Cytokine engineered NK-92 therapy to improve persistence and anti-tumor activity

Abstract

Natural killer (NK) cells are an attractive cell source in cancer immunotherapy due to their potent antitumor ability and promising safety for allogenic applications. However, the clinical outcome of NK cell therapy has been limited due to poor persistence and loss of activity in the cytokine-deficient tumor microenvironment. Benefits from exogenous administration of soluble interleukin-2 (IL-2) to stimulate the activity of NK cells have not been significant due to cytokine consumption and activation of other immune cells, compromising both efficacy and safety. Methods: To overcome these drawbacks, we developed a novel membrane-bound protein (MBP) technology to express IL-2 on the surface of NK-92 cells (MBP NK) inducing autocrine signal for proliferation without IL-2 supplementation. Results: The MBP NK cells exhibited not only improved proliferation in IL-2 deficient conditions but also stronger secretion of cytolytic granules leading to enhanced anti-tumor activity both in vitro and in vivo. Furthermore, the experiment with a spheroid solid tumor model exhibited enhanced infiltration by MBP NK cells creating higher local effector-to-target ratio for efficient tumor killing. These results suggest MBP technology can be an effective utility for NK-92 cell engineering to increase anti-tumor activity and reduce potential adverse effects, providing a higher therapeutic index in clinical applications.

Keywords: interleukin-2; membrane-bound protein (MBP); microwell array chip; natural killer cell; self-activation; tumor-infiltrating lymphocytes.

Figure 1

Schematic diagram of the MBP technology in NK cells and expected role in the tumor site. (A) Schematic illustration of cytokine engineered MBP NK cells and functional difference between parental NK-92 cells and MBP NK cells. Because of self-stimulation via MBP IL-2, MBP NK cells can survive IL-2 deficient environment as well as can reduce systemic risks of hyperimmune activation induced by exogenous treatment of IL-2. (B) The fabrication of the microwell array sheet to generate a hematological tumor model with K562 cancer cells. (C) In the solid tumor spheroid model, A549 spheroid and NK-92 cells or MBP NK cells are co-cultured and observed using confocal microscopy. The chambers are made with a guide block insert, and the bottom surface is coated with poly hydroxyethyl methacrylate (MEMA).

Figure 2

MBP IL-2 engineering in NK-92 cells. (A) Schematic representation of MBP IL-2 expression cassette. The ORF encoding MBP IL-2 was cloned in lentiviral vector to deliver to NK-92 cells. (B) Representative confocal image of parental NK-92 (left) and MBP NK (right). MBP NK cells were labeled with CellTracker (green), and nuclei were stained with Hoechst (blue). The MBP IL-2 structure was identified as red color expressing of mCherry at the C-terminal (red). Scale bar, 10 μm. (C) The MBP IL-2 positive cells according to different MOI analyzed by flow cytometry. After transduction, cells were labeled with PE-conjugated anti-flag tag antibody. The histogram of blue color is shown as MBP IL-2 expressing cells. The isotype control is shown as a grey color in all figures. (D) The expression of mCherry in NK-92 (left) and MBP NK cells (right) were analyzed by flow cytometry. (E) NK-92 cells and MBP NK cells were labeled PE-conjugated anti-IL-2, the expression of mCherry and IL-2 in NK-92 (left) and MBP NK cells (right) were analyzed by flow cytometry.

Figure 3

In vitro proliferation and expansion of NK cells by MBP IL-2 engineering and the IL-2 signal transmission in a cis-acting mechanism. (A) Proliferation of NK-92 cells (gray color) and MBP NK cells (red color) in the presence or absence of IL-2 were investigated by an MTS assay. Means ± SDs of triplicate determinations are shown. (B) Viability of the parental and the MBP NK cells over time in the absence of IL-2 supplement in growth medium. The MBP NK cells were continuously maintained, with approximately 70% viability. (C) Expansion of parental NK-92 and MBP NK cells without exogenous IL-2 addition. While the parental NK-92 cells did not proliferate and exhibited a reduced cell number, the MBP NK cells proliferated appropriately, with an estimated doubling time of 42 hours. (D) Evaluation of the autocrine effect of MBP IL-2 engineering. IL-2 signal reporter HEK cells were purchased from InvivoGen (# hkb-il2) to monitor the activation of the JAK-STAT pathway induced by IL-2. The level of secreted embryonic alkaline phosphatase (SEAP) from reporter HEK cells did not increase proportionally even when the proportion of cocultured MBP NK cells was increased, which confirmed that the MBP IL-2 on the MBP NK cells mainly acts in a cis manner dominantly.

Figure 4

MBP NK cells exhibit enhanced anti-tumor effects compared to parental NK cells. (A) MBP NK mediated cancer cell killing. K562 (left) and A549 cells (right) that stably expressed luciferase were co-cultured with NK-92 or MBP NK cells at an E:T ratio of 5:1. The percentage of dead tumor cells was determined by quantifying bioluminescence. The statistical significance was determined by one-way ANOVA with Tukey's multiple comparison test. *, p = 0.0280, **, p = 0.0018; ****, p < 0.0001. (B) The expression level of B7H6 on the surface of K562 (top) and A549 cells (bottom), which is one of the major activating immune ligands on cancer cells. NKp30 expressed by NK cells can recognize and kill B7H6 expressing tumor cells. (C) The quantitative measurement of lytic granules released from NK-92 and MBP NK cells by ELISA. The concentrations of secreted granzyme B (left) and perforin (right) from NK-92 cells (absence or presence of IL-2) and MBP NK cells were measured in the supernatant from the co-culture with K562 cells. The statistical significance was determined by ordinary one-way ANOVA with Tukey's multiple comparison test. *, p = 0.0138; **, p = 0.0091; ***, p = 0.001. (D) Fold change in CD107a expression analyzed by flow cytometry. The plot was generated by Prism 8 software. The statistical significance was determined by ordinary one-way ANOVA with Tukey's multiple comparison test. **, p = 0.0075; ***, p = 0.0003; ****, p < 0.0001. From all the analyses, statistically insignificant (ns) results are not shown.

Figure 5

Quantitative analysis of cytotoxicity mediated by parental NK and MBP NK cells at the single-cell level in hematological cancer cells (K562). (A) Representative time-lapse images of the NK cells (top) and the MBP NK cells (bottom) cocultured with K562 cells were obtained with a confocal microscope. Both types of NK cells were stained with CellTracker^TM blue, and K562 cells were labeled with calcein AM. Compared to the parental NK cells, the MBP NK cells killed the myelogenous leukemia cells very rapidly at all E:T ratios. The red arrows indicate that the hematological cancer cells were still alive, although the NK cells were actively interacted with the cancer cells and seemed to destroy their membranes. (B) Comparison of target lysis time between the NK and MBP NK cells at different E:T ratios. The time was recorded when the first target cell died. The MBP NK cells killed the target cells more efficiently, with less variation, and the E:T ratio did not affect the time required for target lysis. (C) Probability of target cell lysis by the NK and MBP NK cells at different E:T ratios. Compared to the parental NK cells, the MBP NK cells showed superior lysis performance, and the E:T ratio affected lysis performance. Lysis was evaluated by analyzing 319 microwell chambers. The statistical analysis was performed with an unpaired two-tailed t-test, ***, ρ < 0.001; ns = not significant.

Figure 6

The infiltration and cytolysis abilities of the MBP NK cells in the lung tumor spheroid model. (A) Time-lapse confocal images of A549 tumor spheroids stained with calcein AM and cocultured with NK cells or MBP NK cells. Red arrows indicate clusters of MBP NK cells around the tumor spheroid. The cluster of MBP NK cells exhibited effective cytolysis, creating a massive dead zone in the tumor spheroid (see the yellow arrow). (B) Number of infiltrating cells for the NK and MBP NK cells after co-cultivation with A549 spheroids for 24 hours. The number of infiltrating cells was dramatically increased in the MBP NK cells compared to the parental NK cells. (C) Representative confocal images of A549 spheroids alone and spheroids co-cultured with NK cells or MBP NK cells after 48 hours. Scale bar, 100 μm. The MBP NK cells showed more infiltration inside the tumor spheroid. (D) Quantitative analysis of live cancer cells. The cancer spheroids presented significantly weaker fluorescent signals when cocultured with the MBP NK cells. Fluorescence was analyzed by reviewing the confocal images captured after 48 hours of cocultivation (n=5). The statistical analysis was performed with an unpaired two-tailed t-test, **, ρ < 0.01; ***, ρ < 0.001.

Figure 7

In vivo anti-tumor efficacy of MBP NK in a solidified K562 tumor mouse model. (A) Experimental scheme of K562 xenograft mouse model. Female NIG mice, aged 7-8 weeks (n = 7 / group) were inoculated subcutaneously in their right flank with K562 tumor cells. When tumor volumes reached ~50 mm³, after 7 days of tumor inoculation, mice were intravenously administered MBP NK or NK-92 at a dose of 5 × 10⁶ cells / mouse (3 doses total). (B) Tumor volume analysis to quantitatively determine tumor growth inhibition. Overall and individual responses are shown. Treatment of MBP NK resulted in potent tumor growth inhibition compared to NK-92 or IL-2 co-administrated NK-92. (C) A picture of tumors from each mouse after study termination. (D) On day 26, blood samples were collected from each mouse and IFN-γ level was analyzed by ELISA. (E) Body weight changes (%) in mice after injection of the indicated NK cells. All groups showed body weight change within the 5% range. The statistical significance was determined by ordinary one-way ANOVA with Holm-Šidák's multiple comparison test, ****, p < 0.0001. From all the analyses, statistically insignificant (ns) results are not shown.