Using engineered immune cells from cord blood to target triple-negative breast cancer with strong tumor-killing ability

Animal studies were conducted under protocols approved by the UCLA Division of Laboratory Animal Medicine. Healthy donor PBMCs were obtained from the UCLA/CFAR Virology Core Laboratory and HemaCare under informed consent and in compliance with federal and state regulations; no identifying information was provided. Primary TNBC samples were collected at the Cedars-Sinai Medical Center from consented patients through an IRB-approved protocol (IRB #3032 Resistance Mechanisms in Breast Cancer) and processed.

NOD.Cg-Prkdc^SCID Il2rg^tm1Wjl /SzJ (NOD/SCID/IL-2Rγ^−/−, NSG) mice were purchased from The Jackson Laboratory (Strain #:005557; RRID: IMSR_JAX:005557), and maintained in the animal facilities of UCLA under the following housing conditions: temperature ranging from 68 °F to 79 °F, humidity maintained at 30–70%, a light cycle of On at 6:00 am and Off at 6:00 pm, and room pressure set to negative. 6–10 weeks old female mice were used for all experiments unless otherwise indicated. All animal experiments were approved by the Institutional Animal Care and Use Committee of UCLA (ARC-2013-054). All mice were bred and maintained under specific pathogen-free conditions, and all experiments were conducted in accordance with the animal care and use regulations of the Division of Laboratory Animal Medicine at the UCLA. Experimental mice were randomly assigned to treatment groups to avoid statistically significant differences in the baseline tumor burden.

The X-VIVO 15 Serum-Free Hematopoietic Cell Medium (cat. no. 04418Q) was purchased from Lonza. The StemSpan™ T Cell Generation Kit (cat. no. 09940), comprising the StemSpan™ SFEM II Medium (cat. no. 09605), the StemSpan™ Lymphoid Progenitor Expansion Supplement (cat. no. 09915), the StemSpan™ LPMS (cat. no. 09930), the StemSpan™ Lymphoid Progenitor Differentiation Coating Material (cat. no. 09925), and the ImmunoCult™ Human CD3/CD28/CD2 T Cell Activator (cat. no.10970), and MethoCult™ H4330 MethycelluloseBased Medium (cat. no. 04330) were purchased from StemCell Technologies. The CTS™ OpTmizer™ T-Cell Expansion SFM (no phenol red, bottle format, cat. no. A3705001), the RPMI 1640 cell culture medium (cat. no. MT10040CV), and the DMEM cell culture medium (cat. no. MT10013CV) were purchased from Thermo Fisher Scientific. The CryoStor ^® Cell Cryopreservation Media CS10 (cat. no. C2874) and Iscove's Modified Dulbecco's Medium (cat. no. I3390) was purchased from MilliporeSigma. The C10 medium was made of RPMI 1640 cell culture medium, supplemented with FBS (10% vol/vol), P/S/G (1% vol/vol), MEM NEAA (1% vol/vol), HEPES (10 mM), Sodium Pyruvate (1 mM), Beta-Mercaptoethanol (β-ME) (50 µM), and Normocin (100 µg/ml). The homemade D10 medium was made of DMEM supplemented with FBS (10% vol/vol), P/S/G (1% vol/vol), and Normocin (100 µg/ml). The homemade D10 medium was made of DMEM supplemented with FBS (10% vol/vol), P/S/G (1% vol/vol), and Normocin (100 µg/ml).

α-Galactosylceramide (αGC, KRN7000, cat. no. 867000) was purchased from Avanti Polar Lipids. Recombinant human IL-2 (cat. no. 200-02), IL-3 (cat. no. 200-03), IL-7 (cat. no. 200-07), IL-15 (cat. no. 200 − 15), IFN-γ (cat. no. 300-02), Flt3 ligand (Flt3L, cat. no. 300 − 19), macrophage colony stimulating factor (M-CSF, cat. no. 300 − 25), stem cell factor (SCF, cat. no. 300-07), and thrombopoietin (TPO, cat. no. 300 − 18) were purchased from Peprotech. Fetal Bovine Serum (FBS, lot no. 2087050) were purchased from Gibco and β-ME (cat. no. 1610710) were purchased from Bio-Rad. Penicillin Streptomycin-Glutamine (P/S/G, cat. no. 10-378-016), MEM nonessential amino acids (NEAA, cat. no. 11-140-050), HEPES Buffer Solution (cat. no. 15630080), and Sodium Pyruvate (cat. no. 11360070) were purchased from Gibco. Normocin was purchased from InvivoGen (cat. no. NC9390718).

A parental lentivector, pMNDW, was utilized to construct the lentiviral vectors employed in this study [38, 39]. The 2 A sequences derived from foot-and-mouth disease virus (F2A), porcine teschovirus-1 (P2A), and thosea asigna virus (T2A) were used to link the inserted genes to achieve co-expression. The Lenti/iNKT-MCAR-IL-15 vector was generated by inserting into the pMNDW parental backbone a synthetic tetracistronic gene encoding human iNKT TCRα-F2A-iNKT TCRβ-P2A-MCAR-T2A-IL-15 (MCAR denotes an MSLN-specific CAR, and IL-15 represents the secreted form of human interleukin-15). The Lenti/iNKT-MCAR vector was generated by inserting into the pMNDW parental backbone a synthetic tricistronic gene encoding human iNKT TCRα-F2A-iNKT TCRβ-P2A-MCAR. The Lenti/iNKT-CAR19-IL-15 vector was generated by inserting into the pMNDW parental backbone a synthetic tetracistronic gene encoding human iNKT TCRα-F2A-iNKT TCRβ-P2A-CAR19-T2A-IL-15 (CAR19 denotes a CD19-specific CAR). The Lenti/iNKT-IL-15 vector was generated by inserting into the pMNDW parental backbone a synthetic tricistronic gene encoding human iNKT TCRα-F2A-iNKT TCRβ-P2A-IL-15. The Lenti/MCAR vector was generated by inserting into the pMNDW parental backbone a synthetic gene encoding MCAR. The Lenti/MCAR-IL-15 vector was generated by inserting into the pMNDW parental backbone a synthetic bicistronic gene encoding MCAR-F2A-IL-15. The Lenti/FG vector was generated by inserting into the pMNDW parental backbone a synthetic bicistronic gene encoding Fluc-P2A-EGFP. The Lenti/MSLN vector was generated by inserting into the pMNDW parental backbone a synthetic gene encoding human MSLN. All synthetic gene fragments were obtained from GenScript (Piscataway, NJ, USA) and Integrated DNA Technologies (IDT; Coralville, IA, USA). Lentiviral particles were generated utilizing HEK 293 T cells by employing a standardized transfection procedure with the Trans-IT-Lenti Transfection Reagent (Mirus Bio) [38, 39]. Subsequently, a concentration protocol was applied using Amicon TM Ultra Centrifugal Filter Units in accordance with the manufacturer's specifications (MilliporeSigma).

Human multiple myeloma cell line MM.1 S (cat. no. CRL-2974) and human TNBC cell lines HCC1806 (cat. no. CRL-2335) and MDA-MB-231 (cat. no. CRM-HTB-26) were purchased from the American Type Culture Collection (ATCC). To establish stable tumor cell lines that overexpress firefly luciferase and green fluorescent protein dual reporters (FG), the parental tumor cell lines were transduced with lentiviral vectors carrying the specific genes of interest (i.e., Lenti/FG). 72 h after lentiviral transduction, the cells underwent flow cytometry sorting to isolate the genetically modified cells (as identified as GFP⁺ cells) necessary for creating stable cell lines. The artificial antigen presenting cell line (aAPC) was generated by engineering the K562 human chronic myelogenous leukemia cell line (ATCC, cat. no. CCL-243) to overexpress human CD80/CD83/CD86/41BBL co-stimulatory receptors [24]. The aAPC-MSLN cell lines were generated by further engineering the parental aAPC line to overexpress human MSLN.

Purified human CD34⁺ HSPCs derived from cord blood (CB) were purchased from HemaCare. Healthy donor PBMCs were provided by the UCLA/CFAR Virology Core Laboratory without identification information under federal and state regulations. Upon receipt, both HSPCs and PBMCs were promptly aliquoted and cryopreserved in liquid nitrogen for subsequent experimental use.

Fluorochrome-conjugated antibodies specific for human CD1d (Clone 51.1, PE-Cy7 or APC-conjugated, 1:50, cat. no. 350310 or 350308), CD3 (Clone HIT3a, Pacific Blue, PE, or PE-Cy7-conjugated, 1:500, cat. no. 300330, 300308, or 300316), CD4 (Clone OKT4, PE-Cy7, PerCP or FITC-conjugated, 1:500, cat. no. 317414, 317432 or 317408), CD8 (Clone SK1, PE, APC-Cy7, or APC-conjugated, 1:300, cat. no. 344706, 344714 or 344722), CD14 (Clone HCD14, Pacific Blue-conjugated, 1:100, cat. no. 367122), CD19 (Clone HIB19, APC-Cy7-conjugated, 1:200, cat. no. 302218), CD25 (Clone BC96, PE-conjugated, 1:100, cat. no. 302606), CD34 (Clone 581, PerCP-conjugated, 1:500, cat. no. 343520), CD31 (Clone WM59, FITC-conjugated, 1:100, cat. no. 989002), CD45 (Clone HI30, PerCP, FITC, or Pacific Blue-conjugated, 1:500, cat. no. 982318, 982316, or 982306), CD69 (Clone FN50, PE-Cy7 or PerCP-conjugated, 1:50, cat. no. 310912 or 310928), CD112 (Clone TX31, PE-conjugated, 1:250, cat. no. 337410), CD155 (Clone SKII.4, PE-Cy7-conjugated, 1:250, cat. no. 337614), CD11b (Clone ICRF44, FITC-conjugated, 1:500, cat. no. 982614), MICA/MICB (Clone 6D4, PE or APC-conjugated, 1:25, cat. no. 320906 or 320908), 41BBL (Clone 5F4, PE-conjugated, 1:500, cat. no. 311504), CD83 (Clone HB15e, APC-Cy7-conjugated, 1:500, cat. no. 305330), CD86 (Clone IT2.2, APC-conjugated, 1:500, cat. no. 305412), PD-1 (Clone A17188A, PE or FITC-conjugated, 1:25, cat. no. 379210 or 379206), TIM-3 (Clone A18087E, APC-conjugated, 1:25, cat. no. 364804), CTLA-4 (Clone L3D10, APC-conjugated, 1:50, cat. no. 369606), TIGIT (Clone A15153G, PE-conjugated, 1:50, cat. no. 372706), LAG-3 (Clone 7H2C65, PE-Cy7-conjugated, 1:25, cat. no. 369208), NKG2D (Clone 1D11, PE-Cy7-conjugated, 1:50, cat. no. 320812), DNAM-1 (Clone 11A8, APC-conjugated, 1:50, cat. no. 338312), NKp30 (Clone P30-15, APC-conjugated, 1:50, cat. no. 325210), NKp46 (Clone 9E2, PE-conjugated, 1:50, cat. no. 331908), CD158 or KIR2DL1/S1/S3/S5 (Clone HP-MA4, PE-Cy7-conjugated, 1:50, cat. no. 339512), IFN-γ (Clone B27, PE-Cy7-conjugated, 1:50, cat. no. 506518), Granzyme B (Clone QA16A02, APC-conjugated, 1:2000 or 1:5000, cat. no. 372204), Perforin (Clone dG9, PE-Cy7-conjugated, 1:50 or 1:100, cat. no. 308126), TNF-α (Clone MAb11, APC-conjugated, 1:4000, cat. no. 502912), IL-2 (Clone MQ117H12, APC-Cy7-conjugated, 1:50, cat. no. 500342), β2-microglobulin (B2M) (Clone 2M2, FITC or APC-conjugated, 1:2000, cat. no. 316304 or 316311), HLA-DR (Clone L243, APC-Cy7-conjugated, 1:200, cat. no. 307618), HLA-DR, DP, DQ (Clone Tü39, FITC-conjugated, 1:200 or 1:500, cat. no. 361706), TROP2 (Clone NY18, PE-conjugated, 1:200, cat. no. 363804), EGFR (Clone A21043C, FITC-conjugated, 1:200, cat. no. 386306), and EpCAM (Clone 9C4, PE or Pacific Blue-conjugated, 1:500, cat. no. 324206 or 324218) were purchased from BioLegend. Fluorochrome-conjugated antibodies specific for human iNKT TCR Vɑ24-Jβ18 (Clone 6B11, PE-conjugated, 1:20, cat. no. 552825) were purchased from BD Biosciences. Fluorochrome-conjugated antibodies specific for human fibroblast activation protein FAP (Clone 427819, PE-conjugated, 1:100, cat. no. FAB3715P), ULBP-1 (Clone 170818, PE-conjugated or unconjugated, 1:25, cat. no. FAB1380P or MAB1380), ULBP-2,5,6 (Clone 165903, APC-conjugated, 1:25, cat. no. FAB1298A), MSLN (Clone 420411, APC-conjugated, 1:20, cat. no. FAB32652A), and Nectin-4 (Clone 337516, PE-conjugated, 1:200, cat. no. FAB2659P), were purchased from R&D Systems. A goat anti-mouse IgG F(ab')2 secondary antibody (cat. no. A-11001) was purchased from ThermoFisher. Fixable Viability Dye eFluor506 (e506, 1:500, cat. no. 65-0866-14) was purchased from Affymetrix eBioscience; mouse Fc Block (anti-mouse CD16/32, cat. no. 553141) was purchased from BD Biosciences; and human Fc Receptor Blocking Solution (TrueStain FcX) was purchased from BioLegend (cat. no. 422302). In our study, note the use of antibodies with identical clones but differing conjugated fluorochromes, with one typical antibody listed herein.

All FACS staining was performed following manufacturers' provided protocols. Appropriate isotype staining controls were used for all staining procedures. Stained cells were analyzed using a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotech), following the manufacturer's instructions. FlowJo software version 9 (BD Biosciences) was used for data analysis.

The ELISAs for measuring cytokines or biomarkers were conducted according to a standard protocol provided by BD Biosciences. Samples were collected and analyzed to quantify cytokines or biomarkers. The capture and biotinylated antibodies used for cytokine detection were sourced from BD Biosciences, while the streptavidin-HRP conjugate was obtained from Invitrogen. Human and mouse cytokine standards were purchased from eBioscience, and the Tetramethylbenzidine (TMB) substrate was acquired from Thermo Scientific (cat. no. PI34021). Human IL-17 A ELISA Kit was purchased from Invitrogen (cat. no. BMS2017). Urea Nitrogen (BUN) Colorimetric Detection Kit was purchased from ThermoFisher Scientific (cat. no. EIABUN). Mouse AST ELISA kit was purchased from Abcam (cat. no. ab263882). Mouse ALT ELISA kit was purchased from Abcam (cat. no. ab282882). Mouse Bilirubin ELISA Kit was purchased from MyBioSource (cat. no. MBS3805359). Mouse Glutamate dehydrogenase (GLDH) ELISA Kit was purchased from MyBioSource (cat. no. MBS761948). Absorbance of the samples was measured at 450 nm using an Infinite M1000 microplate reader (Tecan).

^Allo15MCAR-NKT, ^U15MCAR-NKT, ^AlloMCAR-NKT, ^Allo15NKT, and ^Allo15CAR19-NKT cells were generated by differentiating gene engineered human cord blood CD34⁺ HSPCs in a 5-stage clinically guided Ex Vivo HSPC-Derived NKT Cell Culture method. The complete methodology and step-by-step protocols have been described in detail in previously published studies [24, 40]. Here, we provide a summary of the key steps involved in the culture and generation of ^Allo/U15MCAR-NKT and ^AlloMCAR-NKT cells.

At Stage 0, the frozen stock of human CD34⁺ HSPCs was thawed and cultured in T cell X-VIVO 15 Serum-Free Hematopoietic Stem Cell Medium supplemented with human Flt3L (50 ng/ml), SCF (50 ng/ml), TPO (50 ng/ml), and IL-3 (20 ng/ml) for 24 h. Lentiviral transduction was subsequently carried out for an additional 24 h using the Lenti/iNKT-MCAR-IL15 or Lenti/iNKT-MCAR vectors. For ^U15MCAR-NKT cell generation, HSPCs were further electroporated with a CRISPR-Cas9/B2M-CIITA-gRNAs complex, following an established protocol [33]. The gRNA sequences are CGCGAGCACAGCUAAGGCCA (B2M) and GAUAUUGGCAUAAGCCUCCC (CIITA). Both gene knockouts were performed simultaneously to achieve dual disruption of B2M and CIITA. Following electroporation, the mixed CD34⁺ HSPC population was moved to the next stage of cell culture without further purification or sorting.

At Stage 1, gene-engineered HSPCs harvested were cultured in the feeder-free StemSpan™ SFEM II Medium supplemented with StemSpan™ Lymphoid Progenitor Expansion Supplement for 14 days. HSPCs were cultured in CELLSTAR^®24-well Cell Culture Nontreated Multiwell Plates (VWR, cat. no. 82050-892). StemSpan™ Lymphoid Differentiation Coating Material (500 µl/well, diluted to a final concentration of 1X from a stock dilution of 100X), which includes immobilized Delta-like ligands to provide Notch signaling in the absence of feeder cells [41, 42], was applied to the plates and left for 2 h at room temperature or overnight at 4°C. Subsequently, 500 µl of the transfected CD34⁺ HSPC suspension, with a density of 2 × 10⁴ cells/ml, was added to each pre-coated well. Half of the medium in each well was removed and replaced with fresh medium twice per week.

At Stage 2, the Stage 1 cells were harvested and cultured in the feeder-free StemSpan™ SFEM II Medium supplemented with StemSpan™ Lymphoid Progenitor Maturation Supplement for ~ 7 days. StemSpan™ Lymphoid Differentiation Coating Material (1 ml/well, diluted to a final concentration of 1X) was applied to Non-Treated Falcon™ Polystyrene 6-well Microplates (Thermo Fisher Scientific, cat. no. 140675); 2 ml of the harvested Stage 1 cells, resuspended with a density of 1 × 10⁵ cells/ml, was added into each pre-coated well. The cell density was maintained at 1–2 × 10⁶ cells per well during the Stage 2 culturing. Cells were passaged 2–3 times per week with the addition of fresh medium for each passage.

At Stage 3, the Stage 2 cells were harvested and cultured in the feeder-free StemSpan™ SFEM II Medium supplemented with StemSpan™ Lymphoid Progenitor Maturation Supplement, CD3/CD28/CD2 T Cell Activator, and human recombinant IL-15 (20 ng/ml) for ~ 7 days. StemSpan™ Lymphoid Differentiation Coating Material (1 ml/well, diluted to a final concentration of 1X) was applied to Non-Treated Falcon™ Polystyrene 6-well Microplates (Thermo Fisher Scientific, cat. no. 08-772-49); 2 ml of the harvested Stage 2 cells, resuspended with a density of 5 × 10⁵ cells/ml, was added into each pre-coated well. The cell density was maintained at 1–2 × 10⁶ cells per well during the Stage 3 culturing. Cells were passaged 2–3 times per week with the addition of fresh medium for each passage.

At Stage 4, the Stage 3 cells were harvested and verified by flow cytometry to confirm their status as mature ^Allo15MCAR-NKT cells or their derivatives; then the cells underwent expansion stage via an aAPC-based expansion. aAPC-MSLN cells were irradiated at 10,000 rads using a Rad Source RS-2000 X-Ray Irradiator (Rad Source Technologies). The Stage 3 mature ^Allo/U15MCAR-NKT cells and derivatives were co-cultured with the irradiated aAPC-MSLN cells (with a ratio of 1:1). The cells were resuspended in expansion medium (the CTS™ OpTmizer™ T-Cell Expansion Serum Free Medium (Thermo Fisher Scientific), or the homemade C10 medium) supplemented with human IL-7 (10 ng/ml) and IL-15 (10 ng/ml) at a density of 0.5-1 × 10⁶ cells/ml; 2 ml cell suspension was seeded into each well of the Corning™ Costar™ Flat Bottom Cell Culture 6-well Plates. The cell density was maintained at 0.5-1 × 10⁶ cells/ml during the expansion stage. Cells were passaged 2–3 times per week with the addition of fresh medium for each passage. The expanded ^Allo/U15MCAR-NKT cells were aliquoted and cryopreserved in CryoStor^® Cell Cryopreservation Media CS10 using a Thermo Scientific™ CryoMed™ Controlled-Rate Freezer 7450 (Thermo scientific) for stock.

Of note, at the research scale, we typically generated ^Allo15MCAR-NKT cells starting from 10⁴ CD34⁺ HSPCs and successfully produced over 10¹⁰ mature cells. Cell cultures were maintained using either 150 mm cell culture dishes (Thermo Fisher Scientific) or G-Rex 6 M well plates (Wilson Wolf) to support large-scale expansion.

PBMCs from healthy donors were used to generate conventional αβ T cells (hereafter referred to as T cells). T cell activation was achieved using one of two methods: (1) stimulation with Dynabead™ Human T-Activator CD3/CD28 (Thermo Fisher Scientific, Cat. No. 11131D) according to the manufacturer's instructions, or (2) plate-bound activation. For the latter, non-treated 24-well tissue culture plates (Corning, Cat. No. 3738) were coated with Ultra-LEAF™ purified anti-human CD3 antibody (Clone OKT3, BioLegend, cat. no. 317325) at 1 µg/ml (500 µl/well) for 2 h at room temperature or overnight at 4 °C. PBMCs were then resuspended in C10 medium supplemented with 1 µg/ml Ultra-LEAF™ purified anti-human CD28 antibody (Clone CD28.2, BioLegend, cat. no. 302933) and 30 ng/ml IL-2, and seeded into the pre-coated plates at a density of 1 × 10⁶ cells/ml (1 ml/well). Following activation, cells were maintained in C10 medium supplemented with 20 ng/ml IL-2 and cultured for 2–3 weeks. T cells generated by both methods were subsequently tested in the in vitro tumor cell killing assays and showed comparable efficacy, with neither demonstrating tumor cell killing activity.

PBMCs from healthy donors were utilized to generate conventional MCAR-T cells. To produce these cells, non-treated tissue culture 24-well plates (Corning, cat. no. 3738) were coated with Ultra-LEAF™ Purified Anti-Human CD3 Antibody (Clone OKT3, BioLegend) at 1 µg/ml (500 µl/well), at room temperature for 2 h or at 4 °C overnight. PBMCs were resuspended in the C10 medium supplemented with 1 µg/ml Ultra-LEAF™ Purified Anti-Human CD28 Antibody (Clone CD28.2, BioLegend) and 30 ng/ml IL-2, followed by seeding in the pre-coated plates at 1 × 10⁶ cells/ml (1 ml/well). After 2 days, the cells were transduced with either Lenti/MCAR viruses for a period of 24 h. The conventional MCAR-T cells were expanded for about 2 weeks in C10 medium and then cryopreserved for future applications.

Healthy donor PBMCs were sorted with MACS via Anti-iNKT Microbeads (Miltenyi Biotech, cat. no. 130-094-842) labeling to enrich NKT cells, following the manufacturer's instructions. The enriched NKT cells were mixed with donor-matched irradiated αGC/PBMCs at a ratio of 1:1–1:2, followed by culturing in C10 medium supplemented with 10 ng/ml IL-7 and IL-15. On day 3, NKT cells were transduced with Lenti/MCAR-IL15 viruses. The resulting CAR-NKT cells were expanded for about 2 weeks in C10 medium supplemented with 10 ng/ml IL-7 and IL-15 and cryopreserved for future use.

Human tumor cells (e.g., MM.1 S-FG and HCC1806-FG; 1 × 10⁴ cells per well in 96-well plate) were co-cultured with the indicated therapeutic cells (i.e., T, MCAR-T, and ^Allo15MCAR-NKT cells) in Corning 96-well clear bottom black plates for 24 h in C10 medium. The effector cell to target cell (E: T) ratio is indicated in the figure legends. At the end of culture, viable tumor cells were quantified by adding D-luciferin (150 µg/ml; Fisher Scientific, cat. no. 50-209-8110) to cell cultures, followed by the measurement of luciferase activity using an Infinite M1000 microplate reader (Tecan). To test NK receptor-mediated tumor cell killing, 10 µg/ ml Ultra-LEAF™ purified anti-human NKG2D (Clone 1D11, BioLegend, cat. no. 320813) or anti-human DNAM-1 antibody (Clone 11A8, BioLegend, cat. no. 338302) was added to co-cultures to investigate the tumor cell killing mechanism by ^Allo15MCAR-NKT cells, and LEAF™ purified mouse lgG2b κ isotype control antibody (Clone MG2b-57, BioLegend, cat. no. 401202) was included as an isotype control.

In vitro long-term tumor cell killing assay.

A total of 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\times\:$$\end{document} 10⁴ non-engineered tumor cells (e.g., HCC1806 cells; referred to as stimulator cells) was co-cultured with 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\times\:$$\end{document} 10⁵ effector cells in a Corning 96-well clear bottom black plate in C10 medium. Cultures were supplemented with a dose of 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\times\:$$\end{document} 10⁴ stimulator cells every 2 days. 24 h prior to luminescent tumor killing readout, stimulator cells were substituted with 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\times\:$$\end{document} 10⁴ of FG-engineered tumor cells (e.g., HCC1806-FG cells; referred to as indicator cells). To quantify the remaining live indicator cells, 100mL of D-luciferin (10 mg/mL) was added to cell cultures on the day of imaging and the luciferase activities were measured through readout with an Infinite M1000 microplate reader (Tecan).

Healthy donor PBMC-derived, M2-polarized macrophages or MDSCs were used in this assay. PBMCs were resuspended in serum-free RPMI 1640 medium (Corning Cellgro, cat. no. 10-040-CV) at 1 × 10⁷ cells/ml, plated in 10 cm dishes (10–15 ml per dish), and incubated at 37 °C with 5% CO₂ for 1 h. Non-adherent cells were removed, and adherent monocytes were utilized to generate M2-polarized macrophages or MDSCs. To generate M2-polarized macrophages, monocytes were cultured in C10 medium supplemented with recombinant human M-CSF (10 ng/ml, Peprotech, cat. no. 300 − 25) for 6 days. On day 6, macrophages were detached using 0.25% Trypsin/EDTA (Gibco, cat. no. 25200-056), collected, and reseeded in 6- or 12-well plates (0.5-1 × 10⁶ cells/ml) for another 48 h with recombinant human IL-4 (10 ng/mL, Peprotech, cat. no. 214 − 14) and IL-13 (10 ng/ml, Peprotech, cat. no. 214 − 13) to induce M2 polarization. To generate MDSCs, monocytes were cultured in C10 medium supplemented with human GM-CSF and IL-6 (10 ng/mL) for 6 days.

To generate tumor organoids, a 1:1 mixture of 1 × 10⁵ HCC1806-FG tumor cells and 1 × 10⁵ M2 macrophages/MDSCs were resuspended in C10 medium at a concentration of 1 × 10⁵ cells/µl. Cell aggregates were prepared by dispensing 5–10 µl of the cell suspension onto microporous membrane inserts (EMD Millipore, cat. no. PICM0RG50) placed in 6-well plates containing 1 ml of C10 medium per well [43, 44]. After a 2-day incubation period to allow organoid formation, 1 × 10⁶ therapeutic cells (i.e., MCAR-T or ^Allo15MCAR-NKT cells) were resuspended in 100 µl of C10 medium and added on top of each organoid. Co-culture was maintained for 24 h. Following the co-culture period, organoids were mechanically dissociated with a 1 ml pipette and passed through a 70-µm nylon strainer to generate single-cell suspensions for downstream flow cytometry analysis.

In one assay, the primary TNBC patient samples were analyzed for tumor cell phenotype and the TME composition using flow cytometry. TNBC tumor cells identified as CD45⁻CD31⁻FAP (fibroblast activation protein)⁻ cells [45, 46], T cells were identified as CD45^highCD3⁺ cells, CD4 T cells were identified as CD45^highCD3⁺CD4⁺ T cells, CD8 T cells were identified as CD45^highCD3⁺CD8⁺ T cells, B cells were identified as CD45^highCD19⁺ cells, NK cells were identified as CD45^highCD56⁺CD3⁻ cells, monocytes were identified as CD45^highCD11b⁺CD14⁺ cells, tumor-associated macrophages (TAMs) were identified as HLA-DR^highCD206^high monocytes, and myeloid-derived suppressor cells (MDSCs) were identified as HLA-DR^lowCD206^low monocytes. Surface expression of CAR targets and NK ligands on tumor or/and immune cells were also analyzed using flow cytometry.

The clinical patient survival data were obtained from the Tumor Immune Dysfunction and Exclusion (TIDE) database, illustrating the association between MSLN expression in tumors and patient survival in a breast cancer cohort (Prediction of Clinical Outcomes from Genomic Profiles: GSE20486, n = 97). The expression data of CAR target antigens (i.e., MSLN, TROP2, EGFR, and Nectin-4) on TNBC tumor cells and normal tissues were derived from previously published studies [11, 47 –52].

In another assay, the primary TNBC patient samples were used to study tumor cell killing by ^Allo15MCAR-NKT cells. Tumor cells were pre-sorted using a Human Tumor Cell Isolation Kit (Miltenyi Biotec, cat. no. 130-108-339), followed by co-culturing with various therapeutic cells (E: T ratio 1:1) in C10 medium in Corning 96-well Round Bottom Cell Culture plates for 24 h. At the end of culture, cells were collected and live TNBC tumor cells (identified as CD45⁻CD3⁻6B11⁻ cells) was analyzed using flow cytometry. A total of 5 primary OC patient samples were included in this assay.

In another assay, the primary TNBC patient samples were used to study the TME targeting by ^Allo15MCAR-NKT cells. Patient samples were directly co-cultured with ^Allo15MCAR-NKT cells (ratio 1:1) in C10 medium in Corning 96-well Round Bottom Cell Culture plates for 24 h. At the end of culture, cells were collected, and the TNBC TME targeting of ^Allo15MCAR-NKT cells was assessed using flow cytometry by quantifying live human TAM (identified as 6B11⁻CD45⁺CD14⁺CD11b⁺HLA-DR^highCD206^high cells), MDSCs (identified as 6B11⁻CD45⁺CD14⁺CD11b⁺HLA-DR^lowCD206^low cells), T cells (identified as 6B11⁻CD45⁺CD3⁺ cells), B cells (identified as 6B11⁻CD45⁺CD3⁻CD19⁺ cells or 6B11⁻CD45⁺CD3⁻CD20⁺ cells), and NK cells (identified as 6B11⁻CD45⁺CD3⁻CD56⁺ cells). A total of 5 primary TNBC patient samples were included in this assay.

PBMCs of multiple healthy donors were used as responders, to study the T cell-mediated allorejection of ^Allo15MCAR-NKT and ^U15MCAR-NKT cells as stimulators (irradiated at 2,500 rads). MCAR-T cells were included as stimulator controls. Stimulators (5 × 10⁵ cells/well) and responders (2 × 10⁴ cells/well) were co-cultured in 96-well round bottom plates in C10 medium for 4 days; the cell culture supernatants were then collected to measure IFN-γ production using ELISA.

BLI was conducted using the Spectral Advanced Molecular Imaging (AMI) HTX system (Spectral Instrument Imaging). Live animal images were captured 5 min after intraperitoneal (i.p.) administration of D-Luciferin, with doses of 1 mg/mouse for tumor cell (e.g., HCC1806-FG cell) visualization. For tissue imaging, experimental mice received an i.p. injection of D-Luciferin with doses of 10 mg/mouse. Mice were then euthanized, and tissues were collected for BLI. The imaging data were processed and analyzed using AURA imaging software (Spectral Instrument Imaging, version 3.2.0).

On Day 0, NSG mice received orthotopic inoculation of human TNBC tumor cells (HCC1806-FG, 5 × 10⁵ or 3 × 10⁶ cells per mouse). The number of tumor cells used for implantation was selected based on preliminary optimization experiments to ensure consistent and measurable tumor growth within a defined timeframe. This cell dose reliably produces tumors of appropriate size and progression dynamics, allowing for accurate assessment of therapeutic efficacy while avoiding excessive tumor burden that could confound interpretation of treatment effects.

On Day 4, the experimental mice received peritumor (p.t.) injection of vehicle (100 µl PBS per mouse), ^Allo15MCAR-NKT cells (5 × 10⁶ cells in 100 µl PBS per mouse), conventional MCAR-T cells (5 × 10⁶ cells in 100 µl PBS per mouse), ^Allo15NKT cells (5 × 10⁶ cells in 100 µl PBS per mouse), or ^Allo15CAR19-NKT cells (5 × 10⁶ cells in 100 µl PBS per mouse). Notably, MCAR-T cells were enriched to > 95% MCAR⁺ purity prior to injection. Over the experiment, mice were monitored for survival and their tumor load were measured using BLI and tumor size measurement. At the study endpoint, experimental mice were euthanized, and survival data were collected to generate Kaplan–Meier survival curves. To evaluate the GvHD risk, experimental mice were assessed for clinical GvHD scores on days 40 and 50. A score ranging from 0 to 2 was assigned for each clinical GvHD sign, which includes body weight, activity, posture, skin thickening, diarrhea, and dishevelment [53]. In addition, on Day 1, 20, and 50, serum samples were collected from the experimental mice, and the organ damage biomarkers were measured using ELISA. Urea Nitrogen (BUN) Colorimetric Detection Kit was purchased from ThermoFisher Scientific (cat. no. EIABUN). Mouse AST ELISA kit was purchased from Abcam (cat. no. ab263882). Mouse ALT ELISA kit was purchased from Abcam (cat. no. ab282882). Mouse Bilirubin ELISA Kit was purchased from MyBioSource (cat. no. MBS3805359). Mouse Glutamate dehydrogenase (GLDH) ELISA Kit was purchased from MyBioSource (cat. no. MBS761948).

On Day 0, NSG mice received orthotopic inoculation of human TNBC tumor cells (MDA-MB-231-FG, 3 × 10⁶ cells per mouse). On Day 4, the experimental mice received peritumor (p.t.) injection of vehicle (100 µl PBS per mouse), ^Allo15MCAR-NKT cells (5 × 10⁶ cells in 100 µl PBS per mouse), or conventional MCAR-T cells (5 × 10⁶ cells in 100 µl PBS per mouse). Notably, MCAR-T cells were enriched to > 95% MCAR⁺ purity prior to injection. Over the experiment, mice were monitored for survival and their tumor load were measured using BLI.

On Day 0, 5 × 10⁵ HCC1806-FG cells were intravenously injected into the NSG mice to establish a metastatic TNBC model. On Day 4, vehicle (100 µl PBS per mouse), ^Allo15MCAR-NKT cells (10 × 10⁶ cells in 100 µl PBS per mouse), ^U15MCAR-NKT cells (10 × 10⁶ cells in 100 µl PBS per mouse), or conventional MCAR-T (10 × 10⁶ cells in 100 µl PBS per mouse) cells were administered intravenously. Notably, MCAR-T cells were enriched to > 95% MCAR⁺ purity prior to injection. Over the experiment, mice were monitored for survival and their tumor load were measured using BLI. At the study endpoint, experimental mice were euthanized, and survival data were collected to generate Kaplan-Meier survival curves.

Freshly collected samples were immediately delivered to the UCLA TCGB Core for library construction and scRNA-seq. Cells were quantified using a Cell Countess II automated cell counter (Invitrogen/Thermo Fisher Scientific). A total of 10,000 cells from each experimental group were loaded on the Chromium platform (10X Genomics), and libraries were constructed using the Chromium Next GEM Single Cell 3' Kit and the Chromium Next GEMChip G Single.

Cell Kit (10X Genomics), according to themanufacturer's instructions. Library quality was assessed using the D1000 ScreenTape on a 4200 TapeStation System (Agilent Technologies). Libraries were sequenced on an Illumina NovaSeq using the NovaSeq S4 Reagent Kit (100 cycles; Illumina). AddModuleScore was used to calculate module scores of each list of gene signatures, and FeaturePlot function was used to visualize the expression of each signature in the UMAP plots [54, 55].

Graphpad Prism 9 software (Graphpad) was used for statistical data analysis. Student's two-tailed t test was used for pairwise comparisons. Ordinary 1-way ANOVA followed by Tukey's or Dunnett's multiple comparisons test was used for multiple comparisons. Log rank (Mantel-Cox) test adjusted for multiple comparisons was used for Meier survival curves analysis. Data are presented as the mean ± SEM, unless otherwise indicated. In all figures and figure legends, "n" represents the number of samples or animals used in the indicated experiments. A P value of less than 0.05 was considered significant. ns, not significant.

One of the major challenges in developing effective cell-based immunotherapies for TNBC is the lack of a comprehensive profile of immune target biomarkers expressed on tumor cells and within the tumor microenvironment (TME) throughout disease progression [10, 56]. To address this gap, we conducted an immunophenotypic analysis of primary TNBC tumor cells and their corresponding TME using flow cytometry (Fig. 1A). Primary tumor samples were obtained from five TNBC patients, and both tumor cells and infiltrating immune populations were profiled (Table S1). Flow cytometric analysis enabled the identification of multiple cell types, including tumor cells, tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs), T cells, B cells, and NK cells (Fig. S1A). All five primary tumor samples exhibited high levels of EpCAM expression, confirming their epithelial identity (Fig. S1B) [57].

We first evaluated the expression of tumor-associated antigens that are potential CAR targets in TNBC, including MSLN, TROP2, EGFR, and Nectin-4. These markers exhibited heterogeneous expression across patient samples (Fig. 1B and C). Among them, MSLN showed consistently high expression levels on TNBC tumor cells and was associated with poorer overall survival, indicating its role in disease progression (Fig. 1B–1D). Importantly, MSLN expression was low in normal tissues, highlighting it as a promising and tumor-selective CAR target (Fig. 1E) [18]. Based on these findings, MSLN was selected for subsequent CAR engineering efforts. In addition to conventional CAR targets, we also observed robust and stable expression of natural killer receptor (NKR) ligands, such as MIC-A/B and ULBPs (NKG2D ligands), and CD112/CD155 (DNAM-1 ligands), on TNBC tumor cells across all samples, suggesting the potential for NKR-mediated recognition and killing (Fig. 1B and C).

Flow cytometry of the TME revealed a significant enrichment of immunosuppressive myeloid populations, particularly TAMs and MDSCs, alongside other immune cells such as T cells, B cells, and NK cells (Fig. 1F). Notably, a biomarker screening identified high expression of CD1d, a non-polymorphic, MHC class I-like molecule, on TAMs and MDSCs, but not on other immune subsets (Fig. 1G and H). As CD1d is the canonical ligand for the invariant TCR expressed by NKT cells, these findings suggest that CD1d^hi TAMs and MDSCs are natural targets for NKT cell-mediated cytotoxicity [45, 58 –60].

Consistent with this, we detected the presence of NKT cells within the TNBC TME, with frequencies ranging from 0.1 to 1% across samples (Fig. S1C), indicating that NKT cells are capable of infiltrating the TME and may contribute to modulating the tumor immune environment. NKT cells uniquely combine TCR-mediated recognition of CD1d-expressing cells with the expression of multiple activating NKRs, such as NKG2D and DNAM-1, which are capable of recognizing ligands commonly upregulated on TNBC tumor cells [25, 26, 28, 32, 38]. Furthermore, NKT cells can be engineered to express additional CARs, such as MSLN-targeting CARs, further enhancing their tumor-targeting capabilities. In summary, our immune profiling of primary TNBC samples has revealed that TNBC is particularly susceptible to CAR-NKT cell therapy (Fig. 1A). This vulnerability arises from the concurrent expression of tumor-associated CAR targets, NKR ligands, and CD1d-expressing immunosuppressive myeloid cells in the TME (Fig. 1A). These findings support a triple-targeting strategy, leveraging the TCR, CAR, and NKR axes of NKT cells, as a promising immunotherapeutic approach for overcoming the challenges of TNBC treatment.

NKT cells are rare in human peripheral blood (0.001-1%), posing a significant challenge for generating clinically relevant quantities of allogeneic NKT cells suitable for CAR engineering [31, 61, 62]. Moreover, allogeneic NKT cells expanded from blood may be contaminated with bystander conventional αβ T cells, raising the risk of GvHD [63 –65]. To overcome these limitations, we developed a stem cell-based approach and a clinically guided culture method to produce "off-the-shelf" CAR-NKT cells to treat TNBC with high yield, purity, and functional robustness (Fig. 2A) [24, 40].

We constructed a lentiviral vector, Lenti/iNKT-MCAR-IL-15, enabling the co-expression of a human invariant NKT TCR, a mesothelin-specific third-generation CAR (MCAR), and secreted human IL-15 (Fig. 2B). The MCAR construct incorporates a clinically validated SS1 anti-mesothelin scFv and a 28-BBz intracellular signaling domain (Fig. 2B) [66, 67]. Human CD34⁺ HSPCs from cord blood (CB) were transduced with this vector and differentiated into CAR-NKT cells using our feeder-free, 6-week Ex Vivo HSPC-Derived CAR-NKT Cell Culture system [24, 40].

CD34⁺ HSPCs from seven independent CB donors (obtained from commercial vendors such as HemaCare) were successfully transduced and utilized to generate allogeneic IL-15-enhanced MCAR-NKT (^Allo15MCAR-NKT) cells (Fig. 2C and S2A). By the end of the differentiation process, the engineered HSPCs gave rise to a homogeneous population of ^Allo15MCAR-NKT cells, with > 97% co-expression of NKT TCR, CD3, and MCAR (Fig. 2D and E). These cells followed a developmental trajectory consistent with natural NKT cell ontogeny, transitioning from CD4⁻CD8⁻ double-negative (DN) to CD4⁺CD8⁺ double-positive (DP), and subsequently maturing into CD8⁺ single-positive (SP) or DN subtypes (Fig. 2D) [39, 68, 69]. The final cell population was predominantly CD4⁻CD8^+/−, a phenotype known to be pro-inflammatory and cytotoxic, thus favorable for cancer immunotherapy [28, 65, 70, 71].

We next evaluated IL-15 expression and production by ^Allo15MCAR-NKT cells. Upon stimulation with α-GalCer (αGC), a potent NKT cell agonist, ^Allo15MCAR-NKT cells produced robust levels of IL-15, confirming successful transgenic expression and functional secretion of IL-15 (Fig. S2B). To assess the potential for autonomous or dysregulated proliferation, we conducted an in vitro cell proliferation assay in the absence of exogenous cytokine support (Fig. S2C). ^Allo15MCAR-NKT cells failed to survive or expand under cytokine-free conditions, indicating that despite their ability to secrete IL-15, they remain dependent on exogenous IL-7 and IL-15 for survival and proliferation. These results suggest a low risk of cytokine-driven, uncontrolled cell growth (Fig. S2D).

Importantly, the ^Allo15MCAR-NKT cell products were free of contaminating αβ T cells, eliminating the risk of GvHD (Fig. 2D and F). Across all CB donors, cell products consistently demonstrated > 97% purity and clonality (Fig. 2F and G). Calculated theoretical yield estimates indicated that from a single quality CB donor (~ 5 × 10⁶ CD34⁺ HSPCs), approximately 10^{12 Allo15}MCAR-NKT cells could be generated, that is sufficient for formulating an estimated 1,000–10,000 clinical doses (at 10⁸-10⁹ cells per dose) (Fig. 2H) [6, 72, 73]. Vector copy number (VCN) analysis revealed an average of 3–4 copies per ^Allo15MCAR-NKT cell, within the clinically accepted threshold (< 5 copies/cell), supporting the safety of the product with minimal risk of insertional mutagenesis (Fig. 2I) [6, 72, 73].

In summary, we established a robust, clinically relevant platform for generating high-yield, high-purity ^Allo15MCAR-NKT cells from CB HSPCs. These cells will be evaluated for their therapeutic potential against TNBC and represent a promising candidate for off-the-shelf allogeneic cell therapy.

We next evaluated the phenotype and functional attributes of ^Allo15MCAR-NKT cells and performed a direct comparison with conventional MCAR-engineered T cells (MCAR-T), which were generated from healthy donor peripheral blood mononuclear cells (PBMCs) and served as a benchmark control (Fig. S2E-S2I). While MCAR-T cells typically exhibited CAR expression in over 70% of the population, variability in manufacturing and purification steps can lead to inconsistencies in CAR expression across production batches (Fig. S2F and S2H). In contrast, ^Allo15MCAR-NKT cells consistently achieved near-uniform (> 97%) MCAR expression, eliminating the need for additional purification or CAR-enrichment procedures (Fig. 2F and G). This uniformity underscores their off-the-shelf therapeutic potential and facilitates robust quality control during manufacturing. In addition, neither ^Allo15MCAR-NKT cells nor conventional MCAR-T cells expressed MSLN, similar to their non-CAR-engineered counterparts, thereby avoiding self-fratricide mediated by the MSLN-specific CAR (Fig. S2I).

Flow cytometry analysis revealed that ^Allo15MCAR-NKT cells possessed a phenotype closely aligned with endogenous human NKT cells, enriched for NK-like characteristics and cytotoxic functionality. Compared to conventional MCAR-T cells, ^Allo15MCAR-NKT cells expressed higher levels of NK-associated markers such as CD56, along with activating NK receptors (NKRs) including NKG2D, DNAM-1, NKp30, and NKp46 (Fig. 2J). Notably, they lacked expression of inhibitory NKRs, such as killer-cell immunoglobulin-like receptors (KIRs) (Fig. S2J and S2K). Functionally, ^Allo15MCAR-NKT cells demonstrated enhanced secretion of key effector cytokines (e.g., IFN-γ, TNF-α, and IL-2) and elevated levels of cytolytic molecules such as Perforin and Granzyme B (Fig. 2J). These phenotypic and functional features, consistent with their CD8 SP or DN lineage (Fig. 2D), support their potential utility in cancer immunotherapy by combining potent cytotoxicity with broad receptor-mediated tumor recognition.

To characterize the genomic and molecular features of ^Allo15MCAR-NKT cells, we performed single-cell RNA sequencing (scRNA-seq) analysis, using conventional MCAR-T cells as a benchmark control. ^Allo15MCAR-NKT cells displayed a hybrid T/NK cell transcriptional profile, marked by a higher NK cell gene signature and a comparatively lower T cell signature relative to MCAR-T cells (Fig. 2K). ^Allo15MCAR-NKT cells also exhibited elevated expression of genes associated with proliferation and effector function (Fig. 2K). Notably, ^Allo15MCAR-NKT cells upregulated NKR genes such as CD226 and NCR3, as well as genes encoding the cytotoxic effector molecules including Perforin and Granzyme B, consistent with flow cytometry results (Fig. 2J and K). In addition, they expressed high levels of the NKT-lineage transcription factor ZBTB16 (encoding PLZF) and BCL11B (encoding B-cell lymphoma/leukemia 11B), as well as the Th1-associated transcription factor TBX21 (encoding T-bet) (Fig. S3A and S3B) [74]. Pathway analysis further revealed that, compared to MCAR-T cells, ^Allo15MCAR-NKT cells were enriched for gene signatures associated with cell cycle progression, proliferation, activation, and metabolism (Fig. 2L). Collectively, these findings indicate that ^Allo15MCAR-NKT cells possess potent T/NK-like cytotoxic characteristics, supporting their therapeutic potential for antitumor applications.

In addition to conventional MCAR-T cells, we conducted a side-by-side comparison of ^Allo15MCAR-NKT cells with healthy donor PBMC-derived IL-15-enhanced MCAR-NKT (^PBMC15MCAR-NKT) cells using scRNA-seq (Figure S3C). Both cell types expressed key NKT-lineage and Th1-associated transcription factors, including ZBTB16, BCL11B, and TBX21, indicating their shared identity as NKT cells with Th1-like functional potential (Figure S3D). Notably, compared to ^PBMC15MCAR-NKT cells, ^Allo15MCAR-NKT cells exhibited enhanced expression of NK-associated markers and reduced expression of conventional T cell–associated genes, along with higher proliferative and effector signatures but lower expression of cytotoxic gene programs (Figure S3E). Pathway analysis further revealed that ^Allo15MCAR-NKT cells upregulated pathways involved in peptide synthesis, ATP biosynthesis, ribosome biogenesis, lymphocyte activation, and NK cell–mediated cytotoxicity (Figure S3F). These results suggest that ^Allo15MCAR-NKT cells possess a metabolically active, proliferative, and NK-like effector phenotype that may contribute to their unique therapeutic potential.

To evaluate the in vitro antitumor capacity of ^Allo15MCAR-NKT cells against TNBC, we conducted a series of tumor cell killing assays using both established TNBC cell lines and primary tumor cells derived from patients. These tumor models differed in antigen expression, molecular characteristics, and genetic background, providing a comprehensive platform to assess the breadth of antitumor activity of ^Allo15MCAR-NKT cells.

We first examined both CAR- and NKR-mediated cytotoxicity using four tumor cell lines: two NK cell-resistant multiple myeloma (MM) lines, MM.1 S (MSLN⁻) and MM.1 S-MSLN (engineered to overexpress MSLN), and two NK-sensitive TNBC cell lines, HCC1806 (MSLN^high) and MDA-MB-231 (MSLN^low) (Fig. 3A and B). All tumor lines were engineered to co-express firefly luciferase and green fluorescence protein dual reporters (FG) for quantifiable tracking via bioluminescence and flow cytometry (Fig. 3A).

Three therapeutic immune cell groups were tested: (1) ^Allo15MCAR-NKT cells, (2) conventional PBMC-derived MCAR-T cells, and (3) untransduced T cells from healthy donors (Fig. 3C). Within 24 h, unmodified T cells failed to exhibit cytotoxicity toward any of the tumor cell lines. As expected, MCAR-T cells efficiently lysed MSLN⁺ tumor cells but showed no activity against MSLN⁻ targets, confirming the CAR-dependent nature of their killing (Fig. 3D). In contrast, ^Allo15MCAR-NKT cells robustly killed both MSLN⁺ and MSLN⁻ TNBC cells, indicating the use of both CAR-dependent and CAR-independent mechanisms (Fig. 3D).

Phenotypic analysis revealed that ^Allo15MCAR-NKT cells displayed heightened activation (CD69 upregulation) and produced significantly greater levels of effector cytokines (e.g., IFN-γ) and cytotoxic molecules (e.g., Perforin and Granzyme B) than MCAR-T cells (Fig. 3E – 3G). Importantly, ^Allo15MCAR-NKT cells showed antigen-dependent cytokine secretion, producing high levels of IFN-γ, TNF-α, IL-2, and IL-15 in response to MSLN-positive HCC1806-FG tumor cells—consistent with antigen-specific activation and cytolytic function (Fig. S4A and S4B). In contrast, co-culture with MSLN-negative target cells, including MM.1 S-FG and normal human cells (i.e., B cells), resulted in minimal cytokine production, indicating functional specificity and a low risk of off-target activation (Fig. S4A and S4B).

To dissect the CAR-independent cytotoxic mechanism, we performed NKR-blocking assays using neutralizing antibodies against NKG2D and DNAM-1 (Fig. 3H). Blockade of these receptors significantly impaired the cytotoxicity of ^Allo15MCAR-NKT cells, confirming a reliance on NKR-mediated tumor recognition (Fig. 3I). Therefore, ^Allo15MCAR-NKT cells utilize a dual targeting mechanism mediated by both the MCAR and endogenous NKRs to effectively target TNBC. To further dissect the contribution of each targeting axis, we incorporated two additional therapeutic cell controls: non–CAR-engineered allogeneic IL-15-enhanced NKT (^Allo15NKT) cells and allogeneic IL-15-enhanced CD19-specific CAR-NKT (^Allo15CAR19-NKT) cells (Fig. S5A). In the in vitro tumor cell killing assay, none of the tumor lines mentioned above expressed CD19, allowing us to evaluate CAR-independent functions of ^Allo15NKT and ^Allo15CAR19-NKT cells (Fig. S5B). Both cell types exhibited comparable tumor-killing profiles: no cytotoxicity against MM.1 S-FG or MM.1 S-MSLN-FG cells, but robust killing of HCC1806-FG and MDA-MB-231-FG cells (Fig. S5C). These findings indicate that ^Allo15NKT and ^Allo15CAR19-NKT cells mediate potent tumor cell killing via CAR-independent, NKR-mediated mechanisms, as demonstrated by the NKR blocking assay in which tumor cell killing was markedly reduced upon NKR (i.e., NKG2D and DNAM-1) blockade (Fig. S5D and S5E). In contrast, ^Allo15MCAR-NKT cells showed robust killing of MM.1 S-MSLN-FG cells, which express the MSLN target antigen, and enhanced killing of MSLN-positive HCC1806-FG cells (Fig. S5C). Together, these results indicate that ^Allo15MCAR-NKT cells possess dual-targeting capability via MCAR and NKR, whereas ^Allo15NKT and ^Allo15CAR19-NKT cells mediate killing primarily through NKR recognition.

We then investigated the long-term killing potential of ^Allo15MCAR-NKT cells using repeated tumor challenge assays (Fig. 3J). Remarkably, ^Allo15MCAR-NKT cells sustained superior cytolytic activity over five rounds of tumor exposure, whereas MCAR-T cells exhibited diminished function (Fig. 3J). This durable killing of ^Allo15MCAR-NKT cells was associated with low expression of exhaustion markers, including PD-1, CTLA-4, TIM-3, LAG-3, and TIGIT (Fig. 3K), along with sustained high levels of IFN-γ production (Fig. S6A), suggesting enhanced persistence and resistance to functional decline. Furthermore, in a parallel long-term tumor killing assay, ^Allo15MCAR-NKT cells outperformed non–IL-15-engineered allogeneic MCAR-NKT (^AlloMCAR-NKT) cells, maintaining robust and sustained cytotoxic activity for up to 10 rounds of tumor rechallenge (Fig. S6B-S6G). These findings demonstrate that IL-15 expression enhances the durability, persistence, and antitumor efficacy of allogeneic MCAR-NKT cells under repeated antigen stimulation [75, 76].

To directly model tumor antigen escape, a known limitation of conventional CAR-T therapies [77, 78], we generated a mixed population of MDA-MB-231 tumor cells comprising both MSLN⁺ (engineered) and MSLN⁻ populations (Fig. 3L and S6H-S6K). While MCAR-T cells selectively eliminated MSLN⁺ tumor cells, they spared MSLN⁻ cells (Fig. 3M and N). In contrast, ^Allo15MCAR-NKT cells efficiently eliminated both populations, highlighting their capacity to overcome tumor heterogeneity and antigen escape through multifaceted targeting mechanisms (Fig. 3M and N). While this in vitro model provides a useful proxy for tumor heterogeneity and antigen escape, we acknowledge its limitations in fully recapitulating dynamic antigen loss in vivo. Future studies will incorporate long-term in vivo models to more accurately evaluate this phenomenon.

Lastly, we extended these findings to primary patient-derived TNBC tumor cells (Fig. 3O). Across five independent samples, which varied in their expression of MSLN and NKR ligands (Fig. 1B and C), ^Allo15MCAR-NKT cells consistently outperformed MCAR-T cells in cytotoxicity assays (Fig. 3P). To further dissect antigen-specific versus antigen-independent killing, tumor cells were FACS-sorted into MSLN^high and MSLN^low populations and co-cultured with the two therapeutic cell products (Fig. S6L and S6M). MCAR-T cells effectively eliminated MSLN^high tumor cells but spared MSLN^low tumor cells, confirming their reliance on CAR antigen recognition for cytotoxicity (Fig. S6N). In contrast, ^Allo15MCAR-NKT cells exhibited cytotoxicity against both MSLN^high and MSLN^low populations, with enhanced killing of MSLN^high tumor cells (Fig. S6N). These results indicate that ^Allo15MCAR-NKT cells mediate both CAR-dependent and CAR-independent cytotoxicity, likely via their native NKR pathways. This multiple-targeting capability enables effective tumor cell killing across heterogeneous antigen landscapes and supports their potential utility in addressing interpatient variability and antigen escape in TNBC.

Collectively, these results demonstrate that ^Allo15MCAR-NKT cells exhibit potent and durable tumor cell killing through CAR and NKR pathways. Their ability to target antigen-diverse and antigen-low tumor populations supports their potential as a next-generation immunotherapeutic strategy for treating heterogeneous and aggressive tumors such as TNBC.

The immunosuppressive TME represents a major barrier to effective cancer immunotherapy, particularly in solid tumors such as TNBC [79 –82]. TNBC is characterized by a highly immunosuppressive TME that promotes tumor progression, facilitates immune evasion, and contributes to resistance to immunotherapeutic approaches [10, 83, 84]. Immune profiling of primary TNBC samples has revealed that TAMs and MDSCs comprise over 50% of the total immune cell population, underscoring the critical importance of targeting these immunosuppressive components to improve therapeutic outcomes (Fig. 1F).

TAMs and MDSCs in the TNBC TME express high levels of CD1d, a nonpolymorphic MHC class I-like molecule that is the natural ligand for the invariant TCR expressed by NKT cells (Fig. 1G and H) [31, 71]. Thus, these cells are promising targets for elimination by ^Allo15MCAR-NKT cells (Fig. 1A). To investigate this therapeutic potential, we evaluated the interaction of ^Allo15MCAR-NKT cells with TNBC TME components using three complementary models: primary patient-derived TNBC samples (Fig. 4A–4C), in vitro-generated healthy donor PBMC-derived macrophages and MDSCs (Fig. 4D–4H and S7A-S7E), and 3D TNBC tumor organoids (Fig. 4I–4L and S7F-S7I).

In the first approach, co-culture of ^Allo15MCAR-NKT cells with dissociated primary metastatic TNBC tumor samples demonstrated selective depletion of TAMs and MDSCs, attributed to their high CD1d expression (Fig. 4A–4C). Importantly, ^Allo15MCAR-NKT cells spared other immune cell populations, including T cells, B cells, and NK cells, which expressed little to no CD1d, highlighting the specificity and safety of this approach (Fig. 4B).

In the second model, M2-polarized macrophages and MDSCs were generated from healthy donor monocytes and confirmed to express high levels of CD1d (Fig. 4D–4F and S7A-S7C). In co-culture assays, ^Allo15MCAR-NKT cells efficiently eliminated both cell types (Fig. 4G and H, S7D, and S7E). This cytotoxicity was enhanced in the presence of α-GalCer, an NKT cell agonist, and was significantly reduced when CD1d was blocked by a neutralizing antibody, confirming a CD1d-dependent, NKT TCR-mediated killing mechanism (Fig. 4H and S7E). Notably, even in the absence of αGC, ^Allo15MCAR-NKT cells retained cytotoxic function, suggesting an additional contribution from NKR-mediated recognition (Fig. 7H and S7E). Consistently, these cells did not harm CD1d⁻ healthy donor-derived T or B cells, further demonstrating their favorable safety profile (Fig. 4H).

In the third model, 3D TNBC tumor organoids were generated by co-culturing TNBC tumor cells with M2 macrophages or MDSCs, better mimicking the structural and immunological complexity of the TME (Fig. 4I and S7F) [43, 85]. Within these organoids, conventional MCAR-T cells exhibited limited antitumor activity, which was correlated with reduced expression of activation markers such as CD25 and diminished production of effector cytokines, including IFN-γ, TNF-α, and IL-2, indicating the functional suppression of MCAR-T cells by the immunosuppressive myeloid cells (Fig. 4J–4L, and S7G-S7J). In contrast, ^Allo15MCAR-NKT cells maintained robust tumor cell killing and high activation status despite the presence of suppressive M2 macrophages and MDSCs (Fig. 4J–4L, and S7G-S7J). This may be due to their ability to selectively deplete these immunosuppressive cells, thereby remodeling the TME to support an effective immune response (Fig. 4K and M).

To evaluate whether NKT TCR-mediated cytotoxicity against M2-polarized macrophages and MDSCs is dependent on CAR engagement, we co-cultured these myeloid cell populations with three allogeneic NKT cell products mentioned above: ^Allo15NKT, ^Allo15CAR19-NKT, and ^Allo15MCAR-NKT cells (Fig. S7K). Despite the absence of a CAR or the expression of a CD19-targeting CAR, all three NKT cell products exhibited comparable cytotoxicity against both macrophages and MDSCs (Fig. S7L). These results indicate that NKT TCR-mediated killing of CD1d⁺ myeloid cells by allogeneic NKT cells occurs independently of CAR engagement.

In summary, ^Allo15MCAR-NKT cells exhibit a unique capacity to simultaneously target both TNBC tumor cells and key immunosuppressive components of the TME through a multifaceted CAR/TCR/NKR-mediated recognition mechanism (Fig. 4M). This tri-modal targeting strategy enables not only direct cytotoxicity toward tumor cells but also TME reprogramming, positioning ^Allo15MCAR-NKT cells as a highly promising off-the-shelf immunotherapeutic candidate for the treatment of TNBC.

To assess the in vivo efficacy and safety of ^Allo15MCAR-NKT cells, we utilized a human TNBC orthotopic xenograft model in which tumor cells were implanted into the mammary fat pad of NSG mice (Fig. 5A). This model closely recapitulates the tumor growth dynamics and local tissue microenvironment of TNBC in its native anatomical context (Fig. 5A).

Both ^Allo15MCAR-NKT cells and conventional MCAR-T cells were administered directly into the tumor site. While both therapies demonstrated significant tumor suppression, as evidenced by reduced tumor size and extended overall survival, only ^Allo15MCAR-NKT cells achieved complete tumor elimination in all treated mice, resulting in 100% long-term survival (Fig. 5B–5E). These findings underscore the superior antitumor efficacy of ^Allo15MCAR-NKT cells.

Notably, despite initial tumor control, several mice treated with conventional MCAR-T cells succumbed to GvHD, suggesting safety concerns associated with this approach (Fig. 5E). Flow cytometry of peripheral blood revealed that ^Allo15MCAR-NKT cells were detectable for 2–3 weeks post-injection but cleared from circulation following tumor eradication (Fig. 5F and G). In contrast, conventional MCAR-T cells rapidly expanded and occupied ~ 50% of peripheral blood leukocytes, correlating with elevated clinical GvHD scores and histological evidence of tissue damage (Fig. 5F–5I). This expansion was associated with morbidity and mortality due to GvHD and CAR-T cell-associated toxicity (Fig. 5H and I). The seemingly shorter persistence of ^Allo15MCAR-NKT cells may be due to their preferential localization to tumor sites with limited circulation in peripheral blood, as well as their inability to trigger GvHD-related expansion because their TCR recognizes the non-polymorphic MHC-like molecule CD1d [70, 71]. Collectively, these results demonstrate that ^Allo15MCAR-NKT cells combine potent antitumor activity with a favorable safety profile, offering a promising therapeutic strategy for treating TNBC.

To investigate the in vivo antitumor mechanisms of ^Allo15MCAR-NKT cells, we employed the same orthotopic TNBC model but with an increased tumor burden and compared the therapeutic efficacy of three allogeneic NKT cell products: ^Allo15MCAR-NKT, ^Allo15NKT, and ^Allo15CAR19-NKT cells (Fig. S8A). Among these, only ^Allo15MCAR-NKT cells achieved complete tumor eradication, whereas both ^Allo15NKT and ^Allo15CAR19-NKT cells significantly delayed tumor progression but were less effective overall (Fig. S8B and S8C). These findings suggest that while ^Allo15NKT and ^Allo15CAR19-NKT cells rely primarily on NKR-mediated cytotoxicity, the superior efficacy of ^Allo15MCAR-NKT cells is likely due to the combined action of CAR-dependent and NKR-mediated tumor killing.

To further evaluate the antigen-independent antitumor activity of ^Allo15MCAR-NKT cells, we utilized an orthotopic xenograft model with MDA-MB-231-FG human TNBC cells, which exhibit low or absent MSLN expression (Fig. S8D). This model better represents the heterogeneity and CAR antigen-low context of TNBC. In this setting, mice bearing orthotopic MDA-MB-231-FG tumors were treated with either conventional MCAR-T cells or ^Allo15MCAR-NKT cells (Fig. S8D). As anticipated, MCAR-T cells failed to control tumor growth due to the absence of their target antigen (Fig. S8E and S8F). In contrast, ^Allo15MCAR-NKT cells exhibited potent antitumor effects despite the lack of CAR antigen engagement, indicating their ability to eliminate tumors through antigen-independent mechanisms, likely mediated by innate-like NKRs (Fig. S8E and S8F). These results underscore the multifaceted cytotoxic potential of ^Allo15MCAR-NKT cells and their capacity to overcome tumor antigen escape.

TNBC is notorious for its aggressive clinical behavior and propensity to metastasize rapidly to visceral organs, most commonly the lungs, liver, bone marrow, and brain, typically within 3–5 years of diagnosis [83]. Unlike other breast cancer subtypes, TNBC lacks actionable molecular targets and presents an immunosuppressive TME, contributing to poor therapeutic responses and high mortality [83]. To evaluate the therapeutic potential of ^Allo15MCAR-NKT cells in metastatic TNBC, we employed a HCC1806-FG human metastatic TNBC xenograft model (Fig. 6A).

In this model, tumor cells were intravenously injected into NSG mice, initially seeding in the lungs and subsequently disseminating to distant sites, particularly the bone marrow (Fig. 6B–6G, and S9A). This recapitulates the aggressive metastatic progression observed in TNBC patients [86]. Both ^Allo15MCAR-NKT cells and conventional MCAR-T cells were administered intravenously. While both treatments suppressed tumor progression to some extent, ^Allo15MCAR-NKT cells achieved markedly superior tumor control and significantly prolonged survival (Fig. 6B–6G). Strikingly, bone marrow tumor burden was comparably reduced by both cell therapies; however, MCAR-T cells failed to effectively control lung-resident tumor cells. In contrast, ^Allo15MCAR-NKT cells demonstrated robust tumor suppression in both the lungs and bone marrow, indicating broad and consistent antitumor efficacy (Fig. 6E–6G). Despite the potent initial antitumor response, tumor relapse was observed approximately 30 days following ^Allo15MCAR-NKT cell administration, ultimately leading to animal mortality (Fig. 6B–6D). This relapse may be attributed to the limited in vivo persistence of ^Allo15MCAR-NKT cells beyond 30 days in this model. However, given the off-the-shelf nature of this cell product, repeated dosing is feasible and may represent a viable strategy to overcome relapse and achieve sustained tumor control.

Tissue biodistribution analysis revealed that conventional MCAR-T cells predominantly localized to peripheral lymphoid organs, including the spleen and blood, with limited infiltration into the lungs, which are the primary site of TNBC tumor implantation in this model (Fig. 6H and I, and S9B). In contrast, ^Allo15MCAR-NKT cells demonstrated preferential homing to tumor-bearing tissues, particularly the lungs, with reduced accumulation in lymphoid compartments such as the spleen and blood (Fig. 6H and I, and S9B). Additionally, a higher number of ^Allo15MCAR-NKT cells were detected in the bone marrow, potentially due to the dissemination of metastatic HCC1806-FG tumor cells into this compartment (Fig. 6E–6I). These findings highlight the superior tumor-homing capacity of ^Allo15MCAR-NKT cells, a critical advantage for targeting solid and metastatic tumors.

Phenotypic characterization of tumor-infiltrating therapeutic cells further supported this advantage. ^Allo15MCAR-NKT cells displayed a highly activated and cytotoxic profile, marked by elevated expression of activation markers (i.e., CD25 and CD69), effector cytokines (i.e., IFN-γ), and cytolytic molecules (i.e., Granzyme B and Perforin) (Fig. 6J and K). Importantly, these cells maintained low expression of exhaustion markers (i.e., PD-1, CTLA-4, TIM-3, LAG-3, and TIGIT), suggesting sustained functional capacity within the immunosuppressive TME (Fig. 6L).

Collectively, these results demonstrate that ^Allo15MCAR-NKT cells offer potent, durable antitumor efficacy in a challenging metastatic TNBC setting by combining superior tumor-homing, cytotoxicity, and resistance to functional exhaustion. These attributes position ^Allo15MCAR-NKT cells as a promising therapeutic platform for treating metastatic TNBC.

In addition to the risk of GvHD, a major limitation of allogeneic cell therapy is host cell-mediated allorejection. Specifically, host CD8⁺ and CD4⁺ T cells can recognize HLA class I and class II molecules, respectively, on allogeneic cells, leading to their depletion and rejection [23, 64]. This immune response can result in the premature elimination of therapeutic cells and a significant reduction in overall efficacy. To overcome this barrier, we further engineered the ^Allo15MCAR-NKT cell product to completely eliminate surface expression of both HLA class I and class II molecules, thereby rendering the cells resistant to recognition by host CD8⁺ and CD4⁺ T cells, respectively (Fig. 7A).

This was accomplished through targeted disruption of two key genes: B2M, which encodes β2-microglobulin and is essential for the surface expression of all HLA class I molecules, and CIITA, the master transcriptional regulator of HLA class II gene expression [33, 38, 87]. Using CRISPR-Cas9-mediated gene editing with guide RNAs (gRNAs) targeting B2M and CIITA, we generated IL-15-enhanced, MCAR-NKT cells lacking both HLA-I and HLA-II expression, termed ^U15MCAR-NKT cells (Fig. 7A).

To generate these universal CAR-NKT cells, CB-derived CD34⁺ HSPCs were first transduced with the Lenti/iNKT-MCAR-IL-15 vector (Fig. 7A and B). Simultaneously, the HSPCs were electroporated with CRISPR-Cas9 ribonucleoprotein complexes containing B2M and CIITA gRNAs. This dual genetic modification strategy consistently achieved > 50% transduction efficiency and > 50% HLA-I/II double-knockout rate (Fig. 7A and B). The engineered HSPCs were then cultured in the 6-week Ex Vivo HSPC-Derived CAR-NKT Cell Culture to generate ^U15MCAR-NKT cells (Fig. 7C).

Importantly, ^U15MCAR-NKT cells retained similar phenotypic characteristics, CAR expression levels, developmental trajectories, and cytotoxic profiles compared to their parental ^Allo15MCAR-NKT counterparts (Fig. 7C–7H). The absence of HLA-I/II did not impair their differentiation or functional maturation. Furthermore, the double HLA knockout population could be further enriched through magnetic-activated cell sorting (MACS) or fluorescence-activated cell sorting (FACS) (Fig. 7E). Interestingly, even unedited ^Allo15MCAR-NKT cells inherently expressed low levels of HLA-I and negligible HLA-II molecules, likely due to their epigenetically and transcriptionally regulated hypoimmunogenic nature within our culture system (Fig. 7E and I) [24, 32]. This feature suggests that ^Allo15MCAR-NKT cells may already be intrinsically less immunogenic than conventional allogeneic CAR-T cells.

To evaluate immunogenicity, we performed in vitro mixed lymphocyte reaction (MLR) assays in which therapeutic cells were co-cultured with allogeneic donor PBMCs (Fig. 7J). Conventional MCAR-T cells elicited strong alloreactive T cell responses (Fig. 7K). In contrast, ^Allo15MCAR-NKT cells induced significantly reduced T cell activation, while ^U15MCAR-NKT cells triggered minimal to no detectable alloresponse (Fig. 7K). These data confirm that HLA-ablated ^U15MCAR-NKT cells possess extremely low immunogenicity and are highly resistant to T cell-mediated rejection (Fig. 7L).

Importantly, despite the complete absence of HLA-I/II, ^U15MCAR-NKT cells maintained robust antitumor activity. In vitro cytotoxicity assays demonstrated potent killing of TNBC tumor cells, and in vivo efficacy was validated in a metastatic HCC1806-FG TNBC xenograft mouse model (Fig. 7M–7R). ^U15MCAR-NKT cells effectively suppressed tumor progression and extended survival, with comparable efficacy to ^Allo15MCAR-NKT cells and superior performance over conventional MCAR-T cells (Fig. 6A–6D, and Fig. 7O–7R).

In summary, we have successfully developed ^U15MCAR-NKT cells with complete ablation of HLA class I and II expression, which retain potent antitumor activity while demonstrating resistance to host T cell–mediated allorejection. In our current experimental models, including in vitro tumor cell killing assays and in vivo NSG mouse studies, both ^Allo15MCAR-NKT and ^U15MCAR-NKT cells exhibited comparable antitumor efficacy (Figs. 3D, 6A–6D and 7M–7R), likely due to the lack of functional host T cells in these systems. However, under physiological conditions in human patients, host CD4⁺ and CD8⁺ T cells may recognize HLA class II and class I molecules on ^Allo15MCAR-NKT cells, respectively, leading to their immune-mediated elimination. In contrast, ^U15MCAR-NKT cells, which lack surface HLA expression, are expected to evade such rejection. To further assess their efficacy in the presence of host immunity, advanced preclinical models, such as NSG mice reconstituted with human T cells or fully humanized NSG models, could be employed [33, 64]. Collectively, these findings underscore the therapeutic potential of ^U15MCAR-NKT cells as a universal, off-the-shelf immunotherapy platform for the treatment of TNBC and potentially other solid tumors.