Low Energy Electron Irradiation Is a Potent Alternative to Gamma Irradiation for the Inactivation of (CAR-)NK-92 Cells in ATMP Manufacturing

Background With increasing clinical use of NK-92 cells and their CAR-modified derivatives in cancer immunotherapy, there is a growing demand for efficient production processes of these “off-the-shelf” therapeutics. In order to ensure safety and prevent the occurrence of secondary tumors, (CAR-)NK-92 cell proliferation has to be inactivated before transfusion. This is commonly achieved by gamma irradiation. Recently, we showed proof of concept that low energy electron irradiation (LEEI) is a new method for NK-92 inactivation. LEEI has several advantages over gamma irradiation, including a faster reaction time, a more reproducible dose rate and much less requirements on radiation shielding. Here, LEEI was further evaluated as a promising alternative to gamma irradiation yielding cells with highly maintained cytotoxic effector function. Methods Effectiveness and efficiency of LEEI and gamma irradiation were analyzed using NK-92 and CD123-directed CAR-NK-92 cells. LEE-irradiated cells were extensively characterized and compared to gamma-irradiated cells via flow cytometry, cytotoxicity assays, and comet assays, amongst others. Results Our results show that both irradiation methods caused a progressive decrease in cell viability and are, therefore, suitable for inhibition of cell proliferation. Notably, the NK-mediated specific lysis of tumor cells was maintained at stable levels for three days post-irradiation, with a trend towards higher activities after LEEI treatment as compared to gamma irradiation. Both gamma irradiation as well as LEEI led to substantial DNA damage and an accumulation of irradiated cells in the G2/M cell cycle phases. In addition, transcriptomic analysis of irradiated cells revealed approximately 12-fold more differentially expressed genes two hours after gamma irradiation, compared to LEEI. Analysis of surface molecules revealed an irradiation-induced decrease in surface expression of CD56, but no changes in the levels of the activating receptors NKp46, NKG2D, or NKp30. Conclusions The presented data show that LEEI inactivates (CAR-)NK-92 cells as efficiently as gamma irradiation, but with less impact on the overall gene expression. Due to logistic advantages, LEEI might provide a superior alternative for the manufacture of (CAR-)NK-92 cells for clinical application.

Background: With increasing clinical use of NK-92 cells and their CAR-modified derivatives in cancer immunotherapy, there is a growing demand for efficient production processes of these "off-the-shelf" therapeutics. In order to ensure safety and prevent the occurrence of secondary tumors, (CAR-)NK-92 cell proliferation has to be inactivated before transfusion. This is commonly achieved by gamma irradiation. Recently, we showed proof of concept that low energy electron irradiation (LEEI) is a new method for NK-92 inactivation. LEEI has several advantages over gamma irradiation, including a faster reaction time, a more reproducible dose rate and much less requirements on radiation shielding. Here, LEEI was further evaluated as a promising alternative to gamma irradiation yielding cells with highly maintained cytotoxic effector function.
Methods: Effectiveness and efficiency of LEEI and gamma irradiation were analyzed using NK-92 and CD123-directed CAR-NK-92 cells. LEE-irradiated cells were extensively characterized and compared to gamma-irradiated cells via flow cytometry, cytotoxicity assays, and comet assays, amongst others.

INTRODUCTION
The natural killer cell line NK-92 is derived from the peripheral blood of a lymphoma patient and is used as an anti-cancer advanced therapy medicinal product (ATMP) due to its high cytotoxic activity against tumor cells and its "off-the-shelf" availability (1). Further enhancement of NK-92 efficacy was achieved by chimeric antigen receptor (CAR)-modification of the cell line (2). Currently, CAR-NK-92 cells are investigated as alternative effector cells in comparison to autologous CAR-T cells. For acute myeloid leukemia (AML), a possible target is the interleukin 3 receptor alpha (IL-3Ra, also known as CD123), which is overexpressed in AML [summarized in (3)] and expression levels were shown to correlate with clinical prognosis and outcome (4)(5)(6). Previously, we demonstrated the in vitro efficacy of CD123-directed CAR-NK-92 cells (7).
Due to the malignant origin of NK-92 cells, cell inactivation prior to their application is indispensable in order to stop cell proliferation and prevent possible tumorigenesis (2,8,9). For all clinical trials published so far, the gold standard has been gamma irradiation of NK-92 cells at a dose of 10 Gy prior to infusion (10,11). Physically, the effect of ionizing radiation (IR) is based on breakage of atomic bonds with subsequent formation of radicals. The damaging effects are therefore categorized into primary reactions, i.e. direct DNA damage, and secondary reactions from water radiolysis products, including reactive oxygen species (ROS) (12,13). Induced structural DNA damage thereby includes singleor double-strand breaks (DSBs), cross-linkage breaks and nucleotide degradation (13)(14)(15). As a result, apoptosis, necrosis and senescence (13,16) as well as a strong impairment of cell membranes and proteins can occur (13,17).
In addition to the conventional gamma irradiation, low energy electron irradiation (LEEI) has emerged as a novel method for NK-92 cell inactivation (18). Since the penetration depth of the accelerated low energy electrons is limited to < 1 mm in water (19), a thin film of liquid has to be generated to ensure complete irradiation (18). Compared to gamma irradiation, LEEI allows delivery of a high dose rate, which leads to a shorter treatment time (18,20). One of the greatest benefits of LEEI is the generation of only small amounts of secondary radiation (the Bremsstrahlung, X-rays) (18). This is the reason why LEEI facilities do not need to be equipped with complex shielding systems and therefore can easily be implemented in basic laboratories, even as an in-line tool in GMP environments (21). Furthermore, the applied LEEI doses can be adjusted by current intensity, resulting in a wellcontrollable on-off-process with a very good reproducibility, a great advantage over gamma radiation, which is caused by the spontaneous decay of radioactive material (18).
Many preclinical and all clinical studies involving NK-92 cells employ IR as a tool for cell inactivation, but few reports are available on cellular effects caused by this treatment. In our present study, an extensive characterization of both gamma-and LEE-irradiated NK-92 as well as CD123-directed CAR-NK-92 cells with regard to descriptive and functional attributes was performed. Our data indicate the potential of LEEI for manufacturing enhanced NK-92-based cellular therapeutics with preserved cytotoxic activity.

Cell Irradiation
For LEEI, cells were irradiated as described previously (18), using a custom-built irradiation device, equipped with an electron emitter (type EBA 300/270/4, ebeam Technologies, Switzerland) and a custom-made bag module. Cell suspensions were placed into sterile disposable bags and processed at a bag transportation velocity of 5 mm/s through the LEEI-area. Irradiation parameters (200 keV and 0.01 mA, 0.03 mA, or 0.05 mA) correspond to doses of 2.2 Gy, 6.6 Gy, or 11 Gy, respectively, which were calculated based on dosimetry at higher doses and a reduction factor of 0.01 for the irradiation with a slit diaphragm, as described previously (18).
Gamma irradiation was performed at a dose of 10 Gy, using a Cs-137 source (GSR C1, serial number 09/13, Gamma Service Medical GmbH), since 10 Gy represents the clinical gold standard. For both LEEI and gamma irradiation, up to 1.5 × 10 8 cells were irradiated in a volume of 20 mL of cell culture medium. Directly after irradiation, the cell concentration was adjusted to 0.3 × 10 6 viable cells/mL with fresh cell culture medium. Afterwards, the cell concentration was adjusted 2 -3 times per week to 0.3 -0.5 × 10 6 viable cells/mL. In order to check for irradiationindependent effects, control cells were processed in parallel without applying irradiation (process control, data not shown).

Cell Number and Viability
In order to monitor cell inactivation after irradiation, cell counts and viability were determined using a Countess II Automated Cell Counter (ThermoFisher Scientific). According to the manufacturer's instructions, cells were stained with trypan blue solution (Thermo Fisher Scientific) at a ratio of 1:2. All counts were performed in technical triplicates. Alternatively, as indicated in the figure legend, a CASY cell counter & analyzer (OMNI Life Science) was used according to the manufacturer's instructions.

Flow Cytometry Analysis
For flow cytometric analysis of cell surface markers or cell death, the following antibodies were used: APC-Cy7 anti-human NKG2D (BioLegend), PE anti-human NKp30, APC antihuman NKp46, BV421 anti-human CD56, APC Annexin V and 7-aminoactinomycin D (7-AAD) (all from BD Bioscience). Up to 5 × 10 5 cells were washed with 1 × phosphate-buffered saline ( P B S ) s u p p l e m e n t e d w i t h 5 % F B S a n d 2 . 5 m M ethylenediaminetetraacetic acid (EDTA). Antibody staining was performed at 1:15 -1:100 dilution for 30 min at room temperature, protected from light, followed by two further washing steps. For Annexin V staining, cells were resuspended in 1 × Annexin V binding buffer (BD Bioscience), Annexin V was added for 20 min, and cells were washed again in Annexin V binding buffer. Ten minutes prior to measurement, 7-AAD was added.
Data acquisition was performed using a BD FACSCanto ™ II and data were analyzed using BD FACSDiva ™ software. CS&T beads (BD Bioscience) were used for quality control. Recorded events were gated for living cells (FSC-A × SSC-A), then for single cells (FSC-H × FSC-A), and finally for the respective marker. At least 10,000 events were recorded in the live gate. For determination of irradiation-induced cell death, cells were gated from all events (FSC-A × SSC-A) excluding debris.

Cytotoxicity Assay
Cellular cytotoxicity was determined by chromium release assay. Target cells were incubated with 50 µCi of chromium-51 radionuclide (Hartmann analytic) per 6 × 10 5 cells for 2 h at 37°C and 5% CO 2 . Cells were washed three times and coincubated with effector cells at an effector to target (E:T)ratio of 5:1 for 4 h at 37°C and 5% CO 2 . To determine spontaneous chromium-51 release, target cells were incubated with medium; to determine maximum release, cells were incubated with 1% Triton-X100. After coincubation, cells were centrifuged and 50 µL of supernatant were added to 150 µL of scintillation mix (Optiphase HiSafe, Perkin Elmer). Scintillation counts were measured for 1 min/well using a Perkin Elmer MicroBeta Trilux 1450 LSC & Luminescence Counter. Specific lysis was calculated using the following formula: Specific lysis = [(test releasespontaneous release)/(maximum releasespontaneous release)] × 100.

Multiplex Bead Assay for Quantification of Soluble Analytes
Released levels of soluble analytes from NK-92 cells and CD123directed CAR-NK-92 cells were quantified by the bead-based LEGENDplex ™ human CD8/NK panel immunoassay (BioLegend). Effector and target cells were coincubated at a ratio of 5:1. To determine basal level of analytes, effector and target cells were cultured separately. After 2 h of incubation at 37°C, the plate was centrifuged and 100 µL supernatant were collected. To avoid contaminations, the supernatant was centrifuged again and final supernatant was removed and stored at -80°C. Afterwards, the assay was performed according to the manufacturer's instructions. For qualification and quantification of collected data, analysis was performed with the LEGENDplex ™ Data Analysis Software.

Analysis of Metabolism
For analysis of metabolic capacity, the colorimetric WST-1 assay (Roche) and the luminescence-based CellTiter-Glo ® assay (Promega) were performed according to the manufacturer's instructions. Plates were incubated at 37°C for 3.5 h and were measured directly by a plate reader (Tecan Infinite M200).

Alkaline Comet Assay
For further assessment of DNA-DSBs, alkaline comet assays were performed 2 h and 24 h after irradiation under red light to avoid UV-induced unspecific DNA damage according to Ziemann et al. (23). Per sample, 150,000 cells were collected and resuspended in 80 µL of 0.75% (w/v) pre-conditioned lowmelting-point agarose (LMA, Sigma-Aldrich), and applied to slides pre-coated with normal-melting-point agarose (NMA, Sigma-Aldrich). These gels were covered with coverslips, allowed to set at 4°C, followed by addition of a second LMA layer and another solidification step. The coverslips were removed and the slides were incubated in lysis solution overnight at 4°C. After three washing steps in electrophoresis buffer (pH > 13), slides were placed in a pre-cooled electrophoresis chamber, and DNA was allowed to unwind for 20 min before electrophoresis was carried out at 26 V and 300 mA for 20 min. Finally, slides were neutralized and stained with ethidium bromide solution (20 µg/mL, Merck-Millipore). Slides were semi-automatically analyzed for occurrence of DNA damage using an Axioskop fluorescence microscope (Carl Zeiss) and the Comet Assay III software (Perceptive Instruments). Tail intensity (TI) of 100 nuclei per slide was determined, precluding so-called 'hedgehogs' or overlapping nuclei/comets from analysis.

Propidium Iodide (PI) Staining
For assessment of cell cycle phase distribution, up to 5 × 10 5 cells were fixed at the indicated time point after irradiation using 70% ethanol. After two washing steps with 0.1% bovine serum albumin in PBS, RNA was digested with RNase A (100 µg/mL final concentration, Thermo Fisher Scientific) and PI was added at a final concentration of 50 µg/mL.

Light Microscopy
Microscopic images were recorded on a Leica DMIL inverse microscope with a Leica EC3 camera and analyzed using Leica LAS EZ software.
Over-representation analyses (ORA) was performed to identify signalling pathways enriched with genes significantly differentially regulated (FDR < 0.01) between LEEI or gamma irradiation vs. control samples. For assignment of genes to pathways the curated database Reactome pathway database (30) was used. Over-representation analyses was conducted using the R-packages ReactomePA (v1.34.0) (31) and clusterProfiler (v3.18.1) (32). Pathways with less than 10 or more than 500 genes were excluded. All pathways with an adjusted p-value < 0.01 for the hypergeometric test were reported as enriched pathways.

Statistical Analysis
GraphPad Prism version 6 or 8 (GraphPad Software, Inc., USA) was used for statistical analysis of all data. GraphPad QuickCalcs was used for determination of outliers with a significance level of a = 0.05. Normal distribution was determined using D'Agostino and Pearson omnibus normality test. In case of normally distributed data, analysis of variance (ANOVA) was used; in case of not normally distributed data, Kruskal-Wallis test or Mann-Whitney test was used, as indicated. P-values were adjusted for multiple comparisons by Dunn's multiple comparisons test and are symbolized by asterisks as follows: * for p ≤ 0.05, ** for p ≤ 0.01, *** for p ≤ 0.001, and **** for p ≤ 0.0001.

NK-92 and CD123-Directed CAR-NK-92 Cell Proliferation Is Fully Inhibited by LEEI and Gamma Irradiation
In our previous proof of concept work, it was shown that LEEI is suitable for growth inhibition of NK-92 cells at a dose of 11 Gy (calculated dose for the process parameters 200 keV and 0.05 mA) (18). We here aimed to determine the lowest dose parameters able to efficiently block proliferation of CD123directed CAR-NK-92 cells by applying LEEI at 200 keV and 0.01 mA, 0.03 mA, or 0.05 mA and analyzing the cell number and viability for 4 days after irradiation ( Figure 1A). Cells irradiated at an amperage of 0.01 mA showed reduced proliferation and viability, compared to non-irradiated cells, however, inhibition was only partial as cell numbers increased over time. In contrast, cells irradiated at an amperage of 0.03 mA and 0.05 mA showed a decrease in cell number and viability. Based on these results, an amperage of 0.05 mA was used for further experiments, as this dose showed consistent inhibition of cell growth and gave a safety margin to the parameter of 0.01 mA, which is important for implementing LEE irradiation into perspective clinical studies and uses.

LEE-Irradiated Cells Show High In Vitro Functionality
A crucial necessity for the therapeutic efficacy of NK cells is the maintenance of the cytotoxic potential. Based on the application regimens used in clinical trials of NK-92 and CAR-NK-92 cells, we defined a therapeutic window of up to three days postirradiation. Therefore, cytotoxic activity mediated by NK-92 cells against K562 target cells was determined on days 1, 2, and 3 after gamma irradiation or LEEI by chromium release assays (Figure 2A). For the first two days after irradiation, the cytotoxic activity of the NK-92 cells against K562 cells was not significantly impaired (specific lysis = 97.7 ± 8.7% for non-irradiated cells, 82.6 ± 9.6% for LEE-irradiated cells and 80.2 ± 7.9% for gammairradiated cells on day 2). Three days after irradiation, however, a significant reduction of cell-mediated cytotoxicity was detected in gamma-irradiated NK-92 cells (50.3 ± 7.5%, p = 0.0098 compared to non-irradiated cells, Kruskal-Wallis test). In contrast, LEE-irradiated cells showed a preserved cytotoxic activity on day 3 compared to the non-irradiated control (94.9 ± 3.5% for non-irradiated cells, 67.8 ± 7.0% for LEEirradiated cells, p = 0.2327, Kruskal-Wallis test). For CD123directed CAR-NK-92 cells, cytotoxicity against the CD123expressing AML cell line KG-1 was measured for 3 days postirradiation ( Figure 2B). As for NK-92 cells, no significant

Irradiation Causes Decrease in CD56 Surface Expression
In order to determine the effects of radiation on surface protein expression, flow cytometric analysis of CD56, a marker of mature functional NK-92 cells (33,34), was performed ( Figure 3). Overall surface expression of CD56 in NK-92 ( Figure 3A) and CD123-directed CAR-NK-92 cells ( Figure 3B) was not affected by gamma irradiation or LEEI, however, a population with considerable downregulation of CD56 (termed CD56 low ) of 20-50% emerged. To clarify, why the CD56 surface expression was downregulated after irradiation, the CD56 low ( Figure 3C) and CD56 high (Supplemental Figure S2B) populations were analyzed regarding cell death distribution using Annexin V/7-AAD staining (re-analysis of data shown in Figure 1D). A schematic representation of the gating strategy is provided in Supplemental Figure S2A. The CD56 low subpopulation of NK-92 and CD123-CAR-NK-92 cells revealed a significantly higher proportion of Annexin V/7-AAD double positive cells ( Figure 3C), compared to the CD56 h i g h population (Supplemental Figure S2B) (p < 0.0001, Mann Whitney test). Still, only one third of the CD56 low population represented dead cells, whereas the majority of CD56 low cells were viable (negative for both 7-AAD and Annexin V).
Additional markers of NK cell activation (NKp46, NKG2D and NKp30) were analyzed and showed no differences in surface expression levels after irradiation in NK-92 and CD123-directed CAR-NK-92 cells (Supplemental Figures S3A, B).
In order to determine whether the CD56 downregulation of irradiated cells correlated to reduced cell metabolism, the metabolic activity of irradiated NK-92 and CD123-CAR-NK-92 cells was investigated by determining the bioreductive activity for generation of adenosine triphosphate (ATP), as well as cellular ATP levels (Supplemental Figures S3C, D). Nonirradiated as well as irradiated cells of both cell lines maintained their metabolic capacity up to 3 days posttreatment, except for a significant decrease in ATP levels of gamma-irradiated NK-92 cells on day 1 (63.4 ± 9.5%, p = 0.0257, compared to non-irradiated cells, Kruskal-Wallis test).

Irradiation Causes DNA Damage and an Accumulation in G2/M Phase
Since the induction of DNA damage is a key effect of radiation and necessary to guarantee proliferation inhibition, the alkaline comet assay was performed to investigate DNA strand break formation and a potential DNA repair response after gamma irradiation or LEEI. CD123-directed CAR-NK-92 cells were investigated 2 h ( Figure 4A, left) and 24 h ( Figure 4A, right) post-irradiation. Both gamma-and LEE-irradiated cells showed higher TIs than the untreated controls, thus indicating induction of DNA damage (non-irradiated control 2 h: 1.34 ± 0.23%; LEEirradiated 2 h: 5.37 ± 0.91%, p = 0.0004; gamma-irradiated 2 h: 5.79 ± 0.56%, p < 0.0001; non-irradiated control 24 h: 1.10 ± 0.28%; LEE-irradiated 24 h: 6.94 ± 2.25%, p = 0.0017; gamma-irradiated 24 h: 9.04 ± 2.50%, p = 0.0094;Kruskall-Wallis test compared to non-irradiated cells). Comparison of the mean TI values 2 and 24 h after irradiation indicated no substantial DNA repair, but rather progressive DNA damage. TI values after LEEI and gamma irradiation were statistically not significantly different at either time point.
These data led to the hypothesis that accumulation of DNA damage and activation of DNA damage checkpoints cause irradiated cells to arrest in the G2/M phases (35). Therefore, the distribution of cell cycle phases was analyzed via PI staining and, indeed, revealed an increasing amount of cells in G2/M phases after irradiation (Figures 4B, C). The percentage of cells in the G1 phase decreased from day 0 after irradiation to day 3, whereas the percentage of cells in the G2/M phases increased from day 2 (up to 58%). Increased numbers of apoptotic cells in the Sub-G1 phase were detected after LEEI and gamma irradiation (up to 20% on day 6). Both irradiation methods also led to the generation of cells with an increased DNA content (up to 56% on day 6), indicating occurrence of polyploid cells. Morphologically, both LEE-and gamma-irradiated cells showed a drastic increase in cell size as well as dissolution of cell clusters accompanied by vast amounts of debris ( Figure 4D).

LEEI Results in Decreased Differential Gene Expression Compared to Gamma Irradiation
In order to analyze the impact of the irradiation-induced DNA damage on gene expression, NGS-based measurement of the transcriptome of CD123-directed CAR-NK-92 cells was performed 2 h after LEEI or gamma irradiation ( Figure 4E). In these initial experiments, we found a striking difference in the number of differentially regulated genes between both irradiation types: 2 h post-irradiation, gamma-irradiated cells showed 3441 differentially regulated genes (FDR < 0.01), whereas LEE-irradiated cells showed only 265 differentially regulated genes (FDR < 0.01), when compared to non-irradiated control cells ( Figure 4E). There were 193 common genes, which were differentially expressed after LEEI as well as after gamma irradiation.
Over-representation analysis (ORA) based on the Reactome database identified 35 pathways enriched in both LEEI or gamma irradiation vs. non-irradiated samples (Supplemental Table S1). These included pathways for non-sense mediated decay and pathways involved in translation. Genes differentially expressed in the comparison of gamma-and non-irradiated cells were additionally enriched in pathways related to nucleotide excision repair (NER), DNA damage and cell cycle regulation (Supplemental Table S1). The only pathway exclusively enriched in LEEI-treated cells (and not after gamma irradiation) was the RAF-independent MAPK1/3 activation pathway (Supplemental Table S1).

DISCUSSION
One important obstacle of (CAR-)NK-92 therapy is the safe inactivation of these therapeutic cells prior to application, which is crucial for the prevention of secondary tumor development (2,36). Finding a balance between preventing the proliferation while maintaining a high cytotoxic potential is therefore of paramount importance. Especially on account of its "off-theshelf" availability for allogeneic transfusion (2, 36), CAR-NK-92 therapy is currently used as an alternative to CAR-T cell therapy, and could also be successfully applied as a supplemental therapy. Several publications have previously shown that gamma irradiation at a dose of 10 Gy is suitable for inactivation of (CAR-)NK-92 cells in vitro (37)(38)(39)(40)(41)(42)(43)(44), in mouse models (38,40,(43)(44)(45)(46), as well as in first clinical trials (10,11,47). As an alternative method, we recently described the application of LEEI to inactivate NK-92 cells and demonstrated proof of concept that this technique can indeed render these cells proliferationincompetent (18). Addressing the question whether it might be used as an alternative to gamma irradiation, we here provide a detailed comparison between both irradiation methods for the inactivation of NK-92 or CD123-directed CAR-NK-92 cells. Analysis of cell number and viability showed that both irradiation methods fully cease proliferation. On a functional level, the cytotoxic potential of irradiated NK-92 and CD123-CAR-NK-92 cells was stably maintained for up to two days post-irradiation. Other reports in the literature confirm that in vitro cytotoxicity as well as cytokine secretion are maintained for 24 -48 h after gamma irradiation (37,38,40,41,44). Three days after irradiation, we observed a reduction in cytotoxic activity, which was statistically significant only for gamma irradiation (the difference between gamma irradiation and LEEI was not statistically significant). This maintenance of cytotoxic activity could be due to the short exposure time of electron beams or due to the fact that LEEI initiates more direct damage to the DNA through targeted hits from electrons, while the effect of gamma radiation is predominantly mediated by secondary reactions from ROS, which result in increased oxidative stress. On day three post-irradiation, a decrease in the effector molecules sFasL and perforin was observed, whereas other soluble analytes, including IL-17A, granzyme B, granulysin, and IFNg, were not influenced by irradiation or even showed a slight irradiation-induced increase. In accordance with this data, Nowakowska et al. described a decrease of sFasL levels, an increase of IFNg levels as well as constant levels of granzyme B after gamma irradiation of Her2directed CAR-NK-92 cells (37). Furthermore, irradiationinduced increases of IFNg-levels were described in T cells (48). Interestingly, different cytokine secretion patterns were observed when comparing NK-92-mediated lysis of K562 cells and CAR-NK-92-mediated lysis of KG-1 cells, which could indicate that different combinations of effector molecules are generated in unmodified NK-92 cells and CAR-NK-92 cells upon interaction with target cells.
Irradiation had no impact on the total percentage of CD56 + (CAR-)NK-92 cells, as previously described in literature (18,41), but flow cytometry showed an emerging CD56 low population in NK-92 and CD123-directed CAR-NK-92 cells after irradiation. As far as we are aware, this has never been described before. In our analyses, secretion levels of effector molecules and lysis of target cells were mainly maintained after irradiation, despite the negative regulation of CD56 surface expression. Therefore, we conclude that this CD56 low population is phenotypically and functionally different from the CD56 dim population of primary NK cells, which is characterized by a high cytotoxic potential, compared to CD56 bright NK cells (49,50). Further examination of this CD56 low cell population by Annexin V and 7-AAD staining revealed that decreased CD56 surface expression could not be explained by increased cell death with subsequent downregulation of this marker. As the CD56 level is an indicator of NK cell activation and functional potential (51), the observed phenotype might indicate irradiation-induced exhaustion of the cells. However, improved analyses about this is hampered due to the lack of CD16 in NK-92, compared to primary human NK cells, where CD16 in combination with CD56 allows differentiation between cytotoxic and immunoregulatory NK cells. On the other hand, in our study, the irradiation-induced DNA damage had no impact on the surface expression of the activation receptors NKp46, NKG2D and NKp30, which is also in line with current literature (37,41,43).
Cell cycle analysis revealed an accumulation of irradiated cells in the G2/M phases and might reflect the activation of DNA damage checkpoints (35). First preliminary analyses of gene expression levels revealed a remarkable difference between both irradiation methods with gamma irradiation resulting in a more than 12-fold higher number of differentially regulated genes than LEEI. It might be assumed that the effects on DNA and gene expression levels are based on differences in the dose rate. The required number of targeted hits by LEEI is achieved within a very short time period (< 1 s), whereas the cells experience more secondary attacks over several minutes, which are required to reach 10 Gy with gamma irradiation. Jochems et al. previously analyzed gene expression of gamma-irradiated haNK cells ("high affinity", genetically modified NK-92 cells) and found that the majority of differentially expressed genes was associated with cellular activation (52). In our study, both LEEI and gamma irradiation led to differentially expressed genes enriched in 35 different pathways when compared to nonirradiated cells. However, only one pathway could be identified which was enriched specifically after LEEI, the RAF-independent MAPK1/3 pathway. This enzyme cascade, that plays a central role in intracellular transmission of extracellular signals, is activated by several extracellular stimuli, including UV radiation (53,54). In contrast, 286 pathways enriched for differentially regulated genes were found only after gamma irradiation. Included is the NER system. Given the observation that both LEEI and gamma irradiation led to similar DNA damage via strand breaks, this finding might indicate that lesions targeted by NER are more enriched upon gamma irradiation. Such lesions are mainly bulky photo-adducts or thymine dimers induced by UV light and by ROS (55). They are more likely to be secondary effects of radiation induced radicals in the cells, whereas the direct effects of ionizing radiation are DNA strand breaks. Therefore, although the effects on cellular viability are similar, both irradiation technologies seem to have different impacts on the overall cellular functions. A further in-depth analysis of these gene expression studies, which should involve the validation of candidate genes, might therefore reveal fundamental information on the effects of ionizing radiation on cells.
In summary, LEEI is a suitable alternative to state-of-the-art gamma irradiation as it yields comparably inactivated cellular products with a high cytotoxic activity and moreover provides many logistic advantages. Specifically, LEEI has low shielding requirements, which enables installation of radiation facilities in a standard laboratory surrounding, a very accurate and reproducible irradiation, and fast processing due to the high dose rate (18). Nevertheless, it has to be critically mentioned that the irradiation device used for LEEI is a research-scale prototype and still requires optimization for standardized clinical use. To fully leverage these advantages, the currently existing irradiation modules for LEEI have to be transformed into GMP-compatible devices, including an in-line process parameter control. Thereby, this technique could find applications in the automated processing of ATMPs for clinical trials.
The crucial next step will be to investigate and compare the effects of non-irradiated, LEE-irradiated, and gamma-irradiated CAR-NK-92 cells in vivo. Contrary to the project hypothesis, some studies described in literature have shown that nonirradiated cells can be applied safely without persisting or causing secondary tumors in mice (56,57). This could evoke the impression that irradiation of NK-92 cells might be redundant, however, in this case the large differences between immunodeficient mouse models and the complex human organism have to be kept in mind. The fact that non-irradiated NK-92 cells do not persist in NSG mice might for instance be owed to the absence of cytokines. NK-92 are known to grow IL-2 dependent and show rapid cell death upon IL-2 withdrawal in vitro (58). For this reason, some in vivo models described in literature add IL-2 to the therapeutic regimen in order to enhance NK-92 persistence and efficacy (59)(60)(61). Alternatively, the cell line NK-92MI, which is genetically modified to produce IL-2 (39), can be used as a cytokine-independent effector cell population. Indeed, Liu et al. showed that non-irradiated CD19directed CAR-NK-92MI cells persist and proliferate in NOD-SCID as well as in NSG mice (42). Therefore, it is still likely that non-irradiated (CAR-)NK-92 cells could proliferate in patients. Analyzing the persistence of non-irradiated and irradiated cells in mice after adding cytokines, or in humanized mice could provide interesting information in this context. Furthermore, in the future, the observed high in vitro functionality of LEEirradiated cells should be confirmed by comparing the therapeutic effects of LEE-and gamma-irradiated cells in vivo.

DATA AVAILABILITY STATEMENT
The data presented in the study are online available and are deposited in the Gene Expression Omnibus (GEO) repository, accession number GSE166976. Online available under: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE166976