A TLR3 Ligand Reestablishes Chemotherapeutic Responses in the Context of FPR1 Deficiency

In contrast to widespread belief, chemotherapy with cytotoxic agents does not act only through direct tumor cell killing. Rather, several chemotherapeutic agents that have been particularly successful in the clinics are able to trigger an anticancer immune response, explaining how they can mediate long-term effects that last beyond therapeutic discontinuation (1, 2). For example, anthracyclines [such as doxorubicin (DOXO) and mitoxantrone (MTX)] and cyclophosphamide (CTX; an alkylating agent) can stimulate immunogenic cell death (ICD), hence killing malignant cells in such a fashion that they attract dendritic cells (DC) into their vicinity, engulf tumor-associated antigens, and present them to cytotoxic T lymphocytes for launching an immune attack against residual cancer cells (3–5).

One of the danger-associated molecular patterns that are released from cancer cells succumbing to chemotherapy is the cytoplasmic protein annexin A1 (ANXA1; ref. 6). ANXA1 subsequently binds to formyl peptide receptor 1 (FPR1), a G protein–coupled receptor expressed by myeloid cells such as DCs (5, 7). The dialogue between dying cancer cells and DCs, mediated by the ANXA1–FPR1 interaction, chemotactically guides mobile immature DCs toward the dying cells, ultimately facilitating a docking phase that culminates in the phagocytotic uptake of portions of the tumor cells by DCs. This latter process requires calreticulin exposure on the cancer cell as an “eat-me” signal (6, 8, 9).

The functional interaction between ANXA1 and FPR1 is important for the chemotherapy-induced cancer immune dialogue. Thus, the absence of ANXA1 (in cancer cells) or of FPR1 (in immune cells) abolishes the capacity of anthracycline-based chemotherapy to trigger anticancer immunity and to induce efficient tumor growth control (5, 10). This is suggested by preclinical studies in mice as well as by clinical observations showing that low ANXA1 expression in multiple different human cancer types (breast, colorectal, lung, and kidney) is associated with a scarce infiltration by activated DCs (6). However, the expression levels of FPR1 and ANXA1 correlate with tumor progression in patients with gastric cancer (11, 12), suggesting tumor type–specific effects. A loss-of-function allele of FPR1 (rs867228, E346A) that has a high prevalence across all ethnic groups (between 20% and 30%) is associated with poor prognosis in patients with breast cancer treated with anthracycline-based adjuvant chemotherapy (5, 10). DCs from individuals bearing rs867228 either in heterozygosity or in homozygosity show a reduced interaction with dying tumor cells in microfluidic chambers (5).

Of note, FPR1 is the receptor for Yersinia pestis entry in immune cells, the bacillus causing plague, meaning that FPR1-deficient mice are resistant against this infection (13). Although there is no formal proof for this conjecture, it appears plausible that loss-of-function alleles of FPR1 are prevalent in the population because they favor survival in the context of otherwise lethal epidemics, at the cost though of a relative immune defect. Indeed, a loss-of-function allele of Toll-like receptor 4 (TLR4) that is widespread among Caucasians (but not Africans and African Americans) has been suggested to reduce inflammatory reactions but to increase susceptibility to infection (14). FPR1 is the receptor for several ligands, beyond ANXA1 (5, 7) and Y. pestis (13). Thus, FPR1 binds formylated peptides contained in bacteria (15) and mitochondria (16). In addition, FPR1 binds cathepsin G (17) and family with sequence similarity 19 [chemokine (C-C motif)–like], member A4 (FAM19A4; ref. 18). Hence, this pattern recognition receptor (PRR) may affect multiple distinct immune and inflammatory responses.

Intrigued by these premises, we decided to explore strategies for reinvigorating the deficient anticancer immune response caused by FPR1 deficiency. Here, we show that immunotherapy with a TLR3 ligand can overcome the functional consequence of the FPR1 defect. Moreover, we report the finding that FPR1 loss-of-function allele rs867228 is associated with an earlier diagnosis of human breast, colorectal, esophageal, and head and neck squamous cell carcinomas.

To determine the impact of a variety of candidate genes that might be involved in (immuno) therapeutic efficiency (Supplementary Table S1), we took advantage of an assay for antigen presentation by inducible immortalized DCs (iniDC) obtained by doxycycline (Dox)-inducible expression of a transgene encoding the Simian Virus 40 (SV40) large T-cell antigen, which inactivates TP53 and RB (19). CRISPR/Cas9-expressing iniDCs were transfected with guidance RNAs (gRNA) targeting a variety of candidate genes to cause their inactivation, one by one. Then iniDCs were deimmortalized by the removal of Dox and the synthetic glucocorticoid dexamethasone (Dex) from the culture medium to obtain de-iniDCs. Such de-iniDCs were loaded with the model antigen ovalbumin (OVA), washed, and then evaluated for their capacity to present the OVA-derived SIINFEKL peptide to B3Z hybridoma cells and to induce the secretion of IL2 (Fig. 1A). We found that the knockout of some genes enhanced antigen presentation, as observed for C-type lectin domain family 4, member a2 (Clec4a2), confirming a prior report on the immunosuppressive function of this gene (20), whereas the knockout of other genes reduced DC function. The five strongest loss-of-function phenotypes resulted from the knockout of three genes that have been previously described as implicated in chemotherapy-elicited anticancer immune responses: Tlr3, Tlr4, and Fpr1 (the latter causing the strongest defect; refs. 5, 10, 21) as well as CD1d1 antigen (Cd1d1, reportedly involved in antigen presentation for the education of natural killer T cells; ref. 22) and transmembrane protein 175 (Tmem175, which is important for lysosomal function; ref. 23; Fig. 1B). Although wild-type (WT) de-iniDCs injected into methylcholanthrene-induced fibrosarcomas (MCA205) evolving orthotopically (under the skin) on immunocompetent C57BL/6 mice reduced their growth, Fpr1^−/− de-iniDCs failed to do so (Fig. 1C). The tumor growth–reducing effect of WT de-iniDCs depended on T cells and the local presence of the FPR1 ligand ANXA1 because systemic depletion of CD4⁺ and CD8⁺ T cells with suitable antibodies or local injection of a neutralizing anti-ANXA1 antibody abolished this anticancer effect (Fig. 1D and E).

Concentrating on the defective antigen presentation caused by the Fpr1 knockout, we investigated whether addition of PRR agonists would overcome this functional defect caused by the absence of FPR1 (Fig. 2A). Among six TLR and two nucleotide-binding oligomerization domain 2 (NOD2)agonists, only one, polyinosinic:polycytidylic acid (pIC), which is a TLR3 agonist, was capable of stimulating antigen presentation by Fpr1^−/− DCs in a dose-dependent fashion (with a maximum effect at 0.5 μmol/L), elevating their function to that of control DCs (that only express Cas9 but were not transfected with gRNAs; Fig. 2B; Supplementary Fig. S1A and S1B). pIC failed to stimulate antigen presentation by Fpr1^−/− Tlr3^−/− double-knockout de-iniDCs (Fig. 2C). Moreover, pIC-stimulated Fpr1^−/− de-iniDCs recovered their capacity to reduce tumor growth in vivo (Fig. 2D).

In conclusion, FPR1 deficiency causes a defect in DC-mediated antigen presentation that can be overcome by provision of a synthetic TLR3 agonist.

When immunocompetent mice bearing syngeneic, orthotopic (s.c.) MCA205 were systemically (i.p.) treated with the two ICD inducers MTX (24) and CTX (4), they efficiently controlled tumor growth. This effect was not further improved by systemic injection of pIC (Fig. 3A), but was significantly reduced by the administration of antibodies depleting CD4⁺ and CD8⁺ T cells, thus demonstrating that the effects of chemotherapy (MTX+CTX) alone or in combination with pIC fully depend on the T-lymphocyte–mediated anticancer immune response (Fig. 3B; Supplementary Fig. S1C). As previously reported (5, 6, 25), MCA205 fibrosarcoma or TC-1 non–small cell lung carcinoma developing in Fpr1^−/− or Fpr1^+/− hosts as well as MCA205 or TC-1 engineered to lack the expression of the FPR1 ligand ANXA1 failed to respond to chemotherapy with immunogenic anthracycline MTX and/or CTX or DOXO and/or CTX (Fig. 3C and D; Supplementary Figs. S2A–S2F, S3A–S3D, S4A–S4F, S5A–S5F, and S6A–S6C). However, this defective response could be restored when MTX and DOXO and/or CTX treatments were combined with pIC-based immunotherapy (Fig. 3C and D; Supplementary Figs. S2B, S2C, S2E, S2F, S4B, S4C, S4E, S4F, S5B, S5C, S5E, S5F, S6B, and S6C). In all tested models, the deletion of Fpr1 from the genome of tumor-bearing mice (be it in a homozygous or heterozygous fashion) or the knockout of Anxa1 from the transplanted tumor cell lines led to deficient chemotherapeutic responses that could be rescued by systemic pIC administration.

In synthesis, immunotherapy with the TLR3 ligand pIC fully repairs the immune defect resulting from a deficient ANXA1/FPR1-mediated cancer immune dialogue.

To dissect the compensatory capacity of pIC with respect to the FPR1 defect, we implanted MCA205 fibrosarcomas in WT versus Fpr1^−/− mice and treated the animals with MTX/CTX-based chemotherapy alone or in combination with pIC-based immunotherapy, followed by immunophenotyping analyses. Three days after chemotherapy, when DCs are known to be maximally recruited into the tumor bed (26), the FPR1 deficiency is known to interfere with the preferential positioning of CD11c⁺CD86⁺ DCs in the proximity of dying tumor cells identified by staining for proteolytically mature active caspase-3 (Casp3a⁺; ref. 5). pIC chemo-immuno combination therapy led DCs to preferentially localize in the vicinity of dying (Casp3a⁺) tumor cells (Fig. 4A and B). Indeed, the dying/live cell distance (i.e., the ratio between the distance of CD11c⁺CD86⁺ DCs from Casp3a⁺ cancer cells and the distance of CD11c⁺CD86⁺ cells from Casp3a⁻ cancer cells) was reduced after the combined treatment with pIC and chemotherapy but not pIC or chemotherapy alone (Fig. 4B). Thus chemo-immunotherapy corrected the positional defect of Fpr1^−/− CD11c⁺CD86⁺ cells within the tumor. Chemotherapy (alone or combined with pIC) enhanced the density of tumor-infiltrating CD11c⁺CD86⁺ cells (Fig. 4C), and the combination chemo-immunotherapy was particularly efficient in enhancing the frequency of Casp3a⁺ cells in tumors evolving in Fpr1^−/− mice (Fig. 4D). In addition, 10 days after chemotherapy, when the anticancer immune response is full-blown (5, 26–28) and when chemotherapy and pIC were combined, the ratio of CD8⁺ T lymphocytes over regulatory T cells (Treg; defined as CD3⁺CD4⁺CD25⁺Foxp3⁺) was increased (Fig. 4E; Supplementary Fig. S7A–S7C; Supplementary Table S2), reflecting the pivotal capacity of this cotreatment to improve the immune control of the tumors in Fpr1^−/− mice. High-dimensional spectral flow cytometry of tumors developing in Fpr1^−/− mice indicated that, among conventional DCs (CD45⁺CD3⁻CD11c⁺MHCII⁺), the bona fide DC1 subset (defined as CD11b⁺ CD103⁻, also expressing CD8α, CD24, and CXCR1, but lacking macrophages' markers) increased at the expense of the DC2 subset (CD11b⁻CD103⁺) in cancers that responded to chemo-immunotherapy (Supplementary Fig. S8A–S8H). Altogether, these results indicate that pIC can restore the deficient chemotherapy-induced anticancer immune response of Fpr1^−/− mice.

All the aforementioned experiments were based on syngeneic s.c. orthotopic fibrosarcomas or ectopic lung adenocarcinomas. To investigate the role of FPR1 in a more “realistic” setting, we took advantage of a model of carcinogen-induced hormone receptor–positive (25), highly proliferative luminal B-type breast cancer (29). The combination of a synthetic progesterone receptor agonist medroxyprogesterone acetate (MPA) and the DNA-damaging agent 7,12-dimethylbenz[a]-anthracene (DMBA) was administered to WT or Fpr1^−/− female mice (Fig. 5A). Importantly, mammary carcinomas developed significantly earlier in Fpr1^−/− than in WT mice, suggesting that the absence of Fpr1 may accelerate oncogenesis as a result of insufficient immunosurveillance (Fig. 5B). In The Cancer Genome Atlas (TCGA) database (30), patients with breast cancer bearing the FPR1 SNP rs867228 in homozygosity or heterozygosity were diagnosed significantly earlier (by approximately 2 years) than patients harboring the functional FPR1 allele (Fig. 5C and D). For patients affected by luminal B subtype, diagnosis was established approximately 6 years earlier (Fig. 5C and E), whereas no difference was found for other breast cancer histologic and molecular subtypes (HER2⁺, infiltrating ductal, lobular, luminal A, or triple-negative; Fig. 5C; Supplementary Fig. S9A–S9E). We previously demonstrated that FPR1-relevant SNP E346A correlates with breast and colorectal cancer prognosis, though with the subtle difference that for breast cancer, heterozygosity in rs867228 is sufficient to compromise metastasis-free and overall survival, whereas for colorectal cancer, only homozygosity of rs867228 affects progression-free and overall survival (5). Indeed, in a large cohort of more than 1,500 patients with colorectal cancer, homozygosity (n = 78, ∼4% of the cohort) significantly correlated with earlier diagnosis by approximately 3.5 years (Fig. 6A and B). This was confirmed by subgroup analyses for cancers with proficient mismatch repair (pMMR) microsatellite-stable (80% of all colorectal cancers, in which the difference increases to ∼4 years) as opposed to cancers with deficient mismatch repair (dMMR; Fig. 6A and B). A similar correlation between homozygosity for rs867228 and early diagnosis could be identified for esophageal carcinoma (Figs. 5C and 6C) and head and neck squamous cell carcinoma (Figs. 5C and 6D) in the TCGA, whereas no significant difference was found for other cancer types such as non–small cell lung adenocarcinoma or ovarian carcinoma (Fig. 5C). Of note, the allelic frequency of rs867228 was roughly the same across distinct cancer types, and rs867228 had no significant effect on progression-free or overall survival (Supplementary Table S3).

Altogether, these results suggest that FPR1-dependent immunosurveillance plays an important role in controlling the time at which several tumor types develop.

Patients with breast cancer bearing one or two loss-of-function alleles of FPR1 have a relatively poor prognosis as compared with patients harboring both functional alleles, if they are subjected to anthracycline-based chemotherapy (5, 10). Thus, we investigated the capacity of pIC-based immunotherapy to restore the chemotherapeutic response in MPA/DMBA-induced breast cancer developing in Fpr1^−/− hosts. Carcinomas established in WT mice reduced their growth in response to MTX/CTX chemotherapy, and this effect was not improved by pIC (Fig. 7A, left, and B). However, in Fpr1^−/− mice, breast cancers were resistant to MTX/CTX chemotherapy alone. Significant slowdown of the tumor growth was achieved only upon combination of chemotherapy with pIC injections (Fig. 7A, left, and C). Moreover, the chemo-immunotherapeutic combination treatment was the only one to extend the life expectancy of Fpr1^−/− mice (Fig. 7A, right).

These results indicate that breast cancers developing in an Fpr1^−/− environment are amenable to successful treatment if chemotherapy is combined with pIC-based immunotherapy.

Signaling along the ANXA1–FPR1 axis is important for the success of immunogenic chemotherapy. First, AnxA1^−/− cancers do not reduce their growth rate when they are treated with chemotherapy even in an Fpr1-sufficient host. Second, Anxa1-expressing tumors become resistant to chemotherapy if established in Fpr1^−/− mice. DCs lacking Fpr1 do not only fail to juxtapose to dying cancer cells in vivo, but they were also unable to cross-present the archetypal antigen OVA (5). The underlying DC biology, explaining the profound phenotype linked to FPR1 deficiency, is elusive. TLR3 and TLR4 have previously been shown to contribute to chemotherapy efficiency in a similar fashion as FPR1 does, both in mice (animals deficient for Fpr1 or Tlr4 and WT mice injected with Tlr3^−/− MCA205; refs. 5, 21, 31) and in patients with breast cancer (in which loss-of-function alleles in FPR1, TLR3, and TLR4 are epistatic to each other with respect to prognosis in the context of adjuvant anthracycline-based chemotherapy; refs. 5, 10).

Surprisingly, a screening of putative TLR agonists able to correct this poor chemotherapeutic response and to restore the antigen presentation by Fpr1^−/− DCs led to the identification of a single TLR3 agonist, pIC. This finding is commensurate with the fact that high expression of TLR3 (but not TLR4) protein in breast cancers is a positive prognostic factor (10, 32). Obviously, the signaling pathways ignited by TLR3 (expressed intracellularly on endosomes) and TLR4 (expressed on the cell surface) are quite distinct also with respect to their molecular adaptors and downstream signals (33).

Irrespective of these incognita concerning the peculiar link between FPR1 and TLR3, it is clear that pIC can restore the chemotherapeutic response of tumors evolving in an Fpr1^−/− context at multiple levels. pIC promotes the positioning of DCs in the proximity of dying cancer cells, improves the CD8⁺/Treg ratio in the tumor bed, and—most importantly—ameliorates tumor growth control. This latter result was obtained for three different tumor types, namely syngeneic transplantable fibrosarcoma, lung adenocarcinomas, and carcinogen-induced luminal B-like breast cancers, supporting the contention that it can be generalized.

In a hormone-induced model of breast cancer, we found that the absence of Fpr1 greatly accelerated tumor manifestation, suggesting a role for FPR1 in immunosurveillance. This finding was corroborated by clinical data indicating that individuals harboring loss-of-function alleles of FPR1 manifest some cancer types earlier than patients homozygous for the functional FPR1 allele. This applies to luminal B breast cancers, microsatellite-stable colorectal cancer, but not to ovarian carcinoma and lung adenocarcinomas, perhaps reflecting the variable impact of immunosurveillance on these types of malignancies (10, 34).

From a clinical point of view, it may be interesting to consider FPR1 homozygosity in the loss-of-function variation rs867228 as a risk factor for early colorectal cancer (by ∼5 years for ∼4% of the population homozygous for rs867228), implying enhanced surveillance of the patients and specific lifestyle interventions (promote fiber intake, reduce red meat consumption, control weight, and augment physical exercise). Analogous lifestyle adaptations and increased surveillance may be recommended for patients homozygous for rs867228 that are at risk for head and neck and esophageal cancer, given that these tumors manifest approximately 5 and 8.5 years earlier, respectively, than in the rest of the population. At this point, the impact of rs867228 on the age of breast cancer manifestation may be considered as too small (∼2 years earlier diagnosis for all breast cancer subtypes but ∼6 years for the luminal B subtype) or the population at risk too big (because ∼30% of the population harbor the variant allele) to consider this gene as a clinically relevant risk factor. However, future genome-wide association studies might yield a multigene score that has a more profound influence on the age of breast cancer diagnosis, narrowing down the target population. Irrespective of these uncertainties, it may be interesting to evaluate the therapeutic utility of TLR3 ligands such as pIC when focusing on a selected population of patients who carry FPR1 defects, thus offering them a personalized immunotherapy.

Unless otherwise indicated, media and supplements for cell culture were purchased from Gibco-Invitrogen Life Technologies Inc. Plasticware was purchased from Corning B.V. Life Sciences. DOXO, MTX, CTX, Dex, Dox, and pIC were provided by Sigma Aldrich. OVA, agonists to murine NOD containing 1 (NOD1; D-glutamyl-meso-diaminopimelic acid, C12-iE-DAP) and NOD2 (muramyl dipeptide, MDP), as well as to TLR2/6 (Pam3CSK4), TLR4 (Lipopolysaccharide, LPS-B5), TLR5 (Flagellin, FLA-BS), TLR7 and TLR8 (Resiquimod, R848), and TLR9 (CpG oligonucleotide 1826, ODN1826) were purchased from InvivoGen. Monoclonal anti-CD4 (clone GK1.5), anti-CD8 (clone YTS 169.4 or 2.43), and isotype control (clone LFT-2) antibodies were provided by BioXcell. All cells were maintained in standard culture conditions (at 37°C, under 5% CO₂). Murine fibrosarcoma MCA205 cells (class I MHC haplotype H-2b, syngeneic for C57BL/6 mice) and murine non–small cell lung carcinoma TC-1 cells (class I MHC haplotype H-2b, syngeneic for C57BL/6 mice) were cultured in RPMI 1640 medium supplemented with 10% (v/v) FBS, 2 mmol/L l-glutamine, 100 IU/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, 1 mmol/L sodium pyruvate, 1 mmol/L nonessential amino acids, and 1 mmol/L HEPES buffer. AnnexinA1 (Anxa1^−/−), MCA205, and TC-1 cell lines were generated by means of the CompoZr Zinc Finger Nuclease Technology as reported in the study by Vacchelli and colleagues (5) and were cultured in the same condition as parental cell line. TC-1 cell lines were obtained from the ATCC. The iniDC line was a kind gift from Dr. Cornelia Richter (Technische Universitaet Dresden, Germany). The B3Z hybridoma cell line was a kind gift from Dr. Sebastian Amigorena (Institut Curie, Paris, France). iniDC and B3Z cells were cultured in RPMI 1640 medium (as previously described) supplemented with 50 μmol/L β-mercaptoethanol. Culture medium for the iniDCs was further supplied with 10 ng/mL GM-CSF (PeproTech). The cells were tested regularly for mycoplasma using MycoAlert Detection Kit (Lonza) and authenticated via PCR using nine short tandem repeat markers (IDEXX BioResearch).

Mice were bred and maintained in the animal facilities of the Centre de Recherche des Cordeliers in specific pathogen-free conditions in a temperature-controlled environment with 12-hour light/12-hour dark cycles and received food and water ad libitum. Animal experiments followed the Federation of European Laboratory Animal Science Association guidelines and were in compliance with EU Directive 63/2010. Protocols 03941.02, #3794-201601261525379 v3, #4021-2016020216003761 v2, #4022-2016020215425338 v2, #5500-20 16052517404283 v2, #5089-2016041914249886 v2, and #19893-2019032114117517 v4 were approved by the Ethical Committee of the colorectal cancer (CEEA no 5, registered at the French Ministry of Research). Six- to 7-week-old female WT C57BL/6 mice were obtained from Envigo.

For the establishment of syngeneic solid tumors, 2 × 10⁵ WT or Anxa1^−/− MCA205 or TC-1 cells were inoculated s.c. (near the thigh) into WT, Fpr1–mutated homozygous (Fpr1^−/−) or Fpr1 heterozygous (Fpr1^+/−) C57BL/6 (H-2b) mice, and tumor surface (longest dimension × perpendicular dimension) was routinely monitored using a digital caliper. When the tumor surface reached 35 to 45 mm², mice received either 2.9 mg/kg i.t. DOXO in 50 μL PBS or 5.17 mg/kg i.p. MTX in 200 μL PBS or an equivalent volume of PBS. When appropriate, mice also received 50 mg/kg i.p. CTX in 200 μL PBS and/or 50 μg/mice i.p. pIC in 50 μL of PBS (injected at days 1, 4, and 7 after the other treatments). For antibody-mediated depletion experiments, 100 μg of isotype control or monoclonal anti-CD4 or anti-CD8 antibodies in 100 μL PBS were injected i.p. 2 days before, on the same day, and 1 week after chemotherapy. Animals bearing neoplastic lesions that exceeded 20% to 25% of their body mass were euthanized. All experiments contained at least 5 mice per group and were repeated 3 times, yielding similar results. All data were analyzed using GraphPad Prism software and https://kroemerlab.shinyapps.io/TumGrowth/ (35).

The WT s.c. MCA205 tumor model was established as described above. Tumor cell lysate was prepared from cultured 2 × 10⁶ MCA205 cells by a repeated (5×) freezing–thawing process in liquid nitrogen and a water bath (42°C). Cell lysate was further subjected to sonication to ensure complete disruption of cells, then centrifuged at 2,000 rpm for 5 minutes to collect supernatant as the final tumor lysate. De-iniDCs (4-day deinduction as detailed below) were cocultured with the tumor lysate for 2 hours before being collected for treatments. When tumors became palpable, antigen-pulsed de-iniDCs were injected i.t. every 4 days (2 × 10⁶/mouse in 50 μL PBS). For T-cell depletion, 200 μg of isotype control (clone LTF-2, BioXcell) or monoclonal anti-CD4 or anti-CD8 antibodies (clone GK1.5 and 2.43, BioXcell) were injected i.p. at days 0 (first de-iniDC injection), 4, 7, and 14. For ANXA1 neutralization, antigen-pulsed de-iniDCs were resuspended in the anti-ANXA1 solution (Clone 29/Annexin I, BD bioscience, 250 μg/mL) or equivalent isotype control antibody (clone MOPC-21, BioXcell) for i.t. injection. When applicable, pIC (50 μg/mouse) was injected i.p. at days 0, 1, 4, 7, and 14.

Following a published procedure (25, 36), 6- to 7-week-old female C57BL/6 WT or Fpr1^−/− mice underwent surgical s.c. implantation of slow-release MPA pellets (50 mg, 90-day release; Innovative Research of America) at the back of the animals. 200 μL of 5 mg/mL DMBA (Sigma Aldrich) dissolved in corn oil was administered by oral gavage 6 times during 8 weeks. When tumors became palpable, mice received 5.17 mg/kg i.p. MTX in 200 μL PBS and 50 mg/kg i.p. CTX in 200 μL PBS or an equivalent volume of PBS, alone or in combination with 50 μg/mice i.p. pIC in 50 μL of PBS (injected 1, 4, and 7 days after the other treatments) or an equivalent volume of PBS.

Longitudinal analysis of tumor growth data was performed by linear mixed-effect modeling on log preprocessed tumor sizes (over the whole time course), linear modeling (at a single time point), and log-rank test (survival curves). Wald test was used to compute P values by testing jointly that both tumor growth slopes and intercepts (on a log scale) were the same between treatment groups of interest (https://kroemerlab.shinyapps.io/TumGrowth/; ref. 35). Tumor growth data are represented on the untransformed original scale in two forms: individual curves from all measurements of each mouse and group-averaged tumor size alongside its SEM computed at each time point. For mice euthanized before the selected sampling point, the last measure was retained. Statistical analyses are detailed in Supplementary Table S4.

As previously published by Richter and colleagues (19), iniDCs were immortalized under the induction of Dex (100 nmol/L) and Dox (2 μmol/L). When Dex and Dox were removed from the medium (“deinduction”), the iniDCs stopped proliferating and differentiated into primary DCs (“de-iniDCs”). IniDCs were infected with CRISPR/Cas9 coding lentiviral particles (Horizon Discovery), and single-cell clones expressing high levels of Cas9 protein were selected with blasticidin. The gene-editing functionality of the selected clones was verified by immunoblot of cells transfected with gRNA specific to murine high mobility group box 1.

gRNAs were purchased from the Dharmacon-predesigned gRNA library, as a format of Edit-R CRISPR RNA (crRNA) targeting the genes of interest. crRNA was cotransfected with the Edit-R transactivating CRISPR RNA (tracrRNA). For all transfections, 1 × 10⁶ proliferating iniDCs expressing Cas9 were seeded in 6-well tissue culture–treated plates (without Dex/Dox or antibiotics). For each well, 50 pmol/L of crRNA and 50 pmol/L of tracrRNA were diluted in 150 μL medium, incubated for 5 minutes at room temperature, and then mixed with 150 μL medium containing 5 μL of DharmaFECT 1 transfection reagent (Horizon Discovery) for 20 minutes, and then the transfection mix was added to the cells. After 72 hours, the transfection medium was replaced with fresh culture medium (without Dex/Dox or antibiotics), and transfected cells were let to recover for 24 hours before the beginning of functional assays. For the generation of stable knockout cells, medium was replaced with fresh culture medium containing Dex/Dox, and once the transfected cells recovered normal proliferation rate, they were cloned. Selected clones were then verified by Western blotting before being subjected to in vitro antigen cross-presentation assays.

For the in vitro antigen cross-presentation assay, Dex and Dox were removed from the culture medium for 4 days to obtain the de-iniDCs with the characteristics of primary DCs. De-iniDCs (1 × 10⁵ cells/well) were seeded in Falcon 96-well Clear Round Bottom TC-treated cell culture microplate and pretreated with the specific compounds before incubation with soluble OVA (at a final concentration of 1 mg/mL). Six hours after loading with OVA, the de-iniDCs were washed twice with complete culture medium and cocultured with equal numbers of B3Z cells for an additional 18 hours. Then, the coculture plates were centrifuged at 500 g for 5 minutes, and supernatant was collected for the ELISA quantification of IL2 (ELISA MAX Standard Set Mouse IL2, BioLegend).

WT or Fpr1^−/− mice bearing MCA205 fibrosarcomas (size ∼25–40 mm²) were treated with i.t. injection of 2.9 mg/kg DOXO in 50 μL PBS and 50 mg/kg i.p. CTX in 200 μL PBS, alone or in combination with 50 μg/mice i.p. pIC in 50 μL of PBS (three injections: 12, 24, and 36 hours after chemotherapy injection) or an equivalent volume of PBS. Forty-eight hours after treatment, tumor samples were harvested, rinsed quickly with cold PBS, and immediately fixed with 10% neutral buffered formalin (CellStor pots, CellPath) for 4 hours at room temperature. Samples and tissue sections have been prepared as previously described by Ma and colleagues (26) and Vacchelli and colleagues (5). Tissues were stained with primary antibodies diluted in 10% FCS in a humidified chamber (room temperature, 2 hours) specific for CD11c (223H7; MBL International), CD86 (GL-1; BioLegend), and cleaved caspase-3 (Asp175; Cell Signaling Technology). When required (i.e., for unconjugated primary antibodies, at 5 μg/mL after 5 min 3 washes with 10% FBS), sections were stained with appropriate secondary AlexaFluor conjugates (Molecular Probes-Invitrogen) for 30 minutes. Eventually, slides were washed in PBS, counterstained with 10 μmol/L Hoechst 33342 for 5 minutes, and coverslips were mounted with the Fluoromount-GTM mounting medium (Southern Biotechnologies). For each sample, 10 view fields from different sections were captured with an LSM 710 Confocal Microscope (Zeiss). Image analysis was performed with the Fiji (http://fiji.sc/Fiji) open-source image analysis software (37).

WT and Fpr1^−/− mice were sacrificed, and tumors were harvested and placed in ice-cold RPMI (1 mL) in GentleMACS C tubes (Miltenyi Biotec). Tumors were cut into small pieces and enzymatically digested with Miltenyi Biotec mouse tumor dissociation kit using the gentleMACS Octo Dissociator according to the manufacturer's protocol. Single-cell suspensions were obtained by passing dissociated tumors through SmartStrainers (70 μm; Miltenyi Biotec). Tumor homogenates were extensively washed and then resuspended in ice-cold PBS.

Bulk tumor (50 mg/analyzed condition) cell homogenates were stained with LIVE/DEAD Fixable Yellow dye (Thermo Fisher Scientific), and Fc receptors were blocked with anti-mouse CD16/CD32 (clone 2.4G2, BD Biosciences). Surface staining of immune cells was performed with a combination of fluorochrome-conjugated antibodies as detailed in Supplementary Table S2. Cells were then fixed and permeabilized using eBioscience Forkhead box P3 (Foxp3)/Transcription Factor Staining Buffer (Thermo Fisher Scientific). To complete the “Panel T cells,” an additional intranuclear staining was performed with anti-Foxp3 FITC (clone FJK-16s, Thermo Fisher Scientific). Therefore, stained samples were run through a BD LSR II flow cytometer and data acquired using BD FACSDiva software (BD Biosciences). Samples were then analyzed using FlowJo software (Tree Star, Inc.). Normalization on tumor weight and volume of cell suspension acquired allowed for the precise calculation of the absolute number of cells within the 50 mg of stained starting material.

Bulk tumor cell homogenates were fixed in 1% paraformaldehyde for 20 minutes, washed with PBS, and kept at 4°C until staining. Each sample was stained for 25 minutes at 4°C with 50 μL of a combination of fluorochrome-conjugated antibodies as detailed in Supplementary Table S2, washed, and resuspended in 200 μL ice-cold PBS. Samples were acquired on CyTEK Aurora flow cytometer (Cytek BD Biosciences) and analyzed using FlowJo software (Tree Star, Inc.).

Multiple comparisons of tumor-infiltrating immune cell subsets were conducted using a Holm–Sidak test. Outliers were identified using iterative Grubbs test (α = 0.005).

For the TCGA cohort, germline data were downloaded from Huang and colleagues (38); clinical data (age at diagnosis and PAM50 breast cancer subtypes) were accessed via TCGAbiolinks R package. The groups were compared using two-sided Mann–Whitney U test.

Patients who signed written informed consent for translational research, included in the PETACC-8 trial (NCT00265811), were studied. These patients with a stage III colon cancer were randomized to compare FOLFOX (folinic acid, fluorouracil, and oxaliplatin) with FOLFOX + cetuximab in adjuvant setting after surgery. Among them, 1,785 patients were genotyped for rs867228 using the assay C_3266374_1_ (Thermo Fisher Scientific) after DNA extraction from blood samples collected after randomization. The mean age at diagnosis was compared using one-way ANOVA and compared two by two with Tukey correction for multiple testing. The mismatch repair status of the corresponding tumor was determined using immunochemistry and also for ambiguous cases by genotyping using the microsatellite-instability analysis kit from Promega (for details, see ref. 39). Finally, the Consensus Molecular Subtypes classification was determined using NanoString technology (40).

J. Le Naour reports grants from Ministre de l'enseignement supérieur et de la recherche during the conduct of the study. J. Taieb reports personal fees from Amgen, Merck, MSD, Pierre Fabre, Servier, Sanofi, Lilly, and Roche (speaker/ad boards) outside the submitted work. K. Mangane reports grants from Ministre de l'enseignement supérieur et de la recherche during the conduct of the study. C. Richter reports grants and personal fees from DFG (Deutsche Forschungsgemeinschaft; RI 2082/1-1), grants from DFG (Deutsche Forschungsgemeinschaft; SFB-655 B1 and B6), and non-financial support and other from Center for Regenerative Therapies Dresden (genetics) during the conduct of the study. F. André reports grants and other from AZ, Novartis, Lilly, Pfizer, Roche, and Daiichi Sankyo (speaker/advisory board compensated to the hospital) outside the submitted work; in addition, F. André has a patent for TLR3 expression as biomarker for cancer therapy licensed to Innate Pharma and is Founder of Pegacsy. S. Delaloge reports non-financial support from Roche, non-financial support from AstraZeneca, and non-financial support from Pfizer outside the submitted work. P. Laurent-Puig reports personal fees from AstraZeneca, personal fees from Boehringer Ingelheim, personal fees from Biocartis, personal fees from MSD, personal fees from Merck-Serono, personal fees from Lilly, personal fees from Roche, personal fees from Sanofi, personal fees from Pierre Fabre, and grants and personal fees from Servier outside the submitted work. L. Zitvogel reports grants from ANR (RHU LUMIERE), grants from ANR (ILEOBIOME), grants from H2020 EU (ONCOBIOME), personal fees from LYTIX Biopharma (Consulting SAB), other from Kaleido (research contract), other from 9 Meters (research contract), personal fees and other from Transgene (Board of administration and research contract), and other from Daiichi Sankyo (Research contract) during the conduct of the study; other from Everimmune (Founder of biotech Cie) outside the submitted work. G. Kroemer reports grants from Ligue contre le Cancer, grants from Association (Ruban Rose), grants from Cancéropôle Ile-de-France, grants from Gustave Roussy Odyssea, grants from Institut National du Cancer, grants from Inserm, HTE program, grants from Institut Universitaire de France, grants from Agence Nationale de la Recherche, grants from European Commission, and grants from Université de Paris during the conduct of the study. No disclosures were reported by the other authors.

J. Le Naour: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing-review and editing. P. Liu: Formal analysis, investigation, methodology. L. Zhao: Formal analysis, investigation. S. Adjemian: Formal analysis, investigation. Z. Sztupinszki: Data curation, formal analysis. J. Taieb: Supervision. C. Mulot: Formal analysis. A. Silvin: Conceptualization, formal analysis, investigation. C.-A. Dutertre: Formal analysis, supervision. F. Ginhoux: Supervision. A. Sauvat: Formal analysis. G. Cerrato: Formal analysis. F. Castoldi: Investigation. I. Martins: Investigation. G. Stoll: Formal analysis. J. Paillet: Investigation. K. Mangane: Investigation. C. Richter: Resources. O. Kepp: Supervision. M.C. Maiuri: Supervision. F. Pietrocola: Investigation. P. Vandenabeele: Supervision. F. André: Supervision. S. Delaloge: Supervision. Z. Szallasi: Data curation, supervision. P. Laurent-Puig: Data curation, formal analysis, supervision. J. Zucman-Rossi: Supervision. L. Zitvogel: Conceptualization, data curation, supervision. J.G. Pol: Conceptualization, supervision, investigation, methodology. E. Vacchelli: Conceptualization, data curation, formal analysis, supervision, validation, investigation, methodology, writing-original draft, project administration, writing-review and editing. G. Kroemer: Conceptualization, supervision, funding acquisition, writing-original draft, project administration, writing-review and editing.

G. Kroemer is supported by the Ligue contre le Cancer (équipe labellisée); Agence National de la Recherche (ANR)—Projets blancs; ANR under the frame of E-Rare-2, the ERA-Net for Research on Rare Diseases; Association pour la recherche sur le cancer (ARC); Cancéropôle Ile-de-France; Chancellerie des universités de Paris (Legs Poix), Fondation pour la Recherche Médicale (FRM); a donation by Elior; European Research Area Network on Cardiovascular Diseases (ERA-CVD, MINOTAUR); Gustave Roussy Odyssea, the European Union Horizon 2020 Project Oncobiome; Fondation Carrefour; High-end Foreign Expert Program in China (GDW20171100085 and GDW20181100051), Institut National du Cancer (INCa); Inserm (HTE); Institut Universitaire de France; LeDucq Foundation; the LabEx Immuno-Oncology; the RHU Torino Lumière; the Seerave Foundation; the SIRIC Stratified Oncology Cell DNA Repair and Tumor Immune Elimination (SOCRATE); and the SIRIC Cancer Research and Personalized Medicine (CARPEM). Research in the Vandenabeele group is supported by Flemish grants (EOS MODEL-IDI, FWO Grant 30826052), FWO research grants G.0E04.16N, G.0C76.18N, G.0B71.18N, G.0B96.20N), Methusalem (BOF16/MET_V/007), “Foundation against Cancer” (FAF-F/2016/865), and VIB. S. Adjemian was supported by a postdoctoral fellowship from FWO (Flanders Research Organization). F. Pietrocola is supported by a Karolinska Institute Starting Grant and Starting Grant from the Swedish Research Council (2019_02050_3). Z. Szallasi is supported by the Research and Technology Innovation Fund (KTIA_NAP_13-2014-0021 and NAP2-2017-1.2.1-NKP-0002); Breast Cancer Research Foundation (BCRF-18-159) and the Novo Nordisk Foundation Interdisciplinary Synergy Programme Grant (NNF15OC0016584) and Det Fri Forskningsrad (award number 19#7016-00345B); and Department of Defense through the Prostate Cancer Research Program (award number W81XWH-18-2-0056). We thank the staffs of the CHIC and the CEF platforms of the Cordeliers Research Center (Paris, France). The results shown here are partly based upon data generated by the TCGA Research Network: http://cancergenome.nih.gov.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.