An Antibody Designed to Improve Adoptive NK-Cell Therapy Inhibits Pancreatic Cancer Progression in a Murine Model

Pancreatic cancer is the fourth leading cause of cancer-related death worldwide and has an extremely low survival rate (1). Up to 80% of deaths occur within the first year of diagnosis; the overall 5-year mortality rate is more than 95% (2, 3). Surgical resection is the only potentially curative approach; however, only 10% to 20% of pancreatic tumors are operable (about 40% are locally advanced and unresectable and 45% are associated with metastases; ref. 4). Gemcitabine has been used as the first-line treatment for patients with nonresectable advanced pancreatic cancer. However, gemcitabine must be used with a variety of other drugs in pancreatic cancer patients because it is difficult to overcome pancreatic cancer with gemcitabine alone (5). Although there is no reported benefit, numerous attempts have been made to develop a polychemotherapy strategy that includes gemcitabine. A randomized phase III trial of gemcitabine-based combination treatment in patients with localized pancreatic cancer (565 cases) reported median survival of 6.5 months and progression-free interval of 3.6 months compared with gemcitabine alone (6). Despite treatment with gemcitabine-based therapy, the overall survival (OS) is approximately 6 months, and the 1-year survival rate is lower than 20% (7). Therefore, new treatment strategies for pancreatic cancer are needed.

Immunotherapy is an alternative to chemotherapy that better targets tumors without compromising normal tissues (8). Among available immunotherapeutic strategies, natural killer (NK) cells are considered promising. Indeed, NK-cell–based therapeutic strategies for cancer have been developed. Such strategies include receptor-mediated activation and ex vivo expansion of NK cells, chimeric antigen receptor engineering (CAR-NK), adoptive immunotherapy using donor-transformed NK cells, and increasing antibody-dependent cellular cytotoxicity (ADCC). However, despite the benefits of NK-cell therapy for solid tumors, cancer cells can restrict NK-cell activity through a variety of mechanisms that result ultimately in resistance. The key limitation of NK therapy is that tumor-secreted factors (such as PGE₂) disrupt tumor infiltration by NK cells, thereby limiting the efficacy of antitumor immune responses. NK-cell infiltration of solid tumors improves antitumor activity and prognoses for colorectal cancer, non–small cell lung cancer, and clear cell renal cell carcinoma (9–11). Therefore, a strategy aimed at increasing tumor infiltration by therapeutic NK cells is the key to developing successful immune cell–based therapeutic regimens.

Here, we developed an antibody-based NK-cell–homing protein, named NK-cell–recruiting protein-conjugated antibody (NRP-body), which increased the efficacy of NK-cell therapy in preclinical models of pancreatic ductal adenocarcinoma (PDAC). CXCL16, the NK-cell–recruiting protein component of the NRP-body, increased tumor infiltration by expanded NK cells (referred to as exNK cells). Increased infiltration of NK cells reduced the tumor burden in PDAC models and increased OS. Collectively, the results show that this antibody increased NK-cell trafficking to the tumor site, thereby providing opportunities for targeting both primary and metastatic pancreatic cancers.

The human mesothelin (MSLN)-expressing pancreatic cancer cell line Panc-1, CFPAC-1, and the human MSLN-low expression pancreatic cancer cell line Capan-1 were obtained directly from the ATCC and cultured according to the manufacturer's guidelines. Panc-1, CFPAC-1, and Capan-1 were purchased in 2016. The human MSLN-low expression primary AMCPAC02 cell line was provided by Professor Seung-Mo Hong at Asan Medical Center in 2017 (Seoul), Korea, and was prepared as previously described (12). This cell line was periodically authenticated by monitoring cell morphology, growth curve analysis, and inspection of Mycoplasma contamination using a Mycoplasma detection kit (Lonza). Cells were cultured (37°C in a humidified 5% CO₂ incubator) in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS). Ex vivo–expanded NK cells were generated as described previously, with a modification (13). Briefly, freshly isolated PBMCs (3 × 10⁶ cells) were cocultured in complete RPMI 1640 medium (RPMI 1640 supplemented with 10% FBS, 4 mmol/L glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin) in a 24-well tissue culture plate containing 100 Gy γ-irradiated K562 cells (0.5 × 10⁶) and recombinant human IL2 (10 U/mL). K562 cells were provided by Professor Duck Cho at Samsung Medical Center in 2015. The medium was replaced every 2 to 3 days with fresh medium containing 10 IU/mL human IL2. After 7 days of culture, the concentration of IL2 was increased to 100 IU/mL, and IL15 (10 IU/mL) was added to the medium. The medium was replaced every 2 to 3 days.

A human MSLN-specific single-chain variable region (Meso-scFv) antibody was synthesized by GeneScript (Piscataway), as reported by Chowdhury and colleagues (14), and cloned into the Sfi1 sites of the expression vector pLFG106. This vector comprises a signal sequence, the Fc region of human IgG, and DHFR genes on a pcDNA3.1 backbone (Invitrogen). The Meso-scFv was constructed by PCR amplification using primers For-tttgaattcgatgcaggtacaactgcagca and Rev-aaactcgagcctttttaccctcagtttttcaaagttgattatactccttttcaccctttttatttccaactttgtccc. We cloned soluble CXCL16 with the signal sequence excepted behind the furin cleavage site by PCR using primers For-cacactggcggccgcacgggtgaagcggaacgagggcag and Rev-aatctcgagcggccgcctaaggaagtaaatgcttctggtg (CXCL16: NEGSVTGSCY CGKRISSDSP PSVQFMNRLR KHLRAYHRCL YYTRFQLLSW SVCGGNKDPW VQELMSCLDL KECGHAYSGI VAHQKHLLP; AF337812) (Supplementary Fig. S9). The NRP-body expression plasmid was stably transfected into DHFR- and furin-deficient CHO-mutant cells using lipofectamine 2000 (Invitrogen) transfection reagents, according to the manufacturer's instructions. Stably transfected cells were selected by culture in the presence of G418 (geneticin) for 2 weeks. Stable transfectants were further adapted by exposure to 100 nmol/L methotrexate (Sigma-Aldrich) for 2 weeks. The NRP-body (secreted into the culture medium) was identified by Western blot analysis and enzyme-linked immunosorbent assay (ELISA) using an anti-human Fc (Thermo Fisher Scientific).

NRP-body–mediated ADCC was assessed by flow cytometry (15). Panc-1 cells were the targets and exNK cells were the effectors. Panc-1 cells were labeled for 5 minutes with carboxyfluorescein succinimidyl ester (CFSE) (Dojindo), followed by incubation overnight in a 96-well culture plate at 37°C. ADCC was initiated by addition of NRP-body (1 μg/mL) and exNK cells [effector:target (E:T) ratio, 1:1]. The plate was again incubated overnight at 37°C (5% CO₂, humidified atmosphere). Pacn-1 was harvested and stained with Fixable Viability Dye (FVD; eBioscience), which stains dead cells. After 20 minutes, cells were washed with 1 × phosphate-buffered saline (FBS) containing 2% FBS and subjected to flow cytometry analysis using a BD FACSCalibur system (BD). The mean percentage ± standard deviation (SD) of cells under each condition was calculated from three replicate wells.

Chemotaxis assays were performed in Transwell plates (24-well, 5-μm filters; Costar). The wells of the plates were seeded with a Panc-1 cell monolayer. CFSE-stained NK cells (2 × 10⁵ cells/well) were added to the bottom layer in the presence of PBS and a noncleavable NRP-body, or the NRP-body was added to the lower chamber (500 ng/mL). Plates were centrifuged (for 5 seconds at 200 × g) to spin down the cells onto the filter, and cell migration was allowed to proceed for 4 hours. Migrated cells were harvested from the bottom chamber by washing with 5 mmol/L ethylenediaminetetraacetic acid-containing medium. Absolute NK-cell counts were determined by flow cytometry (FACSCalibur) gated on CFSE-positive NK cells and CountBright absolute counting beads (Invitrogen), according to the manufacturer's recommendations. The results were calculated as follows:

(number of NK-cell events/number of bead events) × number of beads added.

Invasion assays were performed in Transwell plates (24-well, 5-μm filters; Costar) seeded with a Panc-1 monolayer. After 2 hours, vehicle, a noncleavable NRP-body, or NRP-body was added to the bottom chamber. In brief, the Matrigel coating on the bottom layer was rehydrated in 0.5 mL of DMEM for 30 minutes immediately before the experiments. exNK cells (5 × 10⁴) were suspended in 0.5 mL of serum-free medium and then added to the upper chamber. The plates were incubated for 48 hours, and cells on the upper side were removed using cotton swabs. Cells that had migrated into the gel were then fixed and stained with 2% ethanol containing 0.2% crystal violet. Invading cells were counted under a light microscope using a 10× objective lens (PAX cam2+ USB2 digital Camera).

Whole-cell lysates of exNK cells were prepared in radioimmunoprecipitation assay buffer, and protein concentrations were determined using a bicinchoninic acid protein assay kit (Pierce). Samples containing equal amounts of protein were resolved in sodium dodecyl sulfate-polyacrylamide gels and transferred to Hybond-enhanced chemiluminescence (ECL) nitrocellulose membranes (Bio-Rad). The membranes were probed with appropriate primary antibodies diluted in Tris-buffered saline/Tween-20 containing 5% bovine serum albumin, followed by horseradish peroxidase-conjugated secondary antibodies. Bound antibodies were visualized using ECL reagents (Abfrontier).

Panc-1 cells were plated (5 × 10⁴ cells/well) into 24-well tissue culture plates. The NRP-body (500 ng/mL) was added to the wells, and the supernatants were collected 4 hours later. The concentration of CXCL16 was measured in a sandwich ELISA (Human CXCL16 Quantikine ELISA kit; R&D Systems).

Microtiter wells were coated overnight at 4°C with rhMSLN in 50 mmol/L sodium carbonate buffer and then blocked with 2% BSA in PBS. The plates were washed four times in PBST between steps. All incubations were carried out for 2 hours at 37°C. Horseradish peroxidase–conjugated goat anti-human IgG (Pierce) was used to detect bound NRP-body. Color was developed with the TMB substrate reagent set (BD Biosciences), and absorbance was measured at 450 nm using a microtiter plate reader (Emax; Molecular Devices).

NSG mice (The Jackson Laboratory) were used for the implantable tumor experiments. To generate orthotopic xenografts, the pancreas of 5-week-old NSG female mice was exposed surgically and injected with Panc-1 cells (suspended in serum-free culture medium (1 × 10⁶ cells/100 μL). To generate metastatic xenografts, Panc-1 cells (1 × 10⁶ cells/100 μL of serum-free culture medium) were injected into the tail vein. From 2 weeks after implantation, mice were divided into three groups and treated with PBS, a noncleavable NRP-body (1 mg/kg), or the NRP-body (1 mg/kg) via intraperitoneal injection. For the adoptive transfer experiments, NK cells were stained for 15 minutes with 50 μmol/L DiR membrane dye (Invitrogen) and then injected (1 × 10⁷ cells) via the tail vein. Five days after adoptive transfer, tumor-infiltrating exNK cells were assessed by fluorescence imaging of resected lymph nodes using a Xenogen Spectrum in vivo imaging system (IVIS; Caliper Life Sciences) equipped with a 710/760 nm excitation–emission filter set. To block CXCL16, tumor-bearing mice received an intraperitoneal injection of monoclonal rat anti-human CXCL16-neutralizing antibody (100 μg; R&D Systems). Control mice received 100 μg of control IgG (Sigma-Aldrich). All animal studies were performed in compliance with the policy of the KRIBB Animal Care and Use Committee.

All data were obtained from at least three independent experiments performed in triplicate and were analyzed using the Student t test. When calculating two-tailed significance levels for equality of means, equal variance between two groups was assumed. P values < 0.05 were considered significant.

As the efficacy of NK immunotherapy relies on the ability of effector cells to infiltrate the tumor microenvironment, we tracked the distribution of exNK cells in pancreatic cancer model mice following adoptive transfer. Two weeks after injecting luciferase-expressing Panc-1 cells (1 × 10⁶/mouse) into NSG mice, exNK cells (1 × 10⁷/mouse) were injected intravenously and their distribution (5 days later) was monitored using the IVIS system. To monitor in vivo trafficking, we stained exNK cells with DiR dye before adoptive transfer. As expected, most DiR⁺ exNK cells were detected in the liver and spleen, but not in pancreatic cancer tissue (Fig. 1A–C). To clarify this further, we analyzed the absolute number and percentage of NK cells infiltrating pancreas, lung, liver, and spleen by flow cytometry. Consistent with the IVIS data, both the number and percentage of NK cells in the liver and spleen were significantly higher than those in pancreatic tumor tissues (Fig. 1D and E). These results indicate that restricted NK-cell infiltration is a factor limiting immunotherapy of pancreatic tumors. On the basis of these findings, we developed an antibody to enhance the infiltration of NK cells into tumor sites.

To increase infiltration of NK cells to the tumor site, we designed an NRP-body containing a tumor-targeting module fused to a functional cargo domain and a noncleavable NRP-body without a cargo domain (Fig. 2A). The tumor-targeting module of the NRP-body comprised a Meso-scFv because MSLN is overexpressed by pancreatic cancer cells. The cargo domain of the therapeutic antibody comprised the Fc portion of the IgG1 protein and a chemotactic factor flanked by furin cleavage sites. Meso-scFv preferentially shuttles functional cargo to MSLN-overexpressing pancreatic cancer cells. Once bound to the cells, cleavage by furin releases the chemotactic factor, which then recruits NK cells to the tumor microenvironment. As furin selectively cleaves the amino acid sequence RVKR within the cargo domain, the chemotactic factor accumulates in the tumor microenvironment and attracts more NK cells. To identify the chemotactic factor optimal for the cargo domain, we focused on CXCR3 and CXCR6 receptors that are expressed by exNK cells but not by naïve NK cells (16). We aimed to select a ligand from among CXCL9, CXCL10, CXCL11 (all of which are ligands of CXCR3), and CXCL16 (a ligand of CXCR6) that would most effectively induce NK-cell infiltration. To this end, we evaluated NK responses to chemokines in a chemotaxis assay. We found that CXCL16 induced a 4-fold increase in exNK-cell migration, indicating that this cytokine is a candidate chemotactic factor for the cargo domain (Fig. 2B). Next, we constructed the Meso-scFv-Fc-CXCL16 (NRP-body) vector, which was then transfected to furin-deficient FD11 cells to produce the NRP-body. The purified NRP-body was of the predicted molecular weight (70 kDa; Supplementary Fig. S1A); furin-mediated cleavage of CXCL16 was confirmed by Western blotting with an anti-human Fc (Supplementary Fig. S1B). Endotoxin contamination of purified NRP-body was lower than 0.01 EU/mg. The toxicity of NRP-body was evaluated by measuring mitochondrial ATP activity using the CellTiter-Glo luminescent cell viability assay. Viability assays showed that NRP-body was nontoxic to exNK cells (Supplementary Fig. S2). To determine whether NRP-body can selectively bind to human MSLN-expressing pancreatic cancer cells, we stained several pancreatic cancer cell lines and analyzed NRP-body binding by flow cytometry. In this study, we used PANC-1 cells, which express MSLN, as a positive control, and HPDE cells, which do not express MSLN, as a negative control. Flow cytometry analysis showed that the NRP-body bound to Panc-1, CFPAC-1, and MiaPaCa-2 cells, but not to AMCPAC02, Capan-1, BxPC-3, or HPDE cells (Fig. 2C). Additionally, we analyzed NRP-body binding to other cancer cell lines (Supplementary Fig. S3). Collectively, the data show that NRP-body bound MSLN-expressing cells selectively. We then investigated whether binding of NRP-body to Panc-1 and CFPAC-1 cells facilitates cleavage of furin recognition sites within the chimeric protein to release CXCL16 from the NRP-body. Panc-1 cells were exposed to the NRP-body for 4 hours, and CXCL16 concentrations in the culture supernatant measured using a CXCL16-specific ELISA kit. CXCL16 protein was detected only in culture medium from cells incubated with the NRP-body; no CXCL16 was detected in the supernatant from cells treated with a furin cleavage site-deficient NRP-body (Fig. 2D). To study release of CXCL16 from the NRP-body in vivo, we tracked the CXCL16 present in tumor tissues in pancreatic cancer models. Mice received an intraperitoneal injection of noncleavable NRP-body (1 mg/kg) or NRP-body (1 mg/kg) at 2 weeks after inoculation of Panc-1 cells. One day later, we sacrificed the mice and analyzed CXCL16 in tumor lysates by ELISA. The results revealed that CXCL16 cleaved from the NRP-body was present in tumor tissues (Fig. 2E). Taken together, the data suggest that furin-mediated proteolysis of NRP-body releases CXCL16 from IgG1.

To determine whether NRP-body could enhance the chemotactic motility of exNK cells via the CXCL16 gradient, we conducted a migration assay in vitro. The chemotactic responses of exNK cells to NRP-body, MSLN scFv-Fc, and noncleavable antibody (0.5 μg/mL) were tested in a Transwell chemotaxis assay. Migrating cells were harvested from the lower chamber and counted by flow cytometry. Panc-1 cells exposed to the NRP-body attracted much higher numbers of NK cells to the lower chamber of the Transwell than cells exposed to a MSLN scFv-Fc and noncleavable NRP-body (Fig. 3A). Furthermore, we used a Matrigel invasion assay to examine the effects of the NRP-body on the invasive properties of exNK cells. Invasive NK cells on the lower surface of the filter were stained and counted. The NRP-body increased the number of infiltrated exNK cells to a greater extent than MSLN scFv-Fc and the noncleavable NRP-body (Fig. 3B). These results indicate that CXCL16 increases exNK-cell trafficking to tumor cells in vitro. Next, we examined whether the NRP-body increased infiltration of exNK cells into tumor sites in the PDAC model. Four weeks after orthotopic injection of luciferase-expressing Panc-1 cells into NSG mice, mice received an intraperitoneal injection of the NRP-body (1 mg/kg). One day later, we intravenously injected DiR⁺ exNK cells (1 × 10⁷/mouse) and monitored the distribution of DiR⁺ exNK cells using the IVIS system. Five days after injection of DiR⁺ exNK cells, we found that infiltration of NK cells into pancreatic tumor tissues in NRP-body–treated mice was higher than that of exNK cells alone (Fig. 3C, left). We also examined whether the NRP-body induced infiltration of exNK cells into metastatic tumor sites. NSG mice received an intravenous injection of Panc-1 cells into the tail vein. Two weeks later, tumor-bearing mice received an intraperitoneal injection of NRP-body (1 mg/kg). We also injected control groups with the noncleavable NRP-body. One day later, we injected mice intravenously with DiR⁺ exNK cells (1 × 10⁷/mouse). Again, 5 days after injection of DiR⁺ exNK cells, DiR⁺ exNK cells exhibited more infiltration into metastatic tumor sites when exposed to the NRP-body (Fig. 3C, middle). Furthermore, we examined the absolute number of DiR⁺ exNK cells in the tumors at the time of sacrifice by flow cytometry. The number of DiR⁺ exNK cells in the NRP-body–treated group was significantly higher than that in the group treated with noncleavable NRP-body (Fig. 3C, right). These results indicate that NRP-body enhanced migration and infiltration of exNK cells into tumor sites. To clarify whether the effect of NRP-body depends on MSLN expression by pancreatic cancer cells, we injected MSLN-expressing CFPAC-1 cells (1 × 10⁶/mouse) together with nonexpressing AMCPAC02 cells (1 × 10⁶/mouse) into mice. Two weeks later, the mice received NRP-body (1 mg/kg) via intraperitoneal injection. One day later, we adoptively transferred DiR⁺ exNK cells (1 × 10⁷/mouse) into the mice and monitored them using the IVIS system. As expected, we detected only DiR⁺ exNK cells in MSLN-expressing tumor sites (Fig. 3D, left). We estimated the absolute number of DiR⁺ exNK cells in the tumors at the time of sacrifice using flow cytometry. The number of DiR⁺ exNK cells in MSLN-expressing CFPAC-1 tumors was significantly higher than that in nonexpressing AMCPAC02 tumors (Fig. 3D, right). In addition, DiR fluorescence data revealed that infiltrated exNK cells proliferated at the tumor site, which might induce further immune responses to tumor cells (Supplementary Fig. S4A and S4B). Taken together, the data suggest that the NRP-body binds to MSLN-expressing pancreatic cancer cells and, consequently, furin expressed by MSLN-expressing cells releases CXCL16 from the NRP-body to increase exNK-cell infiltration into the tumor.

Next, we asked whether the NRP-body–mediated increase in NK-cell infiltration affected the immune response to pancreatic cancer. NSG mice were injected orthotopically with luciferase-expressing Panc-1 cells (1 × 10⁶/mouse), and tumor growth was monitored for 2 weeks. Two weeks after the injection of tumor cells, the NSG mice received an intraperitoneal injection of the cleavable NRP-body (1 mg/kg) or the NRP-body (1 mg/kg). One day later, we adoptively transferred DiR⁺ exNK cells (1 × 10⁷/mouse) into the mice and monitored them for 2 weeks using the IVIS system. Mice injected with a noncleavable NRP-body were used as controls, and tumor growth was monitored with the IVIS system. We found that the tumor burden in mice receiving the NRP-body was lower than that in control mice, even after a single injection of DiR⁺ exNK cells (Fig. 4A). The OS of NRP-body–administered mice was significantly longer than that of control mice (Fig. 4B). Next, we analyzed the organ (lung, liver, pancreas, and spleen) distribution of injected DiR⁺ exNK cells using the IVIS system. DiR⁺ exNK cells were more abundant in tumor tissues of NRP-body–treated mice than in those of control mice (Fig. 4C). In addition, we estimated the percentage and absolute number of DiR⁺ exNK cells in the organs at the time of sacrifice by flow cytometry. The number of DiR⁺ exNK cells in the NRP-body–treated groups was significantly higher than that in the control groups (Fig. 4D and E). To clarify this, we treated tumor-bearing mice with a neutralizing antibody to block CXCL16 activity. The results revealed that blocking CXCL16 reduced NRP-body–induced NK-cell infiltration (Supplementary Fig. S5). We also performed immunohistochemistry (IHC) to confirm the effect of NRP-body on the infiltration of exNK cells into tumor tissue. The IHC results indicated that NRP-body increased the infiltration of exNK cells into tumor tissues (Supplementary Fig. S6). Taken together, these findings show that treatment with NK cells along with the NRP-body inhibited tumor progression, likely by increasing recruitment of NK cells to the tumor site. Next, we confirmed that the NRP-body increased immune responses in a metastatic model. NSG mice were injected intravenously with luciferase-expressing Panc-1 cells (1 × 10⁶/mouse), and tumor growth was monitored for 2 weeks using the IVIS system. Two weeks after injection of tumor cells, NSG mice received an intraperitoneal injection of the noncleavable NRP-body (1 mg/kg) or the NRP-body (1 mg/kg). One day later, we adoptively transferred DiR⁺ exNK cells (1 × 10⁷/mouse) into the mice (once a week for 2 weeks) and monitored them for 2 weeks using the IVIS system. The metastatic tumor burden in NRP-body–treated mice was lower than that in mice treated with noncleavable NRP-body (Fig. 5A). In addition, the OS of the NRP-body–treated group (Fig. 5B) was longer than that of the control group. Next, we analyzed the organ (lung, liver, pancreas, and spleen) distribution of injected DiR⁺ exNK cells using the IVIS system. NK cells were more abundant in the lungs (lungs are a common site of pancreatic cancer metastasis) than in the spleen (Fig. 5C). We also measured the percentage and absolute number of DiR⁺ exNK cells in the organs at the time of sacrifice by flow cytometry. The number of DiR⁺ exNK cells in the NRP-body–treated groups was significantly higher than that in the control groups (Fig. 5D and E). In addition, when the NRP-body was injected intravenously into tumor-bearing mice, the degree of tumor infiltration by exNK cells was similar to that observed after intraperitoneal injection (Supplementary Fig. S7A and S7B). These results indicate that combined therapy with NK cells plus the NRP-body is effective against metastatic pancreatic cancer.

In T cells, the CXCL16/CXCR6 chemokine axis correlates with liver-specific homing via the ERK-RhoA signaling pathway (17). Therefore, to determine whether CXCL16 regulates NK-cell homing via the ERK-RhoA signaling cascade as in T cells, we examined activation of the ERK-RhoA signaling pathway in rCXCL16-treated NK cells by flow cytometry. The data revealed that rCXCL16 induced phosphorylation of ERK more in NK cells than in vehicle-treated cells (Fig. 6A). In addition, we assessed GTP-RhoA in rCXCL16-treated NK cells by Western blotting. We found that GTP-RhoA expression was about 1.7-fold higher in NK cells exposed to rCXCL16 than in cells exposed to vehicle (Fig. 6B). To determine whether expression of GTP-RhoA depends on rCXCL16-induced activation of ERK, we performed Western blotting to analyze GTP-RhoA expression after treatment of NK cells with the ERK inhibitor PD98059. The GTP-RhoA level in PD98059-treated NK cells fell after ERK inactivation (Fig. 6C and D). To confirm whether rCXCL16-mediated activation of ERK-RhoA increases NK-cell motility, we conducted migration and invasion assays using PD98059-treated NK cells incubated with the NRP-body. As expected, ERK inactivation reduced NK-cell migration and invasion (Fig. 6E and F). Taken together, these results suggest that CXCL16, part of the NRP-body, increases NK-cell migration via ERK-dependent activation of RhoA.

Here, we demonstrate that combined therapy with NK cells plus an NRP-body represents an alternative and efficacious form of cell therapy against highly aggressive pancreatic cancer. The combination therapy described herein increased the number of effector cells targeting the tumor and increased OS in PDAC models. This is promising because limited NK-cell infiltration of tumors is considered a barrier to the efficacy of immunotherapy. Furthermore, we showed that combination therapy reduced tumor progression and increased OS in metastatic models. Together, these preclinical data suggest that this combined strategy may improve NK-cell–mediated immune responses against pancreatic cancer.

Patients with pancreatic cancer often present with severe immune dysfunction, which is characterized by proliferation of immunosuppressive cells and increases in proinflammatory cytokines (18). These factors, which are assumed to be present in the tumor microenvironment, contribute to the limited efficacy of immune-based therapy for pancreatic cancer. Increasing pancreatic cancer infiltration by NK cells should improve immunotherapy. Infiltration of carcinomas such as glioblastoma (19), solid lung metastases, and gastric, colorectal, as well as head and neck cancers by NK cells is associated with a good prognosis (20, 21). However, in patients with pancreatic cancer, the number of circulating NK cells falls. Reduced NK-cell infiltration in pancreatic cancer is attributable to expression of PGE₂; however, the factors that cause decreased NK-cell infiltration in patients with pancreatic cancer remain poorly understood. To increase infiltration by NK cells, it is necessary to understand the mechanism that blocks NK-cell infiltration in pancreatic cancer.

Production of large numbers of cytotoxic NK cells and effective infiltration by expanded NK cells are required for successful adoptive immunotherapy. In particular, chemokines and chemokine receptors play a role in increasing infiltration of lymphocytes into the tumor. NK cells express chemokine receptors CCR1, CCR4, CCR5, CCR6, CCR7, CCR9, CXCR3, CXCR4, CXCR5, and CXCR, which are involved in increasing migration of NK cells (22). However, there is insufficient evidence to show that chemokine receptors induce NK-cell infiltration into tumors, and little is known about adoptive immunotherapy using expanded NK cells. Previous studies show that expanded CXCR6-positive NK cells migrate into an irradiated breast cancer cell line, which was enhanced secretion of CXCL16 in vitro (16). Here, we focused on the ability of CXCL16 and its receptor, CXCR6, on expanded NK cells to increase infiltration of expanded NK cells into tumors. Increased expression of CXCR6 enhances chemotaxis of activated CD8⁺ T cells (23). Irradiated 4T1 tumor cells induce CXCL16-dependent chemotaxis of activated CD8⁺ T cells in vitro, and local irradiation of tumors in vivo increases recruitment of tumor-specific CXCR6-positive CD8⁺ T cells (24). This means that CXCL16 inhibits tumor growth by increasing infiltration of solid tumors by expanded CXCR6-expressing NK cells and CD8⁺ T cells. Although the effect of the NRP-body on activated CD8⁺ T cells has not been confirmed, CD8⁺ T-cell–based therapy may be possible through the effects of the NRP-body on expanded NK cells. In addition, it may be possible to induce tumor infiltration by CXCR6-expressing immune cells, especially activated CD8⁺ T cells, in immunocompetent individuals using NRP-body monotherapy as well as combination therapy with NK cells or T cells.

We found that CXCL16 changes expanded NK cells from CD56^dim to CD56^bright. Subsequent studies should investigate how CXCL16 alters the properties of expanded NK cells. CD56^bright regulates cytokine production, but our results show that the killing activity of expanded NK cells was increased by CXCL16 treatment. This suggests that CXCL16 is not only needed for killing activity but also for cytokine regulation by CD56^bright. Indeed, if CXCL16 increases migration and killing by changing the NK-cell subset to CD56^bright, it may be possible to use it as a monotherapy agent to induce migration and killing activity of existing hepatic NK cells (CXCR6-expressing NK cell) in tumor tissue. We believe that further characterization of NK subsets in the liver may be necessary to observe the changes that occur after treatment with CXCL16. Subsequent studies should investigate how CXCL16 alters the properties of expanded NK cells.

The present study demonstrates that treatment with the NRP-body represents a strategy for antitumor therapy by increasing infiltration of pancreatic cancer by expanded NK cells. We focused on whether enough expanded NK cells infiltrated the tumor site upon exposure to chemokines. We screened chemokine receptors expressed by expanded NK cells and selected CXCL16; we then used this to construct the NRP-body. We confirmed that treatment with the NRP-body increased NK-cell infiltration into the tumor site and increased the antitumor ability of expanded NK cells. On the basis of these results, and the effects of NRP-body in mouse models, we propose that combined treatment with the NRP-body and NK cells could be an effective therapy for pancreatic cancer.

No potential conflicts of interest were disclosed.

Conception and design: J. Lee, T.H. Kang, S.-R. Lee, S.-U. Kim, E.-S. Kwon, S. Kim

Development of methodology: J. Lee, J.-S. Kim, D. Cho, S. Kim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Lee, T.H. Kang, K. Kong, S. Kim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Lee, T.H. Kang, H. Choi, S. Kim

Writing, review, and/or revision of the manuscript: J. Lee, T.H. Kang, W. Yoo, K. Kong, D. Cho, J. Kim, E.-S. Kwon, S. Kim

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Lee, S. Jo, K. Kong, S. Kim

Other (in vivo experiment assist): S. Jo

This research was supported by a grant from the Ministry of Science, ICT and Future Planning, South Korea (KGM4941713), and by a grant 14172MFDS974 from Ministry of Food and Drug Safety in 2016..

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.