Experiments were performed using the tumor cells lines A673 (Ewing sarcoma) and U2OS (osteosarcoma), both purchased from ATCC. For the phagocytosis assays, tumor cells were retrovirally (vector MP-71) transduced with GFP. THP-1 cells were purchased from DSMZ. After informed consent and approval of local government regulatory authorities, peripheral blood mononuclear cells (PBMCs) were purchased from DRK-Blutspendedienst.

Cells were cultured at 37°C with 5% CO₂. RPMI standard medium was produced using 10% Fetal Bovine Serum (FBS), 1% P/S and 1% glutamine. RPMI standard medium was used for A673, U2OS, and THP-1 cell lines. XVIVO medium was used for CD14 monocytes. Experiments with tumor cell lines were performed after negative Mycoplasma testing (MycoAlert Mycoplasma Detection Kit, Lonza) were performed regularly. Tumor cell lines were passaged every 2-5 days to avoid hyerconfluency.

The established protocol from Berges et al. (27) was tested and adapted as follows: THP-1 cells were plated at a density of 200k/well in 6 well plates containing 2ml RPMI medium. 1500 IU/ml IL-4 and 1500 IU/ml GM-CSF were added. Cells were cultured a total of 5-6 days to obtain immature DCs. After 2 days, medium was exchanged with supplementation of fresh cytokines. For experiments, cells were used as immature dendritic cells (imDCs) on day 5.

Healthy donor buffy coats were collected from the DRK-Blutspendedienst. Fresh peripheral mononuclear cells (PBMCS) were extracted by magnetic particles (Anti-Human CD14 Magnetic Particles Kit, BD Biosciences) according to manufactures instructions. The CD14 cells were cultured and supplemented with 1000U/ml IL-4 and 800 U/ml GM-CSF to induce DC maturation. Cytokines were renewed on day 3 and imDCs were used for experiments on day 5-6.

Cells were plated at a density of 600k/well in a 6 well plate in 3ml of RPMI medium. 25nM of PMA (phorbol 12-myristate-13-acetate) was added. 48h of incubation with PMA was followed by a 24h-rest period in RPMI medium before scraping off and usage for experiments (29).

The OV XVir-N-31 (26) or AdWT were used for infection experiments. XVir-N-31 and AdWT were kindly provided by Prof. Per Sonne Holm, produced and tested as described before (25). Tumor cells were seeded in 6-, 24-well plates or 10cm-dishes as described in previous reports (30). Multiplicity of infection (MOI) is indicated at respective figures or legends.

Cells were harvested from the 6 well plates and transferred to a 96 well round bottom plate. Plates were centrifuged, and the medium was discarded. Antibodies were usually used at a dilution of 1:100 (CD11c 1:50). Cells were incubated with antibody dilutions incubated at 4°C for 15min or for 30min for anti-CALR/CD47 staining. Afterwards the cells were washed and refilled with PBS for measuring in the MACSQuant Analyzer 10. Experiments were performed in biological replicates (each dot represents one biological replicate). Analysis of T cell-stimulatory surface receptors on THP-1-derived DCs after exposure to pre-treated A673 tumor cells was also performed using flow cytometry after dead/live exclusion and gating on CD45 positive THP-1 cells. Following FACS-antibodies were used: CD47 anti-human FITC REA 220 (Miltenyi Biotech), CD47 isotype REA control antibody human IgG1 VioBright B515 REA293 (Miltenyi Biotech), Calreticulin mouse anti-human Alexa Fluor 488 MAB38981 (R&D Systems), Alexa Fluor 488 goat anti-mouse IgG A11029 (Thermo Fisher Scientific), CD11c anti-human APC B-ly6 (BD Biosciences) and CD45 anti-human APC REA747 (Miltenyi Biotech), CD86 anti-human VioBlue REA968 (Miltenyi Biotech), CD80 anti-human PE REA661 (DAPI Staining Solution (Miltenyi Biotec).

As phagocytes, THP-1-derived imDCs and macrophages as well as healthy donor-derived monocytic imDCs were used. All tumor cells used for phagocytosis assays were retrovirally transduced to express GFP. Phagocytes and tumor cells were co-cultured in a 96 well plate at a 1:2 ratio. Phagocytosis was started by a short centrifugation up to 1500rpm and followed by an incubation period at 37° and 5% CO₂. The time for phagocytosis varied depending on the experiment between 60-90 min, according to published reports (24). The time of phagocytosis is either indicated in the figures or the respective figure legends. Phagocytosis was stopped by centrifugation and followed by staining with CD45/CD11c antibody. DAPI was used for live/dead cell discrimination. The evaluation of phagocytosis was performed via FACS analysis: phagocytosed tumor cells were double-positive for GFP and the myeloid marker CD11c or CD45 (exemplary gating is shown in Supplementary Figure 2 ). In several experiments the CD47i B6H12.2 (Monoclonal Antibody, MA5-11895, Invitrogen) was used. Initially, ideal concentrations for the highest level of phagocytosis were determined by titration experiments with THP-1 macrophages and THP-1 imDCs (data not shown). For experiments with A673 a concentration of 1μg/ml and with U2OS 10 μg/ml was used (both clinically relevant).

Tumor cell killing by XVir-N-31 and/or THP-1-derived macrophages was studied using the sulforhodamine B (SRB) assay. Tumor cells were seeded in 96-well-plates (5,000 cells per well). After viral infection, tumor cells were infected 48 hours after seeding. Macrophages were added 48 hours post infection and cells were fixed 24 hours afterwards with trichloroacetic acid (10%, at 4°C overnight), stained with SRB (Sigma-Aldrich, 0,5% SRB in 1% acetic acid for 30min) and washed afterwards before quantification of the antitumor effect by dissolving SRB-stained cells in 10 mmol/L Tris buffer (pH 10). Extinction was determined at 556nm (Infinite M Nano, Tecan). Results are presented in percent surviving tumor cells compared to mock controls without macrophages, controlled for macrophage only conditions.

In this work, the Bio-Plex Pro Human Cytokine Screening Panel (48-Plex #12007283) was used according to manufacturer’s recommendations to characterize chemo-/cytokine release of THP-1-macrophages after 24 hours of coculture with pre-treated tumor cells CM. Sample acquisition was performed with a Bio-Plex 200 System (Luminex 200). Depicted results represent statistically significant results above threshold detection without detection in control samples (i.e. all treatment condition but without macrophages).

Tumor cells were lysed on ice using a protein lysis buffer (pH 7.2) containing 10mmol/L Tris, protease inhibitor cocktail (Roche Diagnostics), SDS (1%) and sodium orthovanadate (1mmol/L). 2x10⁶ tumor cells were used and homogenized by sonification (Branson) and sheared (27G needle). The protein concentration was determined using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). The protein concentration was equalized in protein loading buffer (containing 100 μL of DTT 1mol/L to 500 μL puffer) and boiled for 5 minutes at 100°C. Samples were either stored at -80°C or used upfront for analysis. Here, samples were loaded onto SDS gels (10%) and separated at 90 V. Then gels were transferred onto PVDF membranes and blocked with 5% non-fat dry milk in TBS-T. Specific antibodies and the Amersham ECL Western Blotting Reagent Pack were used to visualize the gels on the Gel Logic 1500 luminometer. Following antibodies were used: Human CD47 Antibody, (#AF4670, R&D Systems, 1:500), Adenovirus-Hexon (#AB1056, Merck Millipore, 1:1000), GAPDH (#2118S, Cell Signaling Technology, 1:2000), Mouse anti-rabbit IgG HRP (#Sc-2357, Santa Cruz Biotech, 1:1000), anti-sheep IgG HRP (#HAF016, R&D Systems, 1:1000).

The infection of tumor cells with oncolytic adenoviruses induces markers of an immunogenic cell death (ICD), such as surface expression of CALR, representing a ‘eat-me’-signal for immune cells (31–33). This was also shown for the XVir in different cancer entities (25, 33, 34). Furthermore, we recently showed that combination therapy with XVir also modulated the ‘don’t-eat-me’ marker CD47 in humanized EwS xenograft models (25). We now show that viral infection with XVir significantly increased both calreticulin (CALR) surface expression (‘eat-me’) and CD47 surface expression (‘don’t-eat-me’) on all assessed tumors cells ( Figure 1A ). The effect was most distinct and reproducible at 48 hours post infection (hpi). Next, we assessed whether the induction of CD47 exhibited a dose-dependent effect. Indeed, CD47 was upregulated with higher multiplicity of infection (MOI) in both A673 and U2OS cell line, whereas U2OS showed a maximum effect at MOI 25 ( Figure 1B ). Of note, AdWT did not induce CD47 on whole protein and surface level to such a high extent as XVir ( Supplementary Figure 1 ). To standardize downstream experiments, XVir MOI 50 was used for most of following experiments, if not stated otherwise.

As a first step assessing phagocytosis, we tested whether phagocytosis of XVir-infected tumor cells by either THP-1-derived macrophages or THP-1-derived immature dendritic cells (imDCs) was increased despite the upregulation of CD47. The experimental setup is depicted in Figure 2A and the gating strategy is provided in Supplementary Figure 2 . Interestingly, we observed a significant increase of phagocytosis of A673 and U2OS cell lines after XVir-infection by THP-1 macrophages ( Figure 2B ), whereas phagocytosis by THP-1 imDCs was only increased for U2OS cells after infection ( Figure 2C ). Details of statistical testing and results are provided in Supplementary Table 1 . Since the goal was to combine virotherapy with blockade of the CD47-SIRPa-axis, the impact of phagocytosis was additionally studied in combination with the CD47i B6H12.2.

The combination of XVir and CD47i resulted in significantly improved phagocytosis compared to controls in both A673 and U2OS cell lines when plated with all experimental phagocyte types. In our hands CD47i monotherapy only slightly increased phagocytosis of pediatric sarcoma cell lines A673 and U2OS by THP-1-derived imDCs, -macrophages and healthy donor-derived imDCs, which was not significant for most conditions (compare Figures 3A–C ). Concerning the OS cell line U2OS, the combination therapy was significantly better than all monotherapies (XVir or CD47i) and controls for THP-1-derived imDCs and macrophages ( Figures 3A–C , right panels) apart from healthy donor-derived imDCs where combo was not significantly better than XVir monotherapy ( Figure 3C , right). For A673, the combo therapy was also the setup with the best increase in phagocytosis, significantly outperforming controls ( Figures 3A-C , left panels), apart from healthy donor conditions with XVir monotherapy. Representative FACs blots can be found in Supplementary Figure 3A .

Additional chemo-/cytokine analyses from co-culture supernatants of THP-1-macrophages and pre-treated tumor cells show an increased inflammatory response upon combination treatment, e.g. for chemoattractants CCL4 or CCL5 ( Supplementary Figure 4 ). To further evaluate the herein proposed combination approach with regard to T cell-stimulatory properties of utilized antigen-presenting cells, we performed surface receptor staining of THP-1-derived imDCs after exposure to pre-treated tumor cells (72 hours of coculture with A673). Obtained results further support the rationale of the combo treatment, as both T cell-costimulatory CD80 and CD86 expression was significantly induced compared to control and monotherapy approaches ( Supplementary Figure 5 ).

When analyzing the direct cytotoxic effect on tumor cells, we could demonstrate a dose-dependent effect for both A673 and U2OS at increasing effector (THP-1-macrophages)-to-target (E:T) ratios. In A673, there were no significant differences in-between XVir monotherapy and the combination. But all experimental virus-containing conditions decreased tumor cell survival significantly compared to mock and CD47i monotherapy. In U2OS, we could demonstrate stronger tumor killing with the combination approach compared to all conditions, also compared to XVir monotherapy at E:T = 0.5:1 and 1:1 ( Supplementary Figure 6 ).

Taken together, combo therapy increased phagocytosis the most, rather representing an additive phenomenon of both monotherapies. Therefore, synergy testing was not applied. Another interesting observation, which was not followed in more detail, was the fact, that AdWT and AdWT combination with CD47i did not increase phagocytosis of A673 cells ( Supplementary Figure 3B ), adding up to our observation of less CD47 induction after AdWT-infection as compared to XVir-infection ( Supplementary Figure 1C ).