LC3-associated phagocytosis in myeloid cells promotes tumor immune tolerance

SUMMARY

Targeting autophagy in cancer cells and in the tumor microenvironment are current goals of cancer therapy. However, components of canonical autophagy play roles in other biological processes, adding complexity to this goal. One such alternative function of autophagy proteins is LC3-associated phagocytosis (LAP), which functions in phagosome maturation and subsequent signaling events. Here we show that impairment of LAP in the myeloid compartment, rather than canonical autophagy, induces control of tumor growth by tumor-associated macrophages (TAM) upon phagocytosis of dying tumor cells. Single cell RNAseq analysis revealed that defects in LAP induce pro-inflammatory gene expression and trigger STING-mediated type I interferon responses in TAM. We found that the anti-tumor effects of LAP impairment require tumor-infiltrating T cells, dependent upon the STING and the type I interferon response. Therefore, autophagy proteins in the myeloid cells of the tumor microenvironment contribute to immune suppression of T lymphocytes by effecting LAP.

Graphical Abstract

INTRODUCTION

The role of autophagic processes in tumor cells and in the heterogeneous components of the tumor microenvironment, such as stromal, endothelial, and immune cells, is complex, most likely reflecting the intricate network of the spatiotemporal interactions between the different cell types (Amaravadi et al., 2016; Kimmelman and White, 2017). In early stages of tumorigenesis, autophagy plays a protective role against the emergence of malignant cells (Guo et al., 2013), possibly by different mechanisms, such as suppression by p62 of tumor promoting signals, protection against genomic stability, and control of early infiltration of regulatory T cells (Mathew et al., 2009; Mathew et al., 2007; Rao et al., 2014). However, the same components of the autophagy machinery can facilitate tumor progression at later times (Rao et al., 2014). Autophagy in the tumor microenvironment can provide essential nutrients, amino acids, and nucleotides to the tumor, facilitating tumor growth (Sousa et al., 2016).

Previously, we and others demonstrated that a non-canonical function of autophagy proteins, LC3-associated phagocytosis (LAP), couples the processing of engulfed particles, including pathogens, immune complexes, and dying cells, to regulation of macrophage immune responses (Henault et al., 2012; Martinez et al., 2011; Martinez et al., 2016; Martinez et al., 2015). During LAP, recruitment of components of the autophagy machinery to the phagosome catalyzes the lipidation of LC3/GABARAP-family proteins (herein, “LC3”) to phosphatydilethanolamine in the phagosomal membrane and induces phagosome maturation and cargo degradation (Martinez et al., 2015; Sanjuan et al., 2007). LAP is induced by engagement, at the phagosome, of TLR 1/2, TLR2/6, or TLR4 (Sanjuan et al., 2007), Fc-receptors (Henault et al., 2012; Huang et al., 2009), and the apoptotic cell receptor, TIM4 (Martinez et al., 2011). Unlike canonical autophagy, LAP does not require the components of the pre-initiation complex that trigger autophagy in response to nutrient stress (ULK1, FIP200, ATG13) but requires a unique Beclin-1 (BECN1) and VPS34 initiation complex that lacks ATG14 but contains Rubicon (RUBCN) (Florey et al., 2011; Martinez et al., 2011; Martinez et al., 2015).

LAP occurs in macrophages (Sanjuan et al., 2007), conventional dendritic cells (Martinez et al., 2016), plasmacytoid dendritic cells (Henault et al., 2012), and epithelial cells (Florey et al., 2011; Kim et al., 2013). LAP has been implicated in innate immune responses to different classes of pathogens (Sarkar, et al, 2017; Chu, et al., 2017; Martinez, et al., 2015; Tam, et al., 2014) and in effective recycling of all-trans retinoic acid from shed photoreceptors in the retinal pigment epithelium (Kim et al., 2013). LAP functions in the engulfment of dying (Martinez et al., 2011) and living (Florey et al., 2011) cells, and defects in this activity promote a lupus-like condition (Martinez et al., 2016). Engagement of LAP suppresses inflammatory responses upon engulfment of dying cells in vitro (Martinez et al., 2011) and in vivo (Martinez et al., 2016).

Here we sought to determine the importance of autophagic processes in the myeloid compartment of the tumor microenvironment by genetically ablating several autophagy components. We found that deficiency in components of LAP, but not canonical autophagy, specifically in myeloid cells, restricts tumor growth. We observed that LAP regulates the polarization of tumor-associated macrophages (TAM) towards immunosuppressive functions, and inhibits a STING-dependent type I interferon response in TAM, which is required for anti-tumor effects of LAP impairment. The effects of LAP deficiency on tumor growth were associated with increased activation of T lymphocytes in the tumor microenvironment, and depletion of T cells negated the anti-tumor effects. Therefore, the regulation of myeloid cell function by LAP suppresses T cell function, thereby promoting tumor growth.

RESULTS

LAP in myeloid cells promotes tumor growth

To discriminate between the roles of LAP and canonical autophagy in tumor-infiltrating macrophages, we employed conditional ablation of several genes using lysozyme M (LysM/Lyz2)-Cre-lox recombination, which affects monocytes, macrophages, and CD11b⁺ dendritic cells but not CD11c⁺ dendritic cells or T cells (Abram et al., 2014; Martinez et al., 2016). Previously, we demonstrated that myeloid cells in mice in which Beclin-1 (BECN1), VPS34, ATG5, ATG7, or ATG16L1 were ablated by this approach lack both conventional autophagy and LAP in response to TLR engagement (Martinez et al., 2015) or engulfment of dying cells (Martinez et al., 2016) (Fig. S1A, B). In contrast, myeloid cells deficient in FIP200, ULK1, or ATG14 lack autophagy but are fully competent for LAP, while such cells from animals deficient for Rubicon (RUBCN) or NADPH oxidase 2 (NOX2) lack LAP but not autophagy (Fig. S1A, B) (Martinez et al., 2015).

We observed that the growth of subcutaneously engrafted murine B16F10 melanoma was suppressed in LysM-Cre⁺ Becn1^f/f and Atg5^f/f immunocompetent C57BL/6 mice (Fig. 1 A). This protective effect was also observed in LAP-deficient Rubcn^−/− mice but not in LAP-sufficient LysM-Cre⁺ Fip200^f/f or Atg14^f/f animals (Fig. 1A). Similar results were obtained with engraftment of Lewis lung carcinoma (LLC, Fig. 1B) or MC38 adenocarcinoma (Fig. S1C). Notably, deficiency in several components of LAP (VPS34, ATG5, ATG7, ATG16L1, RUBCN and NOX2) also reduced growth of B16 melanoma in the lungs following intravenous injection of the tumor cells, an effect not observed in the absence of proteins required for canonical autophagy (ULK1, FIP200, ATG14) (Fig. 1C).

Figure 1. LAP in myeloid cells promotes tumor growth.

A. Tumor growth in wild-type and deficient littermates subcutaneously injected with B16F10 cells (Atg14^f/f Cre⁻ n=3, Cre⁺ n=4 mice/ Fip200^f/f Cre⁻ n=4, Cre⁺ n=6 mice/ Becn1^f/f Cre⁻ n=4, Cre⁺ n=5 mice/ Atg5^f/f Cre⁻ n=4, Cre⁺ n=5 mice/ Rubcn^+/+ n=7, Rubcn^{− /−} n=4 mice). Color scheme represents autophagy-deficient, LAP-sufficient (red), autophagy-deficient, LAP-deficient (green) and autophagy-sufficient, LAP-deficient (blue) mice.

B. Tumor growth in wild-type and deficient littermates subcutaneously injected with LLC cells (Fip200^f/f Cre⁻ n=7, Cre⁺ n=10 mice/ Becn1^f/f Cre⁻ n=4, Cre⁺ n=4 mice/ Atg7^f/f Cre⁻ n=4, Cre⁺ n=5 mice/ Rubcn^+/+ n=3, Rubcn^−/− n=4 mice). Colors as in (A).

A-B. Data are expressed as mean ±SEM. Significance was calculated with ANOVA. Data are representative of two independent experiments per genotype.

C. Micrometastatic lesions on the lungs of wild-type and deficient littermate mice intravenously injected with B16F10 cells. Data are pooled from two separate experiments. Dots represent the total number of lesions per mouse, line indicates mean ±SEM. Significance was calculated with Student’s t-test. Colors as in (A).

D. Schematic of the K-ras^LSL-G12D, p53^f/f bone-marrow chimeric mice transplanted with bone-marrow cells from either Rubcn^+/+ or Rubcn^−/− donor mice.

E. Quantification of tumor burden by MRI (expressed as % of total lobe volume) in K-ras^LSL-G12D, p53^f/f chimeras inoculated with Adv-cre. Data are pooled from two separate experiments. Dots represent data for each individual mouse (n= 10 mice per genotype), line indicates mean ±SEM. Significance was calculated using Student’s t-test.

F. Representative MRI images from the left lung lobe of the indicated K-ras^LSL-G12D, p53^f/f bone-marrow chimera.

G. Representative micrographs of H&E staining of the right lung cranial lobe for the indicated K-ras^LSL-G12D, p53^f/f bone-marrow chimera. Scale bar = 1 mm.

*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

See also Figure S1.

Because some components of the autophagy pathway, not involved in LAP, failed to compromise tumor growth, we examined the impact of the deletions on canonical autophagy, using accumulation of p62 as a marker of defective autophagy (Pankiv et al., 2007) (Fig. S1B). Further, we confirmed that LysM-Cre reduced protein expression of each conditionally targeted gene (Fig. S1B). Other than RUBCN, which, as expected, is not required for autophagy (Matsunaga et al., 2009; Zhong et al., 2009), myeloid cells lacking any components required for autophagy displayed elevated p62 levels. Therefore, the failure of deletion of some autophagy components (e.g., FIP200, ATG14) to impact tumor growth is not a result of a failure to compromise canonical autophagy in myeloid cells.

To extend these findings to a genetic model of tumorigenesis, we employed animals expressing one copy of a loxp-STOP-loxp (LSL) oncogenic mutant K-ras (K-ras^LSL-G12D) transgene in combination with conditional ablation of p53 (p53^f/f) to induce non-small-cell lung carcinoma (NSCLC) by Cre mediation recombination (DuPage et al., 2009; Johnson et al., 2001). Mice were transplanted with Rubcn^+/+ or Rubcn^−/− bone marrow, and subsequently subjected to inhalation of adenovirus particles expressing Cre (Adv-Cre) (Fig. 1D). As in the heterotopic tumor models, ablation of RUBCN in the hematopoietic compartment compromised tumor growth in this system (Fig. 1E–G).

Several components of the autophagy pathway (e.g., BECN1, VPS34, UVRAG) have roles in endosome trafficking (Itakura et al., 2008; Jaber et al., 2012; Jaber et al., 2016; McKnight et al., 2014) . As RUBCN is a component of this complex in LAP (Martinez, et al., 2015), we examined several types of endosome formation and function. Lack of RUBCN in macrophages had no effect on endocytosis of transferrin or immune complexes (Fig. S1D–F), macropinocytosis of dextran (Fig. S1G), endosome maturation (Fig. S1H), or degradation of EGFR (Fig. S1I). Therefore, based on the role of RUBCN and the pattern of requirement (or not) of autophagy proteins, our results support the conclusion that LAP, but not canonical autophagy, in myeloid cells promotes solid tumor growth.

LAP regulates TAM function

Tumor-infiltrating macrophages (TAM) are a heterogeneous population mostly derived from recruited Ly-6C^hi Ly-6G⁻ circulating monocytes (also known as monocytic myeloid-derived suppressor cells, Mo-MDSC (Movahedi et al., 2008)) that express MHC-II upon maturation (Movahedi et al., 2010) (Fig. S2). LAP or autophagy deficiency in the myeloid compartment did not alter overall infiltration of myeloid cells (CD45⁺ CD11b⁺), granulocytic MDSC (PMN-MDSC (Movahedi et al., 2008), CD11b⁺ Ly-6G⁺ Ly-6C^int) or Mo-MDSC (CD11b⁺ Ly-6G⁻ Ly-6C^high) (Fig. 2 A–C). However, ubiquitous RUBCN deficiency or ablation of LAP in myeloid cells increased surface expression of the inflammatory marker CD86 (Murray et al., 2014) in mature Ly6C⁻ MHC-II⁺ TAM (Fig. 2D). In contrast, surface expression of CD206 (Mrc1), a marker associated with anti-inflammatory, or M2 (Stein et al., 1992) (Murray et al., 2014), macrophage activation was reduced in the LAP-deficient mature TAM (Fig. 2 E). A similar effect was observed in the K-ras^LSL-G12D p53^f/f chimeras (Fig. 1D), where expression of CD206 was decreased in TAM from recipients of RUBCN-deficient versus -sufficient bone marrow, without differences in the number of TAM in the tumors (Fig. 2F–I). These results suggest that LAP regulates TAM polarization into anti-inflammatory M2 cells.

Figure 2. LAP regulates TAM function.

A-C. Percentage of total (A), monocytic (B) and granulocytic (C) myeloid cells within the CD45⁺ population of tumor infiltrating cells isolated from LLC tumors engrafted in Rubcn^−/−, Fip200^f/f Cre⁺, Becn1^f/f Cre⁺, Atg7^f/f Cre⁺ and respective littermate controls, estimated by flow cytometry. Data are expressed as individual data points (n = 3–5 mice per genotype as shown) and mean bar ±SEM.

D-E. MFI of CD86 (D) and CD206 (E) expression on the surface of monocytes (Mo) and mature TAM (TAM) isolated from LLC tumor engrafted into the indicated wild-type and deficient littermates (n=4 per group), measured by flow cytometry. Data are expressed as mean ±SEM. Significance was calculated with ANOVA.

F-G. Absolute number (F) and percentage (G) of macrophages (CD45⁺ Ly-6G⁻ CD11b⁺ F4/80⁺) in the lungs of K-ras^LSL-G12D, p53^f/f bone-marrow chimeras, estimated by flow cytometry.

H-I. Percentage (H) and MFI (I) of CD206 expression on the surface of macrophages in the lungs of K-ras^LSL-G12D, p53^f/f bone marrow chimeras (n= 5 mice per group), measured by flow cytometry.

F-I. Data are expressed as individual data points and mean bar ±SEM. Significance was calculated with Student’s t-test.

A-I. Data are representative of two independent experiments per genotype.

*p<0.05, **p<0.01, ***p<0.001

See also Figure S2.

Type I interferon responses in LAP-deficient TAM promote tumor suppression

To further explore the differences between LAP-competent and LAP-deficient TAM, we sorted CD11b⁺ Ly-6G⁻ (non-granulocytic) myeloid cells from LLC engrafted in Rubcn^+/− and Rubcn^−/− mice, and obtained single cell RNAseq (scRNAseq) profiles. Unsupervised cluster analysis using nonlinear dimensionality reduction (t-distributes stochastic neighbor embedding, t-SNE) revealed differences in the global gene expression profiles of Rubcn^+/− and Rubcn^−/− TAM and identified clusters of cells with unique genetic signatures (Fig. S3 A, B). We identified clusters that were likely to represent TAM, based on consistent expression of F4/80 (Mafb), Ly-6C (Ly6c2), and MHC-II (H2Aa). TAM expressing F4/80, high levels of Ly-6C and low levels of MHC-II were defined as monocytic TAM; F4/80 positive cells expressing lower levels of Ly-6C were defined as mature TAM and further discriminated into MHC-II^hi and MHC-II^lo subpopulations (Movahedi et al., 2010) (Fig. S3 B, C). Two populations of F4/80^lo Flt3⁺, MHC-II^hi cells (likely dendritic cells) were also distinguishable (Fig. S3 B, C). Based on F4/80, Ly-6C and MHC-II, we identified 7 clusters of immature monocytic and mature TAM that were distinct in their distribution of subpopulations in Rubcn^+/− and Rubcn^−/− mice (Fig. 3 A–D, Fig. S3 D, E). Among those, we observed a unique cluster of mature TAM (Ly-6C^lo MHC-II^hi; cluster 5) primarily composed of Rubcn^−/− cells with lower expression of M2 markers (Mrc1 and Arginase-1), supporting the idea that LAP permits anti-inflammatory TAM function (Fig. 3 E). The gene expression profile of this cluster also revealed a prominent type I interferon-induced gene expression signature (Fig. 3F, G). We also observed increased expression of type I IFN target genes in monocytic (Ly-6C^hi MHC-II^lo, cluster 2) Rubcn^−/− TAM (Fig. 3F). These subpopulations were responsible for an overall increase in the expression of type I IFN-inducible genes in LAP-deficient TAM (Fig. 3H). This type I IFN signature was largely absent from other cell populations we analyzed. (Fig. S3F).

Figure 3. LAP-deficiency induces differential gene expression in TAM.

A. tSNE plots of the expression profiles for monocytes and mature TAM from Rubcn^+/− (red) or Rubcn^−/− (blue) tumor-infiltrating myeloid cells isolated from LLC subcutaneous grafts. Each circle represents an individual cell.

B. Same as (A) but colors represent cells clustered together based on similarity of global gene expression.

C. tSNE plots showing the normalized absolute UMI count (log2) of categorizing markers (Mafb, Ly6c2, H2Aa) of monocytes and mature TAM for the clusters in (B).

D. Quantification of proportion of Rubcn^+/− (red) and Rubcn^−/− (blue) cells composing clusters in (B).

E. Expression of Mrc1 and Arginase1 in single myeloid cells for the clusters in (B).

F. Heat map showing the proportion of cells expressing type I IFN pathway target genes (GO term type I interferon pathway, access number GO:0060337, was used for categorization) for the clusters in (B). Clusters are shown in columns and genes in rows.

G. Gene ontology (GO) and Jensen assignments of top 100 uniquely expressed genes in cluster 5 of (B). Dash lines indicate p<0.001.

H. qPCR of type I IFN target genes in isolated Rubcn^−/− (n=4 mice) tumor-infiltrating myeloid cells relative to Rubcn^+/+ (n=5 mice) counterparts. Data are expressed as individual data points and mean bar ±SEM. Significance was calculated with Student’s t-test, *p<0.05. Data are representative of two independent experiments.

See also Figure S3.

Impaired LAP of engulfed dying tumor cells triggers type I IFN signaling in TAM

Tolerogenic signals produced upon engulfment of dying tumor cells have been suggested to be an important driver of macrophage-mediated suppression of adaptive tumor-infiltrating lymphocytes (TIL), ultimately promoting immune tolerance to the tumor (Garg and Agostinis, 2017). We have previously shown that LAP is required for efficient digestion of phagocytosed dying cells by macrophages and that defects in LAP culminate in the production of inflammatory cytokines in response to accumulated cell corpses (Martinez et al., 2011; Martinez et al., 2016). We thus hypothesized that engulfment of dying cells may account for LAP-dependent control of TAM function in the tumor microenvironment. To that end, we generated an LLC line that stably expresses fluorescent spectrin (LLC mCherry-spectrin cells), a component of cytoskeleton anchored to the cell membrane and thus retained in the cytosol upon cell death (Wang et al., 1996). LLC mCherry-spectrin cells were engrafted in mice transgenically expressing green fluorescent protein (GFP)-tagged LC3. Clearance and LAP of dying tumor cells was observed by confocal microscopy in CD11b⁺ tumor-infiltrating cells, in which GFP-LC3 was present at the membranes of phagosomes containing cell corpses (Fig. 4 A). We implanted LLC with mCherry-spectrin in Rubcn^+/− and Rubcn^−/− mice, sorted mCherry⁺ TAM and assessed the expression of inflammatory genes by quantitative PCR. Supporting our findings from scRNAseq, RUBCN-deficient TAM that had engulfed dying tumor cells displayed increased expression of type I IFN target genes, as well as pro-inflammatory II-1β (Fig. 4 B). Expression of Ifnβ was also increased in Rubcn^−/− TAM (Fig. 4 B), supporting the idea that LAP inhibits the production of type I IFN in TAM in response to engulfment of dying tumor cells.

Figure 4. Impaired LAP of engulfed dying tumor cells triggers type I IFN signaling in TAM.

A. Confocal image of harvested subcutaneous mCherry-spectrin⁺ (red) LLC tumor engrafted in GFP-LC3^Tg wild-type mouse. Tumor section was stained with anti-CD11b (blue). Arrows point to GFP-LC3⁺ phagosomes containing LLC debris. Scale bar = 10 μm.

B. qPCR of the expression of type I IFN target genes and Ifn β in tumor-associated myeloid cells that engulfed LLC tumor cells (mCherry-spectrin⁺) isolated from Rubcn^+/− and Rubcn^−/− mice (n=3 mice per genotype). Data are expressed as mean ±SEM.

C. Volcano plot for log₂ fold change (LFC) in the expression of genes in Rubcn^−/− versus Rubcn^+/+ bone marrow-derived macrophages (BMDM) cultured with apoptotic thymocytes for 16 hours. Representative genes associated with M1 TAM phenotype are in blue, type I IFN targets are in green. Horizontal dashed line corresponds to an adjusted p value of 0.01 and vertical dashed line to logFC of 0.33.

D. MFI of the expression of CD206 (CD11b⁺ CD206⁺) on the surface of untreated control BMDM (cntrl) or BMDM stimulated with apoptotic thymocytes for 24 hours, as measured by flow cytometry. Data are expressed as mean ±SEM of triplicate BMDM samples.

E. Quantification of the engulfment of apoptotic mCherry-spectrin⁺ LLC cells by Tim4^+/+ and Tim4^−/− BMDM, measured by flow cytometry. The percentage of engulfment (CD11b⁺ mCherry⁺ double-positive cells) was normalized to Tim4^+/+ BMDM. Data are expressed as mean ±SEM of triplicate BMDM samples.

F. Quantification of LAP in Tim4^+/+ and Tim4^−/− GFP-LC3^Tg BMDM cultured with apoptotic mCherry-spectrin⁺ LLC cells by flow cytometry. The MFI of GFP-LC3 in CD11b⁺ mCherry⁺ double-positive cells was normalized to untreated control BMDM. Data are expressed as mean ±SEM of triplicate BMDM samples.

G. Micrometastatic lesions on the lungs of Tim4^−/− and littermate Tim4^+/+ mice intravenously injected with B16F10 cells. Dots represent the total number of lesions per mouse, line indicates mean ±SEM

H. Tumor growth in Tim4^+/− (n=6 mice) and Tim4^−/− deficient (n=8 mice) littermates injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

I. Tumor growth in Tim4^+/− (n=5), Tim4^−/− (n=5) and Tim4^−/− Rubcn^−/− (n=3) mice injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

B-I. Significance was calculated with Student’s t-test (B, D-G) or ANOVA (H, I).

*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are from one (C, G) or are representative of at least two independent experiments (B, D-F, H-I).

See also Figure S4.

We further investigated a direct link between LAP deficiency and the response to dying cells in vitro using bone marrow-derived macrophages (BMDM). RNA sequencing (RNAseq) analysis identified increased expression of genes associated with a pro-inflammatory M1 macrophage polarization and of type I IFN-inducible genes in Rubcn^−/− BMDM stimulated with apoptotic cells, as compared to responses of wild type BMDM (Fig. 4 C). Gene Set Enrichment Analysis of the differentially expressed genes also revealed overall positive regulation of inflammatory responses in Rubcn^−/− BMDM, as well as induction of IL-6, TNF-α and IFNγ-dependent signaling pathways (Fig. S4). Metabolic reprogramming toward HIF-1α-dependent glycolysis has been identified as a driver of macrophage inflammatory activation (Tannahill et al., 2013). Interestingly, both glycolysis-and hypoxia-associated genes were also induced in Rubcn^−/− BMDM upon engulfment of dying cells, suggesting that LAP may coordinate metabolic reprogramming that drives macrophage inflammatory activation in response to dying cells (Fig. S4). Consistent with our observations of TAMs, engulfment of dying cells also induced expression of CD206, as described (Gonzalez et al., 2017; Soki et al., 2014) which was compromised in RUBCN-deficient BMDM (Fig. 4D).

Efficient engulfment of dying cells by macrophages requires recognition of phosphatidylserine exposed in the surface of dying cells by the T-cell immunoglobulin mucin protein 4 (TIM-4) (Miyanishi et al., 2007), and this participates in the induction of LAP (Martinez, et al., 2011) (Fig. 4 E, F). We observed that TIM4 deletion reduced the growth of metastatic B16F10 melanoma (Fig. 4 G) and engrafted LLC (Fig. 4 H). This effect was not altered by additional deletion of RUBCN (Fig. 4 I), suggesting that TIM4 and RUBCN may function in the same pathway. Overall, these results support a role for LAP in orchestrating an anti-inflammatory phenotype in macrophages in response to the engulfment of dying cells, suggesting that phagocytosis of dying tumor cells is a mechanism through which LAP regulates TAM function in the tumor microenvironment.

A STING-dependent IFN response is required for the anti-tumor effects of LAP deficiency in the tumor microenvironment

We next examined the role of the type I IFN response in LAP-deficient animals. Mice deficient in myeloid ATG5 or ATG7 again displayed delayed growth of implanted carcinoma cells, and this effect was completely lost when these animals were also ubiquitously defective in Interferon Alpha/Beta Receptor 1, IFNAR (Fig. 5A, B). Therefore, the type I IFN response is required for the anti-tumor effects of LAP-deficiency that we observed.

Figure 5. STING and type I IFN promote anti-tumor responses in the absence of LAP.

A. Tumor growth in Atg7^f/f Cre⁻ (n=7), Atg7^f/f Cre⁺ (n=4), Atg7^f/f Cre⁻ Ifnar^−/− (n=10), Atg7^f/f Cre⁺ Ifnar^−/− (n=10) mice injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

B. Tumor growth in Atg5^f/f Cre⁻ (n=6), Atg5^f/f Cre⁺ (n=5), Atg5^f/f Cre⁻ Ifnar^−/− (n=6), Atg5^f/f Cre⁺ Ifnar^−/− (n=9) mice injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

C-E. qPCR of the expression of Ifnβ in Rubcn^+/+, Rubcn^−/− and Rubcn^−/− Tmem173^−/− BMDM untreated (Cntrl), transfected with 1μg/ml LLC tumor cell DNA for 8h (C) or co-cultured with apoptotic LLC cells (BMDM/LLC 1:2 ratio) (D) or apoptotic thymocytes (BMDM/thymocyte 1:10 ratio) (E) for 20 h. Data are expressed as mean ±SEM of triplicate BMDM samples.

F. Tumor growth in Rubcn^+/+ (n=4), Rubcn^−/− (n=4) and Rubcn^−/− Tmem173^−/− (n=7) mice injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

G. Tumor growth in Tmem173^+/+ (n=8), Tmem173^−/− (n=4) and Rubcn^−/− Tmem173^−/− (n=6) mice injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

H. Tumor growth in Atg5^f/f Cre⁻ (n=5), Atg5^f/f Cre⁺ (n=5), Atg5^f/f Cre⁺Tmem173^−/− (n=8), mice injected with LLC cells subcutaneously. Data are expressed as mean ±SEM.

I-J. MFI of CD206 expression on the surface of monocytes (Mo) and mature TAM (TAM), as assessed by flow cytometry. Cells were isolated from subcutaneous LLC tumor engrafted into Rubcn^+/+, Rubcn^−/− and Rubcn^−/− Tmem173^−/− (I) or Atg5^f/f Cre⁻, Atg5^f/f Cre⁺ and Atg5^f/f Cre⁺Tmem173^−/− (n=4–5 mice per group as shown). Data are expressed as mean ±SEM. A-J. Data are representative of two independent experiments.

Significance was calculated with ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

The activation of the cytosolic DNA sensor STING induces type I IFN, but this does not occur upon engulfment of dying cells (Ahn et al., 2018; Corrales et al., 2015; Woo et al., 2014). Strikingly, we found that while transfection of DNA into Rubcn^+/+ or Rubcn^−/− macrophages similarly induced type I IFN expression (Fig. 5C), the engulfment of carcinoma cells or apoptotic thymocytes induced type I IFN expression only in LAP-deficient macrophages (Fig. 5D, E). This effect of engulfment of apoptotic cells in LAP-deficient macrophages was not observed in the macrophages lacking both RUBCN and STING (encoded by Tmem173). Therefore, LAP prevents the activation of STING upon engulfment of apoptotic cells.

We then examined the role of STING in the effects of LAP-deficiency on tumor growth. Mice lacking RUBCN (Fig. 5F) displayed delayed growth of LLC tumors, but this effect was lost upon ablation of STING. Deletion of STING did not affect tumor growth, and further deletion of RUBCN in STING-deficient mice had no effect (Fig. 5G). Similar results were obtained for animals lacking myeloid ATG5 (Fig. 5H). Furthermore, decreased expression of CD206 in the TAMs from these LAP-deficient animals was not observed in the absence of STING (Fig. 5 I, J). Therefore, the effect of LAP-deficiency on tumor growth and TAM polarization are dependent upon STING.

LAP in TAM suppresses T cell function in the tumor microenvironment

We next assessed whether tumor suppression in the absence of LAP required the activation of tumor infiltrating lymphocytes (TIL). We found that CD8⁺ TIL from LLC-bearing LysM-Cre⁺ Atg7^f/f, but not LysM-Cre⁺ Fip200^f/f mice were markedly more activated, producing higher levels of IFN-γ than their wild type mice (Fig. S5 A and Fig. 6 A, B). CD4⁺ T cells also showed higher activation and production of IFNγ in tumors from myeloid ATG5-deficient mice, but not in myeloid FIP200-deficient mice (Fig. S5 A and Fig. 6 C, D). Similar results were obtained for CD8⁺ and CD4⁺ TIL isolated from tumors implanted in Rubcn^−/− in comparison to Rubcn^+/+ animals (Fig. 6 B, D) and in CD8⁺ T cells in the lungs of K-ras^LSL-G12D p53^f/f animals transplanted with Rubcn^−/− bone marrow (Fig. 6 E–G). Moreover, CD8⁺ T cells isolated from tumors implanted in LAP-deficient Rubcn^−/− mice also produced higher levels of granzyme B (Fig. S5 B). CD8⁺ and CD4⁺ TIL isolated from B16F10 tumors similarly displayed an increased activation of effector function in animals lacking LAP (Fig. S6 A, B). Of note, ablation of LAP in TAM did not affect the infiltration or distribution of TIL, but instead modulated their function in the tumor microenvironment (Fig. 6 E, F; Fig. S5 C–E and Fig. S6 C, D). Further, we did not observe any impact of RUBCN deficiency on the infiltration of CD4⁺ Foxp3⁺ regulatory T cells (Fig. S5 F). Despite the increased activation of T cells in the tumor microenvironment of ATG5-deficient mice, we observed no differences in numbers of T cells expressing the exhaustion markers PD1 and TIM3 (Fig. S5 G).

Figure 6. LAP in TAM suppresses T cell function in the tumor microenvironment.

A. Representative flow cytometry plots of CD8⁺ IFNγ⁺ T cells (gated on CD45⁺ NK1.1⁻ B220⁻ CD8⁺), isolated from subcutaneous LLC tumors engrafted in wild-type and deficient mice as indicated.

B. Percentage (upper panels) and MFI (lower panels) of IFNγ⁺ in CD8⁺ TIL (n=4–5 mice per genotype as shown). Data are expressed as individual data points and mean bar ±SEM.

C. Representative flow cytometry plots of IFNγ⁺ CD4⁺ T cells (gated on CD45⁺ NK1.1⁻ B220⁻ CD4⁺), isolated from subcutaneous LLC tumors engrafted in wild-type and deficient mice as indicated.

D. Percentage (upper panels) and MFI (lower panels) of IFNγ⁺ in CD4⁺ TIL (n=4–5 mice per genotype as shown). Data are expressed as individual data points and mean bar ±SEM.

E-F. Absolute number (E) and percentage (F) of CD8⁺ T cells (CD45⁺ TCRβ⁺ CD8⁺) in the lungs of K-ras^LSL-G12D, p53^f/f bone-marrow chimeras, estimated by flow cytometry (n =5 mice per group). Data are expressed as individual data points and mean bar ±SEM.

G. Percentage of IFNγ⁺ CD8⁺ T cells in the lungs of K-ras^LSL-G12D, p53^f/f bone-marrow chimeras, as assessed by flow cytometry (n=5 mice per genotype). Data are expressed as individual data points and mean bar ±SEM.

H. Tumor growth in Atg5^f/f Cre^- and Atg5^f/f Cre⁺ mice upon depletion of T cells. Mice were injected with LLC cells subcutaneously and T cells were depleted with anti-CD4, anti-CD8 or treated with isotype control antibodies (Cre^-, n=4, Cre⁺ n=5, mice per treatment). Data are expressed as mean ±SEM.

I. Tumor growth in Rubcn^+/+ and Rubcn^−/− mice upon depletion of CD8⁺ T cells. Mice were injected with LLC cells subcutaneously and T cells were depleted with anti-CD8 or treated with isotype control antibodies (n=5 mice per treatment). Data are expressed as mean ±SEM.

J. Percentage of IFNγ⁺ production in CD8⁺ T cells isolated from subcutaneous LLC tumors engrafted in Atg7^f/f Cre^-/Cre⁺ and Atg7^f/f Cre^-/Cre⁺ Ifnar^−/− mice, by flow cytometry (n=3–5 mice per genotype as shown). Data are expressed as individual data points and mean bar ±SEM.

K, L. Flow cytometry analysis of IFNγ⁺ (K) and Granzyme B (L) production in CD8⁺ T cells isolated from subcutaneous LLC tumors engrafted in Rubcn^+/+, Rubcn^−/− and Rubcn^−/− Tmem173^−/− mice (n=3–4 mice per genotype as shown). Data are expressed as mean bar and individual data points; error bars represent s.e.m.

M. Percentage of IFNγ⁺ production in CD8⁺ T cells isolated from subcutaneous LLC tumors engrafted in Atg5^f/f Cre⁻, Atg5^f/f Cre⁺ and Atg5^f/f Cre⁺ Tmem173^−/− mice, by flow cytometry (n=4 mice per genotype). Data are expressed as individual data points and mean bar ±SEM.

A-G, J-M. Data are representative of two independent experiments. Significance was calculated using Student’s t-test (B, D-G) or ANOVA (J-M); *p<0.05, **p<0.01, ***p<0.001.

See also Figures S5 and S6.

To test the role of activated T cells in the anti-tumor effects of LAP-deficiency, we depleted T lymphocytes following tumor implantation by injection of anti-CD4 or anti-CD8 antibodies, which effectively depleted CD4⁺ and CD8⁺ T cells, respectively (Fig. S5H). Importantly, antibody-mediated, acute depletion of either CD4⁺ or CD8⁺ T cells impaired tumor suppression in LAP-deficient mice (Fig. 6 H, I and Fig. S5 I). These results indicate that T cell-mediated elimination of tumor cells is required for the tumor suppressive effects of LAP deficiency. Finally, we tested whether type I IFN signaling was required for the regulation of T cell function by LAP-deficient myeloid cells, as STING-dependent type I IF responses (Ahn et al., 2018; Corrales et al., 2015; Woo et al., 2014). We observed that the higher production of IFNγ by CD8⁺ T cells isolate from LLC tumors implanted in LAP-deficient LysM-Cre⁺ Atg7^f/f mice was dependent on IFNAR (Fig. 6 J) and in LAP-deficient Rubcn^−/− or LysM-Cre⁺ Atg5^f/f mice was dependent on STING (Fig. 6 K–M). This result supports the idea that STING-dependent type I IFN responses in LAP-deficient TAM promote tumor elimination by regulating T cell function.

Our findings support a model in which engulfment of dying cells through LAP defines anti-inflammatory polarization of macrophages in the tumor microenvironment, which as a result suppresses T cell function and thus promotes tumor tolerance. In the absence of LAP, TAM polarize toward inflammatory gene expression and trigger a STING-dependent type I IFN response, which promotes T cell effector function to restrict tumor growth.

DISCUSSION

Our studies on the role of LAP in the tumor microenvironment rely on genetic dissection of LAP-associated components in the myeloid compartment. Our conclusion that it is LAP rather than canonical autophagy in TAM that influences anti-cancer T cell responses is based on the pattern of effects, examining ablation of genes involved in both LAP and autophagy (e.g., Becn1, Atg5, Atg7), genes involved in autophagy but not LAP (e.g., Fip200, Atg14), and genes involved in LAP but not autophagy (e.g., Rubcn, Nox2). While our results support a role for LAP rather than autophagy, it remains formally possible that these components function in another biological process that does not involve phagocytosis. Our findings, however, that ablation of the engulfment receptor TIM4 has similar effects, which are not enhanced by additional ablation of RUBCN, and that several endosomal processes are not influenced by ablation of RUBCN, further support our conclusion that it is the engulfment of dying tumor cells, and the subsequent engagement of LAP, that promotes tumor immune tolerance.

In a system of Kras-driven lung oncogenesis, global ablation of ATG7 (required for both canonical autophagy and LAP) in both tumor and normal tissue restricted tumor growth to a greater extent than tumor-specific ablation suggesting a role of host autophagy in promoting tumor growth (Karsli-Uzunbas et al., 2014). Similarly, ablation of autophagy in the microenvironment of pancreatic cancer (by inducible expression of a dominant negative ATG4, required for both autophagy and LAP) impacts tumor growth (Sousa et al., 2016; Yang et al., 2018). While the role of adaptive immunity in these effects was not directly examined, it is possible that ablation of LAP in cells of the tumor microenvironment contributed to these results. Depletion of ATG7 in genetically-induced melanoma was observed to promote anti-cancer T cell responses (Michaud et al., 2014). Although ATG7 is involved in both canonical autophagy and LAP (Sanjuan et al., 2007), this raises the possibility that LAP may also occur and influence immunomodulatory signaling in tumor cells, as suggested for other cancers (Kim and Overholtzer, 2013).

A recent study ablated VPS34 (required for both LAP and autophagy) using CD11c-Cre, which is active in CD8⁺ dendritic cells (CD8⁺ DC) that present tumor antigens to CD8⁺ T cells (Parekh et al., 2017). In this particular setting, they found a general loss of CD8⁺ DC, a reduction in cross-priming, and increased lung metastases in animals injected intravenously with B16F10 cells. We have observed no such loss of CD8⁺ DC in animals lacking RUBCN (Martinez et al., 2016), and therefore this effect was unlikely to be due to loss of LAP. In contrast to CD11c-Cre, LysM-Cre is not active in CD8⁺ DC (Martinez et al., 2016), and therefore it is likely that the effects we observed upon ablation of genes shared by autophagy and LAP pathways (including VPS34) may have depended on “sparing” CD8 ⁺ DC from the depleting effects of autophagy deficiency.

While our studies implicate LAP in TAM in preventing anti-cancer immunity, it remains possible that other cell types can also participate in this effect. LysM-Cre, used in our studies, is penetrant in other myeloid cell types, and our studies do not rule out roles for other myeloid cells. While we failed to observe an IFN signature in tumor-infiltrating CD11b⁺ cells other than monocytes and TAM, it remains possible that other cell types also produce IFN upon LAP ablation. Further, it will be interesting to assess effects of ablation of LAP in other cells of the tumor microenvironment.

While STING function can be restricted by autophagy (Prabakaran et al., 2018), if this mechanism were responsible for increased IFN signaling in our system, we should have expected effects of deletion of autophagy specific components, which we did not observe. Instead, we found evidence that STING is activated in LAP-deficient cells that have engulfed apoptotic cells, and that this is required for the anti-tumor effects. We speculate that the degradation of the DNA of engulfed tumor cells by LAP-dependent lysosomal fusion prevents STING activation. Consistent with this idea, animals lacking lysosomal DNAse II display lethal interferonopathy dependent on STING (Ahn et al., 2012).

Type I IFN signaling promotes CD8⁺ T cell function, expansion, and memory during viral responses (Kohlmeier et al., 2010; Kolumam et al., 2005), and a similar mechanism may function in tumors. Despite the presence of active anti-cancer T cell immunity in LAP-deficient animals, we observed no differences in expression of the expression of the exhaustion markers, PD-1 and TIM-3 in TIL. Of note, a significant portion of the anti-viral responses during LCMV or influenza infection are mediated by CD8⁺ T cells expressing PD-1 and TIM-3 (Rutigliano et al., 2014; Sharma et al., 2011), and thus it is possible that type I IFN signaling promotes activity in these cells, and may represent acute responses of infiltrating T cells prior to exhaustion.

Our study suggests that targeting LAP-specific proteins may be a potential therapeutic strategy. Interfering with LAP may allow a “reprogramming” of the function of tumor-associated macrophages to promote the elimination of cancer cells without interfering with canonical autophagy processes that contribute to tumor suppression by other components of the tumor microenvironment, perhaps including lymphocyte function and antigen presentation.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Douglas Green (douglas.green@stjude.org).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice

Atg7^f/f mice (kindly provided by Masaaki Komatsu at The Tokyo Metropolitan Institute of Medical Science) were bred to LysM-Cre⁺ mice (kindly provided by Peter Murray, St. Jude Children’s Research Hospital). Rubcn^−/− mice were previously generated in C57BL/6 background using CRISPR/Cas9 gene editing technology (Martinez et al., 2015). Ifnar^−/− were purchased from Jackson Laboratories and crossed to Atg7^f/f LysM-Cre^+/−. Tmem173^−/− were purchased from Jackson Laboratories and crossed to Atg5^f/f LysM-Cre⁺ and Rubcn^−/− mice. Nox2^−/− were purchased from Jackson Laboratories. Other mice were kindly provided: Ulk1^−/− (Mondira Kundu, St. Jude Children’s Research Hospital), Becn1^f/f LysM-Cre⁺ (Edmund Rucker, University of Kentucky), Atg5^f/f LysM- Cre⁺ (Thomas A. Ferguson, Washington University), Fip200^f/f LysM-Cre⁺ (Jun-Lin Guan, University of Michigan), Atg14^f/f LysM-Cre⁺ and Atg16l1^f/f LysM-Cre⁺ (Herbert Virgin, Washington University), Vps34^f/f LysM-Cre⁺ (Richard Flavell, Yale University), Tim4^−/− mice (Vijay Kuchroo, Harvard University) and K-ras^LSL-G12D p53^f/f (Mark Hatley, St. Jude Children’s Research Hospital). All mice but K-ras^LSL-G12D p53^f/f were previously crossed to C57BL/6 GFP-LC3^Tg mice (Martinez et al., 2016).

Age and sex matched cohorts of females (8 weeks old) and males (6 weeks old) littermates were used in subcutaneous tumor graft experiments. For tumor metastasis, 12 weeks old females and 10 weeks old males were used. Appropriate littermate controls were used in all experiments. The St. Jude Institutional Animal Care and Use Committee approved all procedures in accordance with the Guide for the Care and Use of Animals. All mice were housed in pathogen-free facilities, in a 12-hour light/dark cycle in ventilated cages, with chow and water supply ad libitum.

Cells

Lewis carcinoma cell line (LLC, derived from C57BL/6, sex information not available) were purchased from ATCC. MC38 adenocarcinoma cell line (derived from female C57BL/6) and B16-F10 melanoma cell line (derived from male C57BL/6) were kindly provided by Hongbo Chi (St. Jude Children’s Research Hospital). RAW264.7 cells (derived from male BALB/C) were obtained from ATCC. Cells were maintained in complete DMEM media (10% fetal bovine serum (FBS), 200 mM L-glutamine and 100 units/ml penicillin-streptomycin). All the cell lines used in tumor studies were confirmed as mycoplasma negative using MycoAlert Mycoplasma Detection kit (Lonza).

For preparation of bone marrow-derived macrophages (BMDM), sex-paired cohorts of male or female mice at 6 to 12 weeks of age were euthanized and bone marrow cells were harvested from the femurs and differentiated in DMEM (Gibco) containing 20% FBS, 200 mM L-glutamine, 100 units/ml penicillin-streptomycin, 20 ng/ml recombinant human M-CSF (Biolegend) for 6 days. BMDM were harvested and seeded on tissue culture plates 1 day before stimulation and maintained in complete DMEM media. All cells used in this study were cultivated at 37° C with 5% CO₂.

METHOD DETAILS

Generation of Cell lines

LLC expressing mCherry-spectrin (LLC-mCherry-spectrin) were produced by retroviral transduction with retroviral plasmid expressing mCherry-spectrin (pMX-mCherry-spectrin-IRES-blasticidin). β-spectrin PH (pleckstrin homology) domain (Wang et al., 1996), a kind gift from Gerry Shaw (University of Florida), was cloned in a pMX-IRES-blasticidin (Cell Biolabs) vector expressing mCherry in tandem using MluI and NotI. The Phoenix-AMPHO cells (ATCC) maintained in complete DMEM media were used to package retroviruses.

RAW264.7 Rubcn^−/− were generated by CRISPR-Cas9 technology by transient transfection with two different pSpCas9(BB)-2A-GFP (PX458) (Addgene, a gift from Feng Zhang) constructs. Oligonucleotide pairs Rubcn-sgRNA1 (position 32858866–32858885 of C57BL/6 chromosome 16) and Rubcn-sgRNA2 (position 32824711–2824730 of C57BL/6 chromosome 16) were annealed and cloned with BbsI. GFP⁺ transfected cells were single-cell sorted, and deletion was confirmed by sequencing.

Heterotopic graft tumor models

1.0 × 10⁵ LLC cells, 2.5 × 10⁵ B16F10 cells and 5.0 × 10⁵ MC38 cells were injected subcutaneously into the middle lower back of the mice. Mice were identified by ear tag and experimenters were blinded to their genotype. Tumor growth was measured using a digital caliper and tumor volume was estimated as length x width² × 0.5 (Euhus et al., 1986; Tomayko and Reynolds, 1989). Tumor metastasis was assessed by tail vein injection of 1.0 × 10⁵ B16F10 cells. Mice were euthanized 21 days post-injections, their lungs excised and all visible tumor spots onto the surface of all lobes were counted.

T cell depletion

200 µg/mouse of rat anti-CD4 (clone GK1.5 BioXCell), rat anti-CD8 (clone 2.43 BioXCell) or rat IgG2b anti-KLH isotype control (clone LTF2 BioXCell) antibodies diluted in phosphate-buffered saline (PBS) were intraperitoneally injected every 5 days, starting at the day prior to inoculation of tumor cells.

Flow cytometry analysis and purification of graft tumor-infiltrating immune cells

For isolation of graft LLC and B16F10 tumor infiltrating immune cells, 14–16 days after implantation mice were euthanized and tumors were harvested. For LLC grafts, extracted tumors were minced and placed in digestion media (Hank’s buffer (Gibco) supplemented with 75 μg/ml Liberase TM (Roche) and 1mg/ml Deoxyribonuclease I (Worthington) for 30 min incubation at 37°C 5% CO ₂. RPMI (Gibco) supplemented with 10% fetal bovine serum and 0.1% β-mercaptoethanol was added to digestion reaction, cells were then dissociated with a syringe plunger and by pipetting up and down and passed through a 70 μM strainer, followed by lysis of red blood cells. B1610 tumors were directly dissociated with a syringe plunger, pipetted up and down and passed through a 70 μM strainer. Tumor cells were then washed, and collected at the 40%−80% interface of a Percoll gradient.

For flow cytometry, single cell suspensions were first blocked with anti-mouse CD16/CD32 (Fc Block BD Biosciences) at room temperature for 5 minutes prior to staining. Surface antigens were stained with antibodies for 10 minutes on ice. Cells were also stained with Zombie Violet Viability kit (Biolegend) according to manufacturer’s instructions for discrimination of dead cells and cell debris. For assessment of IFNγ and granzyme-B production by TIL, purified cells were incubated with ionomycin (1 μg/ml) and phorbol myristate acetate (PMA, 20 ng/ml) for 4 h at 37 °C 5% CO₂ in the presence of protein transport inhibitor (GolgiStop, BD Biosciences). Surface staining was carried out as described above; cells were fixed and permeabilized using Intracellular Fixation & Permeabilization Buffer Set (eBiosciences) and intracellular staining carried out according to the manufacturer’s instructions. V500 anti-CD45 (clone 30-F11), BV570 anti-CD11b (clone M1/70), APC anti-CD86 (clone GL-1), Alexa Fluor anti-MHC II (clone M5/114.15.2), PE Dazzle 594 anti-CD206 (clone C068C2), PE anti-Ly6G (clone 1A8), Pacific Blue anti-Ly6C (cloneHK1.4), PE-Dazzle 594 anti-NK1.1 (clone PK136), BV605 anti-B220 (cloneRA23–6B), Pacific Blue anti-CD4 (clone RM4–5), APC eFluor 780 anti-CD8 (clone 53–6.7), PE anti-Granzyme B (clone 16G6), PerCp-Cy5.5 anti-IFNγ (clone XMG1.2) , PE-Cy7 anti-Foxp3 (clone FJK-16s), PE-Cy7 anti-Tim3 (clone RMT3–23), APC rat anti-PD-1 (clone J43) were used. Flow analysis was performed in a SP6800 Spectral Cell Analyzer (Sony) and data processing was performed using FlowJo software (FlowJo, LLC). MFI values are expressed as the product of the frequency of a subset and its mean fluorescent intensity (Darrah et al., 2007).

For purification of myeloid infiltrating cells, surface antigens were stained with antibodies for 10 minutes on ice (using V450 anti-CD11b, PerCP-Cy5.5 anti-Ly-6G, PerCP-Cy5.5 anti-SiglecF and PerCP-Cy5.5 anti-NK1.1) and sorting of CD11b⁺ Ly-6G⁻ Siglec-F⁻ NK1.1⁻ fraction was performed in an iCyte Reflection system.

Quantitative PCR of myeloid infiltrating cells

For quantitative PCR, total RNA was isolated from purified myeloid infiltrating cells with RNeasy Micro kit (QIAGEN) and cDNA was synthesized with Reverse Transcription Supermix (Bio-Rad). cDNA was pre-amplified using PreAmp Supermix (Bio-Rad). Quantitative amplification was performed with Universal SYBR Green (Bio-Rad) in an ABI Prism 7900-HT system (Applied Biosystems). Primers used (Gbp2, Ifit3, Ifnb, Il1b, Stat1) are list on Key Resources Table.

Imaging of graft tumors

For in situ analysis of graft tumor by microscopy, subcutaneous LLC tumors were excised 14 days after implantation and fixed for 24 hours using 2% PFA in PBS containing 0.3% Triton-100 and 0.1% DMSO. Samples were cryoprotected using 20% sucrose in PBS prior to cryosectioning. Slides were blocked with 1% BSA and 0.05% Tween-20 in Tris-buffered saline (TBS, 20mM Tris, pH 8.0, 100mM NaCl) for 1 hour at room temperature prior to overnight incubation in blocking buffer containing primary antibodies Alexa Fluor 647-conjugate rat anti-CD11b (clone M1/70; Biolegend), rabbit anti-mCherry (Abcam) or rat anti-CD3 (clone 17A2, Fisher) labelled with CF543 (Biotium) as indicated in the legend of the figures, at 4°C. Slides were washed in TBS before application of secondary antibody (Biotium) for 1 hour at room temperature prior to mounting with Prolong gold hardset media containing DAPI (Thermofisher). For Fig. 4A, samples were imaged using a Marianas spinning disk confocal microscope (Intelligent Imaging Innovations) equipped with a 40X 1.3NA objective and EMCCD camera, and analyzed using Slidebook 6 software (Intelligent Imaging Innovations). For Fig. S5B, Images were acquired with Nikon TiE microscope using a 25X, 0.75 NA Plan Apo objective. Multiple images at adjacent positions were acquired with a 10% overlap and stitched in postprocessing to create an enlarged format frame. Analysis was performed using NIS Elements (Nikon Instruments Inc.), wherein the number of CD3 objects were determined by intensity thresholding.

Generation of K-ras^LSL-G12D p53^f/f chimeras

For K-ras^LSL-G12D p53^f/f bone marrow chimeras, 1.0 × 10⁶ bone marrow cells from Rubcn^+/+ or Rubcn^−/− were adoptively transferred through retro-orbital injection into lethally irradiated 4 weeks old female and male K-ras^LSL-G12D p53^f/f littermates. Recipient mice were irradiated at 800 RAD one day before, and an additional 400 RAD on the morning of the transfer. Donor mice were euthanized and bone marrow cells were harvested from the femurs. Red blood cells were lysed, and remaining cells were washed, passed through a 0.45 μm cell strainer, counted and resuspended in sterile PBS prior to transfer. The animals were given amoxicillin-supplemented water (50 mg/ml) for two weeks after irradiation and bone marrow transfer to prevent potential infection. At 8 weeks of age, tumorigenesis was induced in the recipient mice by inhalation of 2.5 × 10⁷ PFU adenovirus-Cre particles (Adv-Cre) (University of Iowa Viral Vector Core).

Analysis of K-ras^LSL-G12D p53^f/f chimeras

For analysis of tumors from K-ras^LSL-G12D p53^f/f chimeras, mice were euthanized at 16 weeks of age (8 weeks after tumor induction by Adv-Cre). The lungs were excised, the left middle lobe isolated for cell isolation and the other lobes inflated by intra-tracheal injection of fixative (2% PFA in PBS containing 0.3% Triton-100 and 0.1% DMSO). The left lobe was then excised for ex vivo MRI, and the right cranial lobe was processed for cryosectioning as described above, followed by standard H&E staining procedure. For Fig. 1G, samples were imaged with a Leica S9i steroscope. Ex-vivo MRI was performed with a 7 T Bruker ClinScan system (Bruker BioSpin MRI GmbH, Germany) equipped with a 12S gradient coil. A 2-channel surface coil was used for MR imaging. Each specimen was placed in 10% agarose gel for MR data acquisition. 3D T1-weighted Fast Low Angle SHot (FLASH) images were acquired (TR/TE = 500/5.04 ms, FOV = 20.6 ×30mm, matrix = 528 × 768, number of acquisitions = 32, thickness = 0.07 mm, scan time = 9.5 – 13.1 hr, depending on the number of slices which varied between 64–88). Segmentation of individual lesions and final tumor volume measurements were obtained using OsiriX MD (Pixmeo, Switzerland). For cell isolation and flow cytometry analysis, excised lung lobes were placed in digestion media and mechanically disrupted using gentleMACS Dissociator (Miltenyi Biotec). Cells were passed through a 70 µM strainer, followed by lysis of red blood cells. Cells were treated and stained for flow cytometry as described above. V500 anti-CD45 (clone 30-F11), BV570 anti-CD11b (clone M1/70), PE-Cy7 anti-Ly6G (clone 1A8), APC anti-F4/80 (clone M1/70), PE Dazzle 594 anti-CD206 (clone C068C2), APC e-Fluor 780 anti-TCRβ (clone H57–597), Pacific Blue anti-CD4 (clone RM4–5), Alexa Fluor 700 anti-CD8 (clone 53–6.7), APC anti-IFNγ (clone XMG1.2) were used. Flow analysis was performed in a SP6800 Spectral Cell Analyzer (Sony) and data processing was performed using FlowJo software (FlowJo, LLC). All cells were initially gated as Total cells/Live cells (Zombie Violet⁻)/Single cells prior to further gating. MFI values are expressed as the product of the frequency of a subset and its mean fluorescent intensity (Darrah et al., 2007).

In vitro experimental procedures

For immunoblot analysis of p62 degradation, BMDM (5.0 × 10⁵) were treated with bafilomycin (50 nM) for 5 hours. The cells were then washed with PBS and lysed in RIPA buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 150 mM NaCl, 1% Nonidet P-40, 1% deoxycholate, and 0.1% SDS) supplemented with protease inhibitor cocktail (Roche). Precleared lysates and supernatants were boiled in Laemmli buffer, resolved by SDS-page, and transferred to 0.45-µm nitrocellulose membranes. Rabbit anti-p62 (Sigma), rabbit anti-ATG14/FIP200/BECN1/VPS34/ATG5/ATG7/RUBCN (all from Cell Signaling), mouse anti-ATG16L1 (MBL) and HRP-conjugated goat anti-actin (Santa Cruz), and species-specific horseradish-conjugated secondary antibodies were used for antigen detection. Immunoblot for detection of RUBCN in RAW264.7 cells were performed likewise.

Trafficking assays in BMDM by imaging were performed as follows. BMDM (5.0 × 10⁵) were plated in 4 well chamber slides (IBIDI). For transferrin endocytosis assay, BMDM were pre-incubated at 37°C in serum free DMEM for 3 0 min. Transferrin Alexa Fluor 568 conjugate (Invitrogen) were added (50 μg/ml) and the cells were incubated at 37°C for 30 min. Cells were washed sequentially with PBS pH 5.5 and pH 7.4 prior to fixation with 4% paraformaldehyde. For IgG complex endocytosis assay, human IgG (Sigma) aggregates were prepared by incubating 10 mg/ml at 62°C for 20 min., followed by removal of aggregated IgG by centrifugation. BMDM were incubated with IgG complexes (1 mg/ml) for 30 min. Cells were washed with PBS, fixed and permeabilized with 0.1% Triton X-100 prior to staining with APC anti-human IgG Fc antibody (clone HP6017, Biolegend). For Dextran pinocytosis assay, BMDM were incubates with Texas Red-labelled Dextran 10,000 mW (Invitrogen) (500 μg/ml) for 30 min, washed with PBS and fixed. All samples were stained with Alexa Fluor 488 wheat germ agglutinin (Invitrogen) and Hoechst 33258 (Invitrogen) prior to imaging. Samples were imaged using a Marianas spinning disk confocal microscope (Intelligent Imaging Innovations) equipped with a 100X,1.45 NA objective and EMCCD camera, and analyzed using Slidebook 6 software (Intelligent Imaging Innovations).

Trafficking assays in wild-type and Rubcn^−/− RAW264.7 cells were performed as follows. For transferrin endocytosis, cells were plated and treated as described above. For endosome maturation assay, RAW264.7 cells (5.0 × 10⁵) plated in chamber slides were loaded with Dextran Oregon Green 488 (500 ng/ml) (Invitrogen) overnight, washed with PBS and chased for 4 h in complete media prior to addition of DQ-Red BSA (Invitrogen) (10 μg/ml) for 1 h. The cells were then washed with PBS, topped with fresh media and immediately imaged by confocal microscopy. For EGF receptor degradation assay, RAW cells were pre-incubated in serum free media overnight. Recombinant EGF (Millipore) diluted in serum free media (100 ng/ml) was added to the BMDM, and samples were washed and lysed in RIPA buffer supplemented with protease inhibitor cocktail (Roche) at the specified time points. Immunoblotting with Rabbit anti-EGFR (clone EP38Y, Abcam) and anti-Actin HRP-conjugated was carried out as described above.

For stimulation of BMDM with apoptotic cells, apoptosis was induced in wild-type C57BL/6 thymocytes by UV irradiation (200 mJ/cm²) or by treatment with staurosporine (1 μM) for 3 hours. For labelling of thymocytes, the cells were incubated with Cell Trace Violet (Invitrogen) prior to apoptotic stimuli, following the manufacturer’s instructions. For LLC and mCherry-spectrin LLC cells, apoptosis was induced by a combination of ABT737 (Selleck Chemicals) and Mcl1 inhibitor S63845 (Cayman) (5 μM) for 2 hours. Apoptotic cells were thoroughly washed with PBS, ressuspended in fresh DMEM complete media and spun down onto macrophages. Prior to sample collection or analysis, wells were washed three times with PBS to eliminate non-engulfed dying thymocytes.

For qPCR analysis, BMDM (5.0 × 10⁵) were stimulated with apoptotic cells or transfected with tumor DNA as described in the legend of the figures. LLC tumor DNA was extracted with DNeasy purification kit (QIAGEN) and transfected with Lipofectamine LTX (Invitrogen). Sample lysis and RNA extraction was performed using RNeasy Mini extraction kit (QIAGEN), and cDNA synthetized with M-MLV reverse transcriptase (Invitrogen) in the presence of RNAse inhibitor (RNAse OUT Invitrogen). Quantitative amplification was performed with SYBR Green PCR Master Mix (Applied Biosystems) in an ABI Prism 7900-HT system (Applied Biosystems).

For flow cytometry analysis, surface antigens were stained for 10 minutes on ice. APC anti-CD11b (clone M1/70) and PE Dazzle anti-CD206 (clone C068C2) were used. For the assessment of LAP by measurement of GFP-LC3 lipidation assessment, the BMDM were incubated with digitonin diluted in PBS (200 μg/ml) for 10 min. on ice after surface staining, then washed in PBS 1% BSA 1mM EDTA. Flow analysis were performed in a a LSFortessa analyzer (BD) and data processing was performed using FlowJo software (FlowJo, LLC).

Single cell RNA seq

Myeloid infiltrating cells from LLC tumors engrafted in Rubcn^+/− and Rubcn^−/− (n=3 per genotype) were purified by cell sorting as described above and polled. The single-cell suspension was resuspended in phosphate-buffered saline containing 0.04% bovine serum albumin at the concentration of 800–1000 cells/μl. In total 8800 cells were loaded into each well of a Chromium single cell capture chip (Chromium Single Cell A Chip Kit, 10X Genomics) with an average recovery of 4000 cells. The captured single cells on droplets were processed for cell lysis, reverse transcription followed by library amplification (Chromium Single Cell 3’ Library and gel Bead Kit v2, 10X Genomics) and indexing as per the manufactures protocol. The libraries were sequenced on an Illumina Nextseq500 (Next Seq 500/550 High Output Kit v2, Illumina). Cell Ranger (10X Genomics) was used to process the raw sequencing data and align the reads to the mm10 reference genome. The resulting UMI count matrices were concatenated. This merged count matrix was filtered as follows. All mitochondrial and ribosomal protein genes were removed, as were all genes with no reads in any cell from either sample. Cells were removed if they had either less than 3500 UMIs, or greater than 15000 UMIs. The data was normalized by dividing the UMI counts of each gene within a cell by the total number of UMIs in the cell, scaled my multiplication of the median total UMI count across cells from the filtered UMI count matrix, and log2-scaled after the addition of 1 to prevent taking logarithms of zero values. The remaining cells were visualized in 2-dimensions using the Barnes-Hut implementation of t-SNE applied to the first 50 principal components of the 1000 most overdispersed genes in the filtered count matrix. Overdispersion was calculated as performed elsewhere (Zheng et al., 2017). Cells were hierarchically clustered using the Ward D2 method applied to the pairwise Euclidean distance between the normalized expression profiles of all cells. Cells were assigned to clusters using the hybrid method of dynamicTreeCut (Langfelder et al., 2008). All primary scRNA seq data are provided in GEO (Accession number GSE118019).

RNA seq

Total RNA was extracted from bulk staurosporine-induced apoptotic thymocytes and from Rubcn^+/+ and Rubcn^−/− BMDM, either unstimulated or stimulated with apoptotic thymocytes for 16 hours. RNA quality was assessed by 2100 Bioanalyzer RNA 6000 Nano assay (Agilent). Libraries were prepared using TruSeq Stranded mRNA kits (Illumina) and subjected to 100 cycle paired-end sequencing on the Illumina HiSeq platform.

Reads were aligned to the mouse genome (GRCm38.p5) first using STAR v2.5.3a (Dobin et al., 2013) with splice junctions inferred from Gencode M15 gene annotations (Mudge and Harrow, 2015). Unmapped reads were then re-aligned using BWA v0.7.12-r1039 (Li and Durbin, 2009). Results from each algorithm were merged, yielding >99.5% of reads mapped for each sample, and featureCounts v1.5.1 (Liao et al., 2014) was used to assign read counts to Gencode gene-name annotations, with predicted genes and processed pseudogenes subsequently excluded. We then used edgeR (Robinson et al., 2010) to filter out genes with fewer than 5 counts per million (CPM) and to normalize counts to trimmed mean of M-values (TMM; (Robinson and Oshlack, 2010). Differential expression was assessed with linear models using the limma package (Ritchie et al., 2015), and the false discovery rate (FDR) was used to correct for multiple testing (Benjamini and Hochberg, 1995). Volcano plots were produced using the ggplot2 package (Wickham, 2009). Differential gene ranks were produced by multiplying the log₂ fold change (LFC) by the inverse of the unadjusted p-value, and Gene Set Enrichment Analysis (GSEA) was performed on ranked lists using GSEAPreranked with the Hallmark gene set database and with the number of permutations set to 5,0000, the enrichment statistic set to “classic”, and all other parameters at default settings (Subramanian et al., 2005). All primary RNA seq data are provided in GEO (Accession number GSE118019).

QUANTIFICATION AND STATISTICAL ANALYSIS

Please refer to the legend of the figures for description of samples (mice, cells or experimental replicates) and sample sizes. No statistical tests were used to estimate sample size. Data were plotted and analyzed with GraphPad Prism 7.0 software (GraphPad Software). Statistical significance was calculated with either unpaired two-tailed Student’s t-test or ANOVA, as specified in the legend of the figures. Welch’s correction was applied for unpaired Student’s t-test if F test indicated significant difference in variance between the compared groups. Differences were considered statistically significant when the p-value was less than 0.05.

• LAP occurs as dead cells are engulfed in the tumor microenvironment.

• TAM lacking LAP display M1 characteristics and compromise tumors.

• TAM lacking LAP engage STING-dependent type I IFN production.

• TAM lacking LAP promote anti-tumor T cell responses.

• Impairment of LC3-associated phagocytosis in myeloid cells of the tumor microenvironment has anti-tumor effects