Ex Vivo Expanded and Activated Natural Killer Cells Prolong the Overall Survival of Mice with Glioblastoma-like Cell-Derived Tumors

Glioblastoma (GBM) is the leading malignant intracranial tumor and is associated with a poor prognosis. Highly purified, activated natural killer (NK) cells, designated as genuine induced NK cells (GiNKs), represent a promising immunotherapy for GBM. We evaluated the anti-tumor effect of GiNKs in association with the programmed death 1(PD-1)/PD-ligand 1 (PD-L1) immune checkpoint pathway. We determined the level of PD-1 expression, a receptor known to down-regulate the immune response against malignancy, on GiNKs. PD-L1 expression on glioma cell lines (GBM-like cell line U87MG, and GBM cell line T98G) was also determined. To evaluate the anti-tumor activity of GiNKs in vivo, we used a xenograft model of subcutaneously implanted U87MG cells in immunocompromised NOG mice. The GiNKs expressed very low levels of PD-1. Although PD-L1 was expressed on U87MG and T98G cells, the expression levels were highly variable. Our xenograft model revealed that the retro-orbital administration of GiNKs and interleukin-2 (IL-2) prolonged the survival of NOG mice bearing subcutaneous U87MG-derived tumors. PD-1 blocking antibodies did not have an additive effect with GiNKs for prolonging survival. GiNKs may represent a promising cell-based immunotherapy for patients with GBM and are minimally affected by the PD-1/PD-L1 immune evasion axis in GBM.


Introduction
Glioblastoma (GBM) is the most common primary malignant tumor of the brain. Although standard treatment is comprised of maximal surgical resection followed by adjuvant radiotherapy and chemotherapy with temozolomide (TMZ), the median survival time remains less than 2 years [1,2]. Bevacizumab is a humanized antibody that inhibits vascular endothelial growth factor, and may improve patient status and reduce the use of corticosteroids; however, a previous clinical trial did not report a significantly longer overall survival time in patients with newly diagnosed GBM [3]. Therefore, there is an urgent need for the identification of novel and effective treatment strategies for GBM. PD-1/PD-L1 pathway using PD-1 blockers following immunotherapy with GiNKs against GBM cell lines, and in a GBM-like xenograft model in vivo.

PD-1/PD-L1 Expression and NCR1 Based on the TCGA Data Set
To determine the expression pattern of the PD-1/PD-L1 pathway and NCR1 in gliomas, we examined the RNA-sequencing data of gliomas from the GlioVis data portal to visualize and analyze brain tumor expression datasets [37]. Our results revealed that PD-1 and PD-L1 were expressed in GBM in The Cancer Genome Atlas (TCGA) database. Compared to WHO grade II and grade III glioma, GBM (grade IV) was associated with the highest level of PD-1/PD-L1 expression (p < 0.05). Furthermore, NCR1, a representative receptor of NK cells, was also expressed in GBM, albeit at quite a low level. Compared to grade II, GBM had significant expression (p < 0.05). The expression of PD-1/PD-L1 and NCR1 was not associated with any differences in the overall survival (OS) (Figure 1). GBM-like xenograft model in vivo. Second, we evaluated the involvement of the PD-1/PD-L1 pathway using PD-1 blockers following immunotherapy with GiNKs against GBM cell lines, and in a GBM-like xenograft model in vivo.

PD-1/PD-L1 Expression and NCR1 Based on the TCGA Data Set
To determine the expression pattern of the PD-1/PD-L1 pathway and NCR1 in gliomas, we examined the RNA-sequencing data of gliomas from the GlioVis data portal to visualize and analyze brain tumor expression datasets [37]. Our results revealed that PD-1 and PD-L1 were expressed in GBM in The Cancer Genome Atlas (TCGA) database. Compared to WHO grade II and grade III glioma, GBM (grade IV) was associated with the highest level of PD-1/PD-L1 expression (p < 0.05). Furthermore, NCR1, a representative receptor of NK cells, was also expressed in GBM, albeit at quite a low level. Compared to grade II, GBM had significant expression (p < 0.05). The expression of PD-1/PD-L1 and NCR1 was not associated with any differences in the overall survival (OS) (Figure 1).

Expression of PD-L1 in GBM Cell Lines
Two glioma cell lines were obtained (U87MG and T98G cells). The relative fluorescence intensity (RFI) of PD-L1 was assessed by comparing the levels of expression to that of a control IgG. The RFI of the U87MG and T98G cells was 7.12 and 2.78, respectively. In both cell lines, PD-L1 was expressed on the cell surface; however, the level of expression was variable (Figure 2a). p Values were determined using Turkey's Honest significant difference test. ** p < 0.001, * p < 0.01, NS: not significant.

Expression of PD-L1 in GBM Cell Lines
Two glioma cell lines were obtained (U87MG and T98G cells). The relative fluorescence intensity (RFI) of PD-L1 was assessed by comparing the levels of expression to that of a control IgG. The RFI of the U87MG and T98G cells was 7.12 and 2.78, respectively. In both cell lines, PD-L1 was expressed on the cell surface; however, the level of expression was variable (Figure 2a).

Expression of PD-1 on GiNKs
Flow cytometry analysis detected PD-1 expression on the surface of GiNKs harvested from healthy volunteers. The frequency of PD-1+/CD56+ NK cells ranged between 1.7% and 8.0% in three healthy individuals (n = 6). The percentage of PD-1-positive cells was highly variable among healthy volunteers, but was less than 10% in each of the samples (Figure 2b,c). A very low frequency of PD-1 expressing NK cells was included in the GiNKs. We demonstrated that γδT cells had a positive expression of PD-1 in Supplementary Figure S1.

Apoptosis-Inducing Effects of GiNKs on GBM Cell Lines In Vitro
The apoptosis detection assays revealed that after 24 h of exposure, a GiNK (effector) to GBM cell (target) ratio (E:T) of 2:1 significantly induced apoptosis in both GBM cell lines compared with an E:T ratio of 0.5:1 (p < 0.05). The GiNKs significantly induced apoptosis in U87MG and T98G cells (p < 0.05); however, treatment with the PD-1 blocker did not enhance or attenuate the apoptosis-inducing effect in both cell lines ( Figure 3).

Expression of PD-1 on GiNKs
Flow cytometry analysis detected PD-1 expression on the surface of GiNKs harvested from healthy volunteers. The frequency of PD-1+/CD56+ NK cells ranged between 1.7% and 8.0% in three healthy individuals (n = 6). The percentage of PD-1-positive cells was highly variable among healthy volunteers, but was less than 10% in each of the samples (Figure 2b,c). A very low frequency of PD-1 expressing NK cells was included in the GiNKs. We demonstrated that γδT cells had a positive expression of PD-1 in Supplementary Figure S1.

GiNKs Produce Cytokines upon Recognizing GBM Cells
Next, we sought to determine the cytokine levels produced by GiNK (effector) cells following exposure to U87MG (target) and T98G (target) cells. The levels of interferon (IFN)γ in the supernatants of U87MG cells co-cultured with GiNKs were 53.4 ± 1.2 pg/mL and 50.3 ± 2.6 pg/mL (treated with the PD-1 blocker), respectively, using the same E:T ratios. Treatment with the PD-1 blocker did not induce the production of the IFNγ on U87MG. The IFNγ levels in the supernatants from T98G cells co-cultured with GiNKs were 38.2 ± 3.7 pg/mL and 32.1 ± 8.0 pg/mL (treated with the PD-1 blocker), respectively, using the same E:T ratios. In addition, treatment with the PD-1 blocker did not induce IFNγ production on T98G. Moreover, IL-6 was abundantly present in supernatants of U87MG cells co-cultured with GiNKs (1:1, 5.6 ± 0.4 ng/mL) and GiNKs treated with the PD-1 blocker (5.2 ± 0.4 ng/mL). IL-6 was also abundantly present in the supernatants derived from T98G cells co-cultured with GiNKs (1:1, 7.7 ± 0.7 ng/mL) and GiNKs treated with the PD-1 blocker (8.3 ± 1.2 ng/mL). Moreover, the addition of the PD-1 blocker did not change the production of the IL-6. Elevated RANTES was detected in the supernatants from the cultures of effector cells alone (0.3 ng/mL). RANTES was also detected in the supernatants of cells with an E:T ratio of 1:1 compared to effector-only conditions. The addition of the PD-1 blocker did not enhance or attenuate the production of the RANTES (Figure 4).

GiNKs Produce Cytokines upon Recognizing GBM Cells
Next, we sought to determine the cytokine levels produced by GiNK (effector) cells following exposure to U87MG (target) and T98G (target) cells. The levels of interferon (IFN)γ in the supernatants of U87MG cells co-cultured with GiNKs were 53.4 ± 1.2 pg/mL and 50.3 ± 2.6 pg/mL (treated with the PD-1 blocker), respectively, using the same E:T ratios. Treatment with the PD-1 blocker did not induce the production of the IFNγ on U87MG. The IFNγ levels in the supernatants from T98G cells co-cultured with GiNKs were 38.2 ± 3.7 pg/mL and 32.1 ± 8.0 pg/mL (treated with the PD-1 blocker), respectively, using the same E:T ratios. In addition, treatment with the PD-1 blocker did not induce IFNγ production on T98G. Moreover, IL-6 was abundantly present in supernatants of U87MG cells co-cultured with GiNKs (1:1, 5.6 ± 0.4 ng/mL) and GiNKs treated with the PD-1 blocker (5.2 ± 0.4 ng/mL). IL-6 was also abundantly present in the supernatants derived from T98G cells co-cultured with GiNKs (1:1, 7.7 ± 0.7 ng/mL) and GiNKs treated with the PD-1 blocker (8.3 ± 1.2 ng/mL). Moreover, the addition of the PD-1 blocker did not change the production of the IL-6. Elevated RANTES was detected in the supernatants from the cultures of effector cells alone (0.3 ng/mL). RANTES was also detected in the supernatants of cells with an E:T ratio of 1:1 compared to effector-only conditions. The addition of the PD-1 blocker did not enhance or attenuate the production of the RANTES (Figure 4).  . The cytokine production in GiNKs cultured with GBM cells. GBM cells (U87MG or T98G cell lines) were co-cultured alone and with GiNKs at the indicated ratios for 24 h. The concentrations of interferon-γ (IFNγ), interleukin-6 (IL-6), and regulated on activation, normal T-cells expressed and secreted (RANTES) in the supernatant were determined. Values are presented as the means ± SD at least two independent experiments (n = 3) in the upper table. Significant differences in lower figures were determined using a one-way analysis of variance (ANOVA), followed by a Tukey's test. The figure represents table data. **** p < 0.0001; *** p < 0.001; ** p < 0.01; * p < 0.05; n.s.: not significant.

Effects of GiNK Treatment in Combination with a PD-1 Blocker against a Subcutaneous Tumor Derived from GBM-like Cells
Next, we studied the anti-tumor effects of GiNKs with and without treatment with a PD-1 blocker on U87MG cells in NOG mice in vivo. To this end, NOG mice were subcutaneously inoculated with U87MG cells followed by three injections into the intravenous via retro-orbital sinus (days 1, 4, and 7) with PBS (control), GiNKs alone, or GiNKs with a PD-1 blocker (Figure 5a).
The mice treated with the GiNKs with and without the PD-1 blocker were associated with a significantly longer survival time compared to the PBS control group (p < 0.05). However, the presence of the PD-1 blocker did not influence the survival time (Figure 5b). The results of the histochemical analysis showed that the tumors from both groups exhibited a pseudopalisading pattern of tumor cells surrounding necrosis, which is a human GBM-like histological feature (Figure 5c). significant.

Effects of GiNK Treatment in Combination with a PD-1 Blocker against a Subcutaneous Tumor Derived from GBM-Like Cells
Next, we studied the anti-tumor effects of GiNKs with and without treatment with a PD-1 blocker on U87MG cells in NOG mice in vivo. To this end, NOG mice were subcutaneously inoculated with U87MG cells followed by three injections into the intravenous via retro-orbital sinus (days 1, 4, and 7) with PBS (control), GiNKs alone, or GiNKs with a PD-1 blocker (Figure 5a).
The mice treated with the GiNKs with and without the PD-1 blocker were associated with a significantly longer survival time compared to the PBS control group (p < 0.05). However, the presence of the PD-1 blocker did not influence the survival time (Figure 5b). The results of the histochemical analysis showed that the tumors from both groups exhibited a pseudopalisading pattern of tumor cells surrounding necrosis, which is a human GBM-like histological feature (Figure 5c).

Current Status of Immunotherapy for GBM
Despite an aggressive standard of care regimen consisting of maximal surgical resection followed by combination radiation and chemotherapy, the prognosis of GBM patients remains poor, with a high recurrence rate and a median OS of less than 2 years [33]. Following the success of ICBs in melanoma and non-small cell lung cancer, interest in the use of immunotherapy as an alternative treatment approach for GBM has rapidly increased in recent years [31]. However, in GBM, immunotherapy, including vaccine therapy, viral therapy, and other cell-based therapies, has had a minimal impact on the OS [38,39]. Moreover, GBM has been proven to be highly resistant to standard treatments due to a combination of tumor heterogeneity, adaptive expansion of resistant cellular subclones, evasion of immune surveillance, and manipulation of various signaling pathways involved in tumor progression and the immune response [38].

Characteristics of GiNKs
NK cells exhibit potent cytotoxic activity against tumor cells via the induction of apoptosis [7] and can remove the abnormal cells as part of the innate immune system [8,9]. Moreover, NK cells play a role in immune surveillance and inhibit tumor occurrence, proliferation, and metastasis [22,23]. We previously reported the expansion of human peripheral blood NK cells using a novel culture system for clinical application as GiNKs. Our method of selectively expanding autologous human NK cells is associated with the highest purity and largest expansion scale using an easy, chemically defined and feederfree method. Moreover, we expanded NK cells from CD3+ T cell-depleted PBMCs, which both enhanced the purity of NK cells and prevented contamination with regulatory T cells (Tregs). It has been reported that a reduction in the function and number of Tregs is beneficial as an immunotherapy against malignant tumors. In addition, GiNKs were reported to exhibit a strong anti-tumor effect against GBM cell lines through inducing apoptosis in vitro [35]. TMZ is widely used as the most effective anti-cancer drug for GBM in the current standard therapy when used concomitantly with radiotherapy. Moreover, GiNK induces apoptosis in TMZ-sensitive and TMZ-resistant human GBM cells and enhances the TMZ-induced antitumor effects in different mechanisms [35]. Based on our findings, immunotherapy using GiNK might represent a promising novel treatment option for patients with GBM.

Role of the PD-1/PD-L1 Pathway on GBM
The cytotoxic function of NK cells is determined by the balance between activating and inhibitory receptor signals [13,14]. Several activating receptors of NK cells (e.g., NKG2D and DNAM-1) recognize their ligands expressed on GBM [15,16,20], and the ligation of activating receptors triggers cytotoxicity in NK cells [17]. In addition, the ligands of NK inhibitory receptors, including PD-1, NKG2A, and KIR2DL, are also associated with NK cell cytotoxicity against tumor cells [18,19,40]. In the present study, we focused specifically on the PD-1/PD-L1 axis as an immune check point. The level of PD-L1 expression in GBM patients from the TCGA data set was associated with the WHO grading of glioma. In addition, Grade 4 glioblastoma was associated with significantly higher PD-L1 expression compared to Grades 2 and 3. According to the aggressive character of glioma, the PD-1/PD-L1 pathway might exhibit higher activation under conditions of immunosuppression. In this study, the level of PD-1 expression on GiNKs harvested from healthy volunteers was extremely low. Pesce et al. reported that NK cells in ovarian cancer patients had a significantly higher level of PD-1 expression compared to that of healthy volunteers [41]. The expression of PD-1 might also be higher in the NK cells of cancer patients that were activated under cancer immuno-surveillance [42,43].
The role of the PD-1/PD-L1 pathway in GBM remains controversial. Huang et al. reported that a blockade of the PD-1/PD-L1 pathway enhanced the anti-tumor effect of mouse NK cells [44]. However, in a recent Phase 3 randomized clinical trial for recurrent GBM treated with bevacizumab or nivolumab, nivolumab failed to demonstrate superiority. However, patients with a methylated MGMT promoter, glioblastoma, and no baseline corticosteroid use may potentially derive a benefit from treatment with immune checkpoint inhibition. [34].

Antitumor Effect of GiNKs against GBM In Vitro and In Vivo
To the best of our knowledge, less attention has been paid to the use of an NK cellbased treatment for GBM. We previously expanded human peripheral blood NK cells harvested from healthy volunteers using the novel culture system for clinical application as GiNKs and reported a strong anti-tumor effect in vitro [35]. The results of the present study showed that GiNKs significantly induced the apoptosis of U87MG and T98G cells in the apoptosis detection assays; however, the presence of a PD-1 blocker did not induce apoptosis in either cell line. We also determined the levels of cytokines produced by GiNK cells upon exposure to U87MG and T98G cells. GiNKs exposed to GBM cells produced significantly higher levels of IFNγ, IL-6, and RANTES. These results support the previous finding that GiNKs have significant anti-tumor effects for GBM cells through the induction of apoptosis and production of several cytokines.
RANTES accumulates in T cells and regulates inflammation in several diseases, playing an active role in recruiting a variety of leukocytes (e.g., T cells, macrophages, eosinophils, and basophils) into inflammatory sites. In collaboration with certain cytokines released by T cells (e.g., IL-2 and IFNγ), RANTES also induces the activation and proliferation of specific NK cells to generate chemokine-activated killer cells [45].
IL-6 was substantially increased when GiNKs were co-cultured with GBM cells. In the cytokine assay with GiNKs and GBM cells, it was not possible to determine which cells produced IL-6. The study by Goswami et al. reported that U87MG cells expressed IL-6 and IL-6 receptors, whereas U87MG cells have an autocrine growth loop [46]. Previously, IL-6/signal transducer and activator of transcription 3 (STAT3) signaling was reported to support GBM cell growth and migration [47]. Moreover, an IL-6/STAT3/hypoxia-inducible factor 1 subunit alpha autocrine loop has been observed in GBM. In addition, GBM cancer stem cells have been found to respond to perturbations caused by hypoxia, the inhibition of STAT3 phosphorylation, and IL-6 stimulation [48]. Thus, IL-6 is involved in the formation and progression of GBM. Based on our findings, U87MG cells may have released IL-6; however, a flow cytometry-based analysis of cytokine-producing cells is required to verify this finding. On the other hand, we observed that GiNKs elicit direct cytotoxicity and release cellular immunity-related cytokines, a finding that is contrary to the role of IL-6. Therefore, we aimed to verify the antitumor effect of GiNKs in vivo.
In our in vivo experimental GBM model, NOG mice were subcutaneously inoculated with U87MG cells and subsequently injected with GiNKs intravenously via the retroorbital sinus three times (days 1, 4, and 7). Treatment with GiNKs with or without a PD-1 blocker exhibited a significantly longer survival time compared to that of the control group. However, there was no significant additional effects on the survival time in the group that received the GiNKs with the PD-1 blocker. Overall, the findings of our study suggest that the intravenous injection of GiNKs is highly effective against subcutaneously injected GBM cells independent of the PD-1/PD-1 pathway. There is a possibility that the GBM immunosuppression system against NK cells will add increased complexity through evasion via another pathway.
There are several limitations associated with the present study. First, we used peripheral blood harvested from healthy volunteers. Typically, the expansion of NK cells from the blood of cancer patients is challenging due to the possibility of immune exhaustion following various types of chemotherapy. Second, we evaluated the anti-tumor effect of intravenously transferred GiNKs in a subcutaneous injection GBM model. It is possible that the adoptively transferred GiNKs might exhibit limited persistence. Moreover, GiNKs may not infiltrate the tumor through the brain-blood barrier, or the tumors could develop mechanisms to evade NK cell surveillance in the brain. Thus, our findings warrant validation in an intra-cranially injected model of GBM and perform a first in-human trial in future studies. Third, we have to pay attention to a natural immune system in mice before the injection of NK cells, even though we adopted the NOG mice in this experiment. We think that it is ideal to use CD34-humanized mice or PBMC-reconstituted mice, this setting would better inform on the efficacy of the combination since a PD-1 blocker would then target T cells present in the tumor. Moreover, we did not evaluate the effect of a PD-L1 blocker in vitro and in vivo. This might synergize better than PD1 alone to eradicate the tumors in combination with GiNKs.

Conclusions
We successfully demonstrated that ex vivo expanded and activated NK cells had an anti-tumor effect for GBM cells in vitro and vivo assays. Our NK cells prolong the overall survival of NOG mice subcutaneously injected with GBM-like cells. PD-1 blocking antibodies did not have an additive effect with our NK cells for prolonging survival in our xenograft model in GBM. Our findings reveal that our NK cells are less affected by the PD-1/PD-L1 immune evasion axis in GBM. In the future study, the anti-tumor effect of our NK cells in an intra-cranially injected model of GBM should be warranted.

Ethics
This study was approved by the ethics committee of Nara Medical University (approval number: 1058). All procedures in studies involving human participants were performed in accordance with the ethical standards of the institutional and/or national research committee and in line with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed consent was obtained from all healthy volunteers included in the study. We collected 8 mL of heparinized peripheral blood obtained from 3 healthy volunteers.

Reagents
Nivolumab as the PD-1 blockade agent was provided as a gift from Ono Pharmaceutical Co., Ltd., Osaka, Japan.

Apoptosis Detection Assays
We performed apoptosis detection assays using an MEBCYTOTM Apoptosis Kit (MBL), in accordance with the manufacturer's instructions. Briefly, GBM cell lines were exposed to GiNKs at E:T ratios of 0.5:1, 1:1, and 2:1 in the presence or absence of 10 µg Nivolumab for 10 7 GiNKs for 4 h. Following the incubation, the floating cells were washed, and adherent cells were trypsinized with PBS, stained with Annexin V-FITC, and maintained at room temperature for 15 min in the dark. The stained cells were analyzed using a FACSCalibur flow cytometer and CellQuest Pro software ver 6.0. Notably, the NK cells were excluded by electronic gating based on the forward-scatter and side-scatter characteristics. The frequency of the Annexin V-positive populations was defined as apoptotic cells, as previously described [35,50].

Cytokine Detection
A Human Cytometric Bead Array Flex Set System (BD Biosciences, San Jose, CA, USA) was used to determine cytokine production in GiNKs cultured with GBM cells. GBM cells (U87MG or T98G cell lines) were co-cultured at ratios of 1:1 for 24 h with GiNKs. Four tests for cytokine detection were performed in each GBM cell. They included the group with GBM cells alone, with GiNKs alone, with GBM cells/GiNKs/IgG, and with GBM cells/GiNKs/PD-1 blocker. The concentrations of interferon-γ (IFNγ: 558269), interleukin-6 (IL-6: 558276), and regulated on activation, normal T-cells expressed and secreted (RANTES: 558324) in the supernatant were determined. The experiment was performed three times. The assays were performed according to the manufacturer's instructions. Data were acquired on a BD FACSMelody flow cytometer (BD Biosciences).

Histochemical Analysis
The subcutaneous tumors were fixed with 10% neutral-buffered formalin and embedded in paraffin. After sectioning, 5-µm-thick sections were placed on glass slides and stained with hematoxylin and eosin (HE). Photographs were taken using an BX-710 (KEYENCE, Osaka, Japan) at 40× magnification.

Statistical Methods
Data are presented as the mean ± standard error. Statistical analyses were performed using Prism 8 (GraphPad Software Inc., San Diego, CA, USA). Statistically significant dif-ferences were determined using a Turkey's Honest Significant difference test and one-way analysis of variance (ANOVA) followed by Tukey's test and two-way ANOVA followed by Sidak test. p < 0.05 was considered statistically significant. Kaplan-Meier curves was also produced using Prism 8. Statistically significant differences were determined using a log-rank test.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Peripheral blood was collected from a healthy volunteer with the approval of and in accordance with the guidelines of the Ethics Committee of Nara Medical University (number: 1058). Informed consent was obtained in accordance with the tenets of the Declaration of Helsinki.

Data Availability Statement:
The data supporting the findings of this study are available from the corresponding author upon reasonable request.