ErbB2 (HER2)-CAR-NK-92 cells for enhanced immunotherapy of metastatic fusion-driven alveolar rhabdomyosarcoma

Introduction Metastatic rhabdomyosarcoma (RMS) is a challenging tumor entity that evades conventional treatments and endogenous antitumor immune responses, highlighting the need for novel therapeutic strategies. Applying chimeric antigen receptor (CAR) technology to natural killer (NK) cells may offer safe, effective, and affordable therapies that enhance cancer immune surveillance. Methods Here, we assess the efficacy of clinically usable CAR-engineered NK cell line NK-92/5.28.z against ErbB2-positive RMS in vitro and in a metastatic xenograft mouse model. Results Our results show that NK-92/5.28.z cells effectively kill RMS cells in vitro and significantly prolong survival and inhibit tumor progression in mice. The persistence of NK-92/5.28.z cells at tumor sites demonstrates efficient antitumor response, which could help overcome current obstacles in the treatment of solid tumors. Discussion These findings encourage further development of NK-92/5.28.z cells as off-the-shelf immunotherapy for the treatment of metastatic RMS.


Introduction
Rhabdomyosarcoma (RMS) is the most prevalent type of soft tissue sarcoma in children (1)(2)(3).Adolescents and young adults over the age of 9 with advanced metastatic, primary refractory or relapsed (r/r), alveolar subtype RMS involving the bone or bone marrow (BM) and/or exhibiting fusion-positive (PAX3/7-FOXO1) characteristics have a particularly poor prognosis.These patients are likely to succumb to their disease within an average of two years following diagnosis (2,(4)(5)(6).Even for those who survive, long-term debilitating effects can result from substantial treatment-related toxicity, such as cardiotoxicity and nephrotoxicity.
Developing new therapeutic strategies to overcome tumor resistance remains a critical unmet need for these heavily pretreated patients.Chimeric antigen receptor (CAR)-engineered cells, which target and inhibit the proliferation and spread of tumor cells, are revolutionizing the treatment of r/r malignancies.However, this therapy can also lead to severe, albeit treatable, side effects, such as cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS), and secondary hemophagocytic lymphohistiocytosis/macrophage activation syndrome (7).
The lack of suitable CAR target antigens, as well as the immunosuppressive nature of the tumor microenvironment (TME), and advanced-stage disease, renders CAR-T cells nonresponsive or exhausted against many solid tumors (8, 9).However, targeting ErbB2 (HER2) in metastatic alveolar RMS (aRMS) (10)(11)(12) by the patient's autologous CAR-T cells appeared to be safe and feasible in a phase I/II clinical trial (13), with a complete response observed in a case report with one child (13,14).
Here, we employed a molecularly and functionally well-defined clonal derivative of the natural killer (NK) cell line NK-92 (NK-92/ 5.28.z) as effector cells (15)(16)(17), carrying an ErbB2-specific secondgeneration CAR with a composite CD28-CD3z signaling domain.Compared to other CAR-NK or CAR-T-cell products, good manufacturing practice (GMP) production of the off-the-shelf NK-92/5.28.z cell product in clinically relevant doses is affordable and less complex or labor-intensive (17).Intralesional therapy with third party, off-the-shelf NK-92/5.28.z cells is currently being explored in a phase I clinical trial (NCT03383978) in patients with recurrent ErbB2-positive glioblastoma (16,18).
In a proof-of-concept analysis, we previously demonstrated specific ErbB2-CAR-mediated recognition and killing of ErbB2expressing RMS cell lines by NK-92/5.28.z cells in vitro (19).Here, recognition of target cells triggered surface expression of CD107a and degranulation of NK-92/5.28.z cells, but not of parental NK-92 cells.The heterogeneous expression of tumor antigens, hypoxia, and the immunosuppressive TME of solid tumors adversely impact the accessibility, infiltration, stimulation, activation, and persistence of conventional CAR-engineered T cells at the tumor site (20).The cancer immune surveillance of NK-92/5.28.z cells is triggered by the anti-ErbB2-targeted CAR and germline-encoded cell surface receptors, as well as by supporting adoptive antitumor immune responses of T cells in the TME (21), suggesting homing to, infiltration into, and persistence of NK-92/5.28.z cells at RMS tumor sites.
To further develop the NK-92/5.28.z cell therapy towards clinical application as a consolidation treatment for children and young adults with r/r, metastatic, ErbB2-positive RMS tumors after re-induction radiotherapy/chemotherapy, we performed detailed preclinical in vitro and in vivo safety and efficacy analyses of NK-92/ 5.28.z cells in clinically relevant, experimental metastasis models, including a human (PAX3-FOXO1) aRMS xenograft model.

Cell surface staining for ErbB2 in tumor organoid RMS cells
Tumor organoid RMS cells were stained with phycoerythrin (PE)-conjugated antibody (BioLegend, 324406) for ErbB2 and subsequently analyzed with a FACSCanto 10c flow cytometer (BD Biosciences).Cells were stained according to the manufacturer's instructions.In brief, aliquots of 5x10 5 cells were washed in phosphate-buffered saline (PBS) and subsequently stained with ErbB2 antibody for 20 minutes.After washing, 1x10 4 events were recorded by flow cytometry using FACSDiva software (Version 6.1.3,BD Biosciences).Fluorescence minus one (FMO) controls were used as references.The actual ErbB2 receptor number was quantified using BD Quantibrite Beads (BD Biosciences) according to the manufacturer's instructions.ErbB2 expression was determined using FlowJo Software (Version 10.8.1, Tree Star Inc.).

Cytotoxicity of NK-92/5.28.z cells against 2D tumor organoid RMS cells
The short-term toxicity of NK-92/5.28.z cells against tumor organoid RMS cells was determined using a europium release assay as reported elsewhere (19).In brief, the aRMS tumor cells were labeled with bis(acetoxymethyl) 2,2':6',2"-terpyridine-6,6"-dicarboxylate (BATDA) reagent (PerkinElmer) for 30 minutes.Thereafter, the cells were washed four times in probenecid (Sigma-Aldrich)containing medium.Cell numbers were counted by Neubauerimproved hemocytometer.Cells were seeded into round-bottom 96well plates at a seeding density of 5000 cells per well.Parental NK-92 or NK-92/5.28.z cells were added at effector to target (E:T) ratios ranging from 20:1 to 2.5:1.For maximum lysis, 20% Triton X-100 solution was added.After three hours of cocultivation, 20 μl of supernatant was transferred to another 96-well flat-bottom plate, and 200 μl of europium solution (PerkinElmer) was added.After an additional 15 minutes of incubation, the plates were measured with a Victor 3 1420 fluorometer (PerkinElmer).The specific lysis of the target cells was calculated by following formula: % Specific lysis = experimental release (signal) − spontanous release (signal) maximum relase (signal) − spontanous release (signal) x100 2.4 Cytotoxic activity of NK-92/5.28.z cells against 3D aRMS tumor spheroids The integrity of RH30 target and NK-92/5.28.z effector cells was verified.In brief, aliquots of 5x10 5 cells were washed in phosphate-buffered saline (PBS) and subsequently stained with ErbB2 antibody or an IgG1k isotype control (BioLegend, 400114) for 20 minutes according to the manufacturer's instructions.After washing, 1x10 4 events were recorded by flow cytometry using FACSDiva software (Version 6.1.3,BD Biosciences).In addition, ErbB2-CAR surface expression of NK-92/5.28.z cells was assessed.For this purpose 2x10 6 either NK-92/5.28.z or NK-92 cells were washed once with DPBS and then incubated with Human TruStain FcX ™ (Biolegend) for 20 minutes at 4°C.After another washing step in DPBS, the cells were incubated with 10 μg recombinant ErbB2 with human Fc tag for 20 minutes.After two additional washing steps, cells were stained with APC antibody directed against the Fc tag (Biolegend), 410712 cells, washed again, and analyzed using a BD FACS Canto 10c instrument (BD Biosciences).
RH30-derived aRMS spheroids were generated to assess the cytotoxic activity of NK-92/5.28.z cells to best mimic 3D tumor structures in vitro: tumor cells were resuspended and were counted by Neubauer-improved hemocytometer.5000 RH30 GFP/luc+ cells per well were seeded in 200 ml of RPMI +10% FBS into ultralow attachment 96-well round-bottom plates without prior coating (Corning) and spun down.On day four of culture, 100 μl of supernatant was carefully removed, and 1x10 5 NK-92 or NK-92/5.28.z cells resuspended in 100 μl of NK-92 medium were added.Spheroids without effector cells were used as controls.Every three to four days, the medium was replaced by X-Vivo 10 medium with 5% FFP and 100 IU/ml IL-2.Spheroids were imaged on day 4, 5, 6, 8 and 10 after initial seeding using a Celigo Image Cytometer (Nexcelom Bioscience) with F-theta lens and AVT PIKE camera.The size of spheroids was quantified by the GFP signal using Fiji software (Version 2.3.0)(28).

Therapeutic activity of NK-92/5.28.z cells in a metastatic xenograft mouse model
In vivo experiments were approved by the appropriate government committee (Regierungspräsidium Darmstadt, Darmstadt, Germany; Gen.-Nr.TVA FK/1070) and were conducted according to the requirements of the German Animal Welfare Act.Twenty-seven female 10-to 12-week-old nonobese diabetic (NOD)/severe combined immunodeficient (SCID)/Il2receptorgamma −/− (NSG) mice received sublethal irradiation with 2.5 Gy (Biobeam 2000) (d-1) according to the protocol for previously established metastatic RMS xenograft model closely resembling a clinical situation (29).One day later (d0), 1x10 5 RH30 GFP/luc+ cells resuspended in 100 μl of PBS were injected intravenously (iv) via the tail vein.Considering them as having an immanence risk for disease progression with then limited treatment options as previously shown (29), immunotherapies were applied iv one day after tumor cell injection (d+1) and were given six times in total over a period of four weeks.For analysis of the cytotoxic capacity of parental NK-92 and NK-92/5.28.z cells, mice were randomly divided into three different groups and injected with the following: control animals received medium (X-Vivo 10 medium with 5% FFP and 100 IU/ml IL-2, n=5), and treatment groups either received NK-92 cells (10x10 6 cells each, n=12) or NK-92/5.28.z cells (10x10 6 cells each, n=10).Both effector cell types were resuspended in X-Vivo 10 medium with 5% FFP and 100 IU/ml IL-2.Group size was set based on the experience from previous experiments with the same xenograft model.
Tumor growth was monitored weekly by bioluminescence imaging (BLI) using an IVIS Lumina II system (Perkin Elmer).Mice were anesthetized by isoflurane inhalation and received subcutaneous injections of 150 μg in vivo grade VivoGlo luciferin (Promega) in 100 μl of PBS per mouse.Fifteen minutes later, images were acquired in dorsal and ventral positions and subsequently analyzed by Living Image In Vivo Imaging Software (Perkin Elmer).Uniform regions of interest were used for all mice, and total flux (photon/s) was used for measurement.The signals of dorsal and ventral images were combined into a luminoscore, and statistical analysis of the tumor burden was performed by one-way ANOVA (30).
Altogether, tumor-bearing mice were randomly divided into the following three groups: -Control: one course of six serial infusions with X-Vivo 10 media with 5% FFP and 100 IU/ml IL-2, n=5 -NK-92: 10x10 6 cells, one course of six serial effector cell infusions, n=12 -NK-92/5.28.z: 10x10 6 cells, one course of six serial effector cell infusions, n=10 During the experiment, animals were monitored daily for disease symptoms, xenogenic graft versus host disease (GVHD) and other adverse effects of the NK cell therapies for a maximum of 105 days.Mice with visible signs of disease progression, discomfort or physical abnormalities were painlessly sacrificed by isoflurane anesthesia followed by cervical dislocation.

Preparation of single-cell suspensions from murine organs
Peripheral blood (PB), BM, lung, liver, gut and spleen samples were isolated and analyzed for persistence of human tumor or effector cells: BM was flushed out of the femur and tibia of mice.PB and BM were incubated with red blood cell lysis buffer (RBC Lysis Buffer (10x), BioLegend) according to the manufacturer´s instructions and washed once with PBS.The organs were cut in representative halves, and one half was preserved in formaldehyde for fluorescence microscopy.The other half was incubated with collagenase D solution (Roche, Basel, Switzerland), filtered through a 70 μm cell strainer, and washed with PBS.Aliquots of the cell suspensions were analyzed by flow cytometry and quantitative polymerase chain reaction (qPCR).Sample identity was blinded for the executors of the consecutively analyses.
Analyses were performed with FlowJo Software (Version 10.8.1, Tree Star).

qPCR analysis of effector cells and tumor burden in murine organs
Genomic deoxyribonucleic acid (DNA) from murine organs was extracted using the Extractme Genomic DNA Kit (BLIRT S.A.).The percentage of human cells in the different murine tissues was determined by a quantitative real-time approach specifically amplifying the human albumin gene.In a second step, the proportions of NK-92 or NK-92/5.28.z cells and tumor cells within the human cell fraction were determined by a human-specific short tandem repeat (STR) genotyping approach.The tumor burden of each mouse was quantified in organs previously described as RMS metastatic sites (liver, lung, BM) (31,32).For effector cell quantification, the spleen as a secondary lymphatic organ was also examined.Primers and probes were obtained from Eurofins (Eurofins Genomics), and the STR multiplex polymerase chain reaction (PCR) system Powerplex 16 (Promega) was used.Additionally, the presence of NK-92/5.28.z cells within the organs was verified by qPCR targeting the CD8a hinge region of the CAR construct.

Histology and fluorescence microscopy
For histological analyses, organ sections were fixed in 4% buffered formalin, paraffin embedded, cut, and stained with hematoxylin-eosin (HE).Immunohistochemistry (IHC) antibodies targeting CD56, MYOD1, DESMIN and MYOGENIN were used to evaluate the RMS phenotype of the tumors.The stained tissue sections were rated by a pathologist for the expression of each protein (-, negative; +, low staining; ++, medium staining; +++, high staining).
Immunofluorescence staining was used to observe immune cell infiltration in organs and tumors: organ slices were deparaffinized and rehydrated.After antigen retrieval and a blocking step, primary mouse-anti targeting human CD45 antibody (Abcam, ab227741) and rabbit-against GFP antibody (Invitrogen, A11122) were added.For detection, secondary antibodies F(ab`) 2 fragment goat-anti-mouse conjugated to Alexa 647 (Jackson ImmunoResearch, 115-606-003) and goat-anti-rabbit IgG conjugated to Alexa 488 (Invitrogen, A11008) were used, and nuclei were stained with 4´,6-eiamidino-2phenylindole (DAPI) (Sigma-Aldrich).The localization of immune cells in the tumor tissue relative to blood vessels was determined by staining with rat-anti-mouse Meca32 antibody (BD Biosciences, 553849) and secondary goat-anti-rat Alexa546 antibody (Invitrogen, A-11081).The samples were examined with a BZ-X810 All-in-One In addition, the stained tissue sections with immune cell infiltration were analyzed by a pathologist for xenogenic GVHD.

Statistical analysis
For statistical analyses and graphical presentation, GraphPad Prism software (version 9, GraphPad Software, La Jolla, CA, USA) was used.The results are given as the mean ± standard deviation (SD).Differences between different groups were evaluated by two-tailed Student´s t test or one-way ANOVA using the Bonferroni-Dunn (nonparametric) method.Overall survivals are given as median with 95% confidence interval, Differences between the survival of different treatment groups were analyzed by the log-rank (Mantel-Cox) test.Differences with p< 0.05 (*), p< 0.01 (**), p< 0.001 (***) and p<0.0001 (****) were considered statistically significant.
3.2 Cytotoxic activity of NK-92/5.28.z cells against 3D aRMS tumor spheroids However, tumor cell suspensions or monolayers do not reflect the difficult-to-treat clinical situation of metastatic tumors.Tumor spheroids best represent the 3D structural, spatial heterogeneity and the hypoxia in the TME that suppresses antitumor immunity (33) and thus represent an intermediate between the preclinical in vitro and in vivo tumor metastasis models described below.
For this ErbB2 expression of RH30 cells (Figure 2A) and CAR expression of NK-92/5.28.z cells (Figure 2B) was confirmed.The cytotoxic capacity of parental NK-92 and CAR-engineered NK-92.5.28.z cells was assessed against established, ErbB2-positive RH30 GFP/luc+ tumor spheroids (Figures 2C, D).Coculture analyses were performed for an additional period of six days.NK-92/5.28.z cells inhibited tumor growth and effectively killed 3D cell structures.Furthermore, NK-92/5.28.z cells proliferated extensively, as shown by cell clustering, while in contact with tumor targets.In contrast, the proliferative and cytotoxic capacity of parental NK-92 cells was much less pronounced and the cells partially inhibited the growth of the 3D tumor spheroids.Spheroid size quantification showed significant reduced spheroid size after NK-92 cell immunotherapy compared to that of the untreated controls (p = 0.0351) on day 10 (Figure 2D).Of note, NK-92/5.28.z cell treatment led to significantly reduced spheroid size compared to untreated controls (p = 0.0013) and NK-92-treated spheroids (p = 0.0425) (Figure 2D).Comparable but less effective results were observed at lower E:T ratios (data not shown).
NK-92/5.28.z cell line is a well-defined clinical grade NK-cell product, which may be provided as an third party, off-the-shelf product.However, because the cell line was derived from a 50-yearold patient with non-Hodgkin´s lymphoma, irradiation prior to infusion is mandatory for safety reasons.Here, irradiated NK-92/ 5.28.z cells showed inhibited proliferation (Supplementary Figure 1A), but retained their direct antitumor toxicity for at least 24 hours in vitro (Supplementary Figure 1B).We have previously shown that the therapeutic effects of non-irradiated and irradiated NK-92/5.28.z cells are comparable in mice (15,16).Therefore, nonirradiated NK-92 and NK-92/5.28.z cells were used for in vivo assessments.

Therapeutic activity of NK-92/5.28.z cells in a metastatic RMS xenograft mouse model
Next, antitumor potential of serial injections of NK-92/5.28.z cells was assessed and the impact of the CAR was compared to parental NK-92 cells in vivo.Thereby, NSG mice carrying a metastatic aRMS (RH30 GFP/luc+ ) xenograft were used to assess homing, tumor invasion, persistence, and antitumor effects as well as the xenogenic cytotoxicity (GVHD) of parental NK-92 and NK-92/5.28.z cells.Sequential infusions of effector cells were given preemptively during the low tumor burden period, and mice were followed for 105 days (end of experiment) (Figure 3A).In order to include all control mice in the BLI analyses, day 50 was chosen as the cutoff day for quantification of tumor engraftment assessed by BLI (Figures 3B-D).BLI confirmed reduced tumor size in the mice with NK-92/5.28.z cell treatment compared to the untreated (p = 0.0101) and NK-92-treated (p = 0.0159) animals at the time of first in vivo response assessment on day +50 (Figures 3B-D).In contrast, NK-92 cell immunotherapy showed no significant effect compared to that of the untreated controls (p = 0.2120) (Figures 3C, D).
Tumor lesions were primarily observed in the livers and lungs of mice except in those with complete response to immunotherapy with NK-92/5.28.z cells.Multiple large (macroscopic) lesions were more likely to be present in the untreated and NK-92-treated mice vs. single small (microscopic) lesions in the NK-92/5.28.z-treated mice (Figure 4A).Immunofluorescence staining with an anti-hCD45 antibody revealed persistence of NK-92 and NK-92/5.28.zFrontiers in Immunology frontiersin.orgcells in the analyzed organs.Exemplary images of animals sacrificed on day 80 (control), day 72 (NK-92) and day 73 (NK-92/5.28.z) are shown (Figure 4A).Close proximity of immune and tumor cells illustrates the response to CAR-based therapy in our metastatic RMS model.Spleens of the mice with immune cell therapy were enlarged compared to the untreated controls (Figure 4B).Furthermore, all organs were scored regarding potential toxicity of the effector cells against normal tissues by an experienced pathologist.No tissue damage or tissue alterations through immune cell infiltration were detected.Meca-32, a homodimeric glycoprotein, is primarily expressed on endothelial cells in mice.To better understand how immune cells reached and infiltrated tumors, we performed immunofluorescence staining of Meca-32, positive endothelial cells and hCD45 positive NK-92/5.28.z cells.We found colocalization of hCD45 positive effector cells with Meca-32 positive endothelial cells, suggesting the immune effector cells reach distant tumor sites via the circulatory system and extravasation, therewith augmenting their antimetastatic potential (Figure 5).

Flow cytometry and qPCR
The persistence of NK-92 and NK-92/5.28.z cells at tumor sites was further confirmed by flow cytometry using human anti-CD45 antibody staining (Figure 6A).In the treatment groups, immune effector cells were detected in all analyzed organs, including the blood.However, cells were too low in numbers to perform flow cytometry analyses for surface CAR expression.These results were further confirmed by qPCR.High immune effector cell infiltration was shown in secondary lymphoid organs such as the spleen (Figure 6B).Of note, the persistence of immune effector cells in spleens was more pronounced in the NK-92-treated mice than in the NK-92/5.28.z-treated mice.In addition, qPCR of the tissues detecting the presence of the CD8a linker of the CAR showed that NK-92/5.28.z cells were still detectable, but again to low in numbers to perform flow cytometry analyses for surface CAR expression (Figure 6C).
Quantifiable tumor levels were detected by qPCR (Figures 6D,  E).The number of tumor cells as well as overall tumor burden was highest in liver tissues followed by lung tissues and was lowest in the BM.High tumor levels were more likely to be present in the untreated and NK-92-treated mice than in mice treated with NK-92/5.28.z cell therapy.Due to the strong autofluorescence properties of analyzed organs, which interfered with flow cytometry emission signals, number of remaining RMS cells and ErbB2 surface expression was not assessable by flow cytometry.

Discussion
Severe CRS and/or ICANS, modest antitumor activity, antigen escape, restricted trafficking, and limited tumor infiltration due to the host and/or the TME and bulky diseases are limitations of CAR-T therapy in solid tumors (9,34).A group at Baylor College of Medicine, Houston, Texas, USA, reported safety and a durable remission over more than four years after initiating ErbB2-CAR T-cell infusions in combination with checkpoint inhibition to overcome the immunosuppressive TME in a child with metastatic aRMS (14).
Autologous CAR-T cells are made from the cancer patients' own peripheral blood lymphocytes, therefore transduction efficiency, Tcell subtype distribution, and activation state can vary, affecting overall product composition, quality, and toxicity.We suggest that serial administrations of CAR-engineered innate immune cells such as the NK-92 cell line may translate into a feasible, safe, effective and affordable treatment strategy for r/r metastatic aRMS.In a proof-ofconcept analysis, we previously showed that the CAR-engineered cell line NK-92/5.28.z has strong in vitro cytotoxicity against ErbB2positive tumor cell lines, including aRMS (19).Thereby, the intrinsic natural and immunomodulatory cytotoxicity of NK-92 cells was further enhanced through the expression of the appropriate CAR.Intrinsic cytotoxicity is mainly based on signaling through activating NK cell receptors and is less likely to allow tumor escape by antigen loss as may occur in the case of major histocompatibility complex (MHC)-mediated killing by T cells (35).Therefore, NK-92/5.28.z cells harbor additional antitumor capacity against r/r malignancies.The immune infiltration into the TME of aRMS tumors is controversial.The hypoxic conditions of growing tumors are associated with the emergence of resistant tumor cell clones and immune evasion (42).Accordingly, the absence of T and B cells and the presence of an extracellular matrix (ECM) and cancerassociated fibroblasts was reported by different groups (43,44).In contrast to embryonal RMS, CD54-positive microvessels are almost absent in aRMS, resulting in low numbers of tumor-infiltrating T cells and M1 macrophages.This phenomenon is considered one potential reason for the poor prognosis of immunologically cold alveolar subtype RMS (44).Recently published single-cell transcriptomic data indicate higher numbers of T and NK cells, but also of M2 macrophages within the TME, that may support immune escape (45).The lack or suppression of proinflammatory immune cell infiltration suggests that the ECM and hypoxic and   to activation of the adaptive immune system in addition to the intrinsic and CAR-mediated cytotoxicity of NK-92/5.28.z cells is an important aspect when considering systemic and repetitive application of irradiated NK-92/5.28.z cells for the treatment of patients.Neither NK-92 cells nor NK-92/5.28.z cells displayed significant toxicity, indicated by the absence of symptoms of CRS, xenogenic GVHD, or histological damage in GVHD-and tumortargeted organs, sometimes seen against tumor cells with low-level antigen expression.Nevertheless, for safety reasons irradiation of human NK-92/5.28.z cell line derived from a 50-year-old patient with non-Hodgkin´s lymphoma is mandatory.Irradiation with 10 Gy resulted in a limited life span and expansion potential thereby providing safety but not reduced anti-RMS cytotoxic capacity of NK-92/5.28.z cells within 24 hours thereafter (17,48).In summary, our data show that NK-92/5.28.z cells exhibit high cytotoxicity in r/r ErbB2-positive 2D patient-derived tumor organoid RMS cells and 3D aRMS tumor spheroid models in vitro, which was reproduced and confirmed in an experimental metastasis model in NSG mice in vivo not representing the adoptive immunity of bystander cells.NK-92/5.28.z cells were capable of migrating toward and persisting at tumor sites.As long as serial infusions of NK-92/5.28.z cells were applied, tumor growth and development were inhibited, which resulted in increased survival of the mice.Furthermore, tumors again responded to a second course of NK-92/5.28.z cell therapy, suggesting that no treatment-induced antigen loss occurred (data not shown).In vivo treatment of mice with parental NK-92 cells had no significant effect, indicating that the observed therapeutic efficacy of NK-92/5.28.z cells was mainly mediated by the second-generation ErbB2-CAR construct.
NK-92/5.28.z cell therapy is already under safety evaluation in patients with recurrent glioblastoma, which is an advantage when considering to extend clinical application of this ready-to-use CARengineered product to other cancer indications such as metastatic aRMS.We therefore propose to use NK-92/5.28.z cells for consolidation therapy of r/r, metastatic ErbB2-positive RMS following re-induction radiotherapy/chemotherapy to benefit from their multiple endogenous cytolytic pathways in addition to direct CAR-mediated killing.Thereby, the limited life-span and lack of in vivo expansion of NK-92/5.28.z cells upon irradiation may be addressed by repeated dosing and combination with checkpoint inhibition or other anticancer therapies that could further enhance or benefit from the immunomodulatory activity of the cells.

2 FIGURE 1
FIGURE 2 Cytotoxic capacity against ErbB2-positive 3D tumor spheroids.(A) ErbB2 surface expression on RH30 cells was confirmed by flow cytometry.(B) Expression of the anti-ErbB2-targeted CAR on NK-92/5.28.z cells was confirmed by flow cytometry.(C) Exemplary coculture images of NK-92 or NK-92/5.28.z cells added to established 3D RH30GFP/luc+ tumor spheroids on day 4 are shown.Imaging was performed on days 4, 5, 6, 8 and 10. (D) The areas with green fluorescent signals were quantified using Fiji.Data of three independent experiments are shown as mean ± SD.Differences were analyzed with a one-way ANOVA using the Bonferroni method and were considered significant for p< 0.05 (*), p< 0.01 (**) or ns.

3
FIGURE 3 Therapeutic potential of NK-92 and NK-92/5.28.z cells in a metastatic xenograft NSG mouse model.(A) An experimental metastasis model in NSG mice carrying low tumor burden of a human aRMS (RH30GFP/luc+) xenograft was established to evaluate the therapeutic potential of NK-92 and NK-92/5.28.z cells in vivo.(B-D) Tumor development and growth was monitored and quantified by BLI combining both dorsal and ventral signals described by the luminoscore for each animal, shown by five representative examples.The mean ± SD is shown and differences were analyzed with a one-sided ANOVA using the Bonferroni-Dunn (nonparametric) method.Differences with p< 0.05 (*) were considered statistically significant.(E) The overall survival of untreated control (n=5), NK-92 cell-treated (n=12) and NK-92/5.28.z cell-treated (n=10) mice is shown.Differences between the survival of different treatment groups were analyzed by the log-rank (Mantel-Cox) test.Differences with p< 0.05 (*) were considered statistically significant.Black arrows indicate animals selected for exemplary images in Figure 4A.

4
FIGURE 4Macroscopic tumor lesions and immunofluorescence staining of metastatic tumors.(A) Tumor cells (Alexa 488, green) and immune cells (Alexa 647, red, in liver and lung additionally indicated by arrows) are shown; cell nuclei were stained with DAPI.Exemplary animals with tumor engraftment were selected from BLI analysis and were sacrificed on day 80 (control), day 72 (NK-92) and day 73 (NK-92/5.28.z), respectively (black arrows in Figure3D).Immunofluorescence staining showed that NK-92 cells, and to a more pronounced extent NK-92/5.28.z cells, preferentially homed to spleens as well as to metastatic tumor sites in lung and liver, resulting in disseminated tumors (control), localized tumors (NK-92), and small tumor nodules (NK-92/5.28.z).Low level NK-92 and NK-92/5.28.z cell infiltration was observed within metastatic tumor sites.(B) The size of the spleen in each treatment group is shown.Differences were considered significant for p< 0.01 (**) or ns.
Cytokines and chemokines released by activated NK-92/5.28.z cells interferon (INF)-g, granulocyte macrophage-colony stimulating factor (GM-GSF) and tumor necrosis factor (TNF)-a), which are essential signals of both innate and adoptive immunity, are likely to support the antitumor toxicity of bystander cells in vivo but are less likely to induce severe CRS or ICANS upon target engagement than CAR-T cells(17,19, 36,37).Unlike CAR-T cells, NK-92/5.28.z cells may be used as a third-party, ready-to-use, off-the-shelf product.Indeed, the alloreactivity of NK-92 cell-centered cell therapy is known to be low even across human leucocyte antigen barriers(38)(39)(40)(41).Therefore we propose NK-92/5.28.z cells as a consolidation treatment for children and young adults with r/r, metastatic ErbB2-positive RMS tumors after re-induction radiotherapy/chemotherapy.Previous studies showed selective recognition and elimination of ErbB2-positive breast carcinoma(15) and glioblastoma (16) cells with high target antigen expression levels by NK-92/5.28.z cells.Despite low and heterogeneous ErbB2 expression levels, we confirmed sufficient and ErbB2-specific antitumor capacity of NK-92/5.28.z cells against both, r/r 2D patient-derived tumor organoid RMS cells and 3D RMS tumor spheroid models.Here, susceptibility to lysis of NK-92/5.28.z cells was not exclusively related to surface ErbB2 expression of target cells.However, good accessibility of the target antigen on tumor organoid RMS cells and tumor spheroids for NK-92/5.28.z cells led to efficient in vitro stimulation, activation and expansion of the cells.Sufficient infiltration into the TME of 3D tumor spheroids further rendered NK-92/5.28.z cells fit and responsive without the tumor cells evading NK-92/5.28.z cell cytotoxicity redirected toward a single tumor antigen.The in vivo antitumor activity of sequential infusions of parental NK-92 and CAR-engineered NK-92/5.28.z cells was evaluated in an experimental metastasis model in NSG mice carrying an aRMS xenograft.We confirmed that both the metastatic behavior and the histopathological characteristics of the PAX3-FOXO1 positive human aRMS cell line RH30 GFP/luc+ were retained in vivo.Compared to the control and treatment with parental NK-92 cells, sequential infusions of NK-92/5.28.z cells significantly prolonged survival, suppressed tumor development and tumor growth and inhibited tumor engraftment in two out of ten mice.Analyses of tumor-targeted organs by PCR revealed the lowest tumor burden in the NK-92/5.28.z cell-treated animals underscoring the response without immune evasion to CARbased therapy in metastatic RMS.The in vivo trafficking and biodistribution of both NK-92 and NK-92/5.28.z cells were confirmed by flow cytometry, IHC and PCR in various tissue samples, including secondary lymphoid organs such as spleens.Close proximity of NK-92 and NK-92/5.28.z cells to tumor sites in livers and lungs as well as infiltration via the circulatory system was determined by fluorescence microscopy analysis.Detection of NK-92 and NK-92/5.28.z cells suggests intrinsic and CAR-specific immune responses that enable activation, expansion, and persistence of the effector cells in vivo.

FIGURE 5
FIGURE 5Localization of effector cells within the tumor tissue.Immunofluorescence staining of blood vessels (green) and NK-92/5.28.z cells or NK-92 cells (right image) (red, indicated by arrows) in liver tumors.Animals were sacrificed on day 72 or day 73.

TABLE 1
Overview of RMS tumor cells used in this study.