IL-15 superagonist N-803 improves IFNγ production and killing of leukemia and ovarian cancer cells by CD34+ progenitor-derived NK cells

Allogeneic natural killer (NK) cell transfer is a potential immunotherapy to eliminate and control cancer. A promising source are CD34 + hematopoietic progenitor cells (HPCs), since large numbers of cytotoxic NK cells can be generated. Effective boosting of NK cell function can be achieved by interleukin (IL)-15. However, its in vivo half-life is short and potent trans-presentation by IL-15 receptor α (IL-15Rα) is absent. Therefore, ImmunityBio developed IL-15 superagonist N-803, which combines IL-15 with an activating mutation, an IL-15Rα sushi domain for trans-presentation, and IgG1-Fc for increased half-life. Here, we investigated whether and how N-803 improves HPC-NK cell functionality in leukemia and ovarian cancer (OC) models in vitro and in vivo in OC-bearing immunodeficient mice. We used flow cytometry-based assays, enzyme-linked immunosorbent assay, microscopy-based serial killing assays, and bioluminescence imaging, for in vitro and in vivo experiments. N-803 increased HPC-NK cell proliferation and interferon (IFN)γ production. On leukemia cells, co-culture with HPC-NK cells and N-803 increased ICAM-1 expression. Furthermore, N-803 improved HPC-NK cell-mediated (serial) leukemia killing. Treating OC spheroids with HPC-NK cells and N-803 increased IFNγ-induced CXCL10 secretion, and target killing after prolonged exposure. In immunodeficient mice bearing human OC, N-803 supported HPC-NK cell persistence in combination with total human immunoglobulins to prevent Fc-mediated HPC-NK cell depletion. Moreover, this combination treatment decreased tumor growth. In conclusion, N-803 is a promising IL-15-based compound that boosts HPC-NK cell expansion and functionality in vitro and in vivo. Adding N-803 to HPC-NK cell therapy could improve cancer immunotherapy. Electronic supplementary material The online version of this article (10.1007/s00262-020-02749-8) contains supplementary material, which is available to authorized users.

A promising source for allogeneic NK cell therapy are CD34 + hematopoietic progenitor cell (HPC)-derived NK cells, since large numbers of cytotoxic NK cells can be generated from various sources, including umbilical cord blood (UCB). First, CD34 + HPCs are expanded and subsequently differentiated into CD56 + HPC-NK cells, leading to more than 1000-fold expansion and high NK cell purity [9][10][11]. HPC-NK cells are highly functional, since they have high activating receptor expression, degranulation capacity, interferon (IFN)γ production, and tumor cell killing capacity [9][10][11][12][13]. Furthermore, we have shown that HPC-NK cells mediate anti-tumor responses in leukemia and ovarian cancer (OC) models in mice, leading to prolonged survival [11,12,14]. To further maximize the anti-tumor effects of HPC-NK cell therapy, combination treatments can be explored to maintain NK cell proliferation and activation and/or to augment NK cell-mediated killing of tumor cells.
Our study goal was to investigate whether and how N-803 enhances HPC-NK cell functionality in leukemia and OC models, and whether N-803 supports HPC-NK cell persistence and anti-tumor effects in OC-bearing NSG mice. We found that N-803 can increase IFNγ production of HPC-NK cells and augment HPC-NK cell-mediated killing of OC and leukemia cells in vitro. Moreover, N-803 supports HPC-NK cell persistence and limits tumor growth in NSG mice bearing human OC.

Tumor spheroid generation
Spheroids were generated from SKOV-3 and SKOV-3-luc-GFP as described in Hoogstad-van Evert et al. [11] with the following adaptations. Culture medium was not supplemented with bovine serum albumin but with 10% FCS and 1% penicillin/streptomycin (MP Biomedicals, 1670049) and agarose medium with 2% penicillin/streptomycin. Tumor spheroids were used 3-5 days after initial seeding.

NK cell proliferation
NK cells were labeled with eFluor450 (eBioScience, 65-0842-85) and cultured in NK MACS/10% HS with/ without rhIL-15 or N-803 (ImmunityBio). Cytokines were refreshed on day 3 and FCM analysis was performed on day 3 and 6. Dead cells were excluded using Fixable Viability Dye eFluor780 (eBiosciences, 65-0865-18). The proliferation gate was set on 1% in the no cytokine condition on day 3. NK cell numbers were based on CD56 gating (CD56-PE-Cy7, Beckman Coulter, A21692) and measuring for a fixed time.

Killing assay
Targets were plated at 30,000 cells/well in 96-well plates (round-bottom for leukemia cells, flat-bottom for OC cells). Targets or HPC-NK/PB-NK cells were labeled with 0.25-1 µM CFSE, and co-cultured at different E:T ratios with or without 1 nM N-803. Notably, SKOV-3-luc-GFP was not labeled with CFSE. OC cells were plated 3 h in advance to allow for adherence. After overnight (cell line) or 48 h (primary cells) co-culture, supernatants were harvested and stored at − 20 °C for enzyme-linked immunosorbent assay (ELISA). Next, leukemia cells and/or NK cells were collected. OC cells were trypsinized using trypLE (Gibco, 12605028) and collected. Subsequently, viability marker 7-Aminoactinomycin D (7-AAD, Sigma, A9400) was added and targets were analyzed. Percentage of target killing by NK cells was calculated as follows: [1-(number of viable targets after co-culture with NK cells)/(number of viable targets cultured without NK cells) × 100%].

Spheroid killing assay
For spheroid killing assays, SKOV-3-luc-GFP cells were used. For overnight assays, different HPC-NK cell numbers were added with or without 1 nM N-803. After co-culture, supernatant was collected for ELISA. For 7-day assays, 13,000 HPC-NK cells and 0, 0.1 or 1.0 nM N-803 or rhIL-15 was used and after 7 days HPC-NK cells were counted based on CD56 positivity and 7-AAD negativity. Spheroids were washed, disrupted using trypLE and targets were counted based on GFP positivity and 7-AAD negativity.

Infiltration assay
SKOV-3-luc-GFP or SKOV-3 spheroids were co-cultured with 200,000 HPC-NK cells with or without 1 nM N-803. In SKOV-3 experiments, HPC-NK cells were labeled with 1 µM CFSE before or CD56-PE-Cy7 after co-culture. After 3 h co-culture, infiltrated and non-infiltrated NK cells were separated as described [32]. First, supernatant was collected containing non-infiltrated NK cells. Next, spheroids were washed, disrupted using trypLE, and infiltrated NK cells were collected. 7-AAD negative and CD56 or CFSE positive NK cells were counted.

Organotypic 3D collagen matrix assay
Organotypic 3D collagen matrix assays were performed as described [34]. In brief, 7500 SKOV-3-luc-GFP cells were plated on a flat-bottom 96-well imaging plate (Greiner CELLSTAR ® , 655090). After overnight adherence, 7500 HPC-NK cells were added in a collagen solution (75 µl/well PureCol1, Advanced Biomatrix, 5005, 3 mg/ml) containing no or 1 nM N-803. After polymerization, no or 1 nM N-803 was added and cells were imaged by time-lapse bright field microscopy with × 20 objective (BD, Pathway 855) at 37° C, 5% CO 2 . Images were captured every 70s for ~ 24 h and subsequently, manual analysis of single cells was performed. Only serial killers were analyzed, defined as NK cells killing two or more targets. Inclusion criteria for cytotoxic events were (i) contact occurred between a single NK cell to a single target, (ii) the target was visible from the start of the movie.

Statistical analysis
Statistical analysis was performed using Graphpad Prism software version 5.03. Fold changes, lag phase to apoptosis and NK cell numbers in mice were first log transformed. Two-sided Student t tests and one-way and two-way ANO-VAs were used as indicated in the figure legends. Significance was defined as p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***).

N-803 enhances HPC-NK cell proliferation, IFNγ production, and leukemia killing
Previously, we showed that N-803 outperforms rhIL-15 in inducing HPC-NK cell proliferation at 0.1 nM [35]. To confirm the optimal N-803 concentration, we performed proliferation assays with different concentrations of rhIL-15 or N-803 for 6 days. Indeed, N-803 induced HPC-NK cell proliferation in a dose-dependent manner (Fig. 1a, b). In comparison with rhIL-15, N-803 was superior in boosting NK cell proliferation at 0.1 nM (33-64%) and proliferation was similar at 1.0 nM (90-92%). All further experiments were performed using 1.0 nM N-803, which induced the most proliferation.
As increased ICAM-1 expression stimulates NK cellmediated killing due to strengthened interactions of NK cells and targets [36], we next investigated NK cell-mediated tumor killing. Leukemia killing was measured after overnight co-culture with HPC-NK cells and with or without N-803. Correlating with ICAM-1 expression, N-803 did not increase HPC-NK cell-mediated K562 killing, but significantly augmented HPC-NK cell-mediated THP-1 killing (Fig. 1f, g). To compare the killing capacity of HPC-NK cells and PB-NK cells, we co-cultured HPC-NK cells or PB-NK cells with K562 or THP-1 with or without N-803. For MHC-I negative K562, N-803 did not improve HPC-NK cell-mediated killing at all, while it did seem to improve PB-NK cell mediated killing at the second highest NK cell dose ( Supplementary Fig. 1b). With regard to MHC-I positive THP-1, N-803 increased HPC-NK and PB-NK cell-mediated killing at all NK cell doses ( Supplementary  Fig. 1c). For both K562 and THP-1, HPC-NK cells were better killers than PB-NK cells at all NK cell doses, except the highest NK cell dose for K562 at which killing was maximal for both NK cell sources. Next, we evaluated the perforin content and granzyme B release of HPC-NK cells and PB-NK cells after priming with N-803 by intracellular staining and ELISA, respectively. We found that both HPC-NK cells and PB-NK cells upregulate perforin and granzyme B levels upon N-803 priming ( Supplementary Fig. 1d, e). The higher killing capacity of HPC-NK cells did not correspond to perforin content, but did correlate with a higher granzyme B release versus PB-NK cells.
To confirm our findings, we co-cultured HPC-NK cells with primary AML samples from patients (Table 1) for 48 h with/without N-803 and investigated IFNγ production, ICAM-1 expression, and killing.
N-803 significantly enhanced IFNγ production at an E:T ratio of 1:1 and 3:1 (Fig. 1h), upregulated ICAM-1 expression in the presence of HPC-NK cells (Fig. 1i) and most importantly increased primary AML killing by HPC-NK cells (Fig. 1j). Collectively, these data show that N-803 boosts IFNγ production by HPC-NK cells, promotes ICAM-1 expression on leukemia cells and improves HPC-NK cell-mediated leukemia killing.

N-803 enhances serial killing properties of HPC-NK cells against leukemia
To examine whether N-803 improves serial killing properties of HPC-NK cells against leukemia, we performed 12 h 1 3 experiments using microwells for live cell imaging with single cell resolution [33]. Here, a mean of 22 or 31% of HPC-NK cells serially killed (≥ 2 targets with at least five targets present at t = 0 h) K562 and THP-1, respectively (Fig. 2a, b). N-803 seemed to enhance these percentages, most distinct for THP-1 (mean 37%, p = 0.07). Most killing HPC-NK cells killed 1 target, followed by 2, 3, 4 and 5 or more targets (Fig. 2c, d). N-803 seemed to increase the number of targets killed by HPC-NK cell serial killers, most pronounced for THP-1. Spontaneous target death was detected in the minority of wells without HPC-NK cells (mean 15% for K562, 45% for THP-1, Fig. 2e, f) and was not affected by N-803. Notably, the majority of targets was killed by serial killer HPC-NK cells (mean 66%, Fig. 2g, h). N-803 augmented this percentage to a mean of 69% for K562 and 78% for THP-1. RhIL-2 and rhIL-15 displayed similar results as N-803 ( Supplementary Fig. 2). Together, these data demonstrate that N-803 improves serial killing properties of HPC-NK cells against leukemia.
Although N-803 did not improve overnight HPC-NK cellmediated SKOV-3 killing, we next studied whether interaction abilities and serial killing properties of HPC-NK cells against OC were affected by N-803 in an organotypic 3D collagen matrix assay, mimicking interstitial tissue. Lag phase to SKOV-3 apoptosis (time from first contact to kill) due to serial killing by HPC-NK cells was mostly short with a median of 19 min for the first kill and similar times for the second kill (Fig. 3g). N-803 did not change these times for the first or second kill. Altogether, these data indicate that despite slightly enhanced IFNγ production, N-803 could not increase ICAM-1 expression and short-term HPC-NK cellmediated (serial) killing of OC cells.
Collectively, these experiments demonstrate that N-803 increases IFNγ and CXCL10 secretion in co-cultures of OC spheroids and HPC-NK cells. Furthermore, N-803 induces HPC-NK cell expansion and boosts OC spheroid destruction during long-term co-cultures.

HPC-NK cells combined with N-803 and nanogam show anti-tumor effects in mice bearing human OC
To determine whether N-803 promotes HPC-NK cell persistence and anti-tumor effects in a human OC mouse model, we used NSG mice bearing peritoneal SKOV-3-luc-GFP tumor nodules [11]. In experiment 1, mice were treated i.p. with HPC-NK cells in combination with PBS, rhIL-15, or N-803 for two weeks and afterwards peritoneal washes were performed. As expected, HPC-NK cells were present in the rhIL-15 group but surprisingly HPC-NK cells were nearly absent in the N-803 groups (Fig. 5a). We hypothesized that the Fc part of N-803 binds to Fc receptors, resulting in Fcmediated HPC-NK cell depletion, in NSG mice lacking immunoglobulins. Hence, in experiment 2 we used irradiation or nanogam (i.e., total human immunoglobulins) to kill or inactivate immune cells containing Fc receptors present in NSG mice, or to block Fc receptors, respectively, to prevent Fc-mediated HPC-NK cell depletion in the presence of N-803. To determine if there was risk for Fc-mediated fratricide, CD16 expression was determined prior to HPC-NK cell injection, which showed 20% CD16 + HPC-NK cells ( Supplementary Fig. 5). Irradiation could not prevent N-803-mediated depletion but nanogam could, resulting in HPC-NK cell persistence and similar NK cell numbers as rhIL-15 treatment (Fig. 5b). Finally, we evaluated tumor growth (experiment 3) in mice treated with two i.p. HPC-NK cell injections in combination with N-803 or rhIL-15, and nanogam compared to a group only receiving nanogam. This experiment showed that both combination treatments significantly reduced tumor growth, compared to the control group (Fig. 5c, d). To conclude, we demonstrate that nanogam restores HPC-NK cell persistence in OC bearing NSG mice receiving N-803. Importantly, HPC-NK cell, N-803 and nanogam combination treatment has an anti-OC effect in vivo.

Discussion
Allogeneic NK cell therapy is a promising approach for cancer treatment and HPC-NK cells mediate anti-tumor responses in leukemia and OC models [11,12,14]. However, tumor eradication is not complete in xenograft NSG models, indicating room for improvement. Optimizing HPC-NK cell anti-tumor efficacy can be achieved by cytokine co-administration. This study investigated whether and how IL-15 superagonist N-803 improves HPC-NK cell functionality in leukemia and OC models, and whether N-803 supports in vivo HPC-NK cell persistence and anti-OC effects.
First, we confirmed that N-803 dose dependently induces HPC-NK cell proliferation. Compared to rhIL-15, N-803 leads to higher proliferation at 0.1 nM but not 1.0 nM, caused by reaching maximum proliferation, which is in line with previous reports [30,35]. Next, we demonstrated that This effect has been demonstrated in numerous NK cell studies [29,30,[37][38][39][40][41][42]. In addition, N-803 increases ICAM-1 expression on leukemia cells after HPC-NK cell co-culture and improves (serial) leukemia killing. Since HPC-NK cells have high lymphocyte function associated antigen 1 (LFA-1) expression [11,12,35], the receptor for ICAM-1, the interaction strength between HPC-NK cells and targets is dependent on ICAM-1 expression. Increased ICAM-1 expression leads to stronger interactions, resulting in targets being more sensitive to killing [12,36]. Interestingly, these effects were found with primary AML and THP-1, but not K562, which may be attributed to unaffected ICAM-1 expression on K562. Furthermore, K562 is MHC-I negative, making it very sensitive to NK cell-mediated killing.
Since HPC-NK cells are highly potent killers compared to PB-NK cells, this leaves a narrow window for improvement. However, for less susceptible MHC-I positive THP-1 cells HPC-NK and PB-NK cell-mediated killing could be improved by N-803. Importantly, we showed for the first time that N-803 promotes HPC-NK cell serial killing properties and that some HPC-NK cells kill 5 or more leukemia cells within 12 h. This is in line with studies using PB-NK or NK-92 cells, in which up to 6 [43], 7 [44], 8 [33], or 14 [45] serial kills were reported within 6-16 h. Our findings further revealed that a minority of HPC-NK cells is a serial killer, responsible for the majority of killing. This is in accordance with previous studies [33,44]. Moreover, we assessed HPC-NK cell serial killing properties against OC cells. As expected based on OC monolayer killing experiments, N-803 did not improve serial killing against OC. Nevertheless, serial killer HPC-NK cells generally kill quickly (median 19 min for the first kill) after initial contact. This median lag phase is similar as in Vanherberghen's study [44], where the mean lag phase (time to lytic hit + time to death) was 17.5 min for serial killers. In our OC model, serial killer HPC-NK cells kill up to three targets, which is lower than our leukemia model and other studies [33,[43][44][45]. Potential explanations for those differences are that we used a low target density and a high E:T ratio in the OC model, while in our leukemia model and other studies higher target densities and/or lower E:T ratios were used. For low target cell densities, we and others [43,45] observed that NK cells often stay in contact with apoptotic cells, limiting the number of serial kills. Lower E:T ratios allow for better serial killing detection, because every NK cell can kill more targets. Furthermore, intrinsic differences between used targets impact sensitivity to (serial) killing by NK cells [33,45]. For instance, SKOV-3 used in our OC model is more difficult to kill than K562 used in our leukemia model and other studies (Figs. 1, 3).
In OC spheroids, N-803 significantly increases IFNγ and CXCL10 secretion during overnight co-culture with HPC-NK cells. Because HPC-NK cells have high CXCR3 expression [11][12][13][14]35], increased CXCL10 secretion could improve NK cell infiltration. Since the relatively high amount of HPC-NK cells, needed for infiltration assays, destroys OC spheroids after overnight incubation, we measured infiltration after 3 h. In this model, no effect of N-803 on HPC-NK cell infiltration was observed, though 3 h co-incubation is likely too short to increase IFNγ and CXCL10 secretion and impact HPC NK cell infiltration. Importantly, in long-term assays, using less HPC-NK cells, N-803 improves HPC-NK cell expansion, and, therefore, OC spheroid killing at the longer term.
Finally, we showed that in vivo N-803 supports peritoneal HPC-NK cell persistence in the presence of human immunoglobulins (nanogam) in NSG mice bearing human OC and this combination treatment has an anti-OC effect. Similar findings were reported by Felices et al. [30], demonstrating improved OC tumor control in NSG mice treated with Since HPC-NK cells hardly persist without cytokine support, and clinical trials will be conducted with cytokine support, we chose to compare HPC-NK cells plus N-803 (or rhIL-15) treatment to no treatment. Notably, pre-treatment of NSG mice with human immunoglobulins (nanogam) was required to prevent Fcmediated HPC-NK cell depletion by N-803 treatment. In patients, pre-treatment with nanogam will not be necessary, since they have immunoglobulins. In Felices' study sublethal irradiation (2.25 Gy) was sufficient to prevent Fc-mediated depletion of PB-NK cells, while in our study sublethal irradiation (2.25 Gy) did not rescue Fc-mediated depletion of HPC-NK cells. One of the differences in the design of these two studies is the timing of irradiation: we irradiated the mice one day before tumor injection, while they irradiated the mice one day before NK cell injection. It might be that in our study immune cells containing Fc receptors in the NSG mice recovered or repopulated before the first N-803 injection, which could have led to Fc-mediated depletion. Alternatively, it could be that HPC-NK cells are more sensitive to Fc-mediated NK cell depletion than PB-NK cells due to differences in activation status. Around 20% of the HPC-NK cells had CD16 expression before NK cell injection (Supplementary Fig. 5), indicating that Fc-mediated fratricide might have been possible. Moreover, we know from our previous publications that CD16 expression is upregulated in NSG mice in vivo [12,14], increasing the risk for Fc-mediated fratricide. Fortunately, Fc-mediated depletion of HPC-NK cells could be prevented by nanogam injection in NSG mice.
Comparing N-803 with rhIL-15 shows that in vivo OC growth was similar. However, it is important to note that the amount of molecules per dose was ~ 7 × lower for N-803 than rhIL-15 and rhIL-15 was given more frequently. Assuming all N-803 or rhIL-15 was consumed before the next dose administration, this suggests that N-803 may indeed have a higher biological activity compared to rhIL-15. In vivo experiments with leukemia-bearing NSG mice, NK cells, and N-803 have previously been carried out [35,39]. Wagner et al. showed K562 leukemia control by N-803primed PB-NK cells [39] and Cany et al. demonstrated intrafemoral THP-1 leukemia control by HPC-NK cells, N-803, and decitabine [35]. Since we found HPC-NK cell depletion in our i.p. OC model, repeating Cany's leukemia study with nanogam might improve treatment results in mice.
Collectively, our results imply that N-803 is an attractive compound to promote HPC-NK cell expansion and functionality for NK cell therapy. Currently, two phase 1 clinical trials with N-803 are recruiting patients in the US in various cancer types (NCT03054909 and NCT02890758). In addition, N-803 has been shown to enhance antibody-dependent cellular cytotoxicity in vitro [38,41] and checkpoint blockade therapy in cancer-bearing mice [40]. For future studies, it would be interesting to compare N-803 to the standard IL-2 co-administration with NK cell adoptive transfer for anti-tumor efficacy, to evaluate whether IL-2 can be replaced by N-803 to prevent Treg-expansion in cancer patients.
In conclusion, N-803 boosts HPC-NK cell proliferation and IFNγ production in vitro. Furthermore, N-803 improves (serial) leukemia killing and long-term OC spheroid destruction by HPC-NK cells. In vivo, N-803 in combination with human immunoglobulins supports HPC-NK cell persistence in NSG mice and this combination treatment mediates an anti-OC effect. In conclusion, N-803 is a promising IL-15-based compound to improve NK cell-based cancer immunotherapy.

Day 26
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.  for a and b), and a repeated-measures two-way ANOVA with Bonferroni correction for c to test for statistical significance ◂