean. MTMs from MYC3 group are not included due to insufficient cell number.

Extended data fig. 5. Generation and validation of Fcgr1-cre mice.

a, A schematic diagram showing the design of Fcgr1-cre knock-in mice with an IRES-tdTomato-T2A-iCre expression cassette targeted to the Fcgr1 gene locus. b, Validation of the Fcgr1-cre knock-in allele by PCR showing germ line transmission of the targeted allele. PCR results were consistently observed across all generations of mice. c, d, Reporter assay showing the specificity of Fcgr1-cre-mediated recombination of the Rosa26^{LSL-H2B-mCherry} allele. Gating strategy of tumor-infiltrating immune cell populations in (c). Top and bottom rows indicate two separate experiments. Histogram overlays in (d) showing H2B-mCherry expression in various tumor-infiltrating immune cell populations isolated from the indicated genotypes. Chromatin-associated H2B-mCherry, but not tdTomato expressed in the cytosol, was retained after fixation and permeabilization in flow cytometric analysis. MTM: mammary tissue macrophage; TAM: tumor-associated macrophage; DC1: type 1 dendritic cell; DC2: type 2 dendritic cell.

Extended data fig. 6. TAMs are dietarily reprogramed in MYC-PyMT mice.

a, A schematic diagram showing the design of single-nucleus RNA-sequencing experiments in Control-PyMT and MYC-PyMT mice. Siglec-F, B220, Ly6G, and Ly6C were used to exclude eosinophils, B cells, neutrophils, and monocytes, respectively. VCAM1 is a marker for the tumor parenchyma-localized TAMs. b-d, TSNE plots showing unsupervised clustering of TAM nuclei isolated from the low protein (LP) diet (2% protein in weight, 5BT9, TestDiet)-conditioned control-PyMT and MYC-PyMT tumors, and color-coded as sub-clusters in (b) and genotypes in (c). Percentage of each sub-cluster of TAM nuclei from control-PyMT and MYC-PyMT tumors (d). e, A scatter plot showing enriched gene sets from differentially expressed genes (adjusted P < 0.05) in cluster 3 versus other clusters. f, Immunofluorescence images showing expression of F4/80, EpCAM, and LAMP in 6 mm x 6 mm tumor tissue samples from the LP diet-treated control-PyMT and MYC-PyMT mice. Scale bar: 20 μm. Representative TAMs marked by yellow dashed lines. g, Quantification of AFU (arbitrary fluorescence unit) of LAMP1 expressing in F4/80-expressing TAMs from the LP diet-treated control-PyMT and MYC-PyMT mice (n = 3 mice for control-PyMT + LP, n = 4 mice for MYC-PyMT + LP). Data points are average AFU from each captured images. h, j, RT-qPCR analysis of Tfeb and Rptor mRNA expression in TAMs of the indicated genotypes (n = 4 mice for each group). i, A schematic diagram showing the generation of bone marrow (BM) chimera mice. BM cells from control, Fcgr1-cre-Rptor^fl/fl mice were transferred into lethally irradiated MYC-PyMT recipients with a tumor burden around 250–350 mm³ followed by switching to the LP diet two weeks after. All statistical data are shown as mean ± S.D. Over-representation analysis based on hypergeometric test with the Benjamini-Hochberg multiple comparison test correction in (e), two-sided Student t-test in (g, h, j).

Extended data fig. 7. Two modes of mTORC1 signaling promoted or suppressed by Rag GTPases.

a. A schematic diagram showing two modes of mTORC1 signaling in TAMs promoted or suppressed by Rag GTPases. Under a normal protein (NP) diet (15% protein in weight, 5CC7, TestDiet) condition, abundant cytosolic amino acids (AAs) suppress the GATOR1 complex, but activate the FLCN complex, and induce Rag GTPases in a RagA/B^GTP-RagC/D^GDP state that recruits mTORC1 and TFEB/TFE3 to the Rag GTPase complex, supporting or inhibiting their activation, respectively. Under a low protein (LP) diet (2% protein in weight, 5BT9, TestDiet) condition, AA starvation activates GATOR1, but inactivates FLCN, and induces Rag GTPases in a RagA/B^GDP-RagC/D^GTP state that dissociates mTORC1 and TFEB/TFE3 from the Rag GTPase complex, leading to TFEB/TFE3 nuclear translocation and mTORC1 activation through Rag GTPase-independent mechanisms. Depletion of GATOR1 maintains a RagC/D^GDP state under the LP diet condition, while depletion of FLCN maintains a RagC/D^GTP state under the NP diet condition, causing inaction or activation of the TFEB/TFE3-mTORC1 signaling pathway, respectively. Depletion of Rag GTPases also activates TFEB/TFE3-mTORC1 signaling under the NP diet condition. b, c, Mass spectrometry quantification of individual AAs in tumor interstitial fluid extracted from control-PyMT and MYC-PyMT tumors under NP or LP diet conditions. AAs shown changes in the LP diet-treated MYC-PyMT tumors were plotted. Due to the absence of spike-in isotope labeled tryptophan, normalized intensity was shown. n = 2 mice each condition for NP treated control-PyMT and MYC-PyMT, and n = 3 mice each condition for LP treated control-PyMT and MYC-PyMT. d, A schematic diagrams showing the generation of bone marrow (BM) chimera mice. BM cells from control and Fcgr1-cre-Depdc5^fl/fl mice were transferred into lethally irradiated MYC-PyMT recipients with a tumor burden around 250–350 mm³ followed by switching to the LP diet two weeks after. e, BM cells from control, Fcgr1-cre-Flcn^fl/fl, and Fcgr1-cre-Rraga^fl/flRragb^fl/fl mice were transferred into lethally irradiated MYC-PyMT recipients with a tumor burden around 250–350 mm³ and kept under the NP diet condition. f-h, RT-qPCR analysis of Flcn, Rraga, and Rragb mRNA expression in TAMs of the indicated genotypes (n = 4 mice for each group). All statistical data are shown as mean ± S.D. Two-sided Student t-test in (c, f-h).

Extended data fig. 8. Dietarily reprogramed TAMs engulf apoptotic cells in MYC-PyMT tumors.

a, b, Experimental design, flow cytometric analysis and quantification of TAM engulfment of apoptotic cells in a complete or low amino acid (AA) RPMI medium (1: 4 mixture of complete RPMI and AA-free RPMI). Apoptotic PyMT cancer cells were labeled by CellTrace Violet (CTV) and induced by H₂O₂ treatment. TAMs were isolated from tumors of control-PyMT and MYC-PyMT mice (n=5 for each group) treated with a low protein (LP) diet (2% protein in weight, 5BT9, TestDiet). c, Immunofluorescence images showing expression or localization of F4/80, EpCAM, and TUNEL in 6 mm x 6 mm tumors from control-PyMT and MYC-PyMT mice under LP or normal protein (NP) diet (15% protein in weight, 5CC7, TestDiet) conditions. Scale bar: 50 μm. Representative macrophages marked by yellow dashed lines. d, Quantification of percentage of TUNEL-positive cells among all cells. e, Quantification of percentage of TUNEL-positive nuclei inside F4/80-expressing macrophages. Data points are average percentage from random captured images. n = 3 mice for each group in (d, e). f, A schematic diagram showing the generation of bone marrow (BM) chimera mice. BM cells from Fcgr1-cre-Rosa26^{LSL-H2B-mCherry/+} mice were transferred into lethally irradiated MYC-PyMT recipients with a tumor burden around 250–350 mm³ followed by switching to the LP diet two weeks after. g, Immunofluorescence images showing expression or localization of F4/80, EpCAM, H2B-mCherry, and TUNEL in 6 mm x 6 mm tumor tissue samples from BM chimera mice described in (f). Scale bar: 10 μm. All statistical data are shown as mean ± S.D. One-way ANOVA with the Tukey multiple comparison test correction in (b, d, e).

Extended data fig. 9. Apoptotic cancer cells are effectively phagocytosed by reprogramed TAMs.

a, b, Schematic diagrams showing bone marrow (BM) chimera mice to define TAM engulfment of dying cancer cells. BM cells from homozygous H2B-EGFP/H2B-EGFP mice were transferred into lethally irradiated S100a8-cre-Rosa26^{LSL-MYC/LSL-H2B-mCherry}PyMT recipients with a tumor burden around 250–350 mm³ followed by switching to a low protein (LP) diet (2% protein in weight, 5BT9, TestDiet) two weeks after. c, A schematic diagram showing dietary treatment of BM chimera mice. BM cells of control or Fcgr1-cre-Pikfyve^fl/fl mice were transferred into lethally irradiated S100a8-cre-Rosa26^{LSL-MYC/LSL-H2B-mcherry}PyMT recipients with a tumor burden around 250–350 mm³, which were either kept under a normal protein (NP) diet (15% protein in weight, 5CC7, TestDiet) condition or switched to a LP diet two weeks after. d, RT-qPCR analysis of PIKfyve mRNA expression in TAMs of the indicated genotypes (n = 4 mice for each group). e, A schematic diagram showing dietary treatment of BM chimera mice. BM cells from control or Fcgr1-cre-Pikfyve^fl/fl mice were transferred into lethally irradiated S100a8-cre-Rosa26^LSL-MYC/+PyMT or S100a8-cre-Rosa26^LSL-MYC/+Pikfyve^fl/flPyMT recipients with a tumor burden around 250–350 mm³ followed by switching to a LP diet two weeks after. f, Experimental design of an in vivo engulfment assay with intratumoral injection of CellTrace Violet (CTV) and CellTrace Yellow (CTY)-labeled apoptotic cells and apoptotic bodies. g, h, Experimental design and flow cytometry analysis of an ex vivo engulfment assay with TAMs and cancer cells fed with apoptotic cells and apoptotic bodies in a low amino acid RPMI medium (1: 4 mixtures of RPMI and amino acid-free RPMI). n = 3 independent experiments and cells from independent mice each group. All statistical data are shown as mean ± S.D. Two-sided Student t-test in (d), one-way ANOVA with the Tukey multiple comparison test correction in (h).

Extended data fig. 10. Reprogram TAMs to outmatch cancer cell competition.

In S100a8-cre-Rosa26^LSL-MYC/+PyMT mice, MYC^hi cancer cells are fitter than MYC^lo cancer cells and outcompete them in a process dependent on higher mTORC1 signaling. Cell debris from apoptotic cancer cells are acquired by MYC^hi cancer cells as a nutrient source to promote mTORC1 activation and the ‘winner’ status. Feeding mice with a low protein diet or depletion of FLCN or Rag GTPases activates TFEB/TFE3 and mTORC1 signaling in tumor-associated macrophages (TAMs) and enhances TAM fitness supported by engulfment of apoptotic cancer cells and apoptotic cell debris. Such a contest in engulfment-dependent nutrient acquisition dictates cell competition between TAMs and MYC^hi cancer cells, which defines an innate immune tumor suppression pathway that may be targeted for cancer immunotherapy.

Fig. 1. mTORC1 signaling supports a ‘winner’ state of MYC^hi cancer cells.

a, A bar graph showing the distribution of human breast cancer patient cases with or without MYC amplification and/or PIK3CA gain-of-function driver mutation from the METABRIC dataset. b, A Kaplan Meier relapse free survival graph of human breast cancer patients categorized into groups with none, either or both of MYC amplification and PIK3CA driver mutation. P-values of log rank test were calculated by comparing all four groups or three groups excluding cooccurrence of MYC amplification and PIK3CA driver mutation. c, Tumor burden in mice of the indicated genotypes including PyMT (n = 5), S100a8-cre-Rosa26^LSL-MYC/+PyMT (n = 5), S100a8-cre-Rptor^fl/flPyMT (n = 5), and S100a8-cre-Rosa26^LSL-MYC/+Rptor^fl/flPyMT (n = 4). d, Immunofluorescence images showing expression of cleaved caspase 3 (CC3), MYC, phosphorylated S6 ribosomal protein at serine 240/244 (p-S6), and EpCAM in 6 mm x 6 mm tumor tissue samples from mice of the indicated genotypes. Scale bar: 20 μm. Representative ‘loser’ cells marked by magenta dashed lines and ‘winner’ cells marked by yellow dashed lines. e, f, Quantification of percentage of cells with high nuclear MYC expression or positive CC3 expression among EpCAM-expressing cancer cells. g, h, Quantification of percentage of cells with high MYC or p-S6 expression among CC3-positive or −negative EpCAM-expressing cancer cells. In (e-h), n = 10 mice for PyMT and S100a8-cre-Rosa26^LSL-MYC/+PyMT, n = 8 mice for S100a8-cre-Rptor^fl/flPyMT, and n = 6 mice for S100a8-cre-Rosa26^LSL-MYC/+Rptor^fl/flPyMT. All statistical data are shown as mean ± S.D. Two-way ANOVA in (c), one-way ANOVA with the Tukey multiple comparison test correction in (e-h).

Fig. 2. A low protein diet triggers discordant mTORC1 activation in cancer cells and TAMs.

a, A scatter plot with linear correlation trend line showing enriched gene sets overlapped between human and mouse tumor samples. b, Tumor burden in mice of the indicated genotypes under normal protein (NP) or low protein (LP) diet conditions, including control-PyMT + NP (n = 4), control-PyMT + LP (n = 6), MYC-PyMT + NP (n = 5), MYC-PyMT + LP (n = 5). Mice with tumor burden around 250–350 mm³ from each genotype were treated with the LP diet or kept with the NP diet. c, Immunofluorescence images showing expression of phosphorylated S6 ribosomal protein at serine 240/244 (p-S6), EpCAM, and F4/80 in 6 mm x 6 mm tumor tissue samples from mice of the indicated genotypes and dietary treatment conditions. Scale bar: 50 μm. Representative macrophages marked by yellow dashed lines. d, Quantification of percentage of cells with high p-S6 expression among EpCAM-expressing cancer cells. e, Quantification of percentage of F4/80-expressing macrophages in the tumor parenchyma. f, Quantification of percentage of cells with high p-S6 expression among F4/80-expressing macrophages. g, Quantification of cell size of F4/80-expressing macrophages. In (d-g), n = 12 mice for control-PyMT + NP or LP, n = 7 mice for MYC-PyMT + NP, and n = 10 mice for MYC-PyMT + LP. Mice were sacrificed after 4-week LP diet treatment for immunofluorescence analysis. NP: control diet with 15% protein in weight (5CC7, TestDiet), LP: low protein diet with 2% protein in weight (5BT9, TestDiet). All statistical data are shown as mean ± S.D. Simple linear regression and GSEA based on the Kolmogorov Smirnov (K-S) test in (a), two-way ANOVA in (b), one-way ANOVA with the Tukey multiple comparison test correction in (d-g).

Fig. 3. TFEB/TFE3-dependent TAM mTORC1 activation suppresses cancer cell mTORC1 signaling.

a, Immunofluorescence images showing expression of EpCAM, F4/80, and TFE3 in 6 mm x 6 mm tumor tissue samples from mice of the indicated genotypes under a low protein (LP) diet condition (2% protein in weight, 5BT9, TestDiet). Scale bar: 50 μm. T: tumor parenchyma. Representative macrophages marked by yellow dashed lines. b, Quantification of percentage of cells with nuclear TFE3 among F4/80-expressing macrophages in the tumor parenchyma from control PyMT (n = 12) and MYC-PyMT (n = 7) mice. Tumor-bearing mice were sacrificed after 4-week LP diet treatment for immunofluorescence analysis. c, A schematic diagram showing dietary treatment of bone marrow (BM) chimera tumor-bearing mice. BM cells of control or Fcgr1-cre-Tfeb^fl/flTfe3^−/− mice were transferred into lethally irradiated MYC-PyMT mice with tumor burden around 250–350 mm³, followed by switching to the LP diet two weeks after. d, j, Immunofluorescence images showing expression of phosphorylated S6 ribosomal protein at serine 240/244 (p-S6), EpCAM, and F4/80 in 6 mm x 6 mm tumor tissue samples in BM chimera mice of the indicated genotypes under the LP diet condition. Scale bar: 50 μm. Representative macrophages marked by yellow dashed lines. e, k, Quantification of percentage of cells with high p-S6 expression among F4/80-expressing macrophages in the tumor parenchyma. f, l, Quantification of cell size of F4/80-expressing macrophages. g, m, Quantification of percentage of F4/80-expressing macrophages. h, n, Quantification of percentage of cells with high p-S6 expression among EpCAM-expressing cancer cells. n = 5 mice for each group in (e-h); n = 7 mice for control BM, n = 6 mice for Fcgr1-cre-Rptor^fl/fl BM in (k-n). Tumor-bearing mice were sacrificed after 4-week LP diet treatment for immunofluorescence analysis. i, o, Tumor burden in BM chimera mice of the indicated genotypes under the LP diet condition, including MYC-PyMT + control BM (n = 5) in (i), MYC-PyMT + Fcgr1-cre-Tfeb^fl/flTfe3^−/− BM (n = 6) in (i), MYC-PyMT + control BM (n = 4) in (o) and MYC-PyMT + Fcgr1-cre-Rptor^fl/fl BM (n = 5) in (o). All statistical data are shown as mean ± S.D. Two-sided Student t-test in (b, e-h, k-n) and two-way ANOVA in (i, o).

Fig. 4. The Rag GTPase nutrient-sensing pathway controls TFEB/TFE3-mTORC1 signaling in TAMs.

a, h, Immunofluorescence images showing expression of phosphorylated S6 ribosomal protein at serine 240/244 (p-S6), EpCAM, F4/80, and TFE3 in 6 mm x 6 mm tumor tissue samples in bone marrow (BM) chimera mice of the indicated genotypes under low protein (LP) diet (a) or normal protein (NP) diet (h) conditions. Scale bar: 20 μm. Representative macrophages marked by yellow dashed lines. b, i, Quantification of percentage of cells with nuclear TFE3 among F4/80-expressing macrophages in the tumor parenchyma. c, j, Quantification of percentage of cells with high p-S6 expression among F4/80-expressing macrophages. d, k, Quantification of cell size of F4/80-expressing macrophages. e, l, Quantification of percentage of F4/80-expressing macrophages. f, m, Quantification of percentage of cells with high p-S6 expression among EpCAM-expressing cancer cells. n = 6 mice for each group in (b-f), n = 6 mice for control BM in (i-m), n = 8 mice for both Fcgr1-cre-Flcn^fl/fl BM and Fcgr1-cre-Rraga^fl/flRragb^fl/fl BM in (i-m). Tumor-bearing mice were sacrificed after 4-week LP diet treatment for immunofluorescence analysis. g, n, Tumor burden in BM chimera mice of the indicated genotypes under either the LP diet condition, including MYC-PyMT + control BM (n = 6) and MYC-PyMT + Fcgr1-cre-Depdc5^fl/fl BM (n = 7), or the NP diet condition, including MYC-PyMT + control BM (n = 6), MYC-PyMT + Fcgr1-cre-Flcn^fl/fl BM (n = 6) and MYC-PyMT + Fcgr1-cre-Rraga^fl/flRragb^fl/fl BM (n = 9). NP: control diet with 15% protein in weight (5CC7, TestDiet), LP: low protein diet with 2% protein in weight (5BT9, TestDiet). All statistical data are shown as mean ± S.D. Two-sided Student t-test in (b-f), two-way ANOVA in (g, n), and one-way ANOVA with the Tukey multiple comparison test correction in (i-m).

Fig. 5. Engulfment-mediated mTORC1 signaling dictates TAM and cancer cell competition.

a, Left panel: immunofluorescence images showing expression of F4/80, H2B-EGFP, and H2B-mCherry in 6 mm x 6 mm tumor tissue samples from bone marrow (BM) chimera mice made by transferring BM cells from H2B-EGFP/H2B-EGFP mice into lethally irradiated S100a8-cre-Rosa26^{LSL-MYC/LSL-H2B-mCherry}PyMT recipients under normal protein (NP) or low protein (LP) diet conditions. Scale bar: 5 μm. Right panel: quantification of mCherry and DAPI-double positive particles in F4/80-expressing macrophages under NP or LP diet conditions (n = 3 mice for each condition, data points are average numbers from each captured images). b, Flow cytometric analysis of engulfment of H2B-mCherry-positive apoptotic cancer cells by TAMs in BM chimera tumor-bearing S100a8-cre-Rosa26^{LSL-MYC/LSL-H2B-mCherry}PyMT mice with the indicated BM genotypes and NP or LP diet treatment conditions (left panel). Mean fluorescence intensity (MFI) of H2B-mCherry was quantified (right panel). n = 5 mice for each condition. c, Immunofluorescence images showing expression of phosphorylated S6 ribosomal protein at serine 240/244 (p-S6), EpCAM, and F4/80 in 6 mm x 6 mm tumor tissue samples in BM chimera mice of the indicated genotypes under the LP diet condition. Scale bar: 50 μm. d, Quantification of percentage of cells with high p-S6 expression among F4/80-expressing macrophages in the tumor parenchyma. e, Quantification of cell size of F4/80-expressing macrophages. f, Quantification of percentage of F4/80-expressing macrophages. g, Quantification of percentage of cells with high p-S6 expression among EpCAM-expressing cancer cells. n = 6 mice for each group in (d-g). Tumor-bearing mice were sacrificed after 4-week LP diet treatment for immunofluorescence analysis. h, i, Tumor burden in BM chimera mice of the indicated genotypes under the LP diet condition [n = 6 in (h) and n = 8 in (i) for each group]. j. Flow cytometric analysis of engulfment of CellTrace Violet (CTV)- and CellTrace Yellow (CTY)-labeled apoptotic cells and apoptotic bodies by TAMs and cancer cells in control and Fcgr1-cre-Pikfyve^fl/fl bone marrow chimeric MYC-PyMT mice under the LP diet condition. k-n. Quantification of CTV and CTY MFI signals in TAMs and cancer cells (n = 5 mice for each group). NP: control diet with 15% protein in weight (5CC7, TestDiet), LP: low protein diet with 2% protein in weight (5BT9, TestDiet). All statistical data are shown as mean ± S.D. Two-sided Student t-test in (a, k-n), one-way ANOVA with the Tukey multiple comparison test correction in (b, d-g), and two-way ANOVA in (h, i).