Experiments involving mice were approved by the Ethics Committee of the University of Navarra. C57BL/6 mice (5–8 weeks old, female) were obtained from Envigo (Huntingdon, Cambridgeshire, UK) and maintained in the animal facility of Cima Universidad de Navarra. C57BL/6 Batf3^tm1Kmm/J (Batf3KO), Tmem173^gt/J (STINGKO), interferon-α (IFNα)/bR^o/o (IFNARKO), C.129S7(B6)Tag1tm1Mom/J (RAG1), B6.Cg-Thy1^a/CyTg(TcraTcrb)8Rest/J (Pmel-1),24 C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT-I), C57Bl/6 Tg14(act/EGFP)Osby (OT-I-enhanced green fluorescent protein (EGFP)) mice were bred at Cima Universidad de Navarra in specific pathogen-free conditions. Batf3 KO,25 Sting1 KO26 and Ifnar1 KO27 mice were kindly provided, respectively, by Kenneth M Murphy (Washington University, St. Louis, MO), by Gloria González Aseguinolaza (Cima Universidad de Navarra, Pamplona, Spain) and by Matthew Albert (Institut Pasteur, Paris, France). The MC38hEGFR cell line was kindly provided by Pablo Umaña (Roche). This cell line was stably transfected with Lipofectamine 2000 (Thermo Fisher Scientific, San Jose, California, USA) with pCI-neo plasmid expressing membrane-bound ovalbumin (OVA) (#25099, Addgene, Cambridge, Massachusetts, USA). MC38hEGFROVA clones were established by limiting dilution. MC38hEGFROVA was chosen because of suitability for ADCC experiments and convenience for detection but control replicate experiments to those shown in figure 1 with MC38OVA without EGFR were performed rendering comparable results. OVA expression was confirmed by intracellular OVA staining (ab85584, Abcam, Cambridge, UK) and real-time PCR. The MC38hEGFROVA, EG7, MC38, B16OVA, CHO FLT3-L FLAG cell lines were maintained at 37°C in 5% CO₂ and were grown in Roswell Park Memorial Institute medium (RPMI) Medium 1640+Glutamax (Gibco Invitrogen, Carlsbad, California, USA) containing 10% heat-inactivated fetal bovine serum (FBS) (Gibco), 100 IU/mL penicillin and 100 µg/mL streptomycin (Gibco) and 50 µM 2-Mercaptoethanol (Gibco). The MC38hEGFROVA cell line was grown with 6 µg/mL of Puromycin (Gibco) and 400 µg/mL of Geneticin (Gibco). To avoid loss of transgene expression, B16OVA and EG7 were maintained with 400 µg/mL of Geneticin.

The HT29 cell line was cultured as other cells but without 2-mercaptoethanol supplementation in the culture medium.

X-63 granulocyte macrophage-colony stimulating factor (GM-CSF) was grown in Iscove’s modified Dulbecco medium (Sigma-Aldrich, St. Louis, Missouri, USA) supplemented with 1 mg/mL of Geneticin with 5% FCS, 100 IU/mL penicillin and 100 µg/mL streptomycin.

Spleens from euthanized Pmel-1 mice were excised and splenocytes isolated mechanically and cultured at a concentration of 1.5×10⁶/mL for 48 hours with 100 ng/mL of human gp100_25-33 (KVPRNQDWL, RP20344, GenScript, Piscataway, New Jersey, USA). After 48 hours, we added fresh media and 30 IU/mL of IL-2 (Proleukin, Novartis) and kept the culture for 48 hours.

To generate murine activated NK cells, we injected at a high hydrostatic pressure into the tail vein of RAG1 mice 10 µg of Apo-Sushi-IL-15 expressing plasmid28 in 2 mL of physiological saline solution. We harvested spleens after 3 days and isolated murine NK cells by negative selection following the manufacturer’s instructions (NK Cell Isolation Kit II mouse, Miltenyi Biotec, Bergisch Gladbach, Germany).

To obtain T cell-derived tumor debris, MC38hEGFROVA cells were incubated for 48 hours with 15 IU/mL of murine IFNγ (Miltenyi Biotec) to increase expression of MHC-I and 100 ng/mL of human gp100 peptide (KVPRNQDWL). Of important note, freeze and thaw as well as doxorubicin-killed cells were also preincubated with IFNγ. Following activation during 48 hours, Pmel-1 splenocytes were mixed with MC38hEGFROVA at a ratio of 10:1 in the presence of 30 IU/mL of IL-2 and 100 ng/mL of human gp100 and cultured for 3 days.

To obtain NK cell-derived tumor debris, in vivo activated NK cells (following the protocol described above) were cocultured with MC38hEGFROVA cells at a ratio of 3.5:1 with 200 IU/mL of IL-2 for 3 days.

Cell debris was washed twice in ice-cold phosphate buffered saline (PBS) and pellets resuspended in ice-cold PBS to be used in the immunization experiments.

To remove cytotoxic effector cells from the immunizing debris in the indicated experiments, debris was treated for 60 min in 5 mL of distilled H₂O and washed with PBS buffer following centrifugation at 4000 rpm for 10 min at 4°C.

Cytotoxic NK and cytotoxic T lymphocytes (CTLs) were studied by FACS using the following fluorochrome-tagged antibodies: anti mouse CD3 phycoerythrin (PE)-Cy7 (17A2, 1:200, Biolegend, San Diego, CA), CD8 BV510 (53–6.7, 1:200, Biolegend), PD1 FITC (29F.1A12, 1:200, Biolegend), Tim3 PerCP-Cy5.5 (RMT3-23, 1:200, Biolegend), CD137 APC (17B5, 1:200, Biolegend), NK.1.1 PerCP-Cy5.5 (PK136, 1: 200, Biolegend), NKp46 APC (29A1.4, 1:200, Biolegend), KLRG1 BV421 (2F1/KLRG1, 1:200, Biolegend), NKG2D biotinylated (MI-6, 1: 200, eBiosciences, San Diego, California, USA), Streptavidin Alexa Fluor 488 (1:100, Biolegend), DX5 PE (DX5, 1:200, BD Biosciences, San Jose, CA). Dead cells were excluded with Zombie Nir (Biolegend) staining.

We set up cocultures of tumor cells and murine T cells or murine NK cells as described above, and 24 hours later, supernatants were collected for HMGB1 detection by ELISA (IBL International, Hamburg, Germany) or stored at −80°C until use. Tumor cells were analyzed for calreticulin expression. Cells were washed with Staining Buffer (0.5% FBS, 0.5% ethylenediaminetetraacetic acid (EDTA) 0.5 M, 1% penicillin/streptomycin) and stained for 30 min on ice with a mix containing: anti-mouse CD16/32 (S17011E, 1:100, Biolegend), CD45BV510 (30-F11, 1: 200, Biolegend) and Calreticulin Alexa Fluor 647 (1:100, Abcam). Cells were then washed and stained with a mix of AnnexinV FITC (1:20