One‐Step Method for Instant Generation of Advanced Allogeneic NK Cells

Abstract Conventional combinatorial anticancer therapy has shown promising outcomes; still, a significant interest in developing new methods to reinforce and possibly merge chemotherapy and immunotherapy persists. Here, a new one‐step method that immediately modifies immune cells into a targeted form of chemoimmunotherapy through spontaneous and rapid incorporation of hydrophobized antibody–drug conjugates (ADCs) on the surface of immune cells is presented. Therapeutic objectives of this approach include targeted delivery of a potent chemotherapeutic agent to avoid adverse effects, enhancing the mobilization of infused immune cells toward tumor sites, and preserving the intense cytotoxic activities of immune cells against tumor cells. The embedding of hydrophobized ADCs on the immune cell membrane using the strategy in this study provides noninvasive, nontoxic, and homogenous modifications that transiently arm immune cells with highly potent cytotoxic drugs targeted toward cancer cells. The resulting surface‐engineered immune cells with ADCs significantly suppress the tumor growth and drive the eradication of target cancer cells through combinatorial anticancer effects. This novel strategy allows convenient and timely preparation of advanced chemoimmunotherapy on a single immune cell to treat various types of cancer.


DOI: 10.1002/advs.201800447
beneficial synergy for maximizing the clinical antitumor activity. [1,2] Effective chemoimmunotherapy requires that (1) chemotherapeutic agents induce cancer cell death and promote immunomodulation, (2) targeted chemotherapy minimizes the adverse effects on immune cells, and (3) immune effector cells maintain their cytolytic activity against cancer cells. [1,3,4] New strategies to combine chemotherapy and immunotherapy that meet the above criteria would represent a new platform for the development of targeted cancer chemoimmunotherapy.
Monoclonal antibodies (mAbs) have been established as the mainstream mode of immunotherapy in clinical oncology as well as excellent vehicles for targeted delivery of cytotoxic agents. [5][6][7] Over the past two decades, antibody-drug conjugates (ADCs) were developed and continued to show promising clinical responses through increasing the payload of highly potent anticancer drugs and overcoming the offtarget adverse effects. [8] In company with ADCs, immune cells emerged as an innovative immunotherapy as the technologies in isolating tumor-active immune cells from a patient and growing them into sufficient numbers for reinfusion in ex vivo condition started to develop. [9] Genetic engineering enabled the customization of receptors on immune cells to target specific cancer antigens and the resulting chimeric antigen receptor T (CAR-T) cells have received clinical attention and support. [10] Although ADCs and immune cell therapies are the state-ofthe-art technologies, the synergy between ADCs and immune cell therapies has become attractive to achieve greater clinical anticancer response. In this combinatorial approach, ADCs induce immunogenic cancer cell death that makes the dying cancer cells much more obvious for transferred immune cells to eliminate. [11] To attain this potential therapeutic benefit, it is necessary to invite creative methods that support the concurrent delivery of ADCs and immune cells to the tumor site.
We developed a one-step method that spontaneously transforms nonspecific immune cells into a new form of targeted chemoimmunotherapy through the introduction of ADCs on the surface of an active immune cell. The process of embedding ADCs on the immune cell surface exploits the hydrophobic interaction between a polymeric lipid chain and the lipid bilayer of the cell membrane. Likewise, lipid-conjugated Conventional combinatorial anticancer therapy has shown promising outcomes; still, a significant interest in developing new methods to reinforce and possibly merge chemotherapy and immunotherapy persists. Here, a new one-step method that immediately modifies immune cells into a targeted form of chemoimmunotherapy through spontaneous and rapid incorporation of hydrophobized antibody-drug conjugates (ADCs) on the surface of immune cells is presented. Therapeutic objectives of this approach include targeted delivery of a potent chemotherapeutic agent to avoid adverse effects, enhancing the mobilization of infused immune cells toward tumor sites, and preserving the intense cytotoxic activities of immune cells against tumor cells. The embedding of hydrophobized ADCs on the immune cell membrane using the strategy in this study provides noninvasive, nontoxic, and homogenous modifications that transiently arm immune cells with highly potent cytotoxic drugs targeted toward cancer cells. The resulting surface-engineered immune cells with ADCs significantly suppress the tumor growth and drive the eradication of target cancer cells through combinatorial anticancer effects. This novel strategy allows convenient and timely preparation of advanced chemoimmunotherapy on a single immune cell to treat various types of cancer.

Introduction
Advanced cancer therapies have recently focused on combining chemotherapy and immunotherapy to promote therapeutically ADCs (hydrophobized ADCs) can be readily incorporated into the lipid bilayer without disrupting the cell membrane integrity and award new functions to the surface-engineered cells. [12,13] We hypothesized that the surface-engineered immune cells containing ADCs could simultaneously deliver potent chemotherapeutic agents to the target tumor and enhance the homing of adoptively transferred immune cells toward the tumor sites without compromising their cytotoxic activities, ultimately intensifying the combinatorial anticancer efficacy to combat cancer.
These T-DM1 surface-engineered NK (SE-NK/T-DM1) cells, the product generated using our approach, recognized and destroyed human epidermal growth factor receptor 2 (HER2)positive cancer cells through the combined activity of T-DM1 and NK cells. This single-injection formulation chemoimmunotherapy, SE-NK/T-DM1 cells, suppressed the progression of the target tumor significantly compared to the cotreatment of NK cells and T-DM1. Our innovative strategy to instantaneously generate advanced immune cells as "off-the-shelf" chemoimmunotherapeutic reagents is therapeutically advantageous over the conventional chemoimmunotherapeutic strategies because it simultaneously delivers antibodies, cytotoxic agents, and immune effector cells to the target tumor (Figure 1).

Results and Discussion
T-DM1 was generated through the expression of trastuzumab (TZ) in mammalian cells followed by DM1 conjugations. [14,15] Prepared T-DM1 was subsequently hydrophobized by attaching DMPE-PEG-NHS, resulting in the production of DMPE-PEG-T-DM1 ( Figure S1, Supporting Information). T-DM1 synthesized in our lab exhibited similar cytotoxicity compared to the commercial product, Kadcyla ( Figure S2, Supporting Information). Surface engineering of NK cells with various amounts of DMPE-PEG-T-DM1 affects neither the viability nor the proliferative activity of NK cells (Figure 2a-c). Reliable modification of 5 × 10 5 immune cells required 100 µg of DMPE-PEG-T-DM1 that yielded ≈2.1 µg of T-DM1 embedded on the cell membrane of 1 × 10 5 SE-NK/T-DM1 cells ( Figure S3, Supporting Information).
T-DM1 was detected on the surface of SE-NK/T-DM1 cells for over 48 h in complete growth media ( Figure 2d) and two of the key NK cell-specific markers, CD56 and 2B4, were available on the surface of SE-NK/T-DM1 cells. These results demonstrate that T-DM1 is embedded on the NK cell surface without internalization and the surface engineering of NK cells with ADCs does not interfere with the NK cell receptor accessibility, suggesting that the inherent cytolytic activity of NK cells is retained upon the surface engineering ( Figure 2e). Because the allogeneic NK cells used in this study, NK-92 cells, lack CD16, CD32, and CD64 IgG receptors that can initiate antibody internalization and antibody-dependent cellular cytotoxicity (ADCC), [16] the surface engineering with T-DM1 appears to have minimal effects on NK cell metabolism and viability. In cancer cells, T-DM1 internalization occurs through HER2 receptor-mediated endocytosis. [17] NK cells do not express HER2 on their membrane, therefore DMPE-PEG-T-DM1 embedded on the surface of SE-NK/T-DM1 cells were not internalized and showed negligible cytotoxicity. Moreover, PEG spacer between the DMEP and T-DM1 provides a physical barrier for internalization. [18] Studies have reported that longer PEG spacers not only inhibit the internalization of biomolecules but also reduce the membrane insertion efficiency by increasing the steric hindrance. [18][19][20] The aforementioned membrane insertion efficiency of DMPE-PEG-T-DM1 yielded about 10% due to the presence of long PEG spacer ( Figure S3 In order for T-DM1 to exert its anticancer activity on cancer cells, T-DM1 on the SE-NK/T-DM1 cells must transfer to the    Internalization of T-DM1 is crucial for its anticancer efficacy because DM1 acts on intracellular targets in cancer cells. [7,21] In keeping with previously reported observations on cellular uptake of T-DM1, trafficking of T-DM1 to lysosomes, and release of DM1, [22] we focused on confirming the internalization of T-DM1 transferred from SE-NK/T-DM1 cells in the target cancer cells. Cancer cells plated on an eight-chambered cover glass slide were labeled with nuclear stating dye (blue) to observe the location of FITC-labeled T-DM1 (green) transferred from SE-NK/T-DM1-FITC cells (red). Identical to the study above, unbound NK cells were thoroughly removed after 30 min of coincubation. Distinct fluorescent dots representing the internalized T-DM1, following the transfer from the assessed SE-NK/T-DM1 cells, were detected in the cytoplasm of target cancer cells (Figure 3c To validate the therapeutic advantages of SE-NK/T-DM1 cells over the T-DM1+NK cotreatment, we first treated the cancer cells with SE-NK/T-DM1 cells and T-DM1+NK cotreatment for 24 h without the removal of unbound immune cells (Figure 4a,b). Both treatments induced similar levels of cancer cell death, indicating that continuous exposure to the T-DM1+NK cotreatment allows enough time for NK cells to identify dying cancer cells affected by T-DM1 in a confined well system. In MDA-MB-231 cells, only the anticancer activity of NK cells was observed in both treatment groups (Figure 4c). Subsequently, cancer cells were incubated with the identical treatments for 2 h and the unbound effector cells were removed to mimic the in vivo cancer-targeted homing effect. We further incubated the remaining cancer-bound effector cells with the target cells for 24 h and recorded the resulting cancer cell death. In SK-BR-3 cells and Calu-3 cells, we found that the level of cancer cell death induced by SE-NK/T-DM1 cells was greater than that induced by NK cell or T-DM1+NK cotreatment, while no significant cell death was noticed in MDA-MB-231 cells (Figure 4d-f). This is due to the fact that a higher number of SE-NK/T-DM1 cells remained bound to SK-BR-3 cells and Calu-3 cells, resulting in an augmented level of anticancer activity.
Next, we assessed the effect of trastuzumab, DM1, and NK cells, contained in SE-NK/T-DM1 cells on cancer cell viability.
To identify the anticancer effect of DM1, we prepared trastuzumab surface-engineered NK (SE-NK/TZ) cells and compared the cancer cell death induced by SE-NK/TZ cells and SE-NK/T-DM1 cells. Cancer cell death was analyzed 24 h of coincubation after the removal of unbound NK cells 2 h after the treatment. As expected, T-DM1 exhibited a greater cytolytic effect against SK-BR-3 cells than TZ (Figure 5a). The resulting enhanced cancer cell death was due to the addition of DM1. NK cells and TZ (TZ+NK) cotreatment showed slightly improved cytotoxicity compared to the NK cells alone however it was much less compared to the T-DM1+NK cotreatment. The treatments involving T-DM1 further enhanced anticancer activity against HER2-positive cancer cells, and SE-NK/T-DM1 cells exhibited anticancer activity superior to all other treatments. We postulated that DM1 contained in T-DM1 induced an increase of ≈20% in the death of HER2-positive cancer cells. Except for the nonspecific cytolytic activity of NK cells, none of the treatments induced significant cytotoxicity in MDA-MB-231 cells (Figure 5b).
We To determine whether or not NK cells were activated upon incorporation of DMPE-PEG-T-DM1 on their surface, we assessed the level of CD107a expression, a prominent degranulation marker, [23,24] on SE-NK/T-DM1 cells and unmodified NK cells upon engaging the target cancer cells (Figure S4

a-c) Cancer cell death induced by coincubating T-DM1+NK cotreatment or SE-NK/T-DM1 cells with SK-BR-3 cells, Calu-3 cells, or MDA-MB-231 cells. Cancer cells labeled with CMAC (blue) dye were incubated with unmodified NK cells, T-DM1, T-DM1+NK cotreatment, or SE-NK/T-DM1 cells for 24 h without removing the unbound NK cells. Similar level of cancer cell death resulted from both T-DM1+NK cotreatment or SE-NK/T-DM1 cells on SK-BR-3 cells and Calu-3 cells in a confined volume. In MDA-MB-231 cells, only the anticancer activity of NK cells was observed. d-f) Targeted anticancer activity of SE-NK/T-DM1 cells was determined by removing the unbound NK cells after the coincubation of SE-NK/T-DM1 cells with SK-BR-3 cells, Calu-3 cells, and MDA-MB-231 cells. Cancer cells labeled with CMAC (blue) dye were incubated with T-DM1+NK cotreatment, SE-NK/T-DM1 cells, or other corresponding treatments.
Unbound NK cells were removed after 2 h of incubation and the remaining cell mixtures were incubated for additional 24 h. Removal of unbound NK cells allows testing for target cancer homing effect. Enhanced HER2-targeted anticancer activity was observed with SE-NK/T-DM1 cells but negligible cancer cell death was observed in MDA-MB-231 cells. Cancer cell death was measured with flow cytometry using an annexin V Alexa Fluor 488 and propidium iodide kit. Data represent mean ± SD (ns, not significant, *P < 0.05, **P < 0.01, ****P < 0.0001, by two-way ANOVA with Bonferroni post hoc tests). Figure S4b,c, Supporting Information). This increase was absent in the nontarget MDA-MB-231 cells ( Figure S4d, Supporting Information). These results prove the absence of nonspecific activation of NK cells following the surface modification with DMPE-PEGT-DM1 and support the target-specific activation of SE-NK/T-DM1 cells.

SK-BR-3 cells, and Calu-3 cells, (
We compared the in vivo anticancer activity of SE-NK/T-DM1 cells to that of T-DM1+NK cotreatment using HER2positive Calu-3 models and HER2-negative MDA-MB-231 models. Tumor-bearing NOD scid Gamma (NSG, NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice administered with 1 × 10 7 SE-NK/T-DM1 cells received ≈210 µg of T-DM1, which is similar to the recommended dose found in the literature for mice models (7-10 mg kg −1 ). [25] In the HER2-positive tumor model, SE-NK/T-DM1 cells exhibited the strongest anticancer efficacy through the combinatorial effects (Figure 6a). The T-DM1+NK cotreatment inhibited tumor growth when compared to the control group. Treatment of SE-NK/T-DM1 cells demonstrated a substantial suppression in tumor growth compared to the T-DM1+NK cotreatment. In the HER2negative tumor model, no significant difference in the tumor growth suppression was observed among all treatment groups (Figure 6b). The Calu-3 models and MDA-MB-231 models had steady body weights during the study period, indicating that the treatments caused no severe toxicity (Figure 6c,d).
For the biodistribution of SE-NK/T-DM1 cells in Calu-3 tumor models, we observed negligible accumulation of NK cells in the heart, kidneys, and lungs (Figure 6e). While NK cells were detected in the liver and spleen, no significant differences in the number of NK cells were observed among the treatment groups.  Data represent mean ± SD (ns, not significant, **P < 0.01, ****P < 0.0001, by one-way ANOVA followed by Tukey post hoc tests).
Surface engineering of NK cells with T-DM1 enabled simultaneous accumulation of T-DM1 and NK cells in the target tumor tissue. Binding of T-DM1 to HER2-positive cancer cells would inhibit the downstream signaling pathway associated with PI3K and AKT and the chemotherapeutic agent, DM1, disrupts the microtubule networks in the target cells, both of which lead to cell cycle arrest and cell apoptosis. [17,26] SE-NK/T-DM1 cells migrated toward the antigen-expressing cancer cells from physical contact with the target cancer cells, which in turn increased the chance to stimulate the cytolytic function of NK All agents were freshly prepared in 250 µL of PBS for 1 min infusion. Tumor and other vital organs were harvested at 24 h post-treatment. Single-cell suspension was prepared from the harvest tissues and APC-conjugated anti-CD56 antibodies were applied to detect NK cells. Flow cytometer was used to count NK cells among 1 × 10 5 total cells. Data represent mean ± SD (ns, not significant; ***P < 0.001; ****P < 0.0001; two-way ANOVA with Bonferroni post hoc tests).
cells. These NK cells in close contact with the target cancer cells, then, eradicate the cancer cell undergoing apoptosis by recognizing damage-associated molecular patterns (DAMP) expressed on the dying cancer cells. [27][28][29][30][31] Allogeneic immune cells, the cell-of-interest to be weaponized by our one-step method, gained attention as a suitable solution to reinforce the diminishing active immune cell population in cancer patients due to their low occurrence adverse effects, high tumor-specific cytotoxicity, predictable anticancer activity, and ease of ex vivo expanding, maintaining, and activating a large cell population. [32][33][34][35][36] Even the most advanced methods of adoptive immune cell therapy, including chimeric antigen receptor-T (CAR-T) cells, still face challenges such as long production time, labor intensity generation process, high cost for widespread application, and limited efficacy in solid tumors. [37][38][39][40][41][42] Using our one-step method, however, advanced immune cells with a specific tumor-homing capability and potent anticancer activity based on chemoimmunotherapy can be generated instantly at the bedside, greatly reducing the time and cost required to obtain sufficient tumor-reactive immune cells. More importantly, the surface-engineered immune cells with ADCs generated by our approach would have enhanced efficacy even in solid tumors because antibodies, chemotherapeutic agents, and immune cells, all of which comprise our advanced immune cells, work in concert to eradicate the target cancer. One potential problem of expanding the application of hydrophobized ADCs to other immune cells may be the presence of Fc receptors. Fortunately, the conjugation of DMPE-PEGs to ADCs may circumvent the issue by masking the Fc region to increase the steric hindrance that lowers the binding affinity of Fc receptors on immune cells. [43] Other creative antibody engineering methods, such as using single chain variable fragment (scFv) and altering Fc region to reduce Fc receptor binding affinity, can be employed as an alternative strategy for our surface engineering purpose. [44][45][46]

Conclusion
We demonstrated that our one-step method constitutes a new platform to produce advanced form of chemoimmunotherapy that can achieve elevated levels of anticancer efficacy in many types of cancers including solid tumors and the product of our innovative mode of chemoimmunotherapy is promising. Furthermore, this approach enables embedding any type of ADCs on the surface of any class of immune cells, including T cells, DCs, and macrophages, in order to transform them to combat a broad spectrum of cancers. The applicability of surface-engineered immune cells is expected to be high because many new ADCs and new types of allogeneic immune cells currently undergoing discovery and development are viable candidates for our one-step method to generate advanced chemoimmunotherapy that potentially target different types of cancers. Our one-step method, a modular design allowing for the matching of allogeneic immune cells and ADCs based on particular needs, will be utilized to produce a wide range of targeted chemoimmunotherapy in the near future that adheres to the goal of creating "off-the-shelf" reagents. Transfer of T-DM1 from SE-NK/T-DM1 cells to target cancer cells was examined using confocal microscopy. SK-BR-3 cells, Calu-3 cells, and MDA-MB-231 cells labeled with 2 × 10 −6 m of CellTracker Red CMTPX and seeded on a Lab-Tek II eight-chambered cover glass slide at a density of 1 × 10 4 cells/well 24 h prior to treatment. NK cells labeled with 1 × 10 −6 m CellTracker Blue CMAC were modified with 100 µg of DMPE-PEG-T-DM1-FITC. After the modification, 1 × 10 5 SE-NK/T-DM1 cells were coincubated with the cancer cells for 30 min and washed with PBS to remove the unbound effector cells. Coincubated cells were imaged by confocal microscopy and collected images were processed by ImageJ software.

Experimental Section
Internalization of T-DM1 was visualized using a similar procedure. Cancer cells labeled with NucBlue Live ReadyProbe Reagent (Ex/Em = 360/460 nm) were seeded on a Lab-Tek II eight-chambered cover glass at a density of 1 × 10 4 cells/well 24 h prior to treatment.  cells, unmodified NK cells, TZ, T-DM1, T-DM1+JK  cotreatment, T-DM1+NK cotreatment, SE-NK/T-DM1 cells, SE-NK/TZ  cells, or SE-JK/T-DM1 cells at an E:T ratio of 10:1 in 600 µL of complete media. Cancer cells receiving T-DM1 treatment received 2.1 µg of T-DM1 that corresponds to the T-DM1 amount on SE-NK/T-DM1 treated at an E:T ratio of 10:1. All treatments were washed 2 h after the coculture, and the remaining cancer-bound effector cells were further incubated for 24 h. All cells were harvested and labeled with the Annexin V Alexa Fluor 488 and propidium iodide kit (Annexin Ex/Em = 495/520 nm and propidium Ex/Em = 535/617) after 24 h of coincubation. Cancer cell death was analyzed by flow cytometry.
To distinguish the effects of antibodies, chemotherapeutic agents, and immune cells, cancer cells labeled with CMAC were coincubated with SE-NK/T-DM1 cells, SE-NK/TZ cells, SE-JK/T-DM1 cells, or other corresponding treatments at E:T ratio of 10:1 in 600 µL of complete media. Unbound effector cells were removed 2 h after the initial coincubation and the remaining cell mixtures were further incubated for 24 h. Cancer cells receiving T-DM1 treatment received 2.1 µg of T-DM1 that corresponds to the T-DM1 amount on surface-engineered immune cells with ADCs treated at an E:T ratio of 10:1. Resulting cancer cell death was identified with Annexin V Alexa Fluor 488 and propidium iodide kit (Annexin Ex/Em = 495/520 nm and propidium Ex/Em = 535/617) using flow cytometry and analyzed by FlowJo software.
In Vivo Tumor Efficacy and Biodistribution: In vivo studies were conducted with six-week-old female NOD scid gamma (NSG, NOD. Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice purchased from the Jackson Laboratory (Bar Harbor, ME). Each mouse was subcutaneously inoculated with 1 × 10 7 cells of Calu-3 cells or MDA-MB-231 cells on the left flank. Cancer cells were suspended in PBS supplemented with 10% (v/v) Matrigel (Fisher Scientific, Bedford, MA). Tumor volume was recorded three times per week by measuring the length and the width of the tumor with a caliper and calculating the tumor volume on the basis of the following formula: V = 0.5ab 2 , using the longest (a) and shortest (b) diameters of the tumor. When the tumor volume reached ≈100 mm 3 , tumor-inoculated mice were randomly assigned to the experimental groups. Control group (n = 4 for Calu-3 model, n = 3 for MDA-MB-231 model) received no treatment but the study groups (n = 4 per group) were weekly administrated with 0.21 mg of T-DM1, 1 × 10 7 NK cells, 0.21 mg of T-DM1+1 × 10 7 NK cotreatment, or 1 × 10 7 SE-NK/T-DM1 cells through tail vein infusion for two weeks (Day 0 and Day 7). All agents were freshly prepared in 250 µL of PBS and the infusion was completed in 1 min. Tumor growth and body weight were monitored for 14 d and relative tumor volume was calculated by dividing the recorded volume with the initial volume.
For biodistribution, NSG mice-bearing Calu-3 tumors (n = 3) received no treatment, 1 × 10 7 NK cells, 0.21 mg of T-DM1+1 × 10 7 NK cotreatment, or 1 × 10 7 SE-NK/T-DM1 cells through tail vein infusion. All agents were freshly prepared in 250 µL of PBS for 1 min infusion. Tumor and major organs, including heart, kidneys, liver, lungs, and spleen, were harvested 24 h after the treatment. Single-cell suspension of each harvested organ was prepared using the gentleMACS Dissociator and tissue dissociation kits (Miltenyi Biotec, Bergisch Gladbach, Germany) following the instructions provided by the manufacturer. Half of each cell mixture was incubated with 30 µg of an APC-conjugated anti-CD56 (Ex/Em = 650/660 nm) antibody for 1 h at 4 °C. Resulting cell mixtures were washed twice with cold PBS and the presence of NK cells were detected from counting 1 × 10 5 total cells by flow cytometry. Collected results were analyzed by FlowJo.
Statistical Analysis: Statistical analysis was performed in Graphpad Prism 6. All data are presented in mean ± SD. All data were analyzed with one-way ANOVA with Bonferroni post hoc tests or two-way ANOVA with Bonferroni post hoc tests. Statistical significance threshold of each test was set at P < 0.05: ns = not significant, P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
Animal Ethics: All animal experiments were approved by the University of Utah Institutional Animal Care and Use Committee. Described animal procedures were conducted according to guidelines and regulations.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.