CAR T cells secreting NGF-neutralizing scFv enhance efficacy in clear cell renal cell carcinoma by relieving immunosuppression through immunosympathectomy

Abstract

Background

Chimeric antigen receptor (CAR) T cells have demonstrated remarkable breakthroughs in treating hematologic malignancies, yet their efficacy in solid tumors is limited by the immunosuppressive microenvironment. Sympathetic nerves significantly contribute to this immunosuppressive milieu in solid tumors. However, the impact of tumor sympathetic denervation on enhancing CAR T-cell antitumor efficacy remains unclear.

Methods

We screened for sympathetic gene sets in various types of cancers and investigated the association of sympathetic nerves with immunosuppression in renal clear cell carcinoma. Using antibodies to block the nerve growth factor (NGF) pathway, we explored sympathetic nerve distribution in tumor tissues and tumor progression. Additionally, we engineered CAR T cells to secrete NGF single chain fragment variable (scFv) to achieve tumor immunosympathectomy and assessed their antitumor efficacy. Bulk RNA sequencing and single-cell RNA sequencing analyses were conducted to evaluate changes in immune cell phenotypes within the tumor microenvironment.

Results

Blocking the NGF pathway with antibodies effectively reduced sympathetic nerve distribution in tumor tissues and delayed tumor progression. CAR T cells engineered to secrete NGF scFv achieved a similar tumor immunosympathectomy and exhibited enhanced tumor suppression. RNA sequencing analyses revealed that this augmented effect was primarily due to the inhibition of the terminal exhaustion phenotype in tumor-infiltrating CD8 T cells and the prevention of macrophage polarization from M1 to M2. This approach maintained a stronger antitumor immune state at the tumor site. Additionally, splenic T cells also exhibited a more potent immune effector phenotype following the infusion of NGF scFv-secreting CAR T cells.

Conclusions

Our results suggest that immunosympathectomy is a novel approach to weaken tumor microenvironment immunosuppression and synergistically enhance CAR T-cell efficacy against solid tumors.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Chimeric antigen receptor (CAR) T cells have faced limitations in solid tumors due to the immunosuppressive tumor microenvironment. Sympathetic nerves are known to contribute to this immunosuppression, yet the effect of tumor sympathetic denervation on CAR T-cell efficacy in solid tumors remains unclear.

WHAT THIS STUDY ADDS

• This study demonstrates that engineering CAR T cells to secrete nerve growth factor single chain fragment variable effectively reduces sympathetic nerve distribution in tumors, alleviates immunosuppression in the tumor microenvironment, and enhances the antitumor activity of CAR T cells.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings indicate that tumor immunosympathectomy could enhance CAR T-cell antitumor efficacy in solid tumors.

Background

Chimeric antigen receptor (CAR) T-cell therapy is an effective cancer treatment modality, particularly in hematological malignancies.¹ However, their efficacy is limited in solid tumors. Existing strategies, which include enhancing CAR T-cell infiltration,² augmenting their effector functions,³ and improving their persistence in vivo,⁴ have somewhat improved the therapeutic outcomes against solid tumors. However, the extensive immunosuppressive pathways within the tumor microenvironment (TME) of solid tumors still predominantly hinder the functionality of CAR T cells in most scenarios.⁵

The peripheral nervous system (PNS) consists of the autonomic and sensory nervous systems, which help the body maintain homeostasis.⁶ Within the PNS, the autonomic nervous system is further divided into the sympathetic and parasympathetic nervous systems.⁷ Recent studies have shown that the sympathetic nervous system, through the secretion of neurotransmitters, can control certain malignant phenotypes of tumors and influence immune responses.⁸ Immune cells generally express adrenergic receptors (ARs).⁹ Numerous studies indicate that ARs signaling can attenuate the effector functions of T cells and promote the formation of an anti-inflammatory phenotype in macrophages.^{10 11} Given the widespread distribution of sympathetic nerves in solid tumors, intervening tumor-infiltrating sympathetic nerves to affect the TME, thereby creating a favorable environment for CAR T cells to combat solid tumors, is a promising strategy.

Existing strategies for targeting tumor innervation primarily include pharmacological blockade of neural signals and denervation.¹² Studies have shown that pharmacological blockade of adrenergic nerve signals, particularly β-ARs signals, has been proven in many preclinical studies to amplify the effects of immunotherapy.¹³ However, some retrospective studies have found no significant correlation between β-blockers and patient prognosis.¹⁴ Sympathetic denervation through the inhibition of neurotrophins signaling pathways is another approach that has received significant attention. Nerve growth factor (NGF) is essential for the recruitment of sympathetic nerves.¹⁵ Research has found that NGF released by cancer cells seem to be the driving factor for nerve infiltration in the TME.¹⁶ Using gold nanoparticles encapsulating siRNA targeting NGF can effectively reduce the density of sympathetic nerves in pancreatic cancer tissues.¹⁷

Here, we explored a hypothesis that CAR T cells, engineered to secrete NGF single chain fragment variable (scFv), can achieve sympathetic denervation at the tumor site, reducing the immunosuppressive activity within the TME and enhancing the antitumor function of CAR T cells. Our results demonstrated CAR T cells secreting NGF scFv maintain a greater proportion of tumor-associated macrophages in the pro-inflammatory M1 phenotype and inhibit the differentiation of tumor-infiltrating T cells into the terminal exhaustion phenotype through a process of immunosympathectomy, thereby improving their antitumor activity. These findings indicate that sympathetic denervation may be a potential method to enhance the anti-solid tumor function of CAR T cells.

Methods

Cell lines and mice

All cell lines were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium, or Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Biological Industries, Kibbutz Beit-Haemek, Israel), 100 U/mL of penicillin, and 100 µg/mL of streptomycin at 37°C in a 5% CO₂ incubator.

5-week-old female SPF grade Balb/c nude mice were purchased from the Comparative Medicine Center of Yangzhou University (n=44). 6-week-old female SPF grade Balb/c mice were acquired from the Experimental Animal Business Department of the Shanghai Institute for Planned Parenthood Research (n=18). All animal experiments were approved by the Committee for Animal Use of China Pharmaceutical University (SYXK(苏)2023–0019).

Analysis of The Cancer Genome Atlas data sets

In GEPIA2,¹⁸ sympathetic nerve markers of NGF, NEFL, PRPH, TH, and NTRK1 were used to search for expression levels. Survival analysis was conducted on the Kaplan-Meier Plotter website using sympathetic nerve markers as variables. The Cancer Genome Atlas (TCGA)-kidney renal clear cell carcinoma (KIRC) count matrices were transformed using the vst method, then divided into a high and low expression of sympathetic nerve markers based on the median value. The IOBR package¹⁹ was used to analyze immune infiltration. Data visualization was carried out using ggplot2.

Validation of sympathetic denervation in nude mouse xenograft model

6×10⁶ 786-O cells were inoculated into the left axilla of nude mice. Once the tumors reached a size of 100 mm³, the mice were randomly divided into four groups (n=5). Each group received intraperitoneal injections of either phosphate-buffered saline (PBS) (100 µL/day), Propranolol (10 mg/kg/day) (MedChemExpress, New Jersey, USA), NGF antibody (200 µg/kg/ two times per week) (R&D Systems, Minnesota, USA), or 6-OHDA (day 7: 100 mg/kg, day 9: 250 mg/kg)²⁰ (MedChemExpress). Tumor volumes were measured twice a week. Tumor volume was calculated using the formula: length×width²/2.

Expression and purification of NGF scFv

The light and heavy chain variable regions of NGF scFv were obtained from the NCBI’s PDB database (Accession No.: 4EDX). The sequences were synthesized and cloned into the pET-22b vector in the order of VH-(G₄S)₂-VL-6×His. The construct was then expressed using ArcticExpress. Purification was carried out using a nickel column.

Assay for blocking activity of NGF scFv

NGF scFv was diluted in a gradient and co-incubated with TrkA-hFc (Sino Biological) in the Beta-NGF (Sino Biological, Beijing, China)—coated plates. Following the binding of goat anti-human IgG Fc (Invitrogen, Maryland, USA), 3,3',5,5'-Tetramethyl-benzidine (TMB) solution was added. The optical density at 450 nm was measured using Multiskan FC tablet reader (Thermo Science).

Construction of CAR vector and retroviral packaging

The vascular endothelial growth factor receptor 2 (VEGFR2) scFv was obtained from a patent (Patent No.: US 8822196B2). An Myc tag was added after scFv to serve as a detection marker. The hinge region, the transmembrane domain, the intracellular signaling domain of CD28, and the intracellular signaling domain of CD3ζ were acquired from the GenBank database (Accession No.: HM754222). Based on V28z, NGF scFv was linked to CD3ζ through a 2A peptide, resulting in the construct V28z/αNGF. The sequence was synthesized and integrated into the Murine stem cell virus (MSCV) retroviral expression vector. The pMSCV plasmid was co-transfected with the pCL-Eco plasmid into 293 T cells in a 3:1 ratio. After 72 hours, the supernatant was collected.

Transduction and characterization of V28z and V28z/αNGF

Fresh splenocytes from Balb/c mice were isolated using a spleen lymphocyte separation kit (Solarbio, Beijing, China), and mouse T cells were obtained using Untouched Mouse T Cells Kit (Invitrogen) following the manufacturer’s protocol. Dynabeads Mouse T-Activator (Invitrogen) were added to the T cells at a bead-to-cell ratio of 2:1 for 48 hours of activation. The activated cells were then transferred to a 24-well plate coated with RetroNectin (Takara, USA) and infected with the retrovirus at a multiplicity of infection (MOI) of 5. The cells were centrifuged at 1,200×g for 90 min at 31°C. CAR T cells were cultured in RPMI-1640 medium (Gibco) supplemented with 10% FBS (Gibco), 200 U/mL IL-2 (Jslyy Pharmaceutical, Jiangsu, China), and 55 µM 2-Mercaptoethanol (Invitrogen). Three days post-infection, the expression of CAR was detected using anti-Myc antibody (1:100) and Cy3-conjugated secondary antibody (1:100) by flow cytometry.

Lactate dehydrogenase cytotoxicity assay

Target cells bEnd.3, MS1, and Renca were used in this assay. Effector cells were added to these target cells at effector-to-target (E:T) ratios of 1:1, 2:1, 4:1, and 8:1. They were co-incubated for 24 hours. Cytotoxicity was then measured using a lactate dehydrogenase (LDH)-Cytotoxicity Assay Kit (MedChemExpress) according to the manufacturer’s instructions on Multiskan FC tablet reader.

Cell proliferation assay

Effector cells were labeled with CellTrace Violet (Invitrogen) according to the manufacturer’s instructions. Labeled effector cells were then added to the target cells at an E:T ratio of 2:1 and incubated for 72 hours. Fluorescence intensity was detected using CytoFLEX (Beckman Coulter) to assess cell proliferation.

In vivo pharmacodynamics evaluation in immunocompetent mice

1×10⁶ Renca cells were inoculated into the left axilla of Balb/c mice. When tumors reached approximately 80 mm³, mice were intraperitoneally injected with 200 mg/kg cyclophosphamide (Baxter, Illinois, USA). Two days later, the mice were randomly divided into three groups and intravenously infused with either untransduced T (UT), 1×10⁷ V28z, or 1×10⁷ V28z/αNGF cells (n=5). Six hours post-infusion, 500,000 U of human recombinant interleukin (IL)-2 was administered intraperitoneally, followed by continuous injections two times per day for 2 days. Tumor volumes were measured twice weekly.

Tissue cell preparation for flow cytometry

Freshly isolated tumor tissues were minced and digested in DMEM containing 0.5 mg/mL collagenase I, 0.5 mg/mL collagenase IV, 0.1 mg/mL DNase, and 0.1 mg/mL hyaluronidase (all from Sigma-Aldrich, Missouri, USA) at 37°C. The suspension was then filtered through a 70 µm cell strainer. Fixable Viability Stain 780 (BD Biosciences, New Jersey, USA) was added to differentiate dead cells. FcR Blocking Reagent (Miltenyi Biotec, Bergisch Gladbach, Germany) and corresponding surface antibodies (online supplemental table 1) were co-incubated with the cells. Following this, cells were fixed and permeabilized, and corresponding intracellular antibodies were added and incubated at room temperature. Flow cytometry was performed using CytoFLEX, and data were analyzed using FlowJo V.10 software.

Peripheral blood T-cell analysis

Take 50 µL of peripheral blood, add 10 times the volume of red blood cell lysis buffer (Beyotime, Shanghai, China), and lyse at room temperature for 5 min. Subsequently, add corresponding antibodies (online supplemental table 1) and incubate at 4°C before analysis with CytoFLEX.

Immunohistochemical analysis

Freshly isolated tumor tissues were fixed in 4% paraformaldehyde for 48 hours and then embedded in paraffin. Tissue blocks were cut into 5 µm thick sections. Antigen retrieval was performed using Tris-EDTA solution (pH=9.0). Blocking was done with 5% goat serum at room temperature for 1 hour. The sections were then incubated overnight at 4°C with TH antibody (1:500), CD3 antibody (1:100), NEFH antibody (1:200), SLC18A3 antibody (1:200) and CD31 antibody (1:200). This was followed by incubation with horseradish peroxidase (HRP)-conjugated secondary antibody (1:1,000) at room temperature for 30 min. TMB solution was added for color development, followed by counterstaining with hematoxylin. The slides were scanned using NanoZoomer-XR (Hamamatsu). Pictures were processed using Image-Pro Plus V.6.0 software.

Western blot

After sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transfer to a polyvinylidene difluoride (PVDF) membrane, the membrane was blocked with 5% non-fat milk at room temperature for 1 hour. TH antibody (1:3,000), NGF antibody (1:1,000), or His tag antibody (1:1,000) were then added and incubated overnight at 4°C. This was followed by incubation with HRP-conjugated secondary antibody (1:10,000) at room temperature for 1 hour. Imaging was performed using 4600SF Chemiluminescent Imaging System (Tanon).

ELISA analysis of sympathetic neurotransmitters in the tissues

100 mg fresh tumor tissues were homogenized in 150 µL pre-chilled PBS using a tissue homogenizer, and the supernatant was collected for analysis. Epinephrine (EPI) and norepinephrine levels were measured using EPI ELISA kit (Elabscience) or Mouse norepinephrine ELISA kit (CUSABIO), following the manufacturer’s instructions.

Cytometric bead array for peripheral blood cytokine analysis

Collect 50 µL of plasma and use the Ms TNF-α Flex Set C8 and Ms IL-6 Flex Set B4 (BD Biosciences) according to the manufacturer’s instructions to measure cytokine levels in the peripheral blood. Data were processed using FCAP Array V.3 software.

Splenocyte activation experiment

Splenocytes were obtained as previously described. Cells were resuspended at 1×10⁶ /mL, and 50 ng/mL phorbol-12-myristate-13-acetate and 500 ng/mL Ionomycin (MedChemExpress) were added to stimulate the lymphocytes, supplemented with GolgiStop (BD Biosciences). After 8 hours, corresponding antibodies (online supplemental table 1) were incubated as previously described and analyzed using CytoFLEX.

Bulk RNA sequencing

For tumor RNA sequencing (RNA-seq), total RNA is extracted from tumor tissues. For T-cell RNA-seq, T cells are isolated from activated splenocytes using the above method, and total RNA is extracted from the T cells. The libraries were sequenced on a NovaSeq 6000 S4 platform, and 150 bp paired-end reads were generated. Paired FASTQ data were quantified for transcripts by aligning to the mm39 reference genome using Salmon (V.1.10.0). Quantification data were imported into R (V.4.2.3) using tximport (V.1.26.1). Differential gene analysis was performed using pydeseq2 (V.0.4.3) in Python (V.3.10.9). Differentially expressed genes were defined as having a log2 fold change >0.5 and a padj<0.05. Data visualization was achieved using Seaborn (V.0.12.2). Preranked Gene Set Enrichment Analysis (GSEA) was conducted using the GO_Biological_Process_2023 gene set in gseapy (V.1.0.6).

Single-cell sequencing library preparation and sequencing

Freshly collected tumor tissues were pooled (three tumors per group). The tissues were minced and digested using the Tumor Dissociation Kit. The digested cell suspension was filtered through a 70 µm filter. Dead cells were removed using the Dead Cell Removal Kit. Single viable cells were collected, and the library was prepared using the 10x Genomics 3’ V3 kit, following the standard protocol of 10x Genomics. The prepared library was sequenced on the NovaSeq 6000 platform using the PE150 mode.

scRNA-seq analysis

FASTQ files were quantified using CellRanger (V.7.2.0), referencing the mm10 2020A genome. Files generated by CellRanger were imported into scanpy (V.1.9.5), and quality control metrics were calculated. Cells with less than 200 genes and genes expressed in fewer than three cells were filtered out. Highly variable genes were selected using the seurat_v3 parameter. Doublet cells were removed using scvi.external.SOLO from scvi-tools²¹ (V.1.0.4). Additionally, cells with pct_counts_mt<20% were filtered to remove low-quality cells. Counts were normalized per 10,000 reads and log2-transformed. Only highly variable genes were used for principal component analysis (PCA). The number of PCs used for neighborhood graph construction and dimensional reduction was set at 30. Clustering was performed using the Leiden algorithm with the resolution set at 0.7. Clusters were manually annotated using known marker genes (online supplemental table 2). Differential gene detection between groups was performed using the sc.tl.rank_genes_groups function. Gene set scoring was done using the sc.tl.score_genes function.

Statistical analysis

The data were presented as the mean±SD. Statistical significance was determined using one-way analysis of variance. GraphPad Prism V.8.0 software was used for some statistical analyses, while Python (V.3.10.9) was employed for the analysis of RNA-seq and single-cell RNA sequencing (scRNA-seq) data. A p value of <0.05 was considered to indicate statistical significance.

Results

Correlation between sympathetic nerve markers and immunosuppression in clear cell renal cell carcinoma

Initially, we screened the expression levels of sympathetic nerve markers in the GEPIA2 database. The results indicated that the gene set was expressed at higher levels in tumor tissues of pheochromocytoma and paraganglioma, KIRC, sarcoma, head and neck squamous cell carcinoma, kidney renal papillary cell carcinoma, pancreatic adenocarcinoma, and thymoma compared with normal tissues (figure 1A). Through further analysis of the relationship between tumor prognosis and gene set expression, cancer types in which high gene set expression is associated with poor prognosis were selected for subsequent research. We found high expression of this gene set is significantly associated with reduced survival in patients with KIRC (figure 1B). We also found that sympathetic nerve markers peripherin, tyrosine hydroxylase (TH), and NGF were highly expressed in clear cell renal cell carcinoma (ccRCC) xenograft mode (online supplemental figure S1). Analysis of the TCGA-KIRC data set indicated that tumors with high expression of sympathetic nerve markers had reduced infiltration of CD4 T cells, CD8 T cells, and M1 macrophages, while the infiltration of immunosuppressive regulatory T cells (Treg) and M2 macrophages increased (figure 1C). Additionally, the expression of immune checkpoints and immunosuppressive factors were significantly upregulated in the tissues which highly express sympathetic nerve markers (figure 1D–E). GSEA results showed that samples with high expression of sympathetic nerve markers had a higher enrichment score with T-cell exhaustion-related genes22,25 compared with those with low expression (figure 1F). These data suggest that sympathetic nerves in the TME may be associated with the formation of an immunosuppressive milieu in ccRCC.

Figure 1. Relationship between sympathetic nerve gene set expression, cancer prognosis, and immune function. (A) Expression of the sympathetic nerve gene set in the tumor and corresponding normal tissues in the GEPIA2 database. (B) The relationship between the expression of the sympathetic nerve gene set and cancer prognosis is presented on the Kaplan-Meier Plotter website. (C–E) The association of the sympathetic nerve gene set expression in KIRC tumor tissues with (C) immune cell infiltration, (D) immune checkpoint expression, and (E) immunosuppressive factors expression. (F) Gene Set Enrichment Analysis of T-cell exhaustion characteristics with high and low expression of the sympathetic nerve gene set. HNSC, head and neck squamous cell carcinoma; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; SARC, sarcoma; THYM, thymoma.

Blocking NGF signaling inhibit ccRCC cell growth in vivo via sympathetic denervation

To evaluate the impact of different approaches for intervening in the sympathetic nerve pathways within tumor tissues on tumor progression, we employed propranolol (β-AR blockade), NGF blocking antibodies (immunosympathectomy), and 6-hydroxydopamine (6-OHDA) (chemical sympathectomy). No significant differences in cell growth were observed among the groups, except for a slight reduction in the propranolol group compared with the PBS group at 48 hours, indicating that anti-NGF and 6-OHDA do not directly affect tumor cell growth (figure 2A, online supplemental figure S2). However, in the nude mouse xenograft model, both 6-OHDA and anti-NGF treatment groups inhibited tumor growth, whereas the propranolol treatment group showed no significant difference in tumor growth from the PBS group (figure 2B–C, online supplemental figure S3). The densities of sympathetic nerves in the tumor were reduced in the 6-OHDA and anti-NGF treatment groups, but not in the propranolol group (figure 2D). Meanwhile, no significant differences in parasympathetic nerves were observed across the treatment groups (online supplemental figure S4A). Additionally, sensory nerve infiltration was not detected in the tumor tissue (online supplemental figure S4B). Previous studies have demonstrated that sympathetic denervation can also reduce tumor vasculature in prostate cancer.²⁰ We assessed vascular density the following treatment and found that the vascular distribution was significantly decreased in the 6-OHDA and anti-NGF groups compared with the PBS and propranolol groups (figure 2E). Given previous analyses showing a correlation between sympathetic nerve markers and tumor-associated macrophage polarization (figure 1C), we assessed macrophage phenotypes and demonstrated that the M1/M2 ratio in tumor tissues increased in the anti-NGF treatment group compared with the propranolol and PBS groups, although less than the 6-OHDA treatment group (figure 2F). Additionally, James et al found ablation of sympathetic nervous leads to the accumulation of myeloid-derived suppressor cells (MDSCs),²⁶ we assessed tumor-infiltrating MDSCs levels and finding no significant differences in MDSCs infiltration among the different treatment groups (figure 2G). These data demonstrate that, like 6-OHDA, anti-NGF can effectively remove sympathetic nerves from tumor tissues, and that sympathetic denervation can inhibit tumor growth by altering the TME, such as by modifying macrophage polarization.

Figure 2. Antitumor effect of NGF antibody in clear cell renal cell carcinoma tumor-bearing nude mice. (A) CCK8 assay to detect the impact of propranolol (10 µM), 6-OHDA (150 µM), and NGF antibody (0.06 µg/mL) on the proliferation of 786-O cells. (B) In vivo drug administration scheme flowchart. (C) Changes in tumor volume over time. (D) Immunohistochemical detection of tyrosine hydroxylase expression in tumors. (E) Immunohistochemical detection of CD31 expression in tumors. (F) Flow cytometry analysis of macrophage phenotypes in tumor tissues. (G) Immunofluorescence detection of MDSCs infiltration in tumor tissues. Mean±SD, *p<0.05, ***p<0.001, and ****p<0.0001. CCK8, cell counting kit-8; i.p., intraperitoneal injection; MDSCs, myeloid-derived suppressor cells; NGF, nerve growth factor; PBS, phosphate-buffered saline.

NGF scFv-secreted CAR T cells activated and exert cytotoxic effects in an antigen-dependent manner in vitro

Before constructing NGF scFv-secreted CAR T cells, we first evaluated the blocking activity of NGF scFv. The binding between NGF and TrkA-hFc was effectively inhibited by the selected NGF scFv (figure 3A, online supplemental figure S5A-C). We then constructed CAR T cells targeting VEGFR2 (named V28z) and their counterparts secreting NGF scFv (named V28z/αNGF) (figure 3B). The positive rates for both V28z and V28z/αNGF were over 50% (figure 3C), with no significant difference between the CD4 T and CD8 T-cell subtypes (online supplemental figure S6). We also found the presence of NGF scFv in the culture supernatant of V28z/αNGF (figure 3D), and its culture supernatant could block the binding between NGF and TrkA-hFc, which was not observed in the supernatants of UT and V28z (figure 3E). Using MS1 and bEnd.3 (high VEGFR2 expression) and Renca (VEGFR2-negative) as target cells (figure 3F), LDH release assays showed that both V28z and V28z/αNGF exerted effective cytotoxic effects against bEnd.3 and MS1 compared with UT, and their cytotoxicity increased with the rise in E:T ratio. However, for Renca cells, no significant cytotoxic activity was observed with any of the effector cells (figure 3G). CD69 expression on T cells and their proliferation after co-incubation also aligned with the cytotoxicity results (figure 3H–I). Additionally, we observed that in the ccRCC tumor scRNA-seq data, NGF’s primary ligand, TrkA, is expressed in T-cell subpopulations (online supplemental figure S7A), and TrkA expression upregulated following T-cell activation (online supplemental figure S7B). This prompted us to investigate whether NGF would affect the in vitro activity of CAR T cells. However, our results indicated that the addition of NGF during co-incubation did not influence the cytotoxic activity of V28z and V28z/αNGF against target cells (online supplemental figure S7C). These findings indicate that V28z/αNGF can secrete NGF scFv with blocking activity, and the secretion of NGF scFv does not affect the antigen-dependent activation and cytotoxic effects of V28z/αNGF.

Figure 3. Construction, identification, and in vitro activity assessment of V28z and V28z/αNGF. (A) Competitive ELISA to assess the blocking ability of NGF scFv on the binding between NGF and TrkA-hFc. (B) Schematic diagram of the chimeric antigen receptor structures of V28z and V28z/αNGF. (C) Flow cytometry detection of Myc tag expression in UT (untransduced T), V28z and V28z/αNGF. (D) Western blot analysis of NGF scFv in the culture supernatant of V28z and V28z/αNGF. (E) Competitive ELISA to evaluate the blocking ability of the supernatant on the binding between NGF and TrkA-hFc. (F) Flow cytometry analysis of VEGFR2 expression in murine cells. (G) Lactate dehydrogenase release assay to assess the cytotoxicity of effector cells against target cells under different effector-to-target ratios. (H) Flow cytometry detection of CD69 expression on T cells after co-cultivation with target cells. (I) CellTrace method to evaluate the proliferation of effector cells after co-cultivation. Mean±SD, *p<0.05, **p<0.01, and ***p<0.001. NGF, nerve growth factor; scFv, single chain fragment variable; vascular endothelial growth factor receptor 2.

V28z/αNGF exhibits stronger antitumor activity compared with V28z in vivo

The efficacy of V28z/αNGF was assessed in immunocompetent tumor-bearing mice (figure 4A). Tumor growth was inhibited in both V28z and V28z/αNGF groups, with V28z/αNGF demonstrating a stronger inhibitory effect (figure 4B–C, online supplemental figure S8A). To characterize the effect of sympathetic denervation, ELISA results demonstrated that the levels of EPI and norepinephrine in tumor tissues were reduced in the V28z/αNGF group compared with both the V28z and UT groups, with no difference observed between the V28z and UT groups (figure 4D). Immunohistochemistry (IHC) revealed a reduction in sympathetic nerves in the tumors of the V28z/αNGF group, whereas no significant difference was observed between the V28z and UT groups (figure 4E). Additionally, both V28z and V28z/αNGF treatments resulted in reduced vascularization in the tumors, with the V28z/αNGF group showing a greater reduction (figure 4F). Furthermore, proportion of peripheral CD3 T cells, infiltration of CD3 T cells and CAR⁺ T cells in the tumor tissues was increased in the V28z/αNGF group (figure 4G–I). In addition, the infiltrating CD3 T cells in the V28z/αNGF tumors had a higher CD4/CD8 ratio (figure 4J), suggesting potentially better therapeutic efficacy.²⁷ To comprehensively characterize the immune status at the tumor site, RNA-seq analysis was performed. GSEA of differentially expressed genes indicated that inflammation-related signaling pathways were enhanced in the V28z/αNGF group compared with the V28z group (figure 4K). We also found an elevation of key pro-inflammatory cytokines IL-6 and tumor necrosis factor α (TNF-α) in the peripheral blood of mice in the V28z/αNGF group (figure 4L). Further analysis revealed that the tumor tissues in the V28z/αNGF group exhibited stronger immune activation markers,¹¹ both in terms of innate immune responses (such as genes encoding Tlr1, Tlr7, Tlr8, or Cd80) and adaptive immune responses (such as genes encoding Cxcl13, Cd48, Nfatc1, or Nfatc2) (figure 4M). In conclusion, these data demonstrate that V28z/αNGF has stronger tumor suppression capabilities compared with V28z and maintains a more robust antitumor immune state in the tumor tissues.

Figure 4. In vivo antitumor activity evaluation of V28z and V28z/αNGF. (A) Schematic diagram of the cell infusion protocol in vivo. (B) Changes in tumor volume over time. (C) Tumor weight on day 24 after tumor inoculation. (D) ELISA was used to measure the levels of epinephrine and norepinephrine in tumor tissues. (E) IHC detection of the sympathetic nerve marker TH in tissue samples. (F) IHC detection of the vascular marker CD31 in tissue samples. (G) Flow cytometry analysis of the proportion of CD3 T cells in peripheral blood lymphocytes. (H) IHC detection of infiltrating CD3 T cells in the tumor tissues. (I) Flow cytometry analysis of CAR⁺ T-cell proportion in tumor tissues. (J) Flow cytometry assessment of the CD4 and CD8 typing of tumor-infiltrating T cells. (K) Gene Set Enrichment Analysis of differentially expressed genes enriched in signaling pathways, comparing V28z/αNGF versus V28z. (L) Cytometric bead array method to measure the levels of inflammatory cytokines IL-6 and TNF-α in plasma. (M) Heatmap analysis of immune activation genes in tumor tissues. Mean±SD, *p<0.05, **p<0.01, ***p<0.001 and ****p<0.0001. CAR, chimeric antigen receptor; FDR, false discovery rate; IL, interleukin; iv, intravenous; NGF, nerve growth factor; TH, tyrosine hydroxylase; UT, untransduced T.

V28z/αNGF inhibits the formation of exhausted T cells^term phenotype in tumor-infiltrating CD8 T cells

We employed scRNA-seq to characterize the TME after treatment. Following quality control, unbiased clustering of tumor samples from all groups revealed 17 distinct cell clusters (figure 5A). Major cell types were annotated based on classic marker genes, demonstrating clear distinctions between different cell types (online supplemental figure S9A-C). T cells were further categorized into CD4 T cells, CD8 T cells, and CD4 CD8 double-negative T cells (figure 5B–C). Given that CD8 T cells are the primary effector cells in tumor killing, we analyzed CD8 T-cell clusters and found that the proportion of the CD8_T cell_2 subgroup was reduced in the V28z/αNGF treatment group compared with the V28z group (figure 5D). Further analysis indicated that the CD8_T cell_2 subgroup expressed higher levels of exhaustion-related genes (Tox, Pdcd1) and almost no expression of effector molecule genes (Gzmb, Gzma), suggesting that this subgroup represented exhausted T cells (Tex) (figure 5E). Additionally, gene set scores representing different subtypes of CD8 T cells, such as effector memory T cells (Tem),²⁸ Tex,²⁹ and tissue-resident T cells (Trm)³⁰ were analyzed. There was no significant difference in the Tem feature score between the two groups (figure 5F). The Tex feature score in CD8 T cells of the V28z/αNGF treatment group was lower than in the V28z group (figure 5G), while the Trm feature score was higher (figure 5H). Tex can be further divided into Tex^int, with proliferation and effector functions, and terminally differentiated Tex^term.³¹ Analysis of the Tex subgroups revealed that the proportion of Tex^int was not different among treatment groups, but the Tex^term subgroup was reduced in the V28z/αNGF treatment group (figure 5I). This observation is in line with the increased proportion of Ki67-positive CD8 T cells in the V28z/αNGF group (figure 5J). These findings demonstrate that V28z/αNGF treatment reduces the exhaustion level of tumor-infiltrating CD8 T cells, more specifically by decreasing the formation of the Tex^term phenotype in CD8 T cells.

Figure 5. Assessment of immune activity in tumor-infiltrating T cells. (A) Uniform Manifold Approximation and Projection (UMAP) representation of scRNA-seq analysis of whole cells. (B) UMAP analysis of T-cell clusters. (C) UMAP plots show the expression of Cd4 and Cd8a. (D) The relative proportion of T-cell subtypes per cluster. (E) Dotplot illustrating the relative expression of signature genes in CD8 T-cell clusters. (F–H) Gene signature analysis of (F) effector memory T cells, (G) exhausted T cells, and (H) tissue-resident T cells in CD8 T cells. (I) Flow cytometry assessment of the proportion of exhausted phenotypes in tumor-infiltrating CD8 T cells. Tex^prog (Tim3⁻CD101⁻), Tex^int (Tim3⁺CD101⁻), Tex^term (Tim3⁺CD101⁺). (J) Flow cytometry analysis of Ki67 expression in tumor-infiltrating CD8 T cells. Mean±SD, *p<0.05, **p<0.01, and ****p<0.0001. NGF, nerve growth factor; scRNA-seq, single-cell RNA sequencing; Tex, exhausted T cells; UT, untransduced T.

V28z/αNGF affects the phenotype of tumor-infiltrating macrophages and neutrophils

Tumor-infiltrating macrophages were further subdivided into five clusters (figure 6A). Following V28z/αNGF treatment, the Macrophages_1 cluster was reduced compared with V28z treatment, while Macrophages_2 and Macrophages_3 clusters were increased (figure 6B). Analysis of macrophage-related characteristic genes showed that the Macrophage_1 subgroup highly expressed M2 phenotype-related genes (Fn1, Mrc1, Trem2), while the Macrophages_2 subgroup highly expressed antigen-presentation-related genes (H2-Ab1, H2-Aa, H2-Eb1) and the Macrophages_3 subgroup highly expressed pro-inflammatory genes (Ccl2, Ccl3, and Ccl4), indicating an M1 phenotype (figure 6C). GSEA analysis revealed a significant enrichment of immune activation and apoptosis induction pathways in macrophages following V28z/αNGF treatment (figure 6D). Flow cytometry results confirmed the above analysis, showing that the M1/M2 ratio was higher after V28z/αNGF treatment (figure 6E). Additionally, neutrophils were further divided into three subgroups (figure 6F, online supplemental figure S10). It was observed that the proportion of Neutrophils_1 was comparable between V28z and V28z/αNGF groups, but there was a transition from Neutrophils_3 to Neutrophils_2 in V28z/αNGF group (figure 6G). It is difficult to define different populations of neutrophils, as few specific cell surface markers have been identified, except for polymorphonuclear (PMN)-MDSCs.³² Analysis revealed that the Neutrophils_1 subgroup primarily expressed PMN-MDSCs characteristic genes (figure 6H). In agreement with the scRNA-seq results, flow cytometry revealed no significant difference in the proportion of MDSCs (CD11b⁺Ly6C⁻ Ly6G⁺) between V28z and V28z/αNGF groups (figure 6I). These results demonstrate that V28z/αNGF treatment leads to a greater maintenance of tumor-associated macrophages in the M1 phenotype and does not impact the formation of PMN-MDSCs.

Figure 6. Evaluation of immune activity in tumor-infiltrating myeloid cells. (A) UMAP analysis of macrophage cell clusters. (B) The relative proportion of macrophage subtypes per cluster. (C) Dotplot showing the relative expression of signature genes in macrophage clusters. (D) GSEA of different macrophage clusters for relevant pathways, with significant terms indicated by a black circle. (E) Flow cytometry analysis of the phenotyping of tumor-infiltrating macrophages. (F) UMAP analysis of neutrophil clusters. (G) The relative proportion of neutrophil subtypes per cluster. (H) StackedViolin plot displaying the relative expression of PMN-MDSC signature genes in neutrophil clusters. (I) Flow cytometry assessment of the proportion of PMN-MDSCs in tumor tissues. Mean±SD, **p<0.01. GSEA, Gene Set Enrichment Analysis; NGF, nerve growth factor; UMAP, Uniform Manifold Approximation and Projection; UT, untransduced T.

V28z/αNGF enhances the effector function of splenic T cells

To gain an unbiased view of changes in T-cell activity within secondary lymphoid organs after treatment, RNA-seq of splenic T cells was performed. A total of 898 differentially expressed genes were identified, and among the genes upregulated in the V28z/αNGF group compared with the V28z group, T-cell effector genes (Gzma, Gzmb) were found, while immunosuppressive genes (Cd36, Havcr2) were found among the downregulated genes (figure 7A). Further analysis revealed that V28z/αNGF treatment resulted in downregulation of immunosuppressive gene expression and upregulation of effector gene expression³³ (figure 7B–C). GSEA analysis showed genes highly expressed in the V28z/αNGF group were enriched in glycolysis and oxidative phosphorylation pathways (figure 7D), consistent with their enhanced effector function.³⁴ In addition, higher levels of IL-2 secretion and Ki67 expression in CD8 T cells were confirmed after restimulation of splenic lymphocytes in the V28z/αNGF group (figure 7E). These results demonstrate that V28z/αNGF treatment can enhance the activity of T cells in secondary lymphoid organs.

Figure 7. Analysis of immune activity in splenic T cells. (A) Volcano plot displaying differentially expressed genes, comparing V28z/αNGF versus V28z. (B) Heatmap analysis of immunosuppressive genes in splenic T cells. (C) Heatmap analysis of immune effector genes in splenic T cells. (D) Gene Set Enrichment Analysis of differentially expressed T-cell genes enriched in signaling pathways, comparing V28z/αNGF versus V28z. (E) Flow cytometry analysis of IL-2 and Ki67 expression in CD8 T cells after restimulation of splenic lymphocytes. (F) Schematic illustration of the antitumor action of V28z/αNGF in solid tumors. Mean±SD, *p<0.05, **p<0.01. CAR, chimeric antigen receptor; IL, interleukin; NGF, nerve growth factor; scFv, single chain fragment variable; UT, untransduced T; vascular endothelial growth factor receptor 2.

Discussion

CAR T cells, as a form of ‘living drug’, are highly susceptible to the influence of the TME in solid tumors. Immunosuppressive factors within the TME can inhibit the activation of CAR T cells, induce their exhaustion, and foster acquired resistance.³⁵ Current strategies to overcome tumor immunosuppression mainly include blocking immune checkpoints,³⁶ secreting cytokines,³⁷ combining with oncolytic viruses,³⁸ and altering the metabolic state of CAR T cells.³⁹ While these approaches have shown positive therapeutic effects in preclinical studies, there is still no ideal strategy that has achieved a breakthrough in clinical research. Recent studies have highlighted the critical role of sympathetic nerves in the formation of immunosuppressive TME in solid tumors.⁴⁰ Blocking sympathetic nerve signals has been proven effective in enhancing the efficacy of immune checkpoint therapy in mouse models.⁴¹ However, it remains unclear whether sympathetic denervation of tumors affects the efficacy of CAR T cells in solid tumors. This study finds that CAR T cells secreting NGF scFv can reduce sympathetic nerves in tumor tissues. The reduction of sympathetic innervation in the TME potentially alleviates the immunosuppressive barriers that CAR T cells face, enabling them to function more effectively against solid tumors.

Studies by Globig et al have shown that exhausted T cells cluster around sympathetic nerves, and ARs on T cells can weaken their proliferation and cytokine secretion capabilities, promoting differentiation into a terminal exhaustion phenotype.⁴² Kamiya’s research demonstrates a correlation between tumor tissue sympathetic nerve infiltration and higher expression of immune checkpoint molecules, with denervation via AAV-TH-DTA suppressing immune checkpoint expression and tumor progression more effectively than β-blockers.⁴³ Here, our results revealed a significant reduction in the proportion of terminal exhausted CD8 T cells in tumor tissues and an enhancement in CD8 T cell’s proliferation following V28z/αNGF treatment. Research has demonstrated that norepinephrine contributes to programmed death-1 (PD-1) treatment resistance in patients with lung cancer by inhibiting the infiltration and function of CD8+T cells.⁴⁴ Additionally, studies have found that β2-AR signaling on CD8+T cells restricts their egress from lymph nodes, thereby weakening T cell-mediated antitumor immune responses following radiotherapy.⁴⁵ In our study, we observed a significant reduction in norepinephrine levels in tumors from the V28z/αNGF treatment group, accompanied by a marked increase in the number of infiltrating T cells. This phenomenon may be related to the reduction of β2-AR signaling, and future research will aim to uncover direct evidence of this effect. Furthermore, research by Leonardo et al has shown that secondary lymphoid organs surrounded by rich sympathetic nerves secrete NA, which suppresses cytokine secretion and cytotoxic effects via ADRB2 on CD8 T cells.⁴⁶ Our observations also indicate that splenic T cells from mice treated with V28z/αNGF showed significantly higher immune effector functions than those in the V28z group, which might partly be due to the influence of circulating V28z/αNGF on splenic sympathetic nerve signals.

Beyond effects on T cells, denervation also impacts myeloid immune cells. Yun et al found that denervation through 6-OHDA increased the proportion of M1 macrophages and enhanced the secretion of inflammatory cytokine.⁴⁷ Conversely, James et al discovered that tumor tissue denervation led to increased PMN-MDSCs infiltration, suppressing antitumor immunity.²⁶ However, in a separate study, Mohammadpour et al demonstrated that β2-AR signaling enhances the immunosuppressive activity of MDSCs by modulating their metabolic pathways.⁴⁸ These findings underscore the diverse effects of nerves and neurotransmitters on myeloid immune cells. Our results indicate that the proportion of antigen-presenting and pro-inflammatory M1 macrophages increased in tumors treated with V28z/αNGF, while anti-inflammatory M2 macrophages decreased. However, unlike James et al’s findings, we did not observe a significant increase in PMN-MDSCs infiltration following sympathetic nerve signal blockade, which might be attributed to differences in mouse strains and tumor types used. We also found despite no significant differences in PMN-MDSCs between V28z/αNGF and V28z groups, there were phenotype shifts in two distinct neutrophil populations. Due to the difficulty in distinguishing different neutrophils by characteristic markers, further research is required to fully elucidate these shifts.

In addition to immunosuppression, infiltration difficulties and antigen heterogeneity are major obstacles hindering the application of CAR T cells in solid tumors. Traditional CAR T cells primarily target tumor cell surface antigens, and they must overcome the vascular endothelial barrier and infiltrate into tumor tissues. Furthermore, the heterogeneous expression of tumor antigens means that CAR T cells cannot fully eradicate all tumor cells, contributing to tumor escape mechanisms. Tumor vasculature is essential for solid tumor growth, and disrupting tumor microvasculature can effectively destroy tumors.⁴⁹ Since CAR T cells can easily contact vascular endothelial cells, which stably overexpress receptors like VEGFR1 or VEGFR2, using CAR T cells to target solid tumor vasculature represents a promising strategy.⁵⁰ Several preclinical studies have shown that CAR T cells targeting VEGFR2 can effectively inhibit tumor growth.⁵¹ In our study, both V28z and V28z/αNGF demonstrated inhibitory activity against renal cancer tumors in in vivo models. However, the tumors in our study were not completely suppressed, which we speculate may be due to the long culture time of CAR T cells (12 days), leading to reduced in vivo activity. Future work will need to optimize the culture process and infusion protocols for CAR T cells to enhance their in vivo antitumor efficacy.

The safety of CAR T cells is also a critical concern. Due to the low-level expression of VEGFR2 in normal vasculature, there is potential for off-tumor toxicity with CAR T cells. However, no significant in vivo toxicity was reported in previous preclinical studies or in a clinical trial (NCT01218867). Furthermore, for immunosympathectomy, clinical studies have demonstrated that targeting NGF with NGF antibodies is well-tolerated in humans with minimal impact on the nervous system (NCT02609828). In our study, we did not observe any significant adverse effects following the infusion of V28z/αNGF, and there were no significant fluctuations in mouse body weight (online supplemental figure S8B). Future experiments will need to evaluate the pharmacokinetics and toxicology of V28z/αNGF in vivo to ensure its efficacy and safety.

Conclusion

This study demonstrates that the constructed VEGFR2-targeted CAR T cells, through secreting NGF scFv, achieve tumor sympathetic denervation. This process mitigates immunosuppression in the TME through various mechanisms, such as inhibiting the formation of the terminal exhaustion phenotype in T cells, enhancing T-cell activity, and increasing the M1/M2 macrophage ratio (figure 7F). As a result, these CAR T cells exhibit enhanced cytotoxic effects against solid tumors. This novel approach not only broadens the scope of CAR T-cell therapy in solid tumors but also offers a strategy to overcome the immunosuppressive barriers typically encountered in these malignancies.

Data availability statement

Data are available in a public, open access repository. All data relevant to the study are included in the article or uploaded as supplementary information.