Non-Coated Rituximab Induces Highly Cytotoxic Natural Killer Cells From Peripheral Blood Mononuclear Cells via Autologous B Cells

Natural killer (NK) cells are becoming valuable tools for cancer therapy because of their cytotoxicity against tumor cells without prior sensitization and their involvement in graft-versus-host disease; however, it is difficult to obtain highly cytotoxic NK cells without adding extra feeder cells. In this study, we developed a new method for obtaining highly cytotoxic NK cells from peripheral blood mononuclear cells (PBMCs) independently of extra feeder cell addition using rituximab not coated on a flask (non-coated rituximab). We found that rituximab could promote both the activation and expansion of NK cells from PBMCs, irrespective of being coated on a flask or not. However, NK cells activated by non-coated rituximab had much greater antitumor activity against cancer cells, and these effects were dependent on autologous living B cells. The antibody-dependent cellular cytotoxicity effect of NK cells activated by non-coated rituximab was also more substantial. Furthermore, these cells expressed higher levels of CD107a, perforin, granzyme B, and IFN-γ. However, there was no difference in the percentage, apoptosis, and cell-cycle progression of NK cells induced by coated and non-coated rituximab. Non-coated rituximab activated NK cells by increasing AKT phosphorylation, further enhancing the abundance of XBP1s. In conclusion, we developed a new method for amplifying NK cells with higher antitumor functions with non-coated rituximab via autologous B cells from PBMCs, and this method more efficiently stimulated NK cell activation than by using coated rituximab.


INTRODUCTION
Natural killer (NK) cells are important innate immune cells that play essential roles in tumor surveillance (1,2). The function and number of NK cells in patients with cancer are usually diminished compared to healthy individuals (3,4). The abundance of NK cells in tumors is positively correlated with prognosis (5)(6)(7). Adoptive NK cell therapy, especially donor NK cell infusion after allogeneic stem cell transplantation, has exhibited impressive clinical responses (8)(9)(10)(11). The clinical outcomes of NK cell therapies are closely correlated with their antitumor cytotoxicity (12,13). Therefore, enhancing the antitumor function is essential for improving the clinical efficacy of NK cell therapy.
Feeder cells and cytokines are frequently used to enhance NK cell activity (14). Feeder cells can help NK cells obtain powerful antitumor functions and can stimulate their proliferation (15). Genetically modified K562 cells, a lymphoblastoid cell line infected with Epstein-Barr virus, are among the cancer cell lines commonly used as feeder cells (16)(17)(18)(19). Feeder cells (which are typically cancer cells) increase the clinical risks of adoptive NK cell therapies. Cytokines, such as interleukin (IL)-2, IL-15, IL-7, IL-12, and IL-18, can also cause NK cells to acquire powerful antitumor functions and undergo rapid proliferation (20)(21)(22)(23)(24). Cytokines are frequently used to amplify NK cells; however, their effects are weaker than feeder cells. Therefore, other methods for activating NK cells need to be developed.
Rituximab, a chimeric human-mouse monoclonal antibody (mAb) that targets CD20, can activate NK cells and cause B-cell death by antibody-dependent cell-mediated cytotoxicity (ADCC) (25,26). In this study, we investigated the use of B cells as feeder cells to activate and expand NK cells through rituximab-mediated ADCC. We also explored whether the methodology in which rituximab was applied (coated or noncoated on flasks) could significantly affect the antitumor functions of NK cells.

NK Cell Culture
Peripheral blood mononuclear cells (PBMCs) were isolated from the blood of healthy donors by Ficoll (Axis-Shield PoC AS, Oslo, Norway) gradient density centrifugation. This procedure was carried out with the approval of the ethics committee of the First Hospital of Jilin University. B cells were isolated using CD20 Microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). PBMCs (with or without B cells) were adjusted to a concentration of 2 × 10 6 cells/mL using Aly505 medium (Cell Science & Technology Institute Inc., Yamagata, Japan) containing 5% autologous serum, 600 IU/mL IL-2, and 10 ng/ mL IL-15 (both from Miltenyi Biotec). The cells were then added to antibody-coated or non-coated flasks to induced and expanded NK cells. For the NK cells cultured in non-coated rituximab flasks, different concentrations of rituximab (0 mg/mL, 0.25 mg/mL, 0.50 mg/mL, 0.75 mg/mL, 1 mg/mL, 2 mg/mL, and 4 mg/mL) were added directly to the medium at the beginning of culture. In all rituximab treatment groups, anti-CD161 antibodies (HP-3G10, cat # 339902; BioLegend, San Diego, CA, USA) were added to the medium at a final concentration of 1 mg/ mL on the first day of culture. After incubation for 3 d, media was replaced with fresh media containing 600 IU/mL IL-2 and 10 ng/mL IL-15. Every 2-3 d, half of the medium was replaced with fresh medium containing 600 IU/mL IL-2 and 10 ng/mL IL-15. NK cells were also expanded with irradiated K562-IL-21 as feeder cells, as previously described (19). After culturing for 14 d, NK cell function was evaluated.
To determine whether living B cells are necessary for expanding and activating NK cells with non-coated rituximab, B cells and NK cells were sorted from PBMCs by CD20 Microbeads and MACSxpress ® NK Cell Isolation Kit (Miltenyi Biotec). Living and fixed B cells (treated with 4% formaldehyde) were then incubated with NK cells according to the ratio of B cells and NK cells in PBMCs with either rituximab or nivolumab (OPDIVO, Bristol Myers Squibb, New York, NY, USA) for 3 d. Then the cells were transferred to the culture medium without antibodies and culture continued. After 14 d, the cells were harvested to evaluate NK cell cytotoxicity.

NK Cell Cytotoxicity
The cytotoxicity of NK cells was determined by performing the calcein-release test (Dojindo Laboratories, Kumamoto, Japan) as previously reported (28). Target cancer cells were collected and adjusted to a concentration of 1 × 10 6 cells/mL with PBS and then incubated with 1 mM calcein-AM at 37°C. After 30 min, the target cells were washed twice with PBS and adjusted to a concentration of 5 × 10 4 cells/mL in RPMI-1640 medium (Gibco, ThermoFisher Biochemical Products (Beijing) Co., Ltd, Beijing, China) containing 5% fetal bovine serum. The effector cells were adjusted to a concentration of 2.5 × 10 5 cells/mL using the same medium. To detect direct killing by NK cells, 100 mL target cells and 100 mL effector cells were added into the same wells of a 96well plate. For ADCC experiments, except for the target cells and effector cells mentioned above, rituximab was added to the medium at a final concentration of 10 mg/mL. To determine the minimum release, 100 mL target cells and 100 mL medium were added to the 96-well plates. To determine the maximum release, 100 mL target cells and 100 mL medium containing 0.4% TritonX 100 were added into one well of the 96-well plate.

Western Blot Assays
Western blotting was performed using mouse mAbs against AKT, p-AKT, or XBP1s (all from Cell Signaling Technology, Danvers, MA, USA), appropriate secondary antibodies, and an Enhanced Chemiluminescence substrate reagent kit (Beijing Labgic Technology Co., Ltd., Beijing, China).

Statistical Analyses
Data were analyzed by one-way or two-way analysis of variation (ANOVA) or two-tailed paired t-test using GraphPad Prism 8 software (GraphPad Software, San Diego, CA, USA).

NK Cells Induced by Non-Coated Rituximab Have Much Stronger Antitumor Function
We used different concentrations of rituximab to activate NK cells. When the rituximab concentration was below 0.75 mg/mL, the cytotoxicity of NK cells induced by coated rituximab increased with increasing rituximab concentration. However, there was no statistical difference in the cytotoxicity of NK cells among the 0.75 mg/mL, 1 mg/mL, 2 mg/mL, and 4 mg/mL treated groups. This indicates that concentrations of coated rituximab greater than or equal to 0.75 mg/mL do not affect the antitumor function of NK cells. With non-coated rituximab, the cytotoxicity of NK cells cultured with 0.50 mg/mL non-coated rituximab was better than those of 0 mg/mL and 0.25 mg/mL coated rituximab. However, there was no difference with the 0.50 mg/mL, 0.75 mg/mL, 1 mg/mL, 2 mg/mL, and 4 mg/mL non-coated rituximab groups. These results indicated that when the rituximab concentration was greater than or equal to 0.50 mg/ mL, the antitumor function of NK cells did not change with the increase of rituximab concentration. ( Figure 1A). Therefore, in the following experiments, 1 mg/mL rituximab was used to remove differences caused by different concentrations of rituximab. Under the same concentration of rituximab, the antitumor activity of NK cells obtained in the non-coated rituximab group was always significantly stronger than the coated rituximab group ( Figure 1A). We also found that the cytotoxicity of NK cells induced by rituximab was stronger than NK cells induced by control IgG-isotype antibody, regardless of whether the antibodies were coated or not (coated: 31.44% ± 3.50% vs. 17.35% ± 6.37%, p < 0.05; non-coated: 52.85% ± 6.06% vs. 17.35% ± 7.49%, p < 0.001; Figure 1B).
To compare the functions between NK cells induced by noncoated rituximab and NK cells induced by previously established methods, a comparative experiment was carried out. Although, the purity and the number of NK cells induced by non-coated rituximab were lower than NK cells induced by feeder cells, there was no difference in the purity and numbers of NK cells, respectively, obtained by non-coated rituximab, coated rituximab, and coated CD16 mAb. (Figures 1C-E). More importantly, the cytotoxicity of NK cells obtained by our method was significantly stronger than NK cells cultured with feeder cells and coated CD16 mAb, regardless of the presence of NK cell-sensitive K562 cells (45.31% ± 11.11% vs. 37.72% ± 11.46%, p < 0.05 and 45.31% ± 11.11% vs. 28.86% ± 10.27%, p < 0.001, respectively; Figure 1F) or NK cell-resistant Raji cells (18.61% ± 5.56% vs. 14.96% ± 5.37%, p < 0.05 and 18.61% ± 5.56% vs. 8.51% ± 2.36%, p < 0.001, respectively; Figure 1G).

Apoptosis and Cell-Cycle Progression in NK Cells Induced by Coated and Non-Coated Rituximab Did Not Differ
No differences were found in terms of apoptosis and cell-cycle progression for the expanded NK cells between the non-coated rituximab group and coated rituximab group (Figure 7).

Non-Coated Rituximab Activated NK Cells Through the AKT-XBP1s Axis
The AKT-XBP1s axis is an important regulatory pathway for NK cell function and is closely related to the release of cytokines, perforin, and granzyme by NK cells (29). We compared changes in the AKT-XBP1s axis in NK cells induced by coated and noncoated rituximab. Our results showed that AKT phosphorylation and XBP1s expression in NK cells from non-coated rituximab were significantly higher than those obtained in coated rituximab ( Figure 8). These results further indicate that non-coated rituximab was more effective in activating NK cells.  the activity of NK cells and has the potential to improve the safety of NK cell therapy in clinical settings, both of which are crucial for clinical efficacy.

DISCUSSION
There are many ways to enhance the activity of NK cells (24). Using feeder cells is a very effective method for expanding and activating NK cells; however, as feeder cells are usually cancer cells, this approach increases the risks of NK cell therapy in clinical applications. Previously, we reported a method for activating NK cells using an anti-CD16 mAb without feeder cells, where coating the CD16 mAb on the cell culture flask efficiently activated NK cells. We found that coated rituximab can also be used to culture NK cells. In terms of NK cell cytotoxicity, NK cells activated with coated rituximab were similar to those coated with the CD16 mAb. However, the molecular mechanisms involved in NK activation by coated CD16 mAb and coated rituximab are clearly distinct. The CD16 mAb that we used is a mouse IgG1 with low binding capacity for human CD16, while rituximab is a human IgG1 with high affinity for CD16. The NKactivating activity of coated CD16 mAb will mostly be related to crosslinking of human CD16 via Fab binding. On the other hand, coated rituximab can activate NK cells only through an accessible Fc portion binding to CD16.
During in vitro NK cell expansion from PBMCs, B cells die spontaneously over time in the absence of CD40 stimulation or IL-4 (31). Rituximab, a mAb that targets CD20, can reduce the number of normal B cells through ADCC in clinical treatment, as normal B cells also express CD20 (32). In this study, we were interested in determining whether B cells can be used as feeder cells to activate NK cells through rituximab-mediated ADCC. To address this question, we depleted B cells from the NK cellinduction system and found that the antitumor function of NK cells activated by non-coated rituximab decreased sharply. However, the cytotoxicity of NK cells induced by coated rituximab was not affected by B cell elimination. After B cells were fixed, although the antitumor effect of NK cells activated by rituximab group was stronger than that of nivolumab group, it was still significantly lower than that of rituximab combined with living B cells group. Furthermore, rituximab F(ab')2 fragments with living B cells did not elicit enhanced NK cytotoxicity ( Figure S1), which revealed that cytokine secretion by rituximab-triggered B cells apparently plays no independent role. These results indicate that rituximab combined with living B cells can better activate NK cells through ADCC. CD16 is associated with the immunoreceptor tyrosine-based activation motif containing T-cell receptor z chain (33). NK cells activate upon the formation of the z-chain homodimer (34). We speculate that when the ADCC effect occurs among living B cells, rituximab and NK cells, synapses will be formed at the contact sites, and a large number of CD20 molecules will gather at the synapse sites to combine with rituximab and CD16 molecules, enhancing the formation of the z-chain homodimer in NK cells, thus leading to their activation. However, when B cells are fixed, the fluidity of the B cell membrane is weakened, which is not conducive to the formation of the z-chain homodimer in NK cells. Therefore, the activation of NK cells activated by fixed B cells and rituximab is obviously weakened.
We found that NK cells cultured with non-coated rituximab showed stronger antitumor ability and higher expression levels of CD107a, granzyme B, and perforin comparing to NK cells cultured with coated rituximab. Therefore, we speculate that coated rituximab is not conducive to binding B cells and NK cells simultaneously because the Fc fragment is fixed at the bottom of the flask, which affects the exertion of the ADCC effect. Therefore, the effect of activating NK cells is weakened ( Figure 9A). When rituximab is not coated at the bottom of the flask, non-coated rituximab can connect B cells to NK cells and exert the ADCC effect, thereby activating and expanding NK cells ( Figure 9B).
CD57 is a marker of highly differentiated cells (35). The proliferative ability of CD57 + NK cells is lower than CD57 -NK cells (35,36). We found that the expression of CD57 on NK cells activated by rituximab decreased significantly, which indicated that NK cells obtained by rituximab treatment still had strong proliferative ability. We also found that the expression of KIR2DL1, KIR2DL2/3, and KIR3DL1 on NK cells induced by non-coated rituximab was significantly higher than in NK cells induced by coated rituximab. It has been reported that mature anergic NK cells (expressing inhibitory KIR receptors for MHC-I) that are adoptively transferred into an MHC-I-sufficient setting become responsive (37). Therefore, we speculate that non-coated rituximab not only mediates NK cells to exert an ADCC effect on B cells, but also strengthens the interactions between KIR molecules on NK cells and MHC-I on B cells, making NK cells responsive. This may be the reason why the antitumor function of NK cells activated by non-coated rituximab was more robust. CD16 expression on NK cells expanded using non-coated rituximab was lower than NK cells expanded with coated rituximab. Loss of CD16 has recently been shown to be important for NK cell detachment and sequential engagement of multiple target cells (38). Rapid detachment between NK cells and target cells increased engagement of NK cells and multiple targets and enabled a greater proportion of NK cells to perform serial killing. In addition, detachment of NK cells from target cells can avoid excessive activation of NK cells and ensure the survival of NK cells (39). Therefore, the shedding of CD16 may be an important molecule to keep NK cells containing high antitumor effect without over-activation. We also found that the ADCC effect and IFN-g-secretion level in NK cells in the noncoated rituximab group were more robust. This finding further confirms that non-coated rituximab might be more suitable than coated rituximab to promote full NK cell activation. CD161, expressed on the surface of NK cells, combines with lectin-like transcript 1 expressed on the surface of B cells and transmits inhibitory signals to NK cells (40). To improve the rituximabmediated activation of NK cells by B cells, an anti-CD161 mAb was added to the NK cell culture medium to block the transmission of inhibitory signals from B cells to NK cells. Rituximab and the CD161 mAb were only added during the first few days of culture because B cells are gradually consumed by NK cells as feeder cells during this time ( Figure S2). Therefore, there was no need to add these antibodies during subsequent culturing. Although NK cells were activated by B cells and rituximab during the initial days of culture, they maintained high activity and did not undergo apoptosis after 14 d in culture. This finding indicates that the method developed in this study is effective in activating NK cells and promoting the amplification of activated NK cells. This method could be used to avoid excessive activation of NK cells by rituximab and reduce cell culture cost.
The engagement of CD16 molecules by mAbs has been reported to markedly induce AKT activation (41). In this study, we found that AKT phosphorylation in the non-coated rituximab group was higher than the coated rituximab group. Additionally, XBP1s accumulated at higher levels in NK cells induced with non-coated rituximab than in NK cells induced with coated rituximab. The AKT-XBP1s axis plays a critical role in the antitumor function of NK cells (29). Therefore, we speculate that non-coated rituximab efficiently combines with B cells and CD16 molecules, increasing AKT activation and the abundance of XBP1s, which then activate NK cells.
In conclusion, we established a new method for activating and amplifying NK cells with non-coated rituximab via autologous living B cells from PBMCs, that showed a potent antitumor function compared to NK cells induced with coated rituximab. This method not only simplifies technical operation, but also enhances the safety of NK cell therapy, which will benefit the use of this as a treatment in the clinic.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by The ethics committee of the First Hospital of Jilin University. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
JC, WL, and YW conceived the study. CN and SZ designed and carried out part of the experiments and drafted the entire manuscript. YC, ML, LZ, DX, ZL, and JX carried out part of the experiments. All authors contributed to the article and approved the submitted version.