The Oncolytic Adenovirus XVir-N-31 Joins Forces with CDK4/6 Inhibition Augmenting Innate and Adaptive Antitumor Immunity in Ewing Sarcoma

Ewing sarcomas (EwS) are characterized by oncogenic fusion-proteins as well as epigenetic and/or chromosomal rearrangements. These alterations, rather than high somatic mutational loads, are mainly responsible for driving tumorigenesis and maintenance of malignant and metastatic phenotypes (1–4). In this regard, most high-risk pediatric sarcomas are showing high YB-1 expression, and nuclear protein expression was associated with poor survival (5, 6). Standard of care therapy consists of neoadjuvant multidrug chemotherapy followed by surgical resection and/or irradiation and consecutive consolidation chemotherapy, complemented by high-dose chemotherapy and autologous stem cell rescue for very high-risk patients (7). Despite those intense treatment regimens, patients with EwS who relapse and/or present with multifocal bone or bone marrow disease at diagnosis still carry poor prognosis (8). Thus, it is unquestionable that novel therapies are urgently needed.

Ongoing early clinical trials try to exploit (i) targeted therapy by addressing signaling pathways such as PI3K/AKT/mTOR, IGF1/IGFR, Sonic hedgehog, ALK, ERRB or multi-tyrosine kinases, epigenetic modification through histone deacetylase or inhibition of histone demethylase, DNA repair via PARP inhibition, or cell-cycle inhibition. Also, (ii) immunotherapies ranging from allogeneic stem cell transplantation, adoptive T-cell therapies mostly utilizing CAR T cells to oncolytic virotherapy, and immune checkpoint blockade (ICB) are being tested. Furthermore, novel (iii) chemotherapy combinations are also evaluated and often combined with targeted therapy or immunotherapy (9).

With regard to immunotherapy, EwS, such as most pediatric cancers, lack sufficient numbers of tumor-infiltrating lymphocytes (TIL; ref. 10) and are therefore considered to be immunologically cold, that is, non-T-cell-inflamed tumors. Consequently, highly immunogenic natural T-cell epitopes have not been identified for EwS (11), so far and ICB in its current form (as monotherapy) failed to induce convincing response rates compared with many cancers of adults (12).

Besides the low appearance of TILs in EwS, whose induction appears to be a major challenge, most appreciated contributors of immune evasion in this tumor entity are low antigen (HLA) expression (13), as well as high nonclassical HLA-G or HLA-E expression (14), presence of protumorigenic M2 macrophages (15), myeloid-derived and fibroblastic-derived suppressor cells (16), and secretion of immunosuppressive cyto-/chemokines (17) and extracellular vesicles (18). However, for EwS, higher intratumoral levels of the IFNγ-induced cytokine IP10 (CXCL10) were correlated with TILs and a consecutive survival benefit (19). Hence, therapies known to induce interferon signaling or actual pegylated IFN |${{\alpha }}$|2b are supposedly attractive and are already exploited in other tumor entities as combination regimens with ICB (20, 21).

As common inducers of IFN signaling, oncolytic adenoviruses (OAd; ref. 22) are appreciated to (i) mediate direct tumor cell lysis, and (ii) to induce T-cell responses against adenoviral epitopes as well as neoantigens (23). Despite their recognized effect in selectively killing tumor cells and inducing an antitumor response, oncolytic viruses when applied as monotherapy have shown limited clinical effects, indicating the need to identify novel combination approaches (24).

The CDK4/6-RB-E2F pathway plays a major role in cell-cycle regulation, mediating the transition from G₁- to S-phase (25). Because this pathway is often deregulated in cancer, inhibitors directed against CDK4/6 (CDK4/6i) have been developed and were already studied in EwS (26). The three CDK4/6 inhibitors palbociclib (PD), ribociclib (LEE), and abemaciclib (LY) have been approved by the FDA for the treatment of breast cancer. However, as monotherapies, they show very limited efficacy (27), and are currently only applied as combination regimen. Besides their direct effect on the cell cycle, preclinical and clinical evidence indicates that the anticancer activity of CDK4/6i also results not only from their ability to block the cell cycle in malignant cells but also from a range of immunostimulatory effects (28). In this regard, it was shown that CDK4/6i were capable of upregulating HLA-I expression (26) and inhibiting CD4⁺ T-cell populations with a regulatory phenotype (T_Regs; refs. 21, 29). Furthermore, it was shown that CDK4/6i could cause metabolic stress, associated with chemokine-induction of CCL5, CXCL9, and CXCL10, which facilitated tumor infiltration by effector T cells (30). Thus, these findings provide a rationale for new combination regimens using CDK4/6i and other immunotherapies as anticancer treatment (21).

Previously, we could demonstrate that CDK4/6i impressively increased adenoviral replication in general and especially YB-1-based oncolytic virotherapy by enhancing tumor cell lysis and therapy response in vitro as well as in in vivo xenograft mouse models. Here, the RB protein was shown to be an important molecular component in the observed effect (31). Thus, combining CDK4/6i- and YB-1-based oncolytic virus has the potential to address immunotherapeutic hurdles of pediatric sarcomas expressing high YB-1.

In this study, we demonstrate the increase of adenoviral replication-mediated immunoactivation through CDK4/6 inhibition, which we describe as a “Mutually Reinforcing Immune Activation Therapy” (MRIAT), which acts on several levels: (i) tumor cells are primed for enhanced viral replication and (ii) for an immune attack. (iii) Viral replication as well as (iv) cell death induction is increased leading to the enhanced activity of (v) preexisting tumor antigen-specific T-cell responses. Also, (vi) inflammatory stimuli induced by this combination therapy attract (vii) innate immune cells, which in turn show enhanced phagocytic as well as antigen-presenting and co-stimulatory capacities stimulating adaptive immunity and (viii) inducing TILs.

The EwS cell line A673 was purchased from ATCC. SK-N-MC was purchased from the German Collection of Microorganism and Cell Culture (DSMZ). TC32 was a kind gift from Prof. Poul Sorensen (University of British Columbia), which was originally obtained from the Childhood Cancer Repository (CCR, Alex's Lemonade Stand Foundation, Children's Oncology Group, COG). Alveolar rhabdomyosarcoma cell line RH41, LCL (feeder cells for T-cell expansion), and THP-1 (for DC maturation and T-cell proliferation experiments) were purchased from DSMZ. Osteosarcoma cell line SJSA-1 was purchased from ATCC. HEK293 cells for virus production and determination of infectious units (IFU) were purchased from ATCC. The IL15-producing cell line NSO was a kind gift of Prof. Stanley Riddell (University of Washington School of Medicine). Prof. Manual Caruso (Centre de recherche de Québec, Université Laval) provided the packaging cell line (293Vec.) RD114 (for retrovirus production). Peripheral blood mononuclear cells (PBMC) of healthy donors were purchased from DRK-Blutspendedienst after informed consent and approval of local government regulatory authorities. Cell lines were cultures according to provider's recommendations and mycoplasma status was monitored regularly (e.g., before start of in vivo experiments; MycoAlert Mycoplasma Detection Kit, Lonza).

Abemaciclib (LY-2835219, LY, Selleck Chemicals), palbociclib (PD-0332991, PD, Sigma-Aldrich), and ribociclib (LEE011, LEE, Selleck Chemicals) for in vitro experiments were dissolved in sterile water or DMSO as 10 mmol/L stock solutions. Working concentrations were prepared in culture media for immediate use. LEE011 for in vivo studies was purchased from MedChem Express as ribociclib succinate and was dissolved freshly in 0.5% methylcellulose (viscosity: 400 cP, Merck Millipore).

Identification of TCR V |$\alpha $| and V |$\beta $|-chains of the TCR specific for CHM1³¹⁹ in the context of HLA-A*02:01 as well as synthesis of retroviral TCR constructs with TCR modifications (codon optimization, murinization) was described previously (32, 33).

XVir-N-31 is a YB-1-driven oncolytic adenovirus, initially described as Ad-Delo3-RGD (34). Viral deletions were introduced into the CR3-region of E1A, rendering its replication dependent on nuclear localization of YB-1. Furthermore, the E1B19k gene was deleted as well as the E3-region. The fiber knob domain was altered to express an RGD-motif (35). XVir-N-31 was produced in HEK293 cells and purified by a two-step CsCl-gradient centrifugation and size-exclusion chromatography with PD-10 desalting columns (GE Healthcare).

Infections in vitro were performed in 10 cm, 6- or 24-well plates, dependent on the specific assay, with multiplicities of infection (MOI) as indicated. After 24 hours of inhibitor treatment, virus infection was performed in 2 mL, 500 μL, or 150 μL for 1 hour, when not stated otherwise. Afterwards culture medium was added as again containing the diluted inhibitor or mock. Processing of cells for downstream analysis is indicated in the results, for example, 48 hours postinfection (hpi).

Virus-induced cell killing was determined using and sulforhodamine B (SRB) assay. Here, cells were seeded overnight in 24-well plates (20,000 to 30,000 cells per well) and infected as described above. Three to five days after infection (further specified in the the figure legends), cells were fixed with trichloroacetic acid (10%, ice-cold, overnight at 4°C) and stained with 0.5% SRB (Sigma-Aldrich) in 1% acetic acid for 0.5 hours. The cytopathic effect was quantified by dissolving the bound SRB in 10 mmol/L Tris buffer (pH 10) for photometric measurement of extinction at 510 nm (Infinite M Nano, Tecan).

Furthermore, the xCELLigence assays (Roche/ACEA) were used for real-time cytotoxicity measurement. Cytotoxicity of XVir-N-31 was detected by detachments of tumor cells after addition at indicated timepoints at different MOI. Infection was performed at a cell index of 0.5 to 1. Experiments were performed in biological triplicates.

The cell-cycle state of cell lines after CDK4/6i treatment was determined by flow cytometry after propidium iodide (PI, Sigma-Aldrich) staining. Here, tumor cells were seeded in 6-well plates and treated with indicated concentrations of CDK4/6i the next day. After 24 hours, cells were washed and fixed with 70% ethanol (ice-cold, drop-wise addition under vortexing), stored at −20°C for at least 18 hours before staining with PI 50 |$\ {{\mu }}$|g/mL and RNase A 100 U/mL (Thermo Fisher Scientific) for 30 minutes. Acquisition was done on a FACSCalibur flow cytometer. FlowJo (10.8.0, BD Biosciences) was used to analyze data.

Cell lysis was performed on ice using a protein lysis buffer containing 10 mmol/L Tris (pH 7.2), protease inhibitor cocktail (Roche Diagnostics), 1% SDS, 1 mmol/L sodium orthovanadate. Approximately 2 × 10⁶ cells from in vitro experiments and approximately 3 to 5 × 10⁶ cells from explanted xenografts (single cell suspension after mechanical and enzymatic dissociation) were used. Before centrifugation at 30,000 rcf (at 4°C), cell lysates were homogenized by sonification (Branson) and sheared (27G needle) until no viscosity was observed. Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) according to supplier's recommendation. After equalization of protein concentration in protein loading buffer (supplemented with 100 μL 1 mol/L DTT to 500 μL buffer) and boiling (5 minutes, 100°C), samples were stored at −80°C or used for analysis upfront. Therefore, samples were loaded into pockets of SDS gels (10–12%) and separated at 75 to 90 V before transfer onto PVDF membranes. Membranes were blocked with 5% nonfat dry milk in TBS-T. Protein band were detected with specific antibodies (listed in the Supplementary Information) and the Amersham ECL Western Blotting Reagent Pack, visualized with the Gel Logic 1500 luminometer.

Adenoviral IFU were quantified by using a hexon titer test (HTT). Here, formation of infectious particles was quantified by immunocytochemistry staining of infected HEK293 cells. Cells were seeded in 24-well plates and infected simultaneously. Then, cells were fixed at approximately 44hpi with ice-cold methanol before staining with anti-hexon primary and consecutively a horseradish peroxidase-conjugated secondary antibody. Development was performed using Liquid DAB + Substrate Chromogen System (Dako). Positively stained HEK293 cells were counted (10 visual fields per well, magnification 20×), and viral titers were determined applying the formula: titer (IFU/mL) = (average number of positive cells/field × fields/well) / [volume of diluted virus used per well (mL) × dilution factor]. Error bars in respective experiments represent the SD of virus titers.

(i) Genomic DNA (for viral genomic copies) was isolated using the DNeasy Blood & Tissue Kit and (ii) RNA (for RT) was isolated using RNeasy Mini Kit according to manufacturer's protocols (both Quiagen). Respective concentrations were determined by NanoPhotometer measurement before further processing.

For gene expression analysis, RNA was transcribed into complementary DNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to manufacturer's recommendation. The reverse transcription program in the iCycler (Bio-Rad Laboratories) consisted of three steps: (i) initiation at 25°C for 10 minutes, (ii) elongation at 37°C for 120 minutes, and (iii) inactivation at 85°C for 5 minutes.

qPCR was used to determine viral genomic copies (fiber DNA) and to assess gene expression after RT. Adenoviral genomic copies and human genes from xenograft experiments were detected using 50 ng genomic or cDNA per reaction. For the detection of mouse immune cell genes, 375 ng of cDNA was applied. Total reaction volume adjusted in 96-well MicroAmp Fast Optical Reaction Plates was 15 |${{\mu }}$|L, comprised of 7.5 μL GoTaq qPCR master mix, 0.75 |${{\mu }}$|L of forward and reverse primers each, 1 |${{\mu }}$|L of DEPC-H₂O, and 5 |${{\mu }}$|L of genomic or cDNA in DEPC-H₂O. Reaction and recording of fluorescence were done in a StepOnePlus Real-Time PCR System (Applied Biosystems), according to manufacturer's recommendation. A melting curve analysis was performed to confirm specific PCR products. All applied primers are described in the Supplementary Information document. Relative quantification was performed using the ∆∆CT-method. Samples were normalized to actin or GAPDH cDNA levels and/or untreated controls. M1/M2 gene marker analysis were performed according to Castro and colleagues (36) Mouse-specific primers (Supplementary Material) were used to detect M1 marker genes NOS2, CXCL10, IL1B, and M2 marker genes ARG1, TGFB1, and MMP9. Then relative gene expression was assessed compared with mouse-specific GAPDH. In a next step, the product of expression of the three M1 markers was divided by the product of expression of the three M2 markers to obtain M1/M2 ratios.

Acquisition of cells for analysis of surface markers, transcription factors, and T-cell proliferation were done with MACSQuant Analyzer 10.

Cells in single cell suspension were stained in 96-well plates in PBS with primary flurochrome-coupled antibodies. REA antibodies were incubated for a minimum of 15 minutes (at 4°C). Staining concentration for antibodies are given in the Supplementary Information document. A maximum of 2 × 10⁶ cells per well were stained with REA antibodies (xenografts). Cell numbers from in vitro experiments ranged from 5 to 20 × 10⁴ cell per staining. Multicolor staining panels (all in vivo experiments and all in vitro DC maturations and T-cell activation experiments) were auto-compensated with REA-compensation beads according to Miltenyi's recommendation (MACSQuant Comp Bead Kit anti-REA). Dead cell exclusion was done by either DAPI, PI, or the fixable Viobilitiy 405/520 dye (staining together with surface markers, according to manufacturer's protocol). When indicated, isotype staining with respective fluorophore was performed, or fluorescence-minus one (FMO) controls were used. All cells were washed with PBS before acquisition.

Intracellular analysis of transcription factors was done in two steps: (i) surface staining (together with fixable Viobility dye) was performed, as described above. (ii) cell permeabilization and fixation for 30 mintes (4°C) followed by staining for T-cell subset-specific transcription factors. Here, specific kits from Miltenyi containing fixation/permeabilization buffer and specific solution were used according to supplier's recommendation.

All fcs raw data were processed using FlowJo. The following gating strategy was applied in all analyses: doublet discrimination in FSC-H versus FSC-A → identification of cells in SSC-A versus FSC-A → dead cell exclusion using DAPI (V1 vs. V2) PI (B3 vs. B2), or Viobility dye (V2 vs. V1 or B1). When several populations were present, specific cells were identified as follows: human T cells by double-positivity for CD45/CD3 (or CD3/CD4 or CD3/CD8), human tumor cells by mouse and humane double-negativity for CD45, mouse myeloid cells by CD45/CD11b double-positivity, neutrophils by Ly6G positivity in myeloid population, monocytes by Ly6C high-positivity in Ly6G-negative myeloid population, and macrophages by F4/80 positivity in Ly6C/Ly6G double-negative myeloid cells. If a specific marker expression in respective subsets was analyzed, isotype or FMO controls were applied.

The Human Cytokine Screening 48-Plax Panel was used for in vitro screening assays. Chemo-/cytokines present in the TME were analyzed using the Human ProcartaPlex Mix&Match 7-plex (containing IL29, IL8, CXCL10, LAP, MIF, SCGF-β, TNFα) Usage was done according to supplier's recommendations and sample acquisition was done with a Bio-Plex 200 System (Luminex 200). Supernatants from in vitro experiments were prepared identical to tumor-conditioned medium at different MOI and time points, further specified in figure legends. To assess chemo-/cytokine levels in the TME, tumors were mechanically dissected with a scalpel (no. 10) in PBS. Afterwards, cell suspension was centrifuged, and supernatant was stored at −80°C until measurement.

Health donor buffy coats were purchased from DRK-Blutspendedienst. The fraction of PBMCs was isolated by density gradient centrifugation using Ficoll-Paque (GE Healthcare) according to manufacturer's instructions and as decribed previously (37). ACK Lysis Buffer (Thermo Fisher Scientific) was used to minimize red blood cell contamination.

Rag2^−/−|$\gamma $|c^−/− mice (BALB/c background, initially obtained from the Central Institute of Experimental Animals; re-imported from Charles River Laboratories after embryonal transfer) were bred and maintained in our animal facility (Center for Preclinical Research, School of Medicine, TUM) under SPF conditions. Experimental and control animals were bred/co-housed within the same room. Animal experiments were approved by local regulatory authorities according to German Federal Law and in accordance with institution guidelines (permission numbers of Regierung von Oberbayern: 55.2–2532.Vet_02–15–102, 55.2–2532.Vet_02–20–165). Experiments were performed on 12- to 30-week-old mice, both male and female. Mice were injected subcutaneously in the flank(s) with A673 or TC32 tumors cells in PBS (specific numbers are given in the figure legends). Tumor size was measured every 2 days with a caliper and tumor volume was calculated with the formula volume = 0.5 × length × width². After the tumor volume exceeded 150 to 300 mm³, mice were randomly assigned to treatment/control groups: mock (i.e., 0.5% methylcellulose without LEE011 and PBS intratumorally), LEE (i.e., 0.5% methylcellulose containing LEE011 and PBS intratumorally), XVir (i.e., 0.5% methylcellulose without LEE011 and XVir-N-31 intratumorally), and combo (i.e., 0.5% methylcellulose containing LEE011 and XVir-N-31 intratumorally). Then, animals were treated with LEE011 200 mg/kg body weight (dissolved in 0.5% methylcellulose) on day X (DX) until day X+4 (DX+4) or mock control (0.5% methylcellulose without LEE011) via oral gavage. 1 × 10¹¹ viral particles (VP, if not stated otherwise) of XVir-N-31 or PBS (in 50 μL) were injected intratumorally on day X+2 (DX+2) and (if indicated) on day X+4 (DX+4). On day X (start of therapy) +5, 3 to 4 representative animals from each group were euthanized for downstream analysis (viral replication, immunogenicity). Tumor-HLA-matched healthy donor nonspecific PBMCs (matched for HLA-A2) were prepared as described above for humanization, when indicated. Two different healthy adult donors were used for humanization experiments in A673 and TC32 models. IL15-producing NSO cells were irradiated with 80 Gy prior to application. Both PBMCs and NSO cells were injected intraperitoneally. Transferred cell numbers and intervals are given at each section.

The tumor size was measured until the tumor volume reached 1,000 mm³ (i.e., permitted maximal tumor burden from our regulatory authority). All end-point criteria were respected in this study. Numbers of animals are given in respective figures.

Mice were euthanized at the indicated time points or at the end of experiment by isoflurane narcosis and cervical dislocation. Experiments were stopped when subcutaneous tumor reached a volume of |$ \ge $|1,000 mm³ or other endpoint criteria were met: significant weight loss, increased frequency of breathing, decreased movement, changes in flight reactions, and others (as defined in approvals from Regierung von Oberbayern). With regard to planned read-outs, explanted tumors/spleens were cut into half (one part fixed in 4% buffered formaldehyde) or dissected with a scalped mechanically as a whole in 4 mL PBS. Twice 0.5 mL of the tumor cell suspension was stored at −20°C for DNA isolation and qPCR analyses. An aliquot for supernatant analysis (TME) was stored at −80°C after centrifugation (2,200 rpm, 5 minutes, for multiplex chemo-/cytokine analyses) followed by the addition of 1 mL tumor digestion buffer [RPMI1640, collagenase IV 200 U/mL (Sigma Aldrich), DNase I 100 |$\ {{\mu }}$|g/mL (Thermo Fisher Scientific)] for enzymatic dissociation by incubation of samples at 37°C for 30 minutes. Next, samples were washed, filtered (70 μm) and finally washed in standard medium before distribution for further RNA and protein isolation, flow cytometry, and cryopreservation. Spleens for flow cytometry analysis were treated comparably. All steps were performed on ice, if not indicated otherwise.

As described (32), CD14⁺ monocytes were purified from fresh PBMC using the Anti-Human CD14 Magnetic Particles Kit (BD Biosciences). Then, the cell fraction was cultured in X-VIVO 15 (Lonza, containing 1% hAB serum) containing 1,000 U/mL IL4 and 800 U/mL GM-CSF to induce DC differentiation. After 3 days, cytokines were renewed and cells were used on day 5 as immature DCs (imDC). DC maturation was induced by a maturation cocktail (MC) containing: 10 ng/mL IL1β, 1,000 U/mL IL6, 10 ng/mL TNF, and 1 μg/mL PGE₂ for additional 2 days. For THP-1-derived imDC, the identical concentrations for IL4 and GM-CSF were used.

First, tumor cells (1.5 × 10⁵) were seeded in 6-well plates. After 24 hours, cells were treated with inhibitor or infected with virus. In case, the combination of CDK4/6i and XVir-N-31 was assessed, virus infection was performed 24 hours after addition of the inhibitor/mock. Forty-eight hours after infection (hpi), supernatant was collected, centrifuged [2,200 rounds per minute (rpm), 5 minutes] and stored at −80°C until further usage.

For all experiments, tumor cells and phagocytes were co-cultured at a ratio of 2:1 in 96-well-plates (U-bottom, TPP). Tumor cells were virally transduced to express GFP and sorted before usage. Before phagocytosis, tumor cells were seeded and treated in 6-well plates (as described for “Generation of tumor-condition medium”). 48 hpi tumor cells were harvested by scratching, washed, and counted before adding them into the 96-well plate. Phagocytosis was started after adding respective numbers of phagocytes (imDCs or macrophages, both derived from THP-1 cells) and by a short centrifugation (1,500 rpm, Mulifuge 3SR Heraeus), followed by an incubation period of 1 to 2 hours at 37°C and 5% CO₂. Phagocytosis was stopped by centrifugation and a washing step before the cells were stained with CD45-APC. Tumor cells were considered as phagocytosed when staining double-positive for GFP and CD45. They were quantified as percentage of total CD45-positive cells and normalized to mock controls.

THP-1-derived macrophages were generated in 6-well plates (RPMI medium) supplemented with 25 nmol/L phorbol 12-myristate-13-acetate (PMA; Thermo Fisher Scientific). Macrophages were harvested with a cell scraper and used for experiments after a 24-hour rest-period.

Transgenic T cells were generated as published previously (33, 37, 38). In short, CD8⁺ T cells were negatively selected (untouched) from PBMCs using Miltenyi Biotech equipment. TCR transgenes were introduced into T cells using a retroviral infection system. Transduction efficacy was evaluated by flow cytometry using CHM1-specific multimers, compared with irrelevant multimers. Purities |$ \ge $|95% were obtained by anti-PE microbead enrichment (Miltenyi Biotech). Transgenic T cells were expanded with 5 × 10⁶ irradiated LCL (100 Gray, Gy) and 2.5 × 10⁷ irradiated PBMCs (30 Gy, pooled from five different donors) in 25 mL T-cell medium (AIM-V medium 500 mL, 5% hAB serum, penicillin 100 U/mL, streptomycin 100 |$\ {{\mu }}$|g/mL). Cytokines (100 U/mL IL2 and 2 ng/mL IL15) were supplemented every 2 to 3 days. Irradiation of cells was performed with BioBeam8000.

After a 72-hour culture period of THP-1 in conditioned-medium, maturation status was evaluated via flow cytometry. Then THP-1 cells were pulsed with 10 |${{\mu }}$|g/mL peptide and 20 μg/mL β₂-microglobulin for 4 hours at 37°C and washed afterwards. Next, CHM1³¹⁹/HLA-A*02:01-specific TCR-transgenic CD8⁺ T cells were added at a 2:1 ratio in T-cell medium containing 30 U/mL IL2. For peptide pulsation, the CHM1³¹⁹-peptide (VIMPCSWWV) and influenza-peptide (GILGFVFTL, as control, when indicated) were used. T cells were labeled with 10 |${{\mu }}$|M eFluor450 dye according to manufacturer's recommendation and washed three times before addition to peptide-pulsed THP-1 cells. T-cell proliferation/activation was assessed 96h after start of coculture after dead cell exclusion with DAPI or Viobility dye on a MACSQuant Analyzer 10.

Statistical analysis was performed using Prism 9. Normally distributed values were analyzed using Student t tests and ordinary one-way or two-way ANOVA in combination with Tukey multiple comparison methods, as indicated in respective section. For arbitrarily distributed values, Kruskal–Wallis and Mann–Whitney U test were applied. Simple linear and nonlinear regression models were used for correlations. Differences in survival were analyzed using the Mantel–Cox log-rank test. Tumor growth curves assessing an abscopal effect were analyzed using the open-access web tool TumGrowth (https://kroemerlab.shinyapps.io/TumGrowth/), applying selected pairwise comparison for longitudinal tumor growth (with holm adjustment when indicated). Significance levels are given as asterisks. Densitometric analysis of immunoblots was done using ImageJ 1.53k.

All antibodies used are listed in the Supplementary Information section.

All relevant data are available from the authors upon request.

First, we compared XVir-N-31 (XVir) monotherapy (mono Tx) to the combination therapy (combo Tx) with CDK4/6i and respective controls (i.e., mock and CDK4/6i mono Tx). We assessed three EwS cells lines (A673, SK-N-MC, TC32) and additionally the alveolar rhabdomyosarcoma cell line RH41, biphasic synovial sarcoma SYO-1, and osteosarcoma cell line SJSA-1 (Supplementary Fig. S1). Cell survival, measured by SRB assays, was significantly decreased with combo Tx in A673, TC32, SJSA-1, and SYO-1 (Fig. 1A and B; Supplementary Fig. S1A–S1D) but not in SK-N-MC and RH41. A performed xCELLigence assay verified the obtained results, showing that CDK4/6i merely slowed down tumor growth, whereas combo Tx induced an earlier detachment of tumor cells (i.e., cytopathic effect), leading to complete detachment of cells (Fig. 1D). Consequently, all analyzed cell lines, apart from SK-N-MC with a known RB1 mutation and resistance to CDK4/6i treatment (26), responded to combo Tx, leading to a significant increase of new infectious particles (infectious units, IFU) compared with XVir mono Tx (Fig. 1E; Supplementary Fig. S1B). Notably, the lowest increase in particle formation was observed in RH41 cells, which display a rather resistant phenotype to CDK4/6i, with EC₅₀ values excessing 1 μmol/L (26).

G₁ cell-cycle state was increased in a dose-dependent manner accompanied by reduction of total RB, phosphorylated (p-) RB, E2F1 and induction of p21 after 24 hours of treatment (exemplarily shown for A673 and abemaciclib, LY in Fig. 1C).

In a time-dependent assessment of relevant proteins by immunoblotting, combo Tx led to a stronger induction of E1A and hexon. IFNγ-induced CXCL10 was upregulated in both LEE mono Tx and combo Tx in vitro (Fig. 1F). Of particular note was also the inhibition of p21, after initial induction with combo Tx, as compared with the other conditions.

Our previous data have demonstrated the synergistic effect of CDK4/6i on XVir-N-31 replication and cell killing (31). Therefore, we now evaluated the immunologic properties of combo Tx focusing on changes of tumor cells first. The verification of viral genome replication and adenoviral hexon protein level showed increased levels in the combination treatment compared with mono Tx (Fig. 2B and C), which is in line with previously published data (31). Furthermore, MHC class I (MHC-I) was induced on RNA level, as well as on both total protein and surface expression levels (Fig. 2C and D; Supplementary Fig. S2A–S2C). At the same time, total RB protein was reduced in combo Tx (Fig. 2C). Because the applied antibodies against RB and GAPDH both detect human and mouse protein, the observed downregulation of RB should be interpreted with caution. Furthermore, we observed a surface receptor-downregulation of the don't-eat-me signal CD47 in A673 but not in TC32 xenografts, and highest surface expression of PD-L1 in the combination treatment group (Fig. 2D–F; Supplementary Fig. S2D). As an increase in oncolysis of CDK4/6i with another OAd was attributed to induce IFN signaling and phosphorylation of STAT1 in breast cancer cell lines (39), explanted A673 xenografts were also subjected to such analyses, showing highest p-STAT1 levels in LEE mono Tx and combo Tx (Fig. 2E and F).

In a further step towards characterization the antitumor immune response, we analyzed human T-cell distribution and infiltration (Fig. 3A) as well as local tumor control and survival (Supplementary Fig. S3A). Analysis of tumor-infiltrating human T cells and human T-cell engraftment in spleens at the end of experiment revealed a significant increase of T cells in A673 xenografted tumors and spleens. Tumor-infiltrating T cells in TC32 xenografts showed the same tendencies (Fig. 3B and D). In both xenograft experiments (A673 and TC32) combo Tx resulted in a significant survival benefit as compared with mono and mock Txs. Of note, TC32 xenografts responded less than A673, reflected by shorter survival rates and inferior growth inhibition of subcutaneous-implanted tumors (Fig. 3C; Supplementary Fig. S3A). Because of lower numbers of infiltrating T cells in mock and XVir groups, additional T-cell characterization did not produce robust results. For LEE mono Tx and combo Tx, a dominance of tumor-infiltrating CD4⁺ T cells was confirmed (Supplementary Fig. S3A and S3B). Of note, the majority of tumor-infiltrating T cells in all groups expressed the exhaustion marker PD-1 (Supplementary Fig. S3C). Representative intracellular stainings of T-cell transcription factors were performed in tumors with the highest T-cell numbers exemplarily. Here, the master regulator T-bet was upregulated, as opposed to ROR-γ and GATA3, indicating a T_H1 differentiation (Supplementary Fig. S3D). To assess systemic immune cell changes, human T cells from explanted spleens were characterized in more detail. Especially CD4⁺ T-cell engraftment was significantly enhanced and a tendency towards increased CD8⁺ T cells was stated for animals receiving combo Tx (Fig. 3E). Additional phenotyping via transcription factors revealed that most engrafted CD4⁺ T cells in animals treated with LEE mono and combo Tx expressed T-bet, demonstrating a specific engraftment of human T_H1 cells (Fig. 3F). We also performed flow cytometric assessment of regulatory CD4⁺ T cells (T_Regs) and surprisingly found that the inhibition of T_Regs by CDK4/6i was unexpectedly maintained in the combo Tx's group (Supplementary Fig. S3G). Consequently, the ratio between tumor-infiltrating T cells and T_Regs (in the spleen) clearly indicated the superiority of combo Tx (Supplementary Fig. S3H) further supported by a correlation analysis demonstrating increased tumor infiltration by T cells when T_Reg frequencies were low (Supplementary Fig. S3I).

As we observed systemic changes of human T-cell engraftment and tumor infiltration, we performed a second set of experiments to evaluate the potential of combo Tx to induce an abscopal effect. For this purpose, tumor cells were implanted at both flanks, and treated as described in Fig. 4A. The growth analysis revealed that tumor growth of non-injected tumors in animals treated with combo Tx was significantly inhibited compared to mono Tx's and mock (Fig. 4B; Supplementary Fig. S4A).

Next, we were interested in identifying possible determinants associated with tumor infiltration of T cells. Therefore, we performed initial cytokine multiplex screening assays in vitro and observed a striking induction of CXCL10 in both LEE mono Tx and combo Tx. CXCL10 was also induced on protein level in vitro by LEE mono Tx and combo Tx (Fig. 1F).

In accordance with in vitro results, measurement of chemo-/cytokines profiles within the tumor microenvironment revealed most distinct alterations in human CXCL10 levels, which were significantly induced by combo Tx (Fig. 4C). Of note, human T-cell infiltration was very low at this time point of analysis and did not differ significantly in-between groups (Supplementary Fig. S4B). When assessing the interferon-inducible chemokine receptor CXCR3 at the end of experiment however, tumor-infiltrating T cells showed highest surface expression levels in the combo Tx group (Fig. 4D), indicating possible chemo-attraction of T cells via the CXCL10/CXCR3 axis.

When focusing on mouse innate immune cell changes, as a parameter of inflammatory response towards combo Tx, we observed an increase of tumor-infiltrating myeloid populations compared with mono Tx's and mock (Fig. 4E; Supplementary Fig. S4C). Further characterization showed an influx of inflammatory-like monocytes in XVir-injected tumors, in both XVir mono and combo Tx (Fig. 4F; Supplementary Fig. S4D). M1/M2 macrophage marker gene analysis from bulk tumor transcripts indicated an increase of the M1/M2 ratio (Fig. 4G; Supplementary Fig. S4E). The increase of M1/M2 ratio was mainly driven by upregulation of mouse-derived CXCL10 (Supplementary Fig. S4F and S4G). Furthermore, numbers of MHC-II-expressing myeloid cells, indicating increased capacity of antigen presentation, were highest in combo Tx, followed by LEE mono Tx (Fig. 4H and I). Finally, correlation analysis of viral fiber DNA with relative numbers of innate immune cells revealed a strong positive correlation in-between viral genomic copies and especially monocyte infiltration (Fig. 4J).

Because of the known limitation of the herein applied in vivo model to analyze the crosstalk in-between mouse innate and human adoptively transferred adaptive immune cells, we analyzed the potential of the combo Tx to alter DC maturation and to stimulate specific T-cell activity in vitro. First, we screened for differences of calreticulin (CALR) surface expression and observed that XVir mono and combo Tx similarly induced CALR surface levels (Supplementary Fig. S5A). However, phagocytosis of THP-1-derived imDC was increased especially in LEE mono and combo Tx but not in XVir mono Tx whereas macrophage phagocytosis in combo Tx significantly outperformed both mono Txs (Supplementary Fig. S5B).

As we also observed an increase in the antigen-presenting capacities of mouse innate cells by LEE mono Tx and combo Tx in our in vivo models (Fig. 4H and I), we assessed both human monocyte-derived imDC and THP-1 dendritic cell maturation (40) with respective cells being exposed to conditioned medium of mono and combo Txs. Here, we confirmed higher HLA-DR surface levels in the combo Tx settings compared to mono Txs and mock for both models accompanied by HLA-I upregulation and a specific increase of T-cell costimulatory molecules CD80/CD86 by combo Tx (Fig. 5A–D). After loading THP-1 cells (HLA-A2⁺) with an EwS tumor-associated antigenic peptide, derived from the source protein chrondromodulin-1 (CHM1³¹⁹; refs. 32, 41), specific T-cell activity was increased the most in the combo Tx setting, indicated by highest T-cell proliferation rates and CD25-activation marker upregulation (Fig. 5F–H).

Current understanding of immunologic properties of EwS is still limited. Recent advances point towards EwS immune suppression and/or immune exclusion by mechanisms including low to absent antigenicity (42), immunosuppressive nonclassical HLA-G/E expression (14), tumor-derived extracellular vesicles impairing DC maturation (18), protumorigenic M2 macrophage polarization (15, 43), presence and activity of T_Regs, as well as myeloid- and fibrocyte-derived suppressor cells (44). It was also observed that EwS are not rich in TILs, and if enriched without prognostic significance (10). Nevertheless, peripheral blood and bone marrow T cells from EwS patients exhibit PD-1 expression suggestive of an exhausted phenotype (45). These observations combined with the low somatic mutation burden in EwS¹ might serve as a reasonable explanation as to why immunotherapeutic approaches including ICB have not been of success (12). Thus, most likely additional combinations will be needed to enable and maintain T-cell infiltration, ideally inducing an IFN-driven inflammation with the capacity to broaden the antitumor T-cell repertoire thereby mediating long-term tumor control. On the basis of this, CDK4/6 inhibitors and oncolytic viruses which act synergistically in tumor cell killing (31) while possessing strong immunogenic properties (46) are of great interest striving to overcome immunosuppressive hurdles of EwS.

Here, we describe EwS-immunologic changes by a novel combination therapy using the oncolytic YB-1-dependent adenovirus XVir-N-31 in combination with CDK4/6 inhibitors. XVir-N-31 has been previously shown to induce a stronger immunogenic cell death than wild-type adenovirus (35, 46). We confirmed increased viral replication with CDK4/6i in a panel of pediatric sarcoma cell lines with a functional RB/E2F pathway in vitro, being in line with our recently published results (31), demonstrating the broad applicability of this approach. In this regard it is noteworthy that CDK2 activation, which is involved in CDK4/6i resistance, is required for adenovirus replication, indicating the possibility also to target CDK4/6i-resistant cancer cells with our combination approach (47). Furthermore, HLA-I upregulation, on gene, protein, and surface expression levels was observed with CDK4/6i in tested in vivo settings in both CDK4/6i monotherapy as well as in combination with XVir-N-31. Therapeutically relevant systemic changes were induced by the combination approach: abscopal effect, preferable engraftment of T cells, and suppression of T_Regs. In addition, induction of PD-L1 was observed both in vitro and in vivo (only shown for in vivo), underlining a strong rational to render pediatric sarcomas susceptible to ICB by this novel treatment setting.

The combo Tx induced IFNγ-induced protein 10 (IP10, CXCL10) in vivo significantly. This was not entirely unexpected because adenovirus infection also leads to activation of mTOR and cell metabolism (48) comparable with CDK4/6i, which was already described to induce higher levels of IFN-downstream chemo- and cytokines (30). The increase of CXCL10 was also associated with increased STAT1 signaling, known to promote antitumor immune responses by rendering innate and adaptive immune cells fully functional. The intratumorally-virus injection in combo Tx-treated animals also induced the infiltration of monocytes to a greater extend compared with the monotherapies, with upregulation of M1 antitumoral macrophage-associated genes in the TME. At the same time, tumor-associated macrophages (TAM) and neutrophils (TAN) were decreased. Furthermore, infiltrating innate cells upregulated HLA class II expression, possibly indicating increased capacities for conventional antigen presentation, compared with mono Tx's and mock. Respective phenomenon was also confirmed by in vitro phagocytosis and DC maturation assays being in line with the downregulation of the don't-eat-me-signal CD47 in A673 xenograft models. This last observation might be of high translation relevance as CD47-blocking antibodies are currently studied in combination therapies in early clinical trials, but this aspect was not further pursued in this study (e.g., NCT05467670). However, correlation analysis of viral fiber DNA with relative numbers of innate immune cells revealed a strong positive correlation in-between viral genomic copies and monocyte infiltration (Fig. 4J), indicating that efficient virus DNA replication in vivo is a precondition to increase immune cell infiltration. Hence boosting viral replication by CDK4/6i most likely increases the inflammatory stimulus cumulating in superior immune cell infiltration.

Although a concurrent attraction of MDSC in our in vivo setting cannot be excluded, as functional testing of tumor-isolated murine myeloid cells was not performed, it is reasonable that respective monocytes are of inflammatory nature, which were shown to mediate important roles in antigen presentation, antiviral immunity with the capacity to differentiate into M1 macrophages or dendritic cells (49, 50). Additional evidence for antitumoral features of infiltrating monocytes in our models can be derived from the upregulation of mouse-specific CXCL10 transcripts within xenografts (Supplementary Fig. S5E). In this regard, House and colleagues demonstrated that macrophage-derived CXCL9 and CXCL10 are determinants for successful ICB (51).

Because of the limitations of our in vivo model concerning the innate and adaptive immune crosstalk, we additionally addressed this subject with in vitro experiments using condition medium-exposed monocytic cells and evaluated their capacity to stimulate antigen-specific T-cell responses. Here, combo Tx was of utmost superiority, further indicating the immunostimulatory and T-cell activating properties of our combination strategy.

Despite the lacking crosstalk in-between innate and adaptive immunity in our model, combo Tx increased human T-cell engraftment in spleens, and led to a consecutive infiltration of T cells into xenografted tumors, exceeding the respective monotherapies significantly. More detailed analyses revealed the preferential engraftment and tumor infiltration of CD4⁺ T cells. Subset characterization showed a T_H1 differentiation and a decrease of CD4⁺ T cells with regulatory properties (T_Regs). Of note, XVir mono Tx did not suppress T_Reg formation but T_Regs remained suppressed in the combo setting and the ratio of TILs compared with T_Regs strongly increased (Supplementary Fig. S3G and S3H). This was of surprise, because adenovirus infection promotes the generation of T_Regs (52) and CDK4/6i treatment was only administered initially for 5 consecutive days, suggesting long-lasting changes of T-cell phenotypes by CDK4/6i. Also, immunosuppressive effects via recruitment and activation of MDSCs through a mechanism dependent on T_Regs (53) is very likely counteracted by our combination strategy (54).

Also, the cognate CXCL10-receptor CXCR3 was strongly induced in representative analyses. TILs also highly expressed PD-1 when present in the tumor. These results point towards a T_H1 antitumor/antiviral response, which might be explained by an adoptively transferred anti-AdV immunity with consecutive specific T cell expansion, as xenografted tumors and adoptively transferred human T cells were matched for HLA-A2. Another possible or additional explanation would be that human T cell engraftment was facilitated by secreted factors induced by combo Tx and secondarily attracted by CXCL10. Since we did not observe an increase in CXCL9 expression it would be interesting to include CXCL9 (55) or other CXCR3 ligands in the backbone of XVir-N-31, which possess enough space for respective transgenes due to its E3 deletion.

Of course, our in vivo studies possess several limitations: using a subcutaneous cell line-derived xenograft model despite an orthotopic and ideally patient-derived xenograft sarcoma model does reflect the clinical situation less (56), but allowed for smooth and more robust experimental execution given the higher complexity of combination therapy evaluation. Also, we used healthy adult donor PBMCs while researching a pediatric tumor entity whose patients in most cases have thymic education still in place.

In conclusion, this novel combo Tx exhibits strong and beneficial immunologic properties addressing a multitude of immunotherapeutic hurdles in EwS and other pediatric sarcomas, exceeding CDK4/6i and OAd as monotherapy, respectively. Herein proposed combination therapy of adenoviral replication-mediated immunoactivation through CDK4/6 inhibition, which we describe as a MRIAT, primes tumor cells for superior adenoviral replication and an immune attack by HLA-I upregulation, boosts oncolysis, and STAT1-driven inflammatory responses (e.g., induction of CXCL10), which pulls antigen-presenting innate cells with antitumoral M1 macrophage characteristics and adoptively transferred T cells into the tumor. Simultaneously T_Reg-inhibition by CDK4/6i persists in the periphery. This might be of clinical importance, because a decrease in T_Reg levels was significantly greater in responder of patients with breast cancer than in nonresponder patients treated with CDK 4/6i, highlighting the importance of immune activation with regard to tumor shrinkage (57). Indeed, both monotherapies with their respective strengths not only seem to complement each other but further boost the antitumor immunostimulatory component of each mono Tx. These indicate that an immune activation threshold, described by Guram and colleagues (58), is needed to realize successful viro-immunotherapy. The combination of OAd and CDK 4/6i with their opposing role on the cell cycle could even be further supplemented with mAbs addressing immune checkpoint molecules or using an upgraded version of XVir-N-31 to express anti-PD-L1, because both treatments have shown to sensitize gliomas to ICB independently, culminating into an improved antitumor response (24, 46). On the basis of these encouraging data, a clinical phase I study of this treatment approach in combination with pembrolizumab is generously supported by Bundesministerium für Bildung, Wissenschaft und Forschung (BMBF, Federal Ministry of Education and Research).

S.J. Schober reports grants from Technical University of Munich and Wilhelm Sander-Foundation and nonfinancial support from Cura Placida during the conduct of the study. A.J. von Ofen reports grants from Technical University of Munich - Translational Medicine Program during the conduct of the study. P.S. Holm reports nonfinancial support from XVir Therapeutics GmbH during the conduct of the study; in addition, P.S. Holm has a patent for CDK4/6 inhibitor in combination with XVir-N-31 pending. No disclosures were reported by the other authors.

S.J. Schober: Conceptualization, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. C. Schoening: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. J. Eck: Data curation, formal analysis, validation, visualization, methodology. C. Middendorf: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. J. Lutsch: Data curation, formal analysis, validation, investigation, visualization, methodology. P. Knoch: Formal analysis, validation, investigation, methodology. A.J. von Ofen: Formal analysis, validation, investigation, visualization, methodology. H. Gassmann: Formal analysis, methodology, writing–review and editing. M. Thiede: Data curation, formal analysis, investigation, methodology. J. Hauer: Funding acquisition, methodology. A. Kolk: Funding acquisition. K. Mantwill: Formal analysis, methodology. J.E. Gschwend: Funding acquisition, writing–review and editing. S.E. Burdach: Conceptualization, supervision, funding acquisition, methodology, writing–review and editing. R. Nawroth: Funding acquisition, investigation, methodology, writing–review and editing. U. Thiel: Conceptualization, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, writing–review and editing. P.S. Holm: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

S.J. Schober and H. Gassmann were funded by the commission for clinical research (KKF) clinician-scientist program of TUM School of Medicine, the Cura Placida Foundation, and the Wilhelm Sander-Foundation (2021.007.1). U. Thiel is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project number 501830041, as well as by the Dr. Robert Pfleger Foundation, the Wilhelm Sander-Foundation, and the Dr. Sepp and Hanne Sturm Memorial Foundation. The clinical translation of the combination therapy for patients with pediatric sarcoma was supported by the Federal Ministry of Education and Research (BMBF) – no.: 01EN2009.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.