CAR-neutrophil mediated delivery of tumor-microenvironment responsive nanodrugs for glioblastoma chemo-immunotherapy

Abstract

Glioblastoma (GBM) is one of the most aggressive and lethal solid tumors in human. While efficacious therapeutics, such as emerging chimeric antigen receptor (CAR)-T cells and chemotherapeutics, have been developed to treat various cancers, their effectiveness in GBM treatment has been hindered largely by the blood-brain barrier and blood-brain-tumor barriers. Human neutrophils effectively cross physiological barriers and display effector immunity against pathogens but the short lifespan and resistance to genome editing of primary neutrophils have limited their broad application in immunotherapy. Here we genetically engineer human pluripotent stem cells with CRISPR/Cas9-mediated gene knock-in to express various anti-GBM CAR constructs with T-specific CD3ζ or neutrophil-specific γ-signaling domains. CAR-neutrophils with the best anti-tumor activity are produced to specifically and noninvasively deliver and release tumor microenvironment-responsive nanodrugs to target GBM without the need to induce additional inflammation at the tumor sites. This combinatory chemo-immunotherapy exhibits superior and specific anti-GBM activities, reduces off-target drug delivery and prolongs lifespan in female tumor-bearing mice. Together, this biomimetic CAR-neutrophil drug delivery system is a safe, potent and versatile platform for treating GBM and possibly other devastating diseases.

Fig. 1. Schematic of enhanced anti-glioblastoma efficacy using combinatory immunotherapy of CAR-neutrophils and tumor microenvironment responsive nanodrugs.

a Human pluripotent stem cells were engineered with CARs and differentiated into CAR-neutrophils that are loaded with rough silica nanoparticles (SiO₂ NPs) containing hypoxia-targeting tirapazamine (TPZ) or other drugs, as a dual immunochemotherapy. b Systemically administered CAR-neutrophil@R-SiO₂-TPZ NPs first attack external normoxic tumor cells by forming immunological synapses and kill tumor cells via phagocytosis. After apoptosis, CAR-neutrophils could then release R-SiO₂-TPZ NPs, which are uptaken by tumor cells. Afterwards, nano-prodrugs respond to the hypoxic tumor microenvironment and effectively kill tumor cells. TEOS tetraethyl orthosilicate, BTES bis[3-(triethoxysilyl) propyl] tetrasulfide, TPZ tirapazamine, BTZ benzotriazinyl.

Fig. 2. Screening neutrophil-specific chimeric antigen receptor (CAR) structures with enhanced neutrophil-mediated anti-tumor activities.

a Schematic of various CAR structures. b Schematic of CAR #1 construct and targeted knock-in strategy at the AAVS1 safe harbor locus of human pluripotent stem cells (hPSCs). Vertical arrow indicates the AAVS1 targeting sgRNA. Red and blue horizontal arrows indicate primers for assaying targeting efficiency and homozygosity, respectively. HDR: homologous recombination repair. c Schematic of optimized neutrophil differentiation from hPSCs under chemically-defined conditions. d Cytotoxicity assays against U87MG glioblastoma cells were performed at different ratios of neutrophil-to-tumor target using indicated neutrophils. Data are represented as mean ± SD of five independent biological replicates, two-tailed Student’s t test. Reactive oxygen species (ROS) generation (e) and ELISA analysis of TNFα release (f) from different neutrophils after coculturing with U87MG cells were determined. n = 5 biologically independent samples. The data are represented as mean ± SD, two-tailed Student’s t test. Source data are provided as a Source Data file.

Fig. 3. Preparation and characterization of hPSC CAR-neutrophils loaded with tirapazamine (TPZ)-containing SiO₂ nanoparticles.

a–e Transmission electron microscope (TEM) (a) and energy dispersive spectroscopy (EDS) elemental mapping images (b) of rough SiO₂ nanoparticles are shown. c Nitrogen adsorption-desorption isotherm of rough SiO₂ nanoparticles along with Barrett-Joyner-Halenda (BJH) pore size distribution plot is shown. Biological triplicates were performed independently. TPZ loading content in SiO₂ nanoparticles (d) and glutathione (GSH)-responsive TPZ release (e) were measured at the indicated time. n = 3 biologically independent samples. One-way analysis of variance (ANOVA) for (e). Fluorescence images (f) and flow cytometry analysis (g) of neutrophils loaded with smooth and rough SiO₂-TPZ. Biological triplicates were performed independently. h Cellular SiO₂ content in hPSC-derived CAR-neutrophils was measured. n = 5 biologically independent samples, two-tailed Student’s t test. Cellular viability (i), n = 3 biologically independent samples, transmigration (j), n = 5 biologically independent samples, chemoattraction abilities (k, l), n = 20 biologically independent samples, and ROS generation ability (m) of hPSC-derived CAR-neutrophils loaded with or without rough SiO₂-TPZ were shown, n = 5 biologically independent samples, two-tailed Student’s t test. PMA: phorbol myristate acetate. All data in this figure are represented as mean ± SD. Source data are provided as a Source Data file.

Fig. 4. CAR-neutrophils loaded with R-SiO₂-TPZ nanoparticles effectively kill glioblastoma cells.

a Representative images of immunological synapses indicated by polarized F-actin accumulation at the interface between CAR-neutrophils and tumor cells at 6, 12 and 24 h were shown. R-SiO₂-TPZ nanoparticles released from CAR-neutrophils upon tumor cell phagocytosis were up-taken by tumor cells. Triplicates were performed independently. b Schematic of neutrophil-mediated anti-tumor cytotoxicity assay. Cytotoxicity against U87MG glioblastoma cells were performed at different ratios of neutrophil-to-tumor target using indicated neutrophils at 24 h (c), 36 h (d), 48 h (e), and 72 h (f). n = 3 biologically independent samples. Data are represented as mean ± SD, one-way analysis of variance (ANOVA). g Bulk RNA sequencing analysis was performed on U87MG cells under various conditions. Heatmap shows expression levels of selected cytoplasm, membrane, oxidative stress, apoptosis, and proliferation-related genes in the indicated glioblastoma cells. n = 2 biologically independent samples. Source data are provided as a Source Data file.

Fig. 5. Functional evaluation of CAR-neutrophils loaded with R-SiO₂-TPZ nanoparticles using biomimetic glioblastoma (GBM) models in vitro.

a Schematic of our in vitro tumor model of GBM with blood-brain-barrier (BBB), which is composed of endothelial cells on the cell insert membrane and tumor cells in the bottom of the same transwell. b Transwell migration analysis of neutrophils at 12 h is shown. Anti-GBM cytotoxicity of indicated neutrophils at 24 h (c) and 36 h (d) was measured and quantified. e ELISA analysis of IL-6 and TNFα released from indicated neutrophils at 36 h was performed. f Second migration of different neutrophils at 48 h is shown. g Anti-GBM cytotoxicity of indicated neutrophils at 60 h was measured and quantified. h–j Schematic of neutrophil-infiltrated three-dimensional (3D) tumor model in vitro was shown in (h). i Representative fluorescent images of infiltrated neutrophils in the 3D tumor models were shown. DAPI was used to stain the cell nuclear and CD45 was used to stain neutrophils. Scale bars, 200 μm. Biological triplicates were performed independently. j The corresponding tumor-killing ability of indicated neutrophils was measured and quantified using cytotoxicity kit. Data are represented as mean ± SD of five independent biological replicates, one-way analysis of variance (ANOVA). Source data are provided as a Source Data file.

Fig. 6. In vivo distribution of CAR neutrophil-delivered R-SiO₂-TPZ nanoparticles (NPs).

a Schematic of intravenously administered Cy5-labeled CAR neutrophil@R-SiO₂ NPs and R-SiO₂ NPs for in vivo cell tracking study. 5 × 10⁵ luciferase (Luci)-expressing U87MG cells were stereotactically implanted into the right forebrain of NRG mice. After 4 days, mice were intravenously treated with PBS, 5 × 10⁶ Cy5-labeled CAR neutrophil@R-SiO₂ NPs and R-SiO₂ NPs. b Time-dependent biodistribution of Cy5+ neutrophils in whole body, brain, and other organs was determined and quantified by fluorescence imaging at the indicated hours. c Biodistribution of CAR neutrophil@R-SiO₂ NPs and R-SiO₂ NPs in mice at 24 h post-injection was analyzed by inductively coupled plasma-optical emission spectrometry (ICP-OES) based on Si element, and data was expressed as the percentage of injected dose per gram of tissue (%ID/g). n = 5 biologically independent samples. Data are represented as mean ± SD. Source data are provided as a Source Data file. d Representative fluorescence images of CD45 and SiO₂ in the indicated glioblastoma xenografts isolated from tumor-bearing mice were shown. Scale bars, 100 μm. Biological triplicates were performed independently.

Fig. 7. In vivo anti-tumor activities of combinatory CAR-neutrophils and R-SiO₂-TPZ nanoparticles (NPs) were assessed via intravenous injection.

a Schematic of intravenously administered PBS, PB-neutrophils, CAR-neutrophils, and CAR-neutrophil@ R-SiO₂-TPZ NPs for in vivo tumor-killing study. 5 × 10⁵ luciferase (Luci)-expressing U87MG cells were stereotactically implanted into the right forebrain of NRG mice. After 4 days, mice were intravenously treated with indicated neutrophils weekly for a month. Time-dependent tumor burden was determined (b) and quantified (c) by bioluminescent imaging (BLI) at the indicated days. Data are mean ± SD for mice in (b) (n = 5), one-way analysis of variance (ANOVA). d Kaplan-Meier curve demonstrating survival of indicated experimental groups (n = 5) was shown. Released human tumor necrosis factor-α (TNFα) and IL-6 in the peripheral blood (e) and body weight (f) of different mouse groups were measured at the indicated days. Data are mean ± SD, n = 5 biologically independent samples. g, h Anti-tumor activity of increased dosage frequencies of CAR-neutrophils and R-SiO2-TPZ NPs was assessed. g Schematic of intravenously administered CAR-neutrophils, R-SiO₂-TPZ NPs and CAR-neutrophil@ R-SiO₂-TPZ NPs for in vivo tumor-killing study. h Kaplan-Meier curve demonstrating survival of indicated experimental groups was shown (n = 5). Kaplan–Meier curves were analyzed by the log-rank test. Source data are provided as a Source Data file.