Targeting HER2-Positive Solid Tumors with CAR NK Cells: CD44 Expression Is a Critical Modulator of HER2-Specific CAR NK Cell Efficacy

Abstract

Background/Objectives: Monoclonal antibody therapies for HER2-positive tumors frequently encounter resistance, requiring alternative treatment strategies. This study investigates the use of natural killer (NK) cells expressing HER2-specific chimeric antigen receptor (CAR) to address this issue. CAR NK cells have several benefits over CAR T cells: they are less likely to cause severe side effects such as cytokine release syndrome and neurotoxicity, can be sourced from various origins, and do not trigger Graft versus Host Disease, making them ideal for "off-the-shelf" applications. Methods: We have generated NK-92 cell lines expressing first, second and third-generation HER2-specific CARs with CD28 and/or 41BB costimulatory domains using a retroviral transduction system, followed by FACS sorting and expansion to obtain pure HER2-CAR NK-92 cell products for functional benchmarking. Results: In vitro tests showed that these CAR NK cells were effective against both trastuzumab-sensitive (CD44^-) and -resistant (CD44⁺) tumors in monolayer cultures. However, in three-dimensional spheroid models and in vivo xenografts, they were less effective against CD44+ trastuzumab-resistant tumors. Conclusions: This reduced efficacy highlights the significant role of the tumor microenvironment, particularly the extracellular matrix, in hindering the therapeutic potential of CAR NK cells. Despite the promising in vitro performance of CAR NK cells, this study emphasizes the need for improved strategies to enhance their penetration and effectiveness in resistant tumors: optimizing CAR constructs and devising methods to overcome extracellular matrix barriers are crucial for advancing CAR NK cell therapies in oncology.

Keywords: CAR NK cell; CD44; HER2-CAR; extracellular matrix; solid tumor; trastuzumab.

Figure 1

Generation of HER2-specific human CAR NK cells and their proliferation in vitro: (a) Schematic structure of HER2-targeting chimeric antigen receptor molecules (created with BioRender). (b) Representative flow cytometry dot-plots of non-transduced (NT) and HER2.z, HER2.CD28.z, HER2.41BB.z, and HER2.CD28.41BB.z CAR NK cells after FACS sorting. (c) Quantitative cell proliferation data. A total of 1 × 10⁵ HER2-CAR NK or NT NK cells were plated in duplicates in the presence of 400 IU/mL interleukin-2. Every 3.5 days, the effector cell number was determined by flow cytometry, and then the initial effector cell quantity was placed onto new plates in conditions identical to the beginning of the experiment. The expansion rate was the ratio of the cell number measured at the end and at the beginning of the last 3.5-day period. Histograms represent the mean ± SD (n = 3; assay in duplicates); *** p < 0.001.

Figure 2

HER2-specific CAR NK cells successfully recognized and killed HER2-positive target cells: (a) 1 × 10⁵ HER2-CAR NK or NT NK cells ± 10 μg/mL trastuzumab were cocultured with MDA-HER2, JIMT-1, or N87 (HER2⁺) or MDA (HER2⁻) target cells in 1:1 NK cell-to-tumor cell ratio. After 24 h, IFNγ was determined in the culture supernatant by ELISA (n = 3, assay performed in duplicates). (b) Firefly-luciferase-based cytotoxicity assay using HER2-CAR NK or NT NK cells ± 10 μg/mL trastuzumab against 3 × 10⁴ MDA-HER2, JIMT-1, or N87 (HER2⁺) or MDA (HER2⁻) target cells at 1–0.3–0.1–0.03–0.01–0.003:1 NK cell-to-tumor cell ratio (marked as E:T ratio in the figure). Cell culture medium was not supplemented with IL-2 (n = 3; assay was performed in duplicates). Histograms show mean ± SEM; * p < 0.05; *** p < 0.001. All variants of HER2-CAR NK-92 cells were compared to the non-treated (NT control) at the effector-to-target ratio of 1:1. Additionally, the NT + TRAST group was compared to the .z group across the effector-to-target ratio range of 0.003 to 1.

Figure 3

41BB-containing HER2-specific CAR NK cells selectively destroy three-dimensional CD44⁻ N87 tumor spheroids and spare CD44⁺ JIMT-1 spheroids: (a) Schematic representation of the optical sections through the spheroid. (b) Representative images (at 24 h) for detection of cytolytic activity of HER2-CAR NK cells and trastuzumab-targeted NT NK cells against HER2⁺/CD44⁺, trastuzumab-resistant JIMT-1 spheroids (left column), and HER2⁺/CD44⁻, trastuzumab-sensitive N87 spheroids (right column). Target cells are green (GFP expressing); dead cells are visualized by PI uptake (red). The actual scalebar is visible on the image and its dimensions are either 200 µm or 300 µm, depending on the specific image shown.

Figure 4

HER2-CAR NK cells are unable to eliminate CD44⁺, trastuzumab-resistant JIMT-1 xenografts in vivo. Mice were injected s.c. with 3 × 10⁶ JIMT-1.ffLuc cells. Mice on day 14 (arrow) received a single i.v. dose of 5 × 10⁶ NT NK cells (NT + TRAST; n = 10, solid black line and NT, n = 10, dashed black line) or HER2-CAR T cells (CD28.z, n = 10, green; 41BB.z, n = 10; blue and CD28.41BB.z, n = 10, pink). Tumor growth was followed by bioluminescence imaging. (a) Representative images of JIMT-1.ffLuc-injected animals. (b) Quantitative bioluminescence imaging data of JIMT-1.ffLuc xenografts (average total radiance = photons/s/cm²/sr). (c) Kaplan–Meier survival curve. Histograms represent mean ± SD.

Figure 5

HER2-CAR NK cells exerted significant in vivo antitumor effect against CD44⁻. trastuzumab-sensitive N87 xenografts. Mice were injected s.c. with 3 × 10⁶ N87-1.ffLuc cells. Mice on day 14 (arrow) received a single i.v. dose of 5 × 10⁶ NT NK cells (NT + TRAST; n = 10, solid black line and NT, n = 10, dashed black line) or HER2-CAR T cells (CD28.z, n = 10, green; 41BB.z, n = 10, blue; and CD28.41BB.z, n = 10, pink). Tumor growth was followed by bioluminescence imaging. (a) Representative images of N87.ffLuc-injected animals. (b) Quantitative bioluminescence imaging data of JIMT-1.ffLuc xenografts (average total radiance = photons/s/cm²/sr) (NT vs. all other treatments: *** p < 0.001; NT + TRAST vs. HER2-CAR treatments: n.s.). (c) Kaplan–Meier survival curve (NT vs. all other treatments: *** p < 0.001; NT + TRAST vs. HER2-CAR treatments: n.s.). Histograms represent mean ± SD.