Augmenting CAR T Cell Functions with LIGHT

Abstract

Chimeric antigen receptor (CAR) T-cell therapy has resulted in remarkable clinical success in the treatment of B-cell malignancies. However, its clinical efficacy in solid tumors is limited, primarily by target antigen heterogeneity. To overcome antigen heterogeneity, we developed CAR T cells that overexpress LIGHT, a ligand of both LTβR on cancer cells and HVEM on immune cells. LIGHT-expressing CAR T cells displayed both antigen-directed cytotoxicity mediated by the CAR and antigen-independent killing mediated through the interaction of LIGHT with LTβR on cancer cells. Moreover, CAR T cells expressing LIGHT had immunostimulatory properties that improved the cells’ proliferation and cytolytic profile. These data indicate that LIGHT-expressing CAR T cells may provide a way to eliminate antigen-negative tumor cells to prevent antigen-negative disease relapse.

Introduction

Adoptive transfer of T cells engineered to express a chimeric antigen receptor (CAR) on their surface is an area of growing interest in cancer immunotherapy. Six products have been approved, including autologous CD19-directed and BCMA-directed CAR T cells for the treatment of B-cell acute lymphoid leukemia (B-ALL) and multiple myeloma (MM), respectively (1). In these malignancies, CAR T cell targets are typically lineage markers such as CD19, CD22, and BCMA. Developing CAR T cells for solid tumors is more challenging partly because solid tumor antigens are heterogeneous; antigen-specific CAR T cells become ineffective when antigen-negative tumor cells escape CAR T cell–mediated cytolysis and reestablish a tumor mass (2) that is non-responsive to the adoptively transferred CAR T cells, leading to disease relapse. An effective and durable antitumor response might instead be achieved by further modifying the CAR vector to not only express the CAR, but also other biologically active molecules, such as cytokines, chemokines, or costimulatory molecules (3). We investigated the activity of CAR T cells engineered to additionally express LIGHT, also known as tumor necrosis factor superfamily member 14 [TNFSF14], to augment CAR T-cell function.

As a member of the TNF family of cytokines, LIGHT expression is primarily expressed on activated T cells, activated natural killer (NK) cells, and immature dendritic cells (DCs) (4). Like most TNF superfamily ligands, the active form of LIGHT functions as a homotrimer when binding to its two known receptors: Herpes virus entry mediator (HVEM) and lymphotoxin-β receptor (LTβR) (5). LIGHT signaling via these receptors is reported to be cell-type dependent, but both interactions have been implicated in immune-related tumor biology.

LIGHT–HVEM interactions are reported to be responsible for the immune-stimulatory function of LIGHT. HVEM expressed on other lymphocytes, NK cells, smooth muscle cells, and epithelial cells can trigger the co-stimulatory functions of LIGHT to activate and expand T and NK cells, and to induce maturation of DCs (6,7). LIGHT is also responsible for the NK–DC crosstalk that occurs in the priming of de novo antitumor response (8). In addition, the costimulatory effect of the LIGHT–HVEM interaction is independent of CD28-derived costimulation provided in the CD28-based CAR T-cell designs (9,10). Moreover, the LIGHT–HVEM interaction induces T cells and NK cells to produce elevated levels of proinflammatory cytokines, such as interferon-gamma (IFNγ) and granulocyte-macrophage colony-stimulating factor (GM-CSF), a phenotype commonly associated with enhanced antitumor response (11). Therefore, the incorporation of LIGHT within the CAR vector may lead to augmentation of CAR T-cell activation, proliferation, and persistence.

LIGHT is further reported to bind to LTßR found on the surface of a variety of epithelial, stromal, and myeloid cells, as well as immature DCs, but not lymphocytes. Although it is not well characterized, the LIGHT–LTßR interaction may play a role in the development and maintenance of lymphoid structures (12). In the context of antitumor immunity, the LIGHT–LTßR signaling axis appears to be important for pro-apoptotic processes and increases the susceptibility of cancer cells to immune responses (13). More to this effect, LIGHT enhances antitumor efficacy by sensitizing tumor cells to cytokines such as IFNγ (14) and has been reported recently to play a role in repairing chaotic or dysregulated tumor vasculature and assisting effector cell trafficking and infiltration into solid tumors (15,16).

These data suggest that a multifaceted phenotype for LIGHT could enhance the efficacy of CAR T-cell therapy. LIGHT-CAR T cells may sensitize antigen-negative tumor cells to proinflammatory cytokine, eliciting a broader cytolytic effect than CAR T cells targeting one tumor antigen. We report here our study of LIGHT overexpression on CAR T cells, a strategy with the promise of eliminating low-density and antigen-heterogeneous populations of tumor cells and preventing antigen-negative relapse.

Materials and Methods

Animal models

All experiments were performed in accordance with the MSK Institutional Animal Care and Use Committee (IACUC) approved protocols (MSK #00–05-065). For xenogeneic studies, NOD-Prkdc^em26Cd52Il2rg^em26CD22/NjuCrl, Coisogenic Immunodeficient (NCG) mice were purchased from Charles River Laboratories and subsequently bred and housed under specific-pathogen-free (SPF) conditions in the animal facility of MSK. For all xenograft experiments, 6- to 12-week-old gender-matched mice were used. Tumors were engrafted with retrovirally transduced green fluorescent protein (GFP)-firefly luciferase transgene, and mice were imaged via bioluminescence to confirm equal tumor load and randomized to different treatment groups 1 day before CAR T-cell treatment. For the xenograft model of PDAC, 2×10⁶ AsPC1 or MIAPACA2 PDAC cells expressing GFP-firefly luciferase were inoculated subcutaneously 14 days prior to CAR T-cell treatment. Mice were treated with 1×10⁶ CAR T cells for AsPC1 and 2×10⁶ CAR T cells for MIAPACA2 intravenously 14 days after tumor inoculation. Mice were euthanized when tumor volume exceeded 1,500 mm³ by caliper measurements, when tumor growth led to a 20% reduction in body weight, or when mice suffered from hind limb paralysis and other signs of graft-versus-host disease (GvHD). The investigator was blinded when assessing the outcome. For patient-derived xenograft (PDX) models of PDAC, NCG mice were inoculated with 2×10⁵ cells orthotopically (intrapancreatic) 7 days prior to treatment with 4×10⁵ CAR T cells. For syngeneic mouse models, C57BL/6J mice (RRID:IMSR_JAX:000664) were purchased from Jackson Laboratory. 6 to 12 weeks old gender-matched mice were used to assess the toxicity profile of LIGHT-CAR T cells in an immunocompetent model.

Cell Lines

293-Glv9-packaging cells (a gift from Michel Sadelain, MSK, New York, NY, in 2018) were maintained in DMEM (Dulbecco’s Modified Eagle Medium, ThermoFisher) with high-glucose supplemented with 10% heat-inactivated fetal bovine serum (FBS, Atlanta Biologicals), nonessential amino acids (Atlanta Biological Flowery Branch), 2 mM L-glutamine (Invitrogen), and 1% penicillin/streptomycin (Invitrogen). Gpg29 fibroblasts (H29) (a gift from Michel Sadelain, MSK, New York, NY, in 2014) The U937 human acute leukemia line (a gift from Ross Levine, MSK, New York, NY, in 2018, RRID:CVCL_0007); the AsPC1 pancreatic adenocarcinoma cell line, the CAPAN2 pancreatic adenocarcinoma cell line, and the MIAPACA2 pancreatic adenocarcinoma cell line (gifts from Jan Grimm, MSK, New York, NY, in 2021); the PATU8988t and the Panc1 pancreatic adenocarcinoma cell line (gifts from Nancy Du, Weill Cornell, New York, NY, in 2022); the JMN cell line (a gift from Tao Dao, MSK, New York, NY, in 2022); the SW620 and MDA-MB-231 cell lines (Purchased from ATCC, CCL-227 and HTB-26, respectively); and the OCI-AML2 (DSMZ ACC99), SET-2 (DSMZ ACC608), A-375 (ATCC CRL-1619), and HT-29 (ATCC HTB-38) cell lines were modified to express GFP-firefly luciferase using the 293-Glv9-packaging cells, which produce murine-leukeumia gamma-retrovirus (MLV) to transduce the cancer cells. GFP-firefly Luciferase (gLuc) expressing cancer cell lines were used to detect cancer cells in vitro and in vivo by bioluminescence. Overexpression constructs for mesothelin and LTβR (Panc1 and MIAPACA2) were generated with the 293-Glv9-packaging system as well. The mouse cancer cell lines EL4 (ATCC TIB-39) and B16-F10 (CRL-6475) were transduced with VSV-G retroviral supernatants containing Moloney murine leukemia virus-pseudotyped retroviral particle produced by Phoenix-ECO cell line (ATCC CRL-3214). The human endothelial cell line HUV-EC-C (ATCC HUVEC) was maintained in F-12K medium (ThermoFisher, #21127022) supplemented with 5 mL of a 10mg/mL stock heparin solution (Sigma, #H3393), 10% heat-inactivated FBS (Atlanta Biologicals), and 500μL of 30mg/mL endothelial cell growth supplement (Sigma, E2759). All cancer cell lines were maintained in RPMI-1640 medium (Thermo Fisher Scientific) or DMEM with 10% heat-inactivated FBS nonessential amino acids (Atlanta Biological Flowery Branch), 10mM HEPES (hydroxyethyl piperazineethanesulfonic acid, Invitrogen), 2mM L-glutamine (Invitrogen), 1% penicillin/streptomycin (Invitrogen), and 11mM glucose (Invitrogen). Cancer lines were sorted by fluorescence-activated cell sorting (FACS) based on the high expression of GFP. Cell lines were authenticated with short tandem repeat (STR) profiling and routinely tested for potential mycoplasma contamination every month (Lonza Mycoalert Mycoplasma Detection Kit). Cancer cell lines were kept at low-passage number (<30) before discarding and thawing a new vial.

Generation of retroviral constructs

Plasmids encoding the CAR constructs in the SFG gamma-retroviral vector (17) were transfected into gpg29 fibroblasts (H29) with human signaling domains. The calcium phosphate (CaPO4) ProFection Mammalian Transfection System (Promega) was used according to the manufacturer’s suggested protocol to generate vesicular stomatitis virus G-glycoprotein-pseudo typed (VSV-G) retroviral supernatants (RRID:Addgene_164440). The VSV-G retroviral supernatants were then used to transduce stable retroviral producer line 293-Glv9. The SFG gamma-retroviral vector was cloned by Gibson Assembly (New England Biolabs) using a designed gBlocks gene fragment (Integrated DNA Technologies) that includes anti-human CD371 (scFv-B10H4L clone, No. PCT/US2020/050386), anti-human mesothelin ScFv (scFv-MF-T WO2009068204A1), or anti-CD19 (scFv-SJ25C1, PCT/US7446190B2), and Myc-tag sequence (EQKLISEEDL), human CD28 transmembrane and intracellular domain, human CD3ζ intracellular domain without the stop codon, P2A-self cleaving peptide, and the human LIGHT protein. In addition, VSV-G retroviral supernatants were used to construct stable Moloney murine leukemia virus-pseudotyped retroviral particle-producing Phoenix-ECO cell lines.

Isolation and culture of primary human T cells

Regular buffy coats containing peripheral blood from de-identified healthy donors were collected by and purchased from the New York Blood Center under an IRB-exempt protocol. All donors provided informed consent. Peripheral blood mononuclear cells (PBMC) were isolated from buffy coats using Lymphoprep (Stemcell) gradient centrifugation. Purified T cells were isolated by magnetic negative selection using the EasySep Human T Cell Isolation Kit (Stemcell). Immediately after isolation, T cells were resuspended in T-cell medium (RPMI-1640+10% heat-inactivated FBS + 2mM L-Glutamine+ 1% Penicillin/streptomycin) in addition to IL-2 (100IU/mL) (Proleukin, purchased from MSKCC Pharmacy) and CTS CD3/CD28 Dynabeads at a bead to cell ratio of 1:2 (Thermo Fisher Scientific). Forty-eight hours after the initial expansion, T cells were spinoculated with retroviral supernatant collected from 293Glv9 retroviral packaging cells on RetroNectin-coated plates (Takara Bio) for 2 consecutive days. Transduction efficiency was determined by flow cytometric analysis using anti-myc tag. All experiments were normalized for CAR⁺ viable cells.

Mouse T-cell isolation and retroviral transduction

C57BL/6J mice (Jackson Laboratory #000664) were euthanized and their spleens were harvested. Following tissue dissociation and red blood cell lysis. CD3⁺ T cells were enriched via negative selection using the EasySep Mouse T Cell Isolation Kit (StemCell). Cells were then expanded in vitro by culturing in RPMI-1640 supplemented with 10% heat-inactivated FBS, nonessential amino acids, 1mM sodium pyruvate, 10mM HEPES, 2mM L-glutamine, 1% penicillin/streptomycin, 11mM glucose, 2μM 2-mercaptoethanol, 100 IU of recombinant human IL-2 (Prometheus Therapeutics & Diagnostics), and anti-CD3/28 Dynabeads (Life Technologies) at a bead:cell ratio of 1:2. 24 h and 48 h after initial expansion, T cells were spinoculated with viral supernatant collected from Phoenix-ECO cells as described previously (18). After the second spinoculation, cells were rested for one day and then used in adoptively transfer studies.

Short-term quantitative cytotoxicity assay

The short-term cytotoxicity of CAR T cells was assessed by a luciferase-based coculture assay with target cells expressing firefly-luciferase. Target tumor cells (5×10⁴) and effector CAR T cells were incubated with varying (E:T) ratios in triplicates in white/black-walled 96-well plates (Corning) in a total volume of 200 μL of T-cell media. Target tumor cells were plated alone at the same cell density to determine the maximal luciferase signal without CAR T cells. After 24 hours or 72 hours, depending on the target tumor cell type, 75 ng of D-luciferin (Gold Biotechnology) was dissolved in 50 μL of PBS and added to each well using multichannel pipettes. The bioluminescence reading of each well was detected using a Spark plate reader (Tecan) and quantified using the SparkControl software (Tecan). Percent cell lysis was determined by the proportion of the sample signal divided by the max signal (target tumor cells alone).

Repetitive antigen stress-test assay (proliferation assay)

CAR T cells were co-cultured with target tumor cells at an E:T ratio of 4:1 in 12-well triplicates. After 5 days, the cells were collected, and a portion of the population was stained for CAR expression and analyzed via flow cytometry. CAR T cells were counted via myc tag, while tumor cells were evaluated as GFP⁺ populations. CAR T cells were replated back to the original E:T ratio with new tumor target cells for subsequent rounds of coculture. Total fold expansion was calculated by multiplying the fold expansion in each round of stimulation.

Adoptive transfer of CAR T cells

For tracking tumor studies, immunocompromised mice were inoculated intravenously (IV). with firefly luciferase-expressing tumor cells on day 0. Bioluminescence imaging (BLI) used the IVIS Spectrum in vivo imaging system with Living Image software (RRID:SCR_014247) (PerkinElmer) for the acquisition of imaging datasets. Mice with equal tumor burden were randomized to different cohorts at the time of CAR T-cell treatment. CAR T cells (1–2 × 10⁶) were administered IV in subcutaneous flank models after 14 days of PDAC tumor engraftment. 4 X10⁵ CAR T cells were administered IV in orthotopic intrapancreatic PDX models 7 days after tumor engraftment. BLI or caliper measure to track tumor burden was performed weekly after CAR T-cell treatment.

In vitro T cell–stimulation assay

To assess the costimulation potential of LIGHT, ULTRA-LEAF anti-CD3 (OKT3, BioLegend) resuspended in 50 μL of sterile PBS was plated on a non-tissue coated 96-well plate overnight at 4°C. Plates were washed 2x with sterile PBS before the addition of healthy donor T cells isolated from PBMC (EasySep Human T Cell Isolation Kit, Stem Cell). Media containing recombinant human IL-2 (100 IU/mL) and varying concentrations of soluble recombinant LIGHT (RNDsystems) and recombinant hLIGHT-FC (Sinobiological) was added to assess for an increase in cell number and expression of IL2RA/CD69 via flow cytometry.

Cytokine secretion profiling (Luminex)

After co-incubation of cancer cells with CAR T cells for 24 hours at an E:T ratio of 25,000:25,000 in 96-well round bottom plates in 200 μL of T-cell media, culture supernatants were collected and analyzed. Cytokine detection was done using the MILLIPLEX MAP Human Cytokine/Chemokine, Premixed 9 Plex Kit (Millipore) and the FLEXMAP 3D system (Luminex).

Apoptosis assay

Apoptosis induction was tested with the human Proteome Profiler apoptosis antibody array kit (R&D Systems, Wiesbaden, Germany) as described by the manufacturer after coculturing AsPC1 PDAC cells with Meso-DEL, Meso-28z, and Meso-28z-LIGHT CAR T cells for 48 hours at an E:T ratio of 1:1. Interferon-gamma (IFN-γ) was purchased from Peperotech (300–02) to assess the cancer cells’ susceptibility to interferon-mediated killing.

CRISPR–Cas9 knockout of LTβR, HVEM, and CD371 in tumor cells and T cells

U937, AsPC1, and MIAPACA2 cells were transfected by electrotransfer of modified Cas9 mRNA (tri-link) and gRNA (Synthego) using an AgilePulse MAX system (Harvard Apparatus). Cells (2×10⁵) were mixed with 5 μgs of Cas9 mRNA and 5 μgs of gRNA into a 2mm cuvette. Following electroporation, cells with electroporation buffer were transferred into media and incubated at 37°C, 5% CO₂ overnight. Cells were spun down and fresh culture media was added. After 72 hours, the knockout efficiency was assessed by surface expression of the target molecule via flow cytometry. The bottom 10% of the knockout populations were sorted out and expanded to generate a bulk cell line encompassing no expression of the target antigen. The following gRNAs wre

CD371 gRNA: 5’ UGC UGG ACG CCA UAC ATG AG

LTβR gRNA: 5’ UGG UUC UCC GAC GCA UAU GG

mLTβR gRNA: 5’ UGG GCG GUA UAU GUC UGU GG

HVEM gRNA: 5’ AAG GAG GAC GAG UAC CCA GU

Flow cytometry and FACS sorting

Flow cytometric analyses were performed using a Beckman Coulter Gallios or a Thermo Fisher Attune NxT flow cytometer and the data were analyzed with FlowJo_v10.8.1. DAPI (0.5mg/mL, Sigma-Aldrich) or a LIVE/DEAD Fixable Violet Dead Cell Stain Kit (Thermo Fisher) were used to exclude dead cells in all experiments. Human TruStain FcX Fc Receptor Blocking Solution (BioLegend) was used to block the non-specific binding of antibodies via Fc receptors. The following anti-human antibodies were used for flow cytometry: anti-CD3e (OKT3/UCHT1 Biolegend), anti-CD4 (SK3 Biolegend), anti-CD8 (SK1 Biolegend), anti-CD11b (M1/70 Biolegend), anti-CD19 (1D3 Biolegend), anti-CD25 (BC96, M-A251 Biolegend), anti-CD45 (HI30/2D1 Biolegend), anti-CD69 (L78/FN50 Biolegend), anti-CD80 (2D10 Bioelgend), anti-CD86 (IT2.2 Biolegend), anti-CD371 (50C1 Biolegend), anti-HLA DR (L243 Biolegend), anti-myc (9B11 Abcam), anti-TIM3 (F38–2E2 Biolegend), anti-LAG-3 (3DS223H Biolegend), anti-PD1 (J105 Biolegend), anti-LIGHT (T5–39 Biolegend), anti-HVEM (122 Biolegend), anti-LTβR (31G4D8 Biolegend), CellTrace CFSE (C34554 Thermo Fisher), and CellTrace Far Red dye (C34564 Thermo Fisher).

Evaluation of mesothelin surface densities

The surface densities of mesothelin molecules on the cancer cells were examined by standardized flow cytometry using a commercial quantitative analysis kit, QIFIKIT^® (Agilent) following the user manual provided by the manufacturer of the kit. Briefly, we set up and optimized the voltages for forward-scattering (FSC), side-scattering (SSC), and BL1 fluorescence channel (for fluorescein isothiocyanate, FITC) on an Attune NxT Flow Cytometer (Invitrogen) using the set-up beads (QIFIKIT^®). Next, following standard staining and washing procedures, we stained the calibration beads (QIFIKIT^®), a combination of 5 populations of beads bearing different known numbers of mouse antibodies on their surfaces, with FITC-conjugated goat anti-mouse immunoglobulin F(ab’)2 fragment. Then, using flow cytometry, the mean fluorescence intensities (MFIs) of the calibration beads of known surface densities, represented by antigen binding capacity (ABC) values, were recorded, from which a standard curve between the MFI and ABC values was constructed.

ELISA

Supernatants from CAR T cells (3 × 10⁶/well) were collected after 48 hours of culture. ELISA was performed using the LIGHT/TNFSF14 Human ELISA Kit (Invitrogen) following the user manual provided by the manufacturer of the kit.

Sample Preparation for cellular indexing of transcriptomes and epitopes by sequencing (CITEseq)

CAR T cells were sorted for myc expression to generate a pure CAR T-cell population for convenience of downstream sequencing analysis. After 48-hours co-culture with cancer cells, the CAR T-cell population was sorted out and cells were individually resuspended in 100 μL of cell staining buffer (BioLegend) with 5 μL of Human TruStain FcX Fc blocking reagent (BioLegend). Cell suspensions were then incubated at 4°C for 15 minutes, during which the antibody pool was prepared using 1 μg of each TotalSeq-C Human Universal Cocktail V1.0 antibodies (BioLegend). In addition, Human Totalseq Hashtag 1–4 (BioLegend) was used to stain individual donor CAR T cells for each condition. After 30 minutes incubation at 4°C, cells were washed twice with 1 mL of PBS 1X and finally resuspended in PBS at 1 × 10⁶ cells/ml for the downstream loading. For each condition, the 4 individual donors were pooled together at equal proportion to generate a final sample for CITEseq submission. Final concentration for submission was ~1,000 cells/uL, minimum 50 uL sample.

Cells were stained with Trypan blue and a Countess II Automated Cell Counter (ThermoFisher) was used to assess both cell number and viability. Following quality control (QC), the single cell suspension was loaded onto Chromium Next GEM Chip K (10X Genomics PN 1000286) for gel beads in emulsion (GEM) generation, cDNA synthesis, cDNA amplification, and library preparation of 10,000 cells proceeded using the Chromium Next GEM Single Cell 5’ Kit v2 (10X Genomics PN 1000263) according to the manufacturer’s protocol. cDNA amplification included 13 cycles and 50 ng of the material was used to prepare sequencing libraries with 14 cycles of polymerase chain reaction (PCR). Indexed libraries were pooled equimolar and sequenced on a NovaSeq 6000 in a PE28/88 run using the NovaSeq 6000 S4 Reagent Kit (200 cycles) (Illumina). An average of 22,500 reads was generated per cell. Amplification products generated using the methods described above included both cDNA and feature barcodes tagged with cell barcodes and unique molecular identifiers. Smaller feature barcode fragments were separated from longer amplified cDNA using a 0.6X cleanup using aMPure XP beads (Beckman Coulter catalog # A63882). Libraries were constructed using the 5’ Feature Barcode Kit (10X Genomics PN 1000256) according to the manufacturer’s protocol with 8 cycles of PCR. Indexed libraries were pooled equimolar and sequenced on a NovaSeq 6000 in a PE28/88 run using the NovaSeq 6000 S4 Reagent Kit (200 cycles) (Illumina). An average of 82 million paired reads was generated per sample.

CITE-seq analysis

FASTQ sequencing reads were processed using the Cell Ranger pipeline, which extracts cell barcodes, unique molecular identifiers (UMI), and cDNA reads or antibody barcodes, aligns cDNA reads to the human GRCh38 reference genome and generates gene and antibody UMI count matrices. Resultant-filtered sparse count matrices were loaded into R as a Seurat object. HTODemux was performed to assign cell barcodes to specific patient samples and droplet types ((19)), and ambiguously assigned cells (i.e. doublet, unmapped, or negative) were removed.

Cells that passed the following QC filters were included in downstream analysis: (1) singlets identified by cell hashing, (2) cells with >200 and <5,000 detected genes (outliers may represent empty droplets, low-quality cells, doublets, or multiplets), (3) cells with <20,000 cDNA UMI and <4,000 antibody-derived tags (ADT) UMI (outliers may represent doublets or multiplets or cells with aberrant clumps of antibodies), and 4) cells with <5% mitochondrial gene expression (extensive mitochondrial contamination is often found in low quality or dying cells). In total, 28,855 cells across 16 samples (4 conditions with 4 biological replicates each) passed QC and were included in downstream analysis.

The RNA data was normalized using SCTransform (20) and integrated across conditions using Seurat, as previously described (21). The ADT data was normalized using centered log-ratio transformation across cells. Dimensional reduction, identification of multimodal neighbors, UMAP visualization based on a weighted combination of RNA and ADT data, and graph-based clustering were performed using Seurat’s weighted nearest neighbor workflow (22). The clusters were annotated based on conserved gene and surface protein markers identified across conditions using the FindConservedMarkers function as well as previously known markers of T cell type, proliferation, activation, cytotoxicity, cytokines, and dysfunction.

Differential gene and surface protein expression analyses between subsets of interest were performed using Seurat’s FindMarkers function, which utilizes Wilcoxon rank-sum test. Gene set enrichment analysis (Gene Ontology [GO]) was performed using the fgsea package. Density plots grouped by gene sets were generated by using the frequency of fold change values per gene within each set. Additional functional analyses of differentially expressed genes were performed using the Human Molecular Signatures Database (MSigDB) hallmark gene sets from 2020 in the EnrichR (RRID:SCR_001575) package. All significant genes (adjusted P-value <0.05) were further evaluated using the transcription factor enrichment analysis tool ChIP-X Enrichment Analysis 3 (ChEA3) (23).

The expression of a custom gene set associated with cytotoxicity (GZMA, GZMB, GZMH, GZMM, GZMK, NKG7, GNLY, PRF1) was scored in single cells using AddModuleScore, which calculates the average expression levels of all the genes in a given gene set and then subtracts the average expression levels of control gene sets (24). All genes were binned based on their average expression, and 5 control genes were randomly selected from each bin. This method controls expression of the gene set for differences in cell quality and library complexity across single cells. The Seurat’s Featureplot function was used to visualize expression of this cytotoxicity gene set across clusters on the multimodal UMAP.

Establishment of patient-derived tumor xenografts and derivation of PDX cells

Fresh tumor specimens were multiregionally sampled from 14 liver metastases of pancreatic cancer autopsy, as approved by MSKCC institutional review board–approved protocols #15–149 and #15–021. Tissues were immediately rinsed with sterile saline, chopped into 2–4 mm pieces and immediately transplanted subcutaneously in the flanks of NSG mice and followed for PDX development. Primary metastases tumor as well as the PDX tumors were also chopped into 1–2mm pieces with collagenase Type IV (Stem Cell Technologies/ 0.5 mg/mL) for 1 hr at room temperature. The cells were sieved through 100-micron (Fisher Brand) nylon mesh and collected by centrifugation. The cells were washed twice in serum free RPMI medium and were immediately injected (with 1:1 dilution in Matrigel) into the NSG mice flanks subcutaneously or orthotopically into the pancreas under our IACUC-approved protocol (14–02-002). A portion of the cells was also plated on tissue culture coated plates to passage the primary cells for 2–4 passages.

For orthotopic placement of the patients’ cells, an injection of 2 mg/kg of meloxicam was given subcutaneously for pre-emptive analgesia immediately after the animal was anesthetized. A small volume (0.1 cc) of local anesthetic agent, such as bupivacaine (Marcaine 0.25%), was infiltrated into the tissue adjacent to the intended incision line. The skin was then painted with a 10% povidone-iodine (Betadine^®) or chlorhexidine (Nolvasan^®) solution. Sterile scissors were then used to make a 0.5–1 cm parasagittal incision through the abdominal musculature over the spleen. The injection was made into the pancreatic parenchyma below the capsule using a 28G needle injecting a maximum volume of 50 μl. The spleen and pancreas were gently replaced within the peritoneal cavity. The muscle layer is closed using sterile absorbal suture (e.g., Vicryl) of the appropriate diameter in a simple interrupted pattern. Skin edges were closed with sterilized wound clips (Autoclips) or with a monofilament absorbable suture (e.g. Monocryl) of the appropriate diameter in a subcuticular pattern.

Ultrasound Imaging and Quantification

High-contrast ultrasound imaging was performed on a Vevo 2100 System with an MS250 13- to 24-MHz scan head (VisualSonics) to stage and quantify pancreas tumor burden in the orthotopic PDX model. Tumor volume was calculated with the Vevo LAB software using following equation: $\text{V}=\frac{\left(\frac{1}{6}\right)\pi L{W}^2}{1000}$

Necropsy and Histopathology

In the immunocompent model that assesses for potential toxicity associated with LIGHT-CAR T cells, LIGHT-CAR T-cell treated C57BL/6J mice were euthanized with CO₂. Following gross examination all organs were fixed in 10% neutral buffered formalin, followed by decalcification of bone in a formic acid solution (Surgipath Decalcifier I, Leica Biosystems). Tissues were then processed in ethanol and xylene and embedded in paraffin in a Leica ASP6025 tissue processor. Paraffin blocks were sectioned at 5 microns, stained with hematoxylin and eosin (H&E), and examined by a board-certified veterinary pathologist. Images were obtained at 20x magnification 0.40 numerical aperture using a Olympus BX45 (Olympus America Inc, Center Valley, PA, USA) microscope with a Olympus DP27 camera from the Laboratory of Comparative Pathology core at MSKCC. Images were generated using the Olympus cellSens Entry 3.1 software and processed using ImageJ/QuPath software. The scale bar represents 100 microns. The following tissues were processed and examined: heart, thymus, lungs, liver, gallbladder, kidneys, pancreas, stomach, duodenum, jejunum, ileum, cecum, colon, lymph nodes (submandibular, mesenteric), salivary glands, skin (trunk and head), urinary bladder, uterus, cervix, vagina, ovaries, oviducts, adrenal glands, spleen, thyroid gland, esophagus, trachea, spinal cord, vertebrae, sternum, femur, tibia, stifle join, skeletal muscle, nerves, skull, nasal cavity, oral cavity, teeth, ears, eyes, pituitary gland, brain.

Hematologic assays

For hematology, blood was collected from C57BL/6J naïve mice treated with CAR T cells at (Day 30 post-treatment) into tubes containing EDTA (ethylenediaminetetraacetic acid). Automated analysis was performed on an IDEXX Procyte DX hematology analyzer and the following parameters were determined: white blood cell count, red blood cell count, hemoglobin concentration, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, red blood cell distribution width standard deviation and coefficient of variance, reticulocyte relative and absolute counts, platelet count, platelet distribution width, mean platelet volume, and relative and absolute counts of neutrophils, lymphocytes, monocytes, eosinophils, and basophils.

Serum chemistry assays

For serum chemistry, blood was collected from C57BL/6J naïve mice treated with CAR T cells at (Day 30 post-treatment) into tubes containing a serum separator, the tubed were centrifuged and the serum was obtained for analysis. Serum chemistry was performed on a Beckman Coulter AU680 analyzer, and the concentration of the following analytes was determined: alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, creatine kinase, gamma-glutamyl transpeptidase, albumin, total protein, globulin, total bilirubin, blood urea nitrogen, creatinine, cholesterol, triglycerides, glucose, calcium, phosphorus, chloride, potassium, and sodium. Sodium/potassium ratio, albumin/globulin ratio were calculated.

Quantification and statistical analysis

All statistical analyses were performed using GraphPad Prism software (v9.0.0.121), RRID:SCR_002798 (GraphPad). Data points represent biological replicates and the data points for the replications are shown as the mean ± SEM or mean ± SD as indicated in the figure legends. Statistical significance was determined by paired two-tailed Student’s t-test, one-way ANOVA, or two-way ANOVA as indicated in the figure legends. The log-rank (Mantel-Cox) test was used to determine statistical significance for overall survival in mouse survival experiments. Significance was indicated with *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

Data availability.

The CITE-seq data generated in this study have been deposited in the Gene Expression Omnibus (Ascension number: GSE240974). All other data generated in this study are available within the article and its supplementary data files. Further information and requests for resources and reagents should be directed to and will be fulfilled by the corresponding authors, Renier J. Brentjens and David A. Scheinberg.

RESULTS

LTβR is necessary for LIGHT-mediated tumor cytotoxicity.

We used a γ-retroviral vector to engineer healthy donor–derived human T cells to express both a CAR and LIGHT. A CD371-directed single chain variable fragment was fused to a myc tag to allow for CAR detection, and inserted upstream of human CD28, and CD3ζ signaling domains to generate a human second-generation CAR (CD371–28z). To generate the corresponding LIGHT-modified CAR T cells, the CD371–28z construct was inserted upstream of a P2A self-cleaving element, which was itself upstream of a human LIGHT transgene (CD371–28z-LT). A negative control, non-functional CAR T cell lacking the intracellular signaling portion of the CAR and unable to elicit downstream function upon antigen recognition was also generated (CD371-Del; Fig. 1A).

Figure 1. LTβR is necessary for LIGHT-mediated tumor cytotoxicity.

(A) Schematic of CAR design with and without LIGHT. LTR, long terminal repeats; AML-targeting, Anti-CD371 ScFv. DEL construct serves as a negative control without the intracellular signaling domains.

(B) Transgene expression of the CAR constructs after retroviral transduction of primary human T cells.

(C) Healthy human donor-derived CD371-directed CAR T cells were cocultured with tumor cells expressing GFP and firefly luciferase at efferent effector: tumor ratios. 24 hours later, bioluminescence was measured and plotted as a percentage of the signal detected in a coculture of non-functional CD371-DEL CAR T cells. U937, CD371-expressing AML cell line transduced with GFP-firefly luciferase was used for tumor tracking. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(D) In vitro cytotoxicity assay of CAR T cells with U937 CD371KO. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(E) In vitro cytotoxicity assay of CAR T cells with OCI-AML3 (left) and SET2 (right). CD371-low/no expressing AML cell line transduced with GFP-firefly luciferase. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(F) Flow plots of HVEM expression in HVEM KO in U937 CD371KO AML cell lines (top). Flow plots of LTβR expression in LTβR KO in U937 CD371KO AML cell lines (bottom)

(G) In vitro cytotoxicity assay of CAR T cells with U937 CD371KO AML cancer cell lines expressing GFP, and firefly luciferase have either HVEM, LTβR, or HVEM and LTβR double KO. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

We validated the expression of our engineered CAR and LIGHT constructs in primary human T cells (Fig. 1B). LIGHT-CAR T cells were then evaluated for cytotoxicity against acute myeloid leukemia (AML) tumor cells, which endogenously express the AML-associated antigen CD371. In co-culture assays of firefly luciferase-expressing (Luc⁺) tumor cells with CAR T cells, CD371-directed CAR T cells specifically killed CD371⁺ U937 AML cell lines (Fig. 1C), and second-generation CAR T cells were as cytotoxic as LIGHT-CAR T cells against AML. In vitro cytolysis was measured by bioluminescence.

To evaluate any additive cytolysis mediated through LIGHT, which may have been concealed by CAR-mediated killing, LIGHT-CAR T cells were tested against a CD371 knockout (KO) U937 AML cell line (Fig. 1D), and against CD371^low/negative AML cell lines (Fig. 1E). We observed that LIGHT-CAR T cells outperformed second-generation CD371-CAR T cells in killing CD371^low and CD371⁻ AML cell lines. To assess whether this enhanced cytotoxicity was mediated by LIGHT, we used CRISPR to generate knockout cell lines that lacked expression of the known receptors for LIGHT, (LTβR and HVEM), in the CD371KO U937 AML cell line (Fig. 1F). Co-culture assays of these cells with the second-generation CAR T cells and LIGHT-CAR T cells demonstrated that removal of LTβR from tumor cells conferred resistance to the effects of LIGHT, whereas removal of HVEM did not (Fig. 1G). These data suggest that LIGHT-CAR T cells kill tumor cells in an LTβR-dependent manner through an unknown mechanism.

We also evaluated the cytotoxicity of CD371-LIGHT CAR T cells against various CD371⁻ solid tumor cell lines expressing LTßR (A375 melanoma, HT-29 colorectal adenocarcinoma, and PATU8988t pancreatic adenocarcinoma). Consistent with our previous findings in AML, CD371-LIGHT CAR T cells were more effective than second-generation CAR T cells in suppressing the outgrowth of these CD371⁻ cell lines (Fig. 2A). To determine whether another feature of CD371-CAR T cells explains their augmented activity, we developed additional CAR T-cell constructs directed against mesothelina widely overexpressed tumor-associated antigen found on a variety of cancers (25,26). We also developed an additional construct, termed Meso-Del-LT, which lacked the intracellular signaling portion of the CAR but retained the LIGHT transgene (Fig. 2B). CD19-directed CAR T constructs, 1928z and 1928z-LIGHT (Supplemental Fig. S1A) were also designed as irrelevant antigen targeting CAR T-cell controls. The expression of both the mesothelin-directed CAR and LIGHT in primary human T cells was confirmed by flow cytometry (Fig. 2C). As pancreatic ductal adenocarcinoma (PDAC) is composed of heterogeneous populations of mesothelin-positive cells, we used multiple PDAC cell lines with varied mesothelin expression and assessed their sensitivity to CAR and LIGHT-modified CAR T cell–mediated tumor lysis via in vitro cytotoxicity assay. Consistent with our previous findings (Fig. 1), second-generation CAR T cells and LIGHT-CAR T cells had similar cytotoxic effects in a cell line with high target-antigen density (Fig. 2D), but LIGHT augmented CAR T cells outperformed second-generation CAR T cells against cells with low target-antigen density (Fig. 2E, 2F). We tested the mesothelin-directed CAR T cells against a variety of mesothelin-positive cell lines and observed similar results with LIGHT-CAR T cells displaying increased cytotoxicity as compared to the corresponding second-generation CAR T cells (Supplemental Fig. S1B–S1D). PDAC cell lines were stained for HVEM expression; given that the cells lines stained low/negative for HVEM, we concluded HVEM expression on tumor cell is not the mediator of the enhanced cytolysis observed with LIGHT-CAR T cells (Supplemental Fig. S1E). Panc1 was used as a negative control and displayed no killing without expression of mesothelin and LTßR (Fig. 2G). As Panc1 is negative for both mesothelin and LTβR, we generated cell lines that overexpressed mesothelin alone, LTβR alone, and a cell line with double overexpression to better evaluate the LIGHT-mediated killing mechanism (Supplemental Figure S1F). When comparing the cytotoxicity of LIGHT-CAR T cells to second-generation CAR T cells against different PDAC cell lines, it was observed that LIGHT-CAR T cells exhibited superior cell-killing activity when the target antigen for the CAR was not present in high abundance (Fig. 2H).

Figure 2. LIGHT-CAR T cells eliminate an antigen-heterogeneous population of PDAC.

(A) In vitro cytotoxicity of CAR T cells was assessed using a luciferase killing assay. Various solid tumor cancer cells lines (A375- melanoma; HT29 – colorectal; Patu_8988t – pancreatic ductal adenocarcinoma) were transduced with GFP and firefly-luciferase for tumor tracking. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± SEM.

(B) Schematic of CAR design with and without LIGHT. LTR, long terminal repeats; mesothelin-directed CAR T cell, Anti-mesothelin ScFv. DEL construct serves as a negative control without the intracellular signaling domains.

(C) Transgene expression of the CAR construct after retroviral transduction of primary T cells.

(D) In vitro cytotoxicity assay of CAR T cells with CAPAN2 using luciferase killing assay. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(E) In vitro cytotoxicity assay of CAR T cells with AsPC1 using luciferase killing assay. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(F) In vitro cytotoxicity assay of CAR T cells with MIAPACA2 using luciferase killing assay. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(G) In vitro cytotoxicity assay of CAR T cells with Panc1 using luciferase killing assay. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± standard error of the mean (SEM).

(D-G) Expressions of mesothelin and LTβR with the corresponding PDAC cell lines are shown below.

(H) Quantitative determination of cell surface antigen (mesothelin) of various PDAC cell lines are listed with the corresponding tumor lysis % of both second-generation CAR T cell (Meso-28z) and LIGHT-CAR T cell (Meso-28z-LIGHT) at 2:1 effector to tumor ratio.

We also investigated whether LIGHT in its soluble form (sLIGHT) would confer similar effects on CAR T cells as in its membrane-bound isoform. We collected supernatant containing sLIGHT secreted from LIGHT-CAR T cells, quantified it via ELISA (Supplemental Fig. S2A), and added the supernatant to second-generation 1928z or meso-28z CAR T cells when performing a cytotoxicity assay against AsPC1 and MIAPACA2. Supernatant derived from LIGHT-CAR T cells did not enhance the cytotoxicity of second-generation CAR T cells (Supplemental Fig. S2B, S2C).

In case the amount of sLIGHT produced by LIGHT-CAR T cells was insufficient to confer any effect, we also used a concentration of recombinant LIGHT (rLIGHT) previously reported to have a biological effect. This resulted in a slight increase in the cytotoxicity of second-generation CAR T cells against MIAPACA2 that was still significantly lower than that of LIGHT-CAR T cells (Supplemental Fig. S2D), suggesting that sLIGHT is not the primary mediator of augmented cytotoxicity.

LIGHT confers improved co-stimulation and a proliferative advantage to T Cells

We investigated if LIGHT provided additional costimulation to CAR T cells by interrogating its effect on the activation, proliferation, and persistence of CAR T cells. In the resting state and with target cells with high antigen density, there were no noticeable differences in the proliferation profile of second-generation and LIGHT-CAR T cells (Supplemental Fig. S2E) However, when these cells were co-cultured with the PDAC cell line, MIAPACA2, which expresses low level mesothelin, LIGHT-CAR T cells exhibited better proliferation and persistence in a repetitive antigen stimulation assay (Fig. 3A). On day 15, we evaluated these CAR T cells for activation marker IL25RA (CD25) and co-inhibitory receptors (PD-1, LAG-3, and TIM-3) by flow cytometry (Fig. 3B, 3C). Secretion of proinflammatory cytokines (e.g., IFNγ, perforin, GM-CSF, and in some cases TNF-alpha) were significantly higher in LIGHT-CAR T cells, suggesting that cytokine levels contributed to their superior cytotoxicity (Fig. 3D). In contrast, no significant increase in IL-2 or granzyme (GZM) B secretion was observed. GZMA was significantly higher in LIGHT-CAR T cells (Fig. 3D).

Figure 3. LIGHT overexpression improves CAR T cell proliferation and secretion of proinflammatory cytokines while being able to eradicate antigen-negative tumor cells.

(A) Mesothelin-directed LIGHT-CAR T cells exhibited better proliferation in a repetitive antigen stimulation assay with low mesothelin expressing PDAC cell line, MIAPACA2. CAR T cells were cocultured with tumor cells at a 4:1 effector: tumor ratio for 5 days, and then CAR T cells were taken out and put onto new tumor cells at the original E:T ratio. The total fold expansion was quantified from multiplying each round of fold expansion every 5 days. Data is representative of 3 independent experiments of 3 different human donors and data errors were analyzed with mean ± SEM.

(B) Flow cytometric analysis of activation marker, IL2RA (CD25), on various CAR T cell constructs after 15 days of coculture with MIAPACA2 PDAC cell line.

(C) Flow cytometric analysis of co-inhibitory receptors (PD-1, TIM-3, LAG-3) of various CAR T cell constructs after 15 days of coculturing with MIAPACA2 PDAC cell line.

(D) CAR T cells (1928z, Meso-28z, and Meso-28z-LIGHT) were co-cultured with 1:1 (25,000) with PDAC cancer cell lines (CAPAN2, AsPC1, MIAPACA2, and PANC1) in 200μL of media for 24 hours in 96well plates. Cells were pellet down and the supernatant was collected for multiplex cytokine profiling using the LUMINEX FLEXMAP 3D system. Data are representative of 4 separate human donors. Data errors were analyzed with mean ± SEM.

(E) Panc1 PDAC cell line with hMSLN and LTβR overexpression was stained with CellTrace Far Red dye prior to coculture with Meso-DEL, Meso-28z, and Meso-28z-LIGHT CAR T cells at 4 to 1 effector to tumor ratio for 48 hours prior to flow cytometric analysis to assess for viable tumor cells. Panc1 wildtype without hMSLN were stained with CellTrace CSFE dye prior to mixing with the previously mentioned Panc1 cell line to assess for CAR T cell’s ability to eradicate antigen-negative tumor cells. CAR T cells at 4 to 1 effector to tumor ratio were cocultured with the mixed-antigen population for 48 hours prior to flow cytometric analysis to assess for viable tumor cells.

(F) Quantification of the mesothelin negative population (Q1) from 3E. Data is representative of 3 independent experiments from cancer cells harvested after cocultured with CAR T cells from 3 different human donors. Data errors were analyzed with mean ± SEM.

(G) In vitro cytotoxicity of CAR T cells was assessed using a luciferase killing assay. MDA-MB231 cancer cell line was co-cultured with 4 different CAR T cells conditions (Meso-DEL, Meso-DEL-LIGHT, Meso-28z, Meso-28z-LIGHT) with or without 100ng/mL IFNγ at 1 to 1 effector to tumor ratio for 48 hours. Plots represent 3 independent experiments with 3 different human donors. Data errors were analyzed with mean ± SEM.

LIGHT-CAR T cells sensitize cancer cells to cytokine-mediated killing through the reduction of anti-apoptotic genes

Consistent with our previous findings in AML, LTβR KO on MIAPACA2 cell line abolished the killing advantage of LIGHT-CAR T cells (Supplemental Fig. S3A, S3B). To further investigate how the LIGHT-LTβR interaction may be conferring a cytotoxic advantage, we knocked in a truncated version of LTβR lacking the intracellular domain and noticed the killing advantage conferred by LIGHT-CAR T cells was not rescued (Supplemental Fig S3A, S3C). We then tested whether a non-functional CAR with LIGHT (Meso-DEL-LIGHT) would kill cancer cells in the presence of exogenous CD3/28 stimulation via antibody-coated beads (Dynabeads). We found that the CD3/CD28 stimulation enhanced Meso-DEL-LIGHT CAR T cells cytotoxicity against MIAPACA2 compared to meso-DEL CAR T cells. Additionally, when repeated with an LTβR KO of the same cancer cell line, the augmented cytotoxicity was abolished, further indicating that the cytotoxicity is driven by LIGHT-LTβR engagement (Supplemental Fig. S3D). To validate that the cytotoxicity advantage was due to direct engagement of LIGHT to LTβR and not by binding to HVEM on T cells, we knocked out the HVEM receptors on Meso-DEL, Meso-28z, Meso-28z-LIGHT T-cells (Supplemental Fig. S3E). Killing assays with HVEM KO CAR T cells displayed no significant differences between second-generation CAR T cells and LIGHT-CAR T cells when high-density mesothelin expression was present on target cells (Supplemental Fig. S3F). HVEM KO CAR T cells were co-cultured with low-density mesothelin target cell, MIAPACA2, and still displayed enhanced cytolysis compared to second-generation CAR T cells. (Supplemental Fig. S3G). However, HVEM KO CAR T cells seemed to proliferate less compared to wild-type CAR T cells (Supplemental Fig. S3G). To better understand the mechanism by which LIGHT-CAR T cells mediated killing, we performed a human apoptosis array (R&D Systems) on MIAPACA2 cancer cells that had been co-cultured with second-generation CAR T cells and LIGHT-CAR T cells. Among the 35 genes associated with the apoptosis signaling pathways, we looked at the most differentially expressed genes between the cancer cells treated with different CAR T cell conditions (Supplemental Fig. 3H, I, J). MIAPACA2 treated with LIGHT-CAR T cells versus second-generation CAR T cells had strikingly lower levels of Bcl-2 and Bcl-xL, which are critical pro-survival and anti-apoptotic proteins. The results were consistent with previously established research, in which the addition of IFNγ with recombinant LIGHT led to downregulation of Bcl-xL and LIGHT sensitized IFNγ-mediated apoptosis of HT-29 human carcinoma cells(14).

We performed a mixed-antigen assay to assess whether LIGHT-CAR T cells were better able to eradicate antigen-negative tumor cells or rather if they were just killing the antigen-positive cancer cells better. The Panc1 PDAC cell line, which naturally does not express mesothelin and LTβR, was used to generate cell lines with different expression of mesothelin and LTβR (Supplemental Fig. 1F). Panc1 cells expressing both mesothelin and LTβR were stained with CellTrace Far Red dye and cell viability was assessed after co-cultured with CAR T-cells. The cancer cells expressing both mesothelin and LTβR stained with far red were eradicated. In a mixed-antigen killing assay, Panc1 LTβR-expressing cells (no hMSLN) were added in a 1:1 ratio with the Panc1 cells expressing both mesothelin and LTβR. LIGHT-CAR T cells were able to completely eradicate mesothelin-negative Panc1 LTβR-expressing cells (Fig. 3E, 3F). Additionally, we tested whether adding IFNγ with LIGHT-CAR T cells to the co-culture would improve cytolytic capacity of LIGHT-CAR T cells. Indeed, when 100ng/mL of IFNγ was added to the coculture with cancer cells, there was better killing compared to CAR T cells alone. This effect was amplified by LIGHT in which even non-functional CAR T cells (Meso-DEL-LIGHT) were able to kill better than the non-LIGHT CAR T-cell counterparts (Fig. 3G). Altogether, we hypothesized that LIGHT-CAR T cells were killing better partly by sensitizing cancer cells to IFNγ-mediated cytolysis.

Single-cell multiomic profiling reveals higher expression of cytoxicity, cytokine, and chemokine genes.

We performed CITE-seq to evaluate the transcriptional state and surface protein expression of individual CAR T cells before and after co-culture with cancer cells. We generated mesothelin-directed CAR T cells with and without LIGHT (Meso-28z-LT and Meso-28z, respectively) using T cells from 4 healthy human donors and sequenced them post-sorting at baseline (t=0) and 48 hours after co-culture with cancer cells (t=48h). We performed cell hashing (19) to combine samples of the same condition (construct and timepoint) from different donors and used a panel of 130 ADTs to stain cell surface proteins. We obtained 28,855 cells across 16 samples that passed QC for downstream analysis (Supplemental Fig. S4A).

The transcriptional profiles of Meso-28z and Meso-28z-LT cells differed at baseline. LIGHT-CAR T cells expressed significantly higher levels of ACTB, CCND3, JUND, ITGB7, KLF2, and HLA-A, among many other genes (Fig. 4A), and their profiles were enriched for gene ontology (GO) terms such as T-cell activation, signaling receptor binding, and cell secretion/export compared with controls (Fig. 4B). At t=48h after co-culture, LIGHT-CAR T cells expressed higher levels of genes associated with cytotoxicity and inflammatory cytokines/chemokines, including CSF2 (Colony-stimulating factor 2/GM-CSF), GZMB, TNFRSF4 (OX40), IL25RA, CCL4, and IL13 (Fig. 4C). and their profiles were enriched for GO terms such as receptor-ligand activity, cytokine activity, and T-cell migration/chemotaxis compared with control CAR T cells (Fig. 4D) LIGHT-CAR T cells also expressed higher levels of multiple cell surface proteins to control CAR T cells, including CD71 (a marker for T-cell activation and proliferation), and CD272 (BTLA, a co-signaling molecule in the CD28 superfamily that binds to HVEM and influences the LIGHT-HVEM signaling), and expressed higher levels of additional activation markers, including IL25RA and CD69.

Figure 4. Single-cell multiomics profiling of LIGHT-CAR T cells reveals more activated cell states with higher expression of cytotoxicity genes and cytokines/chemokines upon co-culture with cancer cells.

(A) Volcano plot depicting differentially expressed genes between LIGHT-CAR T cells (Meso-28z-LT, blue) and control CAR T cells (Meso-28z, red) at rest (t0). The x-axis indicates log fold-change of the average expression between the groups. The y-axis indicates negative log of the adjusted P-value based on Bonferroni correction using all features in the dataset.

(B) Gene ontology (GO) terms enriched in differentially expressed genes in LIGHT-CAR T cells (Meso-28z-LT) compared to control CAR T cells (Meso-28z) at (t0). LIGHT-CAR T cells are enriched for expression of genes involved in T cell activation, signaling receptor binding, and cell secretion/export.

(C) Volcano plot depicting differentially expressed genes between LIGHT-CAR T cells (Meso-28z-LT) and control CAR T cells (Meso-28z) 48 hours after co-culture with cancer cells (t48).

(D) GO terms enriched in differentially expressed genes in LIGHT-CAR T cells (Meso-28z-LT) compared to control CAR T cells (Meso-28z) 48 hours after co-culture with cancer cells (t48). LIGHT-CAR T cells are enriched for expression of genes involved in receptor-ligand activity, cytokine activity, and T cells migration/chemotaxis.

(E) Heatmap displaying differentially expressed surface protein in LIGHT-CAR T cells (Meso-28z-LT) compared to control CAR T cells (Meso-28z) at rest (t0) and 48 hours after co-culture with cancer cells (t48). LIGHT-CAR T cells show higher expression of activation markers at both timepoints.

(F) Weighted-nearest neighbor (WNN) UMAP of single cells (n=28,855) integrated and clustered based on a weighted combination of RNA and antibody-derived tag (ADT). Clusters or shared cell states were annotated based on conserved gene and surface protein expression across conditions and previously known markers of T cell type, proliferation, activation, cytotoxicity, and cytokines.

(G) Expression of a custom cytotoxicity gene set (GZMA, GZMH, GZMM, GZMK, NKG7, GNLY, PRF1) in LIGHT-CAR T cells (Meso-28z-LT) compared to control CAR T cells (Meso-28z) at rest (t0) and 48 hours after co-culture with cancer cells (t48) projected onto the WNN UMAP.

(H) Violin plots comparing expression of indicated genes in distinct clusters of LIGHT-CAR T cells (Meso-28z-LT) compared to control CAR T cells (Meso-28z) 48 hours of co-culture with cancer cells (t48)

We then investigated features of enhanced cytotoxicity in subsets of LIGHT-CAR T cells. After integrating our CITE-seq data and clustering single cells using a weighted combination of RNA and ADT, we identified 10 clusters (shared cell states) across conditions and visualized them using a weighted-nearest neighbor (WNN) Uniform Manifold Approximation and Projection (UMAP) (Fig. 4E, 4F). Using conserved genes and surface proteins across conditions and previously known markers of T-cell type, proliferation, activation, cytokines, and dysfunction, as well as cytotoxicity (GZMA, GZMB, GZMH, GZMM, GZMK, NKG7, GNLY, PRF1), we annotated 9 clusters of CD4⁺ CAR T cells with variable proliferative, cytotoxic, activated, and inflammatory features and 1 cluster of CD8⁺ CAR T cells with strong cytotoxicity features (Fig. 4F and Supplemental Fig S4B–D). LIGHT-CAR T cells belonging to CD8⁺ cluster (cluster 4) and 2 CD4⁺ clusters (cluster 1 and 3) most highly expressed the predefined cytotoxicity genes at baseline and t=48h compared with controls (Fig. 4G). Specifically, LIGHT-CAR T cells had greater upregulation of GZMB and GNLY at t=48h compared with controls, and this effect was most pronounced in clusters 1, 3, and 4 (Fig. 4H). LIGHT-CAR T cells also demonstrated greater expression of IL2RA and CSF2 (GM-CSF) at t=48h, but this effect was more generalized across clusters. Overall, these findings indicate that LIGHT-CAR T cells are more activated at baseline as well as after co-culture with cancer cells and exhibit features consistent with stronger cytotoxic and effector functions upon encounter with cancer cells. These data corroborate our in vitro data showing improved functionality of LIGHT-CAR T cells compared with standard CAR T cells.

LIGHT-CAR T cells limit PDAC tumor outgrowth and confer survival benefits in xenograft models.

To evaluate the in vivo efficacy of mesothelin-directed LIGHT-CAR T cells, we established a xenograft flank model using the AsPC1 PDAC cell line (Fig. 5A), which was chosen given the excellent expansion and cytotoxic effect of LGHT-CAR T cells in vitro. Caliper measurement and BLI of tumor volume showed that LIGHT-CAR T cells controlled tumor outgrowth; and this was associated with significantly improved overall survival compared with the standard second-generation CAR T cells (Fig. 5B–D). We evaluated the tumor-infiltrating lymphocytes at the tumor site on day 14 post-CAR T treatment and noticed significantly higher numbers of LIGHT-CAR T cells in the tumor as well as overall higher expression of LIGHT (Supplemental Fig. S5A, S5B). Given this success in tumor control, we then tested another PDAC cell line, (MIAPACA2), which expresses low levels of mesothelin but exhibited susceptibility to LIGHT-mediated cytotoxicity (Fig. 2F). Consistent with our in vitro data, LIGHT-CAR T cells demonstrated superior tumor control on tumor BLI in vivo (Fig. 5E–G). Despite this, the MIAPACA2 cell line was not very aggressive, and no mice were euthanized due to tumor volume or significant weight loss. All mice were euthanized on day 64, due to the onset of graft-versus-host disease (GvHD; Fig 5H).

Figure 5. LIGHT-CAR T cells limit PDAC tumor outgrowth and confers survival benefit in subcutaneous xenograft model of PDAC.

(A) Schematics of LIGHT-CAR T cell treating a mesothelin-expressing PDAC cell line, (AsPC1), in a xenograft flank model.

(B) Tumor burden (mm³) quantified by caliper measurement ([L*W*W]/2) post-CAR T cell treatment (days).

(C) Representative bioluminescent images (BLI) show tumor growth of PDAC in untreated and CAR T cell-treated groups (various constructs) at days (D) post-CAR T cell treatment.

(D) Kaplan–Meyer plot showing mouse survival days post CAR T cell treatment (AsPC1).

(E) NCG mice were subcutaneously injected with 2 × 10⁶ MIAPACA2 PDAC cell line. Mice were randomly assigned to groups 14 days later and were infused intravenously with 2 × 10⁶ CAR-T cells. Tumor growth was assessed weekly via BLI using in vivo imaging (IVIS).

(F) Total flux (photons/second [p/s]) shows tumor burden in mice treated with various CAR T cell constructs at days post-CAR T cell treatment. Whole-body BLI via IVIS with standard error of the mean (SEM).

(G) Representative BLI images show tumor growth of MIAPACA2 in CAR T cell treated groups (various constructs) at days (D) post treatment.

(H) Kaplan–Meyer plot showing mouse survival days post-CAR T cell treatment (MIAPACA2).

Evaluation of LIGHT-CAR T cells in an orthotopic PDX model of human PDAC.

To evaluate the efficacy of LIGHT-CAR T cells in a more clinically relevant tumor model, we used an orthotopic, PDX model of PDAC. We first screened for mesothelin expression using immunohistochemistry on a variety of PDX models generated by the Memorial Sloan Kettering Cancer Center (MSK) Antitumor Core (Supplemental Fig. S5C–5E). We measured the surface expression of mesothelin by flow cytometry, performed an in vitro cytotoxicity assay, and assessed the functional activity of our mesothelin-directed CAR T cells in a PDX sample (Supplemental Fig. S5F). Consistent with our previous findings, LIGHT-CAR T cells demonstrated higher cytotoxicity compared with second-generation CAR T cells (Supplemental Fig. S5G).

The PDAC2 PDX was transduced with a luciferase transgene to allow for in vivo tracking of tumor burden by bioluminescence. The tumor was engrafted orthotopically (intrapancreatic), and after 7 days for tumor engraftment, tumor-bearing mice were treated with CAR T cells (Fig. 6A). The CAR T-cell dose was lowered to 400k to keep it consistent with the lower amount of cancer cells being injected in this intrapancreatic model. As previously observed, non-functional Meso-DEL CAR T cells lacked therapeutic efficacy, while LIGHT-CAR T cells induced a delayed progression of tumor outgrowth as compared with second-generation CAR T cells (Fig. 6B, 6C). Complete tumor eradication was not observed, perhaps owing to the lower CAR T-cell dose (400k instead of 1–2×10⁶ CAR T cells used in the previous xenograft models) used in these studies when compared to prior preclinical mouse models. However, size reduction or delayed progression of the tumor correlated with a significant survival benefit for mice treated with LIGHT-CAR T cells (Fig. 6D). We performed additional orthotopic PDX models of PDAC with PAAD_53a. PAAD_53a was engrafted for 28 days orthotopically for sufficient tumor engraftment measured via ultrasound before treatment with 2×10⁶ CAR T cells (Fig. 6E). Tumor burden was monitored weekly with ultrasound and LIGHT-CAR T cells completely eradicated tumor outgrowth and concomitantly led to longer tumor-free survival compared to the groups treated with control and second-generation CAR T cells (Fig. 6F, 6G). In conclusion, LIGHT-CAR T cells had superior antitumor response in clinically relevant models of human PDAC.

Figure 6. Anti-tumor response of LIGHT-CAR T cells in patient-derived orthotopic model of human PDAC.

(A) PDX PDAC2-Luc with Matrigel were engrafted into immunodeficient NCG mice via intra-pancreatic injection, followed by CAR T cell treatment 7 days later. Tumor growth was assessed weekly via bioluminescence imaging (BLI) using in vivo imaging (IVIS). (n=5 mice per group)

(B) Representative BLI show tumor growth in untreated and CAR T cell-treated groups at various days post-CAR T cells treatment.

(C) Total flux (photons/second [p/s]) shows tumor burden in mice treated with various CAR T cell constructs at various days post-CAR T cell treatment. Whole-body BLI via IVIS with mean ± SEM.

(D) Kaplan–Meyer plot showing mouse survival days post CAR T cell treatment (PDAC2).

(E) PDX PAAD_53a with Matrigel were engrafted into immunodeficient NSG mice via intra-pancreatic injection, followed by CAR T cell treatment 28 days days later. Tumor growth was assessed weekly via ultrasound and tumor volume was calculated with Vevo Lab software using the following equation. $\text{V}=\frac{\left(\frac{1}{6}\right)\pi L{W}^2}{1000}$

(F) Tumor burden of mice treated with CAR T cells measured with ultrasound and calculated using Vevo Labs software.

(G) Kaplan–Meyer plot showing mouse survival days post CAR T cell treatment (PAAD_53a). (n=5) mice per treatment group.

LIGHT-CAR T cells display no adverse effect in immunocompetent mouse models and non-cancerous cell types

Given that LIGHT can elicit cytotoxicity in an LTβR-dependent manner, and LTβR expression is not restricted to tumor cells, we evaluated adverse effects, including any overt toxicities or histological abnormalities, of LIGHT-CAR T cells in vivo. We used a fully immunocompetent mouse model and sought to use murine CAR (mCAR) T-cell constructs directed against murine CD19 (mCD19) as these mCAR constructs have been well characterized and validated in the development of other CAR T cell platforms. Moreover, B-cell aplasia derived from CAR T-cell activity can be used as a proxy for in vivo CAR T-cell activity in the absence of tumor. Mouse CD19 CAR T-cell constructs with and without mouse LIGHT (mLIGHT) were designed with myc tag to allow for the detection of CAR T cells (Fig. 7A). m19mt-DEL, mCD19 with myc tag but lacking the intracellular signaling portion of the CAR served as a negative control. We successfully transduced mouse T cells with our constructs and validated both CAR and mLIGHT expression (Fig. 7B).

Figure 7. LIGHT-CAR T cells display no significant toxicity in immunocompetent mouse models and non-cancerous cell types.

(A) Schematics of CAR designs with or without LIGHT. Anti-mCD19 ScFv constructs for the syngeneic model are illustrated. m19mt-DEL construct serves as a negative CAR control without intracellular signaling upon antigen recognition.

(B) Transgene expression of the CAR constructs after retroviral transduction of mouse T cells.

(C) Mouse CD19-CAR T cells were cocultured with tumor cells expressing GFP and firefly luciferase at different effector to tumor cell ratios. 1928z-LIGHT CAR T cells exhibited better cytotoxicity against CD19-negative cancer cell lines. Expression of mesothelin and LTβR with the corresponding cell lines are shown.

(D) Knockout (KO) of LTβR in the B16F10 melanoma cell line abolished the killing advantage of CD19-LIGHT CAR T cells.

(E) mCD19-CAR T cells were engrafted into immunocompetent C57/BL6 mice followed by peripheral blood collection with retro-orbital eye bleed at days 7, 14, and 30. At the day 30 endpoint, the mice were sacrificed for full necropsy for analysis of potential toxicity in various organs and tissues.

(F) Flow cytometry analysis of B cells, T cells, and myeloid cells from peripheral blood of C57/BL6 mice with CAR T cell infusion at D14.

(G) Relative weight change (percent change in initial weight) in the days following injection of various mouse CAR T cell constructs.

(H) Human CAR T cells expressing LIGHT were cocultured for 72 hours with HUVEC stained with Celltrace Far Red dye. Flow cytometry analysis displayed 2 distinct populations and no reduction in HUVEC cell number were observed with LIGHT-CAR T cells.

(I) Quantification of HUVEC count after 72 hours of coculture with CAR T cells with and without LIGHT. Plots represent 3 independent experiments with 3 different human donors.

To evaluate the cytotoxicity of our mouse mCD19 LIGHT-CAR T cells and enhanced cytolytic capacity conferred by mLIGHT, we used EL4, a mouse thymoma line that overexpresses CD19 and B16F10, a mouse melanoma cell line. B16F10 melanoma does not express CD19, but expresses mLTβR, the target receptor for mLIGHT. We observed complete elimination of the CD19⁺ EL4 cell line by bioluminescence and better tumor killing of mouse melanoma B16F10 using LIGHT-CAR T cells compared with second-generation CAR T cells (Fig. 7C). Further validating the effect of LIGHT, when we knocked-out mLTβR from B16F10, the enhanced cytolysis of LIGHT-CAR T cells was abolished (Fig. 7D).

Mouse CD19-directed CAR T cells were injected into C57/BL6 mice and peripheral blood was collected at 7-, 14-, and 30-days following CAR T-cell injection. Full necropsy assessment after day 30 for any abnormal histopathology or serum chemistry demonstrated no significant toxicity associated with LIGHT-CART cells (Fig. 7E, Supplemental Fig. S6A). Flow cytometry analysis of the peripheral blood demonstrated B-cell aplasia following CD19-directed CAR T cells with and without LIGHT resulted in B-cell aplasia, but none associated with m19mt-DEL, the non-functional CAR construct (Fig. 7F). By day 30, no toxicity (significant weight loss, paralysis, and neurotoxicity symptoms) was observed with LIGHT-CAR T cells (Fig. 7G). Lymphocytic infiltration increased slightly in some tissues, and higher globulin levels were associated with a higher inflammation state, suggesting biological effects from LIGHT in the immunocompetent mouse model (Supplemental Fig. S6B, S6C). To evaluate whether the LIGHT-CAR T cells are cytotoxic to normal cells, we stained human umbilical vein endothelial cells (HUVEC), selected as a normal cell with LTβR expression, with CellTrace Far Red (CSFR) dye before co-culture with human CAR T cells. After 72 hours, the cells were collected for flow cytometry analysis, and the amount of CSFR⁺ cells was quantified (Fig. 7H). There was no significant decrease in HUVEC numbers after co-culture with LIGHT-CAR T cells (Fig. 7I). Collectively, these data indicate that LIGHT-CAR T cells displayed no adverse effects in immunocompetent mouse models and non-cancerous cell types.

Discussion

Although CAR T cells have shown great success in the treatment of B-cell malignancies, response rates in patients with solid tumors have been underwhelming. The reasons for this include the lack of suitable target antigens, tumor heterogeneity, poor CAR T-cell trafficking, poor CAR T-cell infiltration of tumors, and an immunosuppressive tumor microenvironment. Importantly, the heterogeneous antigen expression of tumors renders antigen-specific CAR T cells ineffective, leading to antigen escape and antigen-negative disease relapse (27,28). Novel strategies to overcome antigen heterogeneity and antigen escape have the potential to improve responses to CAR T-cell therapies in both solid tumor and hematologic malignancies.

For the past two decades, investigators have explored method to deliver LIGHT as an antineoplastic therapy, including through bacterial and viral vectors, as well as fusion proteins (29–33). Tumor volume reduction and primary tumor elimination have been observed in multiple preclinical models, but LIGHT-induced, direct tumor-specific killing has not been characterized. Here, we describe the overexpression and delivery of LIGHT via CAR T-cell therapy. Prior work has demonstrated that overexpression of LIGHT in transgenic mice results in a hyper-activated T cell state, increasing the risk of spontaneous autoimmunity (4). However, CAR T cells can act as “micropharmacy” tumor-directed delivery systems to minimize systemic toxicity, while providing biologically active molecules directly to the tumor site (34–36). In addition to LIGHT-mediated tumor cytotoxicity, we show that LIGHT-CAR T cells have improved activation, proliferation, and persistence compared with second-generation CAR T cells.

In this study, LIGHT-CAR T cells eradicated antigen-heterogeneous cancer cells via an orthogonal method of tumor cell lysis. We first targeted the AML-associated antigen, CD371 (CLEC12A) on AML blasts and leukemic stem cells, which present with a relatively heterogeneous antigen expression of CD371, versus the relatively uniform expression of CD19 by B-ALL and of BCMA by MM (37). The CD371-directed LIGHT-CAR T cells exhibited enhanced cytotoxicity compared with second-generation CD371-directed CAR T cells. Moreover, augmented cytolysis occurred in both antigen low and antigen knockout settings, demonstrating that LIGHT-CAR T cells can prevent the outgrowth of antigen-negative tumor cells in vitro. We next observed similarly augmented cytotoxicity of LIGHT-CAR T cells in a variety of solid tumor cell lines, where they eliminated antigen-heterogeneous populations of cancer cells and conferred superior tumor control and a concomitant survival benefit in xenograft models of PDAC.

Additionally, we found that the LIGHT/LTßR signaling axis was important for the cytotoxicity of LIGHT-CAR T cells. Systematic deletion of known receptors for LIGHT from tumor cells abolished the augmented cytotoxicity of LIGHT-CAR T cells in LTßR-knockout cell lines. Overexpression of a truncated, non-signaling version of LTßR in the knockout cells also abolished augmented cytolysis, further validating this finding. Furthermore, we demonstrated increased activation and proliferation profiles of LIGHT-CAR T cells compared with second-generation CAR T cells and corroborated these results through single-cell multiomics data. These findings collectively suggest that LIGHT-CAR T cells have greater antitumor potency, expansion, and persistence in the context of antigen-heterogeneous malignancies.

A recent study employing genome-wide screening to identify synthetic drivers of T-cell activation and proliferation demonstrated increased proliferation and effector functions associated with overexpression of LTßR (38). This finding suggests that the signaling effects of the of LIGHT/LTßR axis are cell-type dependent and supports its use to augment T-cell function. Additionally, LIGHT’s co-stimulatory effect may be pleiotropic, acting on both the CAR T cells and other endogenous immune cells to enhance endogenous antitumor responses (39). Prior work has demonstrated that LIGHT plays additional roles in the normalization of tumor vasculature and maintenance of both tertiary lymphoid structures and high endothelial venules at tumor sites; these features could drastically enhance the effects of immune effector cell trafficking to and infiltration of tumor mass (16,40) and are positive indicators of dense lymphocytic infiltrates associated with improved antitumor responses and patient survival (41). In addition, recent work demonstrated that the utilization of LIGHT with CAR T cells enhanced chemokine secretion and led to increased T-cell trafficking and cytotoxicity by reversing the immunosuppressive tumor microenvironment (42). What we have additionally identified is the interaction of LIGHT with LTßR on cancer cells sensitized the cancer cells to cytokine-mediated cytolysis through downregulation of anti-apoptotic genes and this augmented cytotoxicity may be clinically useful in killing antigen-negative tumors cells in the context of antigen-heterogeneous disease. Together, these studies indicate that LIGHT may not only boost the effect of CAR T-cell therapy but also enhance existing endogenous antitumor responses by engaging endogenous immune cells, normalizing tumor vasculature, and facilitating the formation of tertiary lymphoid structures and high endothelial venules.

In summary, we have developed CAR T cells with LIGHT overexpression capable of robust killing of antigen-heterogeneous tumor cells. These LIGHT-CAR T cells are more activated, proliferative and secrete more proinflammatory cytokines upon antigen encounter compared with CAR T cells targeting only one specific tumor antigen. The LIGHT-mediated enhancements correlate with improved tumor control and survival in multiple in vivo models compared to non-LIGHT CAR T cells. To comprehensively interrogate the impact of LIGHT armored CAR T cells in an immunocompetent setting, we plan to employ a B16F10 melanoma model engineered to express human mesothelin and use our established anti-human mesothelin CAR T cells to evaluate the antitumor effect as well as the impact of LIGHT on other endogenous immune cells. mCD19 CAR T cells were only used to assess toxicity and not used to assess antitumor efficacy in vivo because CD19 is widely expressed in the blood compartment by mouse B cells, providing a large antigen sink that may confound the antitumor effect. Overall, this approach represents a potential therapeutic strategy to improve the effectiveness of CAR T cells targeting solid tumors, and its translation to the clinic holds promise for improving CAR T cell–mediated antitumor efficacy in the setting of both hematological and solid tumor malignancies.

Synopsis:

Tumor-antigen heterogeneity limits CAR T-cell efficacy against solid tumors. The authors overcome this by developing CAR T cells that overexpress LIGHT, which provides an antigen-independent mechanism of CAR-mediated cytolysis while simultaneously providing costimulatory properties to the T cells.