Lipid nanoparticles deliver DNA-encoded biologics and induce potent protective immunity

Abstract

Lipid nanoparticles (LNPs) for mRNA delivery have advanced significantly, but LNP-mediated DNA delivery still faces clinical challenges. This study compared various LNP formulations for delivering DNA-encoded biologics, assessing their expression efficacy and the protective immunity generated by LNP-encapsulated DNA in different models. The LNP formulation used in Moderna’s Spikevax mRNA vaccine (LNP-M) demonstrated a stable nanoparticle structure, high expression efficiency, and low toxicity. Notably, a DNA vaccine encoding the spike protein, delivered via LNP-M, induced stronger antigen-specific antibody and T cell immune responses compared to electroporation. Single-cell RNA sequencing (scRNA-seq) analysis revealed that the LNP-M/pSpike vaccine enhanced CD80 activation signaling in CD8⁺ T cells, NK cells, macrophages, and DCs, while reducing the immunosuppressive signals. The enrichment of TCR and BCR by LNP-M/pSpike suggested an increase in immune response specificity and diversity. Additionally, LNP-M effectively delivered DNA-encoded antigens, such as mouse PD-L1 and p53^R172H, or monoclonal antibodies targeting mouse PD1 and human p53^R282W. This approach inhibited tumor growth or metastasis in several mouse models. The long-term anti-tumor effects of LNP-M-delivered anti-p53^R282W antibody relied on memory CD8⁺ T cell responses and enhanced MHC-I signaling from APCs to CD8⁺ T cells. These results highlight LNP-M as a promising and effective platform for delivering DNA-based vaccines and cancer immunotherapies.

Introduction

Lipid nanoparticles (LNPs) are crucial in nucleic acid drug delivery, enabling the transport of mRNAs and siRNAs into target cells [1]. They have gained prominence with the success of mRNA COVID-19 vaccines from Pfizer/BioNTech and Moderna, demonstrating their ability to elicit immune responses [2, 3]. LNPs use four types of lipids—cationic lipids, PEGylated lipids, phospholipids, and cholesterol— to encapsulate genetic materials, allowing for efficient and safe delivery [4]. Cationic lipids encapsulate negatively charged nucleic acids [5], PEGylated lipids provide stability [6], phospholipids maintain structural integrity [7], and cholesterol contributes to the stability and membrane fluidity [8]. Although LNP-delivered mRNA technology excels in high delivery efficiency and adaptability, it has some drawbacks, including the intrinsic instability of mRNAs, strict storage requirements for the final products, and high production costs [9]. LNP delivery of DNA offers advantages such as greater stability, lower storage needs, and reduced production costs, thus enabling broader use of genetic medicine [10, 11].

Studies have shown that LNP-delivered mRNA or siRNA can enhance transfection efficiency and tissue-specific delivery by optimizing lipid components and their proportions [12,13,14]. Zhu et al. report that optimized LNPs for DNA delivery are developed using different helper lipids and component ratios, with about one thousand formulations tested using a combination of DLin-MC3-DMA, cholesterol, DMG-PEG 2000, and one of six helper lipids [15]. LNP/mRNA formulations can induce a broad immune response by optimizing the lipid composition, molar ratios, and structure [14]. These formulations improve delivery efficacy by enhancing cellular uptake and endosomal escape [16, 17]. In this study, we evaluated several LNP types designed for mRNA and siRNA delivery to transport DNA in vitro and in vivo. We used the LNP formulation used in the Moderna COVID-19 mRNA vaccine (LNP-M) to deliver various DNA-encoded biologics, including the SARS-COV-2 spike protein, mouse PD-L1, the mouse p53 mutant R172H (p53^R175H), and two antibodies targeting mouse PD1 and human p53 mutant R282W (p53^R282W). We performed single-cell RNA sequencing (scRNA-seq) to investigate the mechanisms underlying the enhanced immunity mediated by spike immunization or intratumoral delivery of anti-p53^R282W antibodies. These studies demonstrate that LNP-encapsulated DNA-encoded biologics offer a promising approach for infection prevention and cancer treatment.

Materials and methods

Cell lines and cell culture

The following cell lines were used in this study: human embryonic kidney cell line 293T (ATCC, Manassas, USA), mouse B cell lymphoma line A20 (ATCC), mouse metastatic lung cancer cell line 344SQ (University of Texas M.D. Anderson Cancer Center, provided by Jonathan M. Kurie), human pancreatic adenocarcinoma HupT3 (harboring the R282W mutation in the p53 gene, obtained from Sigma-Aldrich), and mouse colon adenocarcinoma cell line MC38 (Sigma-Aldrich, St-Louis, MO, USA). The MC38-p53^KO/R282W cell line, which stably overexpresses human p53^R282W following endogenous p53 knockout (p53^KO), was generated by knocking out the endogenous p53 mutant alleles (G242V & S238I) and introducing the human p53 gene with the R282W mutation. 293T, MC38, and MC38-p53^KO/R282W cells were cultured in DMEM (Gibco, Grand Island, NY, USA) supplemented with 10% Fetal Bovine Serum (FBS, Gibco) and 1 × anti-anti solution (Gibco). A20 cells were cultured in RPMI-1640 Medium (Gibco) supplemented with 10% FBS, 0.05 mM 2-mercaptoethanol (Gibco, 21985023), and 1 × anti-anti solution. 344SQ cells were cultured in RPMI 1640 medium supplemented with 10% FBS and 1 × anti-anti solution. HupT3 cells were cultured in DMEM (Gibco) supplemented with 2 mM Glutamine (Cytiva, Marlborough, MA, USA; SH30034.01), 1% non-essential amino acids (NEAA, Cytiva, SH30238.01), 1% sodium pyruvate (Cytiva, SH30239.01), 10% FBS, and 1 × anti-anti solution.

Plasmid preparation

The following plasmids were used in this study: pmaxGFP (Lonza), gWIZ-Luc, gWIZ-p53-R172H, gWIZ-PD-L1 (Twist Bioscience, South San Francisco, CA, USA), and pαH-Spike (SARS-CoV-2 HexaPro Spike with two Strep-Tag II and a His-Tag; Addgene, Cambridge, MA, USA; 154754). To construct plasmids expressing monoclonal antibodies (mAbs), the heavy chain (HC) and light chain (LC) of the mouse PD1-mAb (RMP1-14) were fused to the Fc region of mouse IgG2a and cloned into the gWIZ vector (Twist Bioscience). Mouse R282W-mAb was generated using hybridoma screening [18]; its HC and LC were fused to the Fc region of human IgG1 and cloned into the pTwist vector (Twist Bioscience). Plasmids were purified using the endotoxin-free ZymoPURE™ II Plasmid Maxiprep Kit (Zymo Research, Irvine, CA, USA; D4203), as previously described [18].

LNP/DNA formulation

Four LNP formulations were used: LNP-B, composed of ALC-0315 (MedChemExpress, Monmouth Junction, NJ, USA; HY-138170), ALC-0159 (MedChemExpress, HY-138300), DSPC (Sigma-Aldrich, P1138), and Cholesterol (Sigma-Aldrich, C3045); LNP-M, comprising SM-102 (Cayman Chemical, Ann Arbor, MI, USA; 33474), DMG-PEG 2000 (Avanti Polar Lipids, Alabaster, AL, USA; 880151), DSPC, and Cholesterol; LNP-P containing DLin-MC3-DMA (Cayman Chemical, 34364), DMG-PEG 2000, DSPC, and Cholesterol; and LNP-n, consisting of DLin-MC3-DMA, PEG-DSPE (Avanti Polar Lipids, NC0668213), DSPC, and cholesterol. These formulations were prepared by diluting the respective lipid components in 40 μl ethanol (Fisher Scientific, Fair Lawn, NJ, USA; A4094) at a molar ratio of cationic lipid: DSPC: Cholesterol: DMG-PEG 2000 = 51: 8: 38: 3 for four components in all LNPs. When needed, NLS (MedChemExpress, HY-P1876B) and/or histones (Sigma-Aldrich, H5505) were incubated with DNA or DSPC for 2 h at room temperature. The preparation of LNP/DNA nanoparticles follow the same standard method as described below.

For DNA encapsulation, 40 μg of DNA vector or antibody-expressing plasmids with HC and LC in a 1:1 ratio, was diluted with water and citrate buffer (pH 3.5; final concentration 25 mM; bioWORLD, Dublin, Ohio, USA; 40320053–1) for a final volume is 80 μl. The lipid and DNA phases were loaded into the Spark Cartridge (Precision Nanosystems, South San Francisco, CA, USA; NIS0013) and formulated using a NanoAssemblr Spark Formulation Device (Precision Nanosystems) according to the manufacturer’s instructions. The resulting LNP/DNA nanoparticles were purified by dialyzing against PBS with free Ca²⁺ and Mg²⁺ at pH 7.4 using a Pur-A-Lyzer Maxi Dialysis Kit (Sigma-Aldrich, PURX60005) overnight and then concentrated to a final pDNA concentration of 0.8 mg/mL using 50 KDa Amicon Ultra-0.5 mL Centrifugal Filters (Merck Millipore, Billerica, MA, USA; UFC505024). The encapsulated efficiency of LNP/DNA was quantified using a Quant-iT Pico-Green dsDNA assay kit (Thermo Fisher Scientific, Waltham, MA, USA; P11496) following the manufacturer’s instructions. Before the experiment, the LNP/DNA nanoparticles were sonicated for 3 s, three times.

Particle size and zeta potential analysis

LNP nanoparticles were diluted with ddH₂O at a 1:100 ratio. Size measurements were taken at 25 °C with a 173° scattering angle using dynamic light scattering (DLS) on the Zetasizer Nano ZS90 (Malvern, UK). Zeta potential determinations, based on the electrophoretic mobility of the nanoparticles in the aqueous medium, were performed using folded capillary cells in automatic mode.

For morphology assessment, LNP nanoparticles were resuspended in ddH₂O and deposited onto a glow-discharged carbon‐coated 400-mesh copper grid for 5 min. Subsequently, the films were negatively stained with 1 wt% uranyl acetate at room temperature for 1 min. After staining, the copper mesh was dried with filter paper and examined using transmission electron microscopy (TEM).

Gel retardation assay

The integrity of LNP/DNA complexes was assessed to investigate the stability of the encapsulated plasmid. Naked plasmid pmaxGFP and LNPs served as controls, while the LNP/DNA complexes were tested in experimental wells, each containing 1 μg of DNA. All samples were loaded onto a gel containing 1% agarose and run in l × Tris–EDTA buffer (pH 8.3) using a gel electrophoresis system (Bio-Rad, Hercules, CA, USA). The gel was then visualized using a gel imaging system (Bio-Rad).

Transfection

293T cells were seeded on a 24-well plate at a density of 2 × 10⁵ cells per well. After overnight cultured, the cells were treated with various transfection reagents, including LNP, lipofectamine™ 2000 (Invitrogen, Carlsbad, CA, USA; 11668019), PEI (Polysciences, Warrington, PA, USA; 23966), TansIT-LT1(Mirus Bio, Madison, WI, USA, MIR 2305), and lentivirus, each containing 2 μg of plasmid DNA per well in Opti-MEM medium (Gibco). This dosage is typical for lipoplexes, resulting in high transfection efficiency with minimal impact on cell viability. After a 3 h incubation at 37 °C and 5% CO₂, DMEM supplemented 20% FBS was added to each well. After various time points post-transfection, 293T cells were harvested, washed, and resuspended in PBS for analysis using a fluorescence activated cell sorter (FACS).

Western blot

Samples were collected and lysed in cold lysis buffer (Invitrogen, 89901) supplemented with Protease and Phosphatase Inhibitor (Invitrogen, 78443). Equal amounts of protein were loaded onto 10% SDS-PAGE gels for each sample, separated by electrophoresis, and transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore). Subsequently, the membrane was then blocked with a 5% BSA solution and probed with primary antibodies against His-tag (Sigma-Aldrich, SAB1306084) or β-actin (Sigma-Aldrich, A5316) at 4 °C overnight. The membranes were then washed with phosphate-buffered saline containing 0.1% Tween 20 (PBST) and incubated with horseradish peroxidase (HRP)‐linked secondary antibodies at room temperature for 2 h. After additional washing, immunoreactive bands were visualized using enhanced chemiluminescence (ECL, Thermo Fisher Scientific, 34076) and detected with an automatic chemiluminescence system (Bio-Rad).

Animal experiments

Six- to eight-week-old female BALB/c, 129S1/SvImJ, C57BL/6J and NOD.Cg-Prkdc (scid) Il2rg (tm1Wjl)/SzJ (NSG) mice were obtained from Jackson Laboratory and the Center for Comparative Medicine (CCM) of Baylor College of Medicine and housed in pathogen-free conditions. All animal procedures and protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Baylor College of Medicine.

For intramuscular experiments, BALB/c mice were injected with LNP-M/pLuc (40 μg/mouse), After five days, tissues expressing detectable levels of luciferase were observed. Anesthesia was induced using isoflurane, and 200 μl of substrates at a concentration of 15 mg/ml was injected into the intraperitoneal cavity. Photon emission from the region of interest (ROI) was measured using the Caliper in vivo imaging system (IVIS Lumina II, Caliper LifeSciences, Hopkinton, MA, USA).

For preventive spike vaccine immunization, mice were intramuscularly administered with LNP-M/Ctrl (LNP-M/gWIZ vector), LNP-M/Spike-mRNA (mRNA-1273), or LNP-M/pSpike. Some mice received electroporation (EP) immunization with pSpike (EP/pSpike) using ICHOR medical systems. Each group was immunized three times biweekly with 40 μg of DNA/mouse or 5 μg of mRNA/mouse. For therapeutic vaccine immunization, subcutaneous tumor models were established by inoculating A20 (1 × 10⁶ cells/mouse), 344SQ (2 × 10⁶ cells/mouse), MC38 (3 × 10⁵ cells/mouse), and MC38-p53^KO/R282W (5 × 10⁵ cells/mouse). Tumor models were intramuscularly immunized with LNP-M/pPD-L1 or LNP-M/pR172H on days 7 and 17 and intratumorally on day 24 after tumor inoculation.

For the MC38 or MC38-p53^KO/R282W subcutaneous tumor model, intratumoral injections of LNP-M/pPD1-mAb, LNP-M/pR282W-mAb, or LNP-M/Ctrl (LNP-M/gWIZ or LNP-M/pTwist vector) (40 μg/mouse) were administered on days 14 and 21. Additionally, combination therapy with 500 μg/mouse of αCD4 (BioXcell, West Lebanon, NH, USA; BE0119), αCD8 (BioXcell, BP0117), or αNK1.1 (BioXcell, BE0036) was administered twice weekly following the first intratumoral injection.

For the lung metastasis model, MC38-p53^KO/R282W cells (1 × 10⁶ /mouse) were injected into the tail vein on day 0. Mice received intravenous injections of LNP-pR282W-mAb (40 μg/mouse) on days 7 and 12. On day 22, the mice were sacrificed, and metastatic nodules in the lungs were quantified. To establish intestinal tumors, mice were intraperitoneally administered 2 × 10⁶ MC38-p53^KO/R282W cells suspended in 100 μl of PBS. Mice were then treated intraperitoneally with LNP-M/pR282W-mAb on days 7 and 14 after tumor inoculation. Tumor growth and survival were evaluated at the end of treatment.

NSG mice were subcutaneously inoculated with HupT3 cells (3 × 10⁶/mouse) resuspended in a mixture of serum-free medium and Matrigel (1:1 volume ratio, Corning, New York, NY, USA; 354230) into the right flank on day 0. On day 10, tumor-bearing mice received an intravenous injection of human PBMCs (1 × 10⁷/mouse). Fourteen days post-tumor inoculation, mice were intratumorally injected with LNP-M/pR282W-mAb (40 μg/mouse) on days 14, 19, and 24. For subcutaneous tumors, the tumor volume was calculated using the formula: V = (length × width²)/2. The tumor inhibition rate was calculated using the following formula: (C − T)/C × 100%, where C represents the volume of the largest tumor in the control group, and T is the tumor volume of each treated mouse.

FACS analysis

A single-cell suspension of spleen or tumor-infiltrating leukocytes (TILs) was prepared by gently homogenizing spleen or digested tumor tissues using a mouse Tumor Dissociation Kit (Miltenyi Biotec, Auburn, CA, USA; 130–096-730). After FcR blocking and removing dead cells with a Zombie Aqua Fixable Viability Kit (BioLegend, San Diego, CA, USA), cell surface staining was performed by incubating with the following antibodies for 30 min at 4 °C, followed by intracellular staining. Mouse antibodies used included CD16/32 (BioLegend, 101302), APC-BrdU (BioLegend, 364114), APC/Cy7-CD45 (BioLegend, 103116), Brilliant Violet 750-CD45 (BioLegend, 103157), PE/Cyanine7-CD45 (BioLegend, 103114), Percp/cy5.5-CD19 (BioLegend, 152406), PE-CD3 (BioLegend, 100206), Pacific Blue CD4 (BioLegend, 100531), PerCP/Cyanine5.5 CD8α (BioLegend, 100734), APC-CD49b (BioLegend, 108910), APC-NK1.1 (BioLegend, 156506), APC-IFN-γ (BioLegend, 505810), Alexa Fluor 488-IFN-γ (BioLegend, 505813), Alexa Fluor 488-TNF-α (BioLegend, 506313), PE-IL-2 (Biolegend, 503808), APC-CD11c (BioLegend, 117310), FITC-CD11b (BioLegend, 101206), PE-CD103 (BioLegend, 121406), PerCP-F4/80 (BioLegend, 123126), Alexa Fluor 647-FOXP3 (BioLegend, 126408), APC-Cyanine7-Ly-6G/Ly-6C (BioLegend, 108424), FITC-CD279 (BioLegend, 135214), FITC-CD107a (BioLegend, 121606), PE-CD44 (BioLegend, 103024), FITC-CD62L (BioLegend, 104406), PE-CD80 (BioLegend, 104708), FITC-CD86 (BioLegend, 105006), Pacific Blue-I-A/I-E (BioLegend, 107620), PerCP/Cyanine5.5-H-2K^d/H-2D^d (BioLegend, 114716). Human antibodies used included TruStain FcX™ (BioLegend, 422302), APC-IgG Fc (BioLegend, 410712), Brilliant Violet 421-CD45 (BioLegend, 304032), APC-CD56 (BioLegend, 362504). Isotype-matched immunoglobulin served as control. Data were acquired on a Cytek® NL-3000 FACS system (Cytek Biosciences, Fremont, CA, USA) and analyzed using FlowJo V10 software (Tree Star Inc, Ashland, OR, USA).

Antibody titers

Enzyme-linked immunosorbent assay (ELISA) Plates were coated with 5 μg/ml of antigen and incubated at 4 °C overnight. The plates were then washed with 0.05% Tween 20 in PBS and blocked with 1% BSA in 0.05% Tween 20/PBS at room temperature for 2 h. Ten-fold serially diluted serum samples were added in duplicate and incubated at 37 °C for 2 h. HRP-conjugated anti-mouse IgG (Cell Signaling Technology, Danvers, MA, USA; 7076), IgG1 (Abcam, Cambridge, MA, USA; ab97240), IgG2c (Cell Signaling Technology, 56970S), and IgG2b (Abcam, ab97250) antibodies were then added for 1 h at room temperature, followed by the addition of TMB substrate. Absorbance at 490 nm was measured using a CLARIOstar Plus microplate reader (BMG LABTECH, Ortenberg, Germany).

T lymphocyte proliferation assay

Splenocytes from vaccine-immunized mice were plated onto 96-well tissue culture plates at a density of 1 × 10⁵ cells per well. The culture medium contained IL-2 (100 U/ml) and recombinant spike peptide pool (10 μg/ml; STEMCELL, Cambridge, MA, USA; 100–0676). Plates were incubated at 37 °C in a humidified incubator with 5% CO2 for 5 days, with medium changed on day 3. The cells were then incubated with 10 μM BrdU (Sigma-Aldrich, 19–160) for 2 h at 37 °C. After this, cells were collected and subjected to surface staining with anti-CD8α or CD4 antibodies for 30 min. After fixation, permeabilization, and washing, intracellular staining was performed using APC-BrdU antibody (BioLegend, 364114) for an additional 30 min. After intracellular staining, cells were washed three times with a permeabilization buffer. The percentage of BrdU⁺ cells within CD4⁺ or CD8⁺ T cell populations was analyzed by FACS, with this proportion defined as the proliferation rate.

Measurement of cytokines

Protein levels of IFN-γ, IL-10, IL-4, IL-6, and TNF-α were measured in the culture supernatants of stimulated splenocytes or in serum using ELISA kits from R&D Systems (Minneapolis, MN, USA), according to the manufacturer’s instructions.

Enzyme-linked immunospot (ELISPOT)

The assay was conducted using the Mouse IFN-γ/TNF-α Double-Color ELISPOT Kit (Cellular Technology Limited (CTL), Cleveland, OH, USA), following the manufacturer’s instructions. Initially, splenocytes isolated from immunized mice were plated at a density of 1 × 10⁶ cells/well and stimulated with a spike peptide pool (10 μg/ml) for 60 h at 37 °C with 5% CO₂. After incubation, cells were washed and incubated with detection antibodies, including biotinylated-TNF-α and FITC-IFN-γ. Subsequently, after another round of washing, streptavidin-AP conjugate or FITC-HRP (1:1000 dilution) was added. The visualization of spots was developed on plates. Spot-forming colonies (SFC) per well were scanned and quantified using a CTL reader. IFN-γ spots appeared as red spots, while TNF-α spots were blue.

Cytotoxic T lymphocyte (CTL) assay

Splenocytes were suspended in RPMI 1640 medium supplemented with 100 U/ml IL-2 and 10 μg/ml spike peptide pool, cultured in a 37 °C humidified incubator with 5% CO2 for 7 days. Subsequently, the splenocytes were washed and resuspended in a medium to be used as effector.

cells. Target cells, over-expressing spike-293T cells, were prepared. Effector and target cells were then titrated in U-bottom 96-well plates at effector-to-target cell ratios of 50:1, 25:1, and 12.5:1. Following this, 1 × 10⁴ target cells were added and incubated at 37 °C for 72 h. Cytotoxicity was assessed using the Cytotoxicity Detection Kit PLUS (Roche, Palo Alto, CA, USA; 4744926001), according to the manufacturer’s instructions.

Pathological analyses

The tissue samples were fixed in 10% formalin and embedded in paraffin. The paraffin-embedded tissues were then sectioned into 5 μm slices and stained with hematoxylin and eosin (H&E) according to the manufacturer's instructions. Photographs of the stained tissues were taken using an Olympus IX51 microscope (Olympus America Inc., Center Valley, PA, USA).

scRNA-seq experiment

A single-cell suspension from the spleens of mice immunized three times with the vaccine was prepared. CD45⁺ cells were purified using the EasySep™ Mouse TIL (CD45) Positive Selection Kit (STEMCELL, 100–0350) according to the manufacturer’s instructions. Mice-bearing subcutaneous tumors were treated intratumorally with either LNP-M/pR282W-mAb or a control. Fourteen days after treatment, tumors were harvested and minced in RPMI-1640 medium supplemented with 10% FBS. The single-cell suspension from the dissected tumor samples was prepared using a mouse Tumor Dissociation Kit with a gentleMACS Octo Dissociator with Heaters (Miltenyi Biotech). CD45⁺ cells were sorted from each tumor sample using a FACSAria II instrument. Cells were collected using a 40 μm cell strainer (Corning, 431750) and centrifuged at 300 g for 10 min. The spleen or tumor cells were resuspended in RPMI1640 with 5% FBS. To create the scRNA-seq libraries, equivalent numbers of single cells from three mice per group were combined at a density of 1 × 10³ cells/μl in RPMI1640 medium with 10% FBS. Approximately 10,000 cells in each group were converted into barcoded scRNA-seq libraries using the Chromium™ Next GEM Single Cell 5′ Kit v2 (10 × Genomics, Pleasanton, CA, USA; 1000263) following the manufacturer’s protocol. For preparing full-length T-cell Receptor (TCR) and B-cell Receptor (BCR)V(D)J libraries, the Chromium Single Cell Mouse TCR Amplification Kit (10 × Genomics, 1000254) and Chromium Single Cell Mouse BCR Amplification Kit (10 × Genomics, 1000255) was used to enrich amplified cDNA from the 5′ libraries. The libraries were completed at the Single Cell Genomics Core at Baylor College of Medicine (BCM), following the manufacturer’s protocol. Sequencing was performed on a NovaSeq 6000 by the BCM Genomic and RNA Profiling (GARP) Core.

scRNA-seq data preprocessing

The CellRanger Count v7.1.0 pipeline (https://cloud.10xgenomics.com) was used to process raw sequence data in FASTQ format. The data were aligned to the mouse reference genome using default barcode assignment and Unique Molecular Identifier (UMI) counting with a 2020-A version of the mouse (mm10) genome reference. Default parameters were used with the mouse mm10 genome reference. For obtaining raw gene expression matrices for TCR and BCR, the pipeline used default parameters with the mm10 reference genome. Further processing, including doublet removal and data filtering, was conducted using the R package scDblFinder (v1.14.0) [19] and the R package Seurat (v4.4.0) [20]. In term of tumors, cells with mitochondrial content greater than 15%, nFeature RNA greater than 7,000, or fewer than 200, nCounts greater than 50,000 identified cells were removed from CD45⁺ cells. Following the workflow of the R package Seurat, we utilized the NormalizeData function to standardize the data and selected the initial 2000 variable genes by using FindVariableFeatures for subsequent analysis. Following the linear transformation of single-cell data via the ScaleData function, we utilize RunPCA funtion and select the top 30 principal components for linear dimensionality reduction. The FastMNN function from the R package SeuratWrappers (v0.3.19) (https://github.com/satijalab/seurat-wrappers) was employed for multi-sample integration to eliminate batch effects. The FindNeighbors and FindClusters functions were employed to delineate neighbors and clusters. Next, dimensionality reduction was executed via the unified manifold approximation and projection (UMAP) through the RunUMAP function, with the dimension parameter configured to 30 and the dimension reduction parameter set to mnn. Subsequently, we employed the FindNeighbors and FindClusters functions to ascertain the edge weights between any two cells and cell clusters. For the analysis of T cells, T cells were subset from the integrated data based on the annotated cell clusters. The R packages STACAS (2.1.3) [21] and ProjecTILs (3.2.0) [22] as well as annotation of cell type database [23] were used to assist with the annotation process.

Downstream analysis of scRNA-seq data

To identify differentially expressed genes (DEGs) within clusters, the FindAllMarkers function in the Seurat package was used with a Wilcoxon test and Bonferroni correction for multiple testing. The DEGs of each cell cluster were determined by comparing them to all other cell clusters, using |log2 fold change|≥ 1 and adjusted p-value < 0.05. The top 100 genes were then selected for enrichment analysis. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis for DEGs within cell clusters were performed using Clusterprofiler (v4.8.3) [24]. UMAP and heatmap visualizations were adjusted using the scRNAtoolVis (v0.0.7) (https://github.com/junjunlab/scRNAtoolVis) and pheatmap (v1.012) (https://github.com/raivokolde/pheatmap) packages. Additionally, single cell Gene Set Variation Analysis (scGSVA) was conducted using the R package scGSVA (0.0.16) (https://github.com/guokai8/scGSVA) to identify pathway activity score of each cell types [25]. Cytokine activity was analyzed using the Python implementation of CytoSig [26].

Receptor and ligand interaction analysis

Cellular interactions were investigated using the CellChatR toolkit (2.2) [27]. The CellChatDB database, which includes ligand-receptor interactions supported by literature and protein–protein interaction (PPI) data from mouse datasets, was used to calculate communication probabilities. Initially, the incoming and outgoing parameters for each sample were filtered using the selectK function. The data from the four cell groups were then merged using the mergeCellChat function. The merged data items were analyzed following the Cellchat protocol “Comparison analysis of multiple datasets”.

Processing scVDJ sequence data

For processing single-cell VDJ sequencing data, the Seurat and scRepertoire (1.11.0) [28] packages were used for scVDJ and scRNA-seq analysis. Within the scRepertoire framework, total abundance of clonotypes was visualized using the cloneAbundance function. The relative abundance of clonotypes was determined by calculating their percentage in the sample. Clonotype distribution of CDR3 nucleotide or amino acid sequences was visualized using the cloneLength function. Clonal homeostasis, reflecting the clonal space occupied by specific proportions of clonotypes, was visualized with the clonalHomeostasis function. The clonalProportion function identified the proportion of clonal space occupied by specific clonotypes. Clonal diversity across groups was quantified using the clonalDiversity function.

Statistical analyses

Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software). Data were presented as the mean ± standard deviation (SD). An unpaired students’ t-test was used to compare two groups, while one-way analysis of variance (ANOVA) was used for comparisons among multiple groups. Survival differences were assessed using the log-rank test (Mantel-Cox). A p value < 0.05 was considered statistically significant, with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Results

Characteristics and transfection efficiency of optimized LNP-M/DNA

To develop an effective LNP formulation for DNA delivery, we first conducted a comparative analysis of several types of FDA-approved LNPs from Pfizer BioNTech (LNP-B), Moderna (LNP-M), Alnylam Patisiran (LNP-P) and LNP-n with a modified molar ratio 51: 8: 38: 3. All particles were prepared using a spark formulation device from Precision Nanosystems, unless otherwise indicated. Utilizing dynamic light scattering at an LNP/DNA mass ratio of 16: 1 of LNP/DNA, we observed the optimal size and uniformly dispersed size intensity curves among the LNP/DNA nanoparticles (Figs. 1A and S1A). Their Z-Average sizes were approximately 184.4 nm, 189.9 nm, 211.9 nm, and 210.5 nm, respectively. The surface zeta potential varied slightly, with LNP-B/pGFP at approximately 5.7 mV, LNP-M/pGFP at 5.9 mV, LNP-A/pGFP at 7.0 mV, and LNP-n/pGFP at 4.4 mV. Notably, the LNP-n/pGFP exhibited a lower surface zeta potential than the other three other LNP/DNAs (Figs. 1B and S1B). In gel blocking analysis, all LNPs demonstrated effective encapsulation of DNA plasmids, as evidenced by their retention within the gel electrophoresis wells, in stark contrast to the mobility of the free naked DNA plasmids (Fig. 1C). Furthermore, the electron microscope revealed a uniform diameter of approximately 200 nm for all four LNP/DNAs (Fig. 1D). Encapsulation efficiency was further determined by measuring fluorescence upon adding PicoGreen to the LNPs and comparing it with the post-lysis value. We found that all four LNPs encapsulated plasmids had an efficiency of > 90% (Fig. 1E). The LNP/DNA complexes exhibited high stability in different conditions, such as incubation with bovine, mouse, or human sera and at temperatures from -20 °C to 37 °C, (Figs. 1F, S1C, and S1d).

Fig. 1

Characterization of LNP/DNA nanoparticles. A Size distribution of the four LNP/DNA formulations. B Surface zeta potential of LNP/DNA nanoparticles. C Agarose gel analysis of LNP/DNA nanoparticles. D Micrographs of LNP‐DNA nanoparticles acquired by TEM with 40,000 × magnification (scale bar, 500 nm). E Encapsulation efficiency of the LNP/DNA formulations using LNP-B, LNP-M, LNP-A, and LNP-n. F The stability of LNP/DNA nanoparticles in different sera. G-N 293T cells transfected with LNP/DNA. Cells were plated in a 24-well plate and transfected with four different LNP/pGFP formulations. FACS of GFP was analyzed after 72 h (G), showing the percentages of GFP-positive cells (H) and MFI (I) in triplicates. The time course of cells transfected with LNP/DNA (2 μg/well) was shown in (J-L). The molar ratios of four components in LNP-M were adjusted in (M) and sonication was added in (N), and the cells were analyzed after 72 h or 24 h. Data presented as the mean ± SD. The different significance was set at *p < 0.05, **p < 0.01, and ****p < 0.0001; ns, not significant

To assess the efficiency of LNP-mediated DNA delivery in vitro, GFP expression was detected in 293T cells transfected with four LNP/pGFP particles. FACS was used to measure the mean fluorescence intensity (MFI). LNP-B/pGFP and LNP-M/pGFP exhibited high transfection efficiency in a dose-dependent manner, whereas LNP-A/pGFP showed lower efficiency 72 h after transfection at a dose of 2 μg DNA per well in 24-well plates. LNP-n/pGFP had the lowest transfection efficiency (Fig. 1G-I). We also evaluated the transfection efficiency of LNP-pGFP at different time points and found that the transfection efficiency of LNP-B/pGFP and LNP-M/pGFP increased over time (Fig. 1J-L).

The LNP-M delivered DNA vaccine demonstrated promising results in inducing protective immunity [29]. Next, we use LNP-M exclusively to optimize its compositions and molar ratios. We first adjusted the molar ratio of the four components of LNP-M. LNP-M/pGFP showed the highest efficiency 72 h after transfection with a ratio of SM102: DSPC: Cholesterol: DMG-PEG 2000 = 51: 8: 38: 3 (Fig. 1M). Adjusting the pH of the citrate buffer while maintaining the same ratio of the four lipids of LNP-M revealed that the transfection efficiency of LNP-M/pGFP in 293T cells was the highest at pH 3.5 (Fig. S2A). Among citrate concentrations and volumes, 25 mM in a final volume of 80 µl provided the best transfection efficiency for LNP-M/pGFP (Fig. S2B and S2C). Additionally, brief sonication (three times for 3 s) of the LNP-M/pGFP particles significantly increased the transfection efficiency (Fig. 1N; 24 h after transfection). These results indicated that the optimized LNP-M formulation is effective for DNA delivery in vitro.

We evaluated whether altering the components of LNP-M impacts its DNA delivery efficiency. Replacing DSPC with DOPE led to a significant decrease in the transfection efficiency of LNP-M/pGFP in 293T cells (Fig. 2A and B). Similarly, substituting DMG-PEG 2000 with ALC0159 or DSPE notably reduced the transfection efficiency of LNP-M/pGFP in these cells (Fig. 2C and D). Nuclear entry efficiency is a crucial factor in gene expression by nuclear localization signal (NLS) or histones [30, 31]. Premixing NLS or histones with DNA plasmids or DSPC was tested for subsequent LNP-M/DNA transfection. When NLS, histones, or both were premixed with DNA before microfluidic formulation with 4 lipids of LNP-M, it was observed that either NLS-bound DNA or histones-bound DNA significantly reduced the expression efficiency of LNP-M/pGFP in 293T cells (72 h post-transfection; Figs. 2E, F and S3A–S3C). We next premixed NLS, histones, or both with DSPC before adding 3 other lipids, then performed microfluidic formulation with DNA; NLS and/or histones significantly enhanced the transfection efficiency of LNP-M/pGFP (24 h post-transfection; Figs. 2G, H, and S3D–S3F).

Fig. 2

Optimization of LNP-M formulations to deliver spike. A and B LNP-M with DSPC or DOPE to deliver pGFP. 293T cells were seeded into a 24-well plate and transfected with LNP-M/pGFP, with DSPC replaced by DOPE. FACS of GFP was analyzed after 20 h, showing GFP expression images (A) and percentages of GFP-positive cells (B). C and D LNP-M with DMG-PEG 2000, ALC-0159, or DSPE to deliver pGFP. E and F LNP-M plus NLS or histones to deliver pGFP. pGFP was pre-incubated with NLS (mass ratio at 20:1), histones (mass ratio at 20:1) or NLS + histones (10:1) before microfluidic mixing with four lipids. 293T cells were transfected for 72 h before GFP expression image acquisition. G and H LNP-M plus NLS or histones to deliver pGFP, with DSPC, was pre-incubated with either NLS (mass ratio at 80:1) or histones (mass ratio at 80:1) or NLS + histones (40:1) before microfluidic mixing and transfection. I and J LNP-M delivers the luciferase (Luc) gene into mice. Mice were intramuscularly injected with optimized LNP-M-encapsulated pLuc (40 μg per animal), and bioluminescence was detected after five days. K LNP-M delivers pGFP intramuscularly into mice. L-N LNP-M delivers pSpike into mice. Spike expression in muscle tissues was detected using Western blot, serum levels of IL-6 and TNF-α assessed by ELISA, and body weights measured every three days. O H&E staining of major mouse organs. Mice were immunized three times on days 0, 14, and 28, and sacrificed on day 42 before organ collection and histological analyses. Data were presented as means ± SD. Statistical significance was set at *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; ns, not significant

LNP-M efficiently delivers reporter genes in vivo with minimal toxicity

While the DSPC-NLS/Histone-LNP system showed higher transfection efficiency in vitro, it raises regulatory concerns, as neither NLS nor histone is a component of any FDA-approved drugs. Therefore, we used the LNP-M/DNA system without NLS or histones for all subsequent in vivo experiments. We administered LNP-M/pLuc (a plasmid expressing firefly luciferase) into the muscle tissue of mice. Bioluminescence imaging showed effective expression of luciferase in the mouse muscle tissue (Fig. 2I and J). Similarly, injection of LNP-M/pGFP into muscle tissue resulted in effective GFP expression (Fig. 2K).

We next used the optimized LNP-M to deliver the pαH-Spike encoding HexaPro spike with a His tag. Western blot analysis showed effective expression of the spike protein in mouse muscle tissues (Fig. 2L). To assess whether LNP-M/DNA exhibited toxicity, we compared LNP-M with several common transfection reagents and a virus in vitro. LNP-M/pGFP demonstrated good transfection efficiency and cell viability in 293T cells compared to other transfection or infection methods (Fig. S3G–S3I). In vivo, mice administered the LNP-M/pSpike vaccine displayed normal weight gain over time (Fig. 2N). ELISA analysis revealed significant higher serum IL-6 and TNF-α levels in LNP-M/pSpike-vaccinated mice than the controls (either PBS or LNP-M/Ctrl) on day 2 post-vaccination; by day 7, the levels of the cytokines were comparable (Fig. 2M). Pathological examination of vital organs including the heart, liver, spleen, lung, and kidney, in immunized mice showed no increased inflammation compared to control animals (Fig. 2O). These results indicate that the optimized LNP-M formulation effectively delivers reporters and spike proteins in vitro and in vivo, with minimal toxicity.

LNP-M/pSpike vaccine induces strong protective immunity by enhancing immune cell activation and enriching TCRs and BCRs

Mice were immunized with Spike vaccines on days 0, 14, and 28. Each mouse received 40 μg of DNA delivered by LNP-M (LNP-M/pSpike) or electroporation (EP/pSpike), or 5 μg of Spike-mRNA delivered by LNP-M (LNP-M/Spike-mRNA). Antigen-specific immune responses were assessed on day 42. Compared to the control, the LNP-M/pSpike vaccine group exhibited a significant increase in serum antibody titers of total IgG and its subtypes, including IgG1, IgG2a, and IgG2c. The antibody titers in the LNP-M/pSpike vaccine group showed no significant difference compared to those in the LNP-M/Spike-mRNA or EP/pSpike vaccine groups (Fig. 3A). Additionally, levels of TNF-α and IFN-γ (Th1 cytokines) appeared to be higher in the LNP-M/pSpike group compared to the LNP-M/Spike-mRNA, EP/pSpike or control groups (Fig. 3B). In contrast, levels of IL-4 and IL-10 (Th2 cytokines) showed only minor differences among the group. To evaluate the vaccine-induced T cell immune response, we assessed the proliferation capacity, cytotoxicity, and Th1 cytokine production of CD4⁺ or CD8⁺ T cells in each group. CD4⁺ or CD8⁺ T cells from the spleens of LNP-M/pSpike-immunized mice displayed increased proliferation compared to those from LNP-M/Ctrl vaccine-immunized mice after stimulation with a spike peptide pool (Fig. 3C and D). Notably, CD8⁺ T cells exhibited more proliferation than CD4⁺ T cells. A similar trend was observed in the level of IFN-γ-secreting T cells, as measured by ELISPOT (Fig. 3E and F). Additionally, effector cells from the LNP-M/pSpike immunization group significantly lysed target cells (293T cells expressing spike) in a coculture assay compared to the LNP-M/Ctrl immunization group (Fig. 3G), indicating potent CTL effects. Intracellular staining further revealed higher percentages of IFN-γ⁺, IL-2⁺, or TNF-α⁺ CD4⁺ or CD8⁺ T cells induced by the LNP-M/pSpike or LNP-M/Spike-mRNA than the LNP-M/Ctrl (Fig. 3H–K).

Fig. 3

Antigen-specific immunity induced by LNP-M-delivered DNA- or mRNA-encoded spike. A Anti-spike antibodies in sera. Mice were immunized intramuscularly with the vaccines three times at 2-week intervals. Two weeks after the final immunization, the levels of serum IgG subtype antibodies were detected using ELISA. B Cytokine secretion by activated splenocytes. Splenocytes from immunized mice were cultured with IL-2 (100 U/ml) and a spike peptide pool (10 μg/ml) for 72 h before ELISA analysis of IFN-γ, TNF-α, IL-10, and IL-4. C and D T cell proliferation. The percentages of BrdU⁺ cells were assessed in gated splenic CD4⁺ or CD8⁺ T cells stimulated with a spike peptide pool. E and F ELISPOT analysis of splenic IFN-γ-secreting T cells. G CTL activity of splenic T cells. H-M Functional analysis of splenocytes. Cells were stimulated with the spike peptide pool (10 μg/ml) protein for 67 h before being treated with 500 ng/ml ionomycin, 50 ng/ml PMA, and 5 μg/ml Brefeldin A for an additional 5 h. FACS analyses were conducted using intracellular cytokine staining to assess the percentages of IFN-γ⁺, IL-2⁺, and TNF-α⁺ CD4⁺ or CD8.⁺ T cells. The experiments were performed with 5 mice per group. Data were represented as means ± SD. Statistical significance was set at ***p < 0.001, ****p < 0.0001

Multifunctional T cells that secret Th1-type cytokines such as IFN-γ, IL-2, and TNF-α are pivotal effector cells for providing protective immunity [32]. We evaluated the induction of multifunctional T cells following immunization. As shown in Fig. 3L and M, the percentages of dual-functional IFN-γ⁺TNF-α⁺CD4⁺, IFN-γ⁺IL-2⁺CD4⁺, TNF-α⁺IL-2⁺CD4⁺, IFN-γ⁺TNF-α⁺CD8⁺, IFN-γ⁺IL-2⁺CD8⁺, TNF-α⁺IL-2⁺CD8⁺ T cells, as well as multifunctional IFN-γ⁺IL-2⁺TNF-α⁺CD4⁺ and IFN-γ⁺IL-2⁺TNF-α⁺CD8⁺ T cells, were significantly higher in the spleens of mice immunized with LNP-M/pSpike compared to the control. Notably, the proportion of multifunctional IFN-γ⁺IL-2⁺TNF-α⁺CD8⁺ T cells, but not IFN-γ⁺IL-2⁺TNF-α⁺CD4⁺ T cells, in the LNP-M/pSpike group was higher than that in the LNP-M/Spike-mRNA group. These results indicate that LNP-M/pSpike vaccine effectively induces robust antigen-specific humoral and cellular immunity in mice.

To investigate the mechanism of the immune response induced by LNP-M/pSpike, we analyzed the immune cell fractions in the spleens of the animals using FACS. The LNP-M/pSpike group exhibited higher levels of CD9⁺ B, CD3⁺ T, CD4⁺ T, CD8⁺ T cells, NK, DC, and macrophage compared to the control group. No differences were observed between the LNP-M/pSpike and the LNP-M/Spike-mRNA groups (Fig. 4A). The proportions of Treg cells were similar among all three groups, while myeloid-derived suppressor cells (MDSCs) were significantly increased in the EP/pSpike group. We then performed scRNA-seq on CD45⁺ immune cells isolated from the spleens, identifying sixteen immune cell subsets (Figs. 4B and S4A-S4D). UMAP analysis revealed a notable increase in B2_B, MZ_B, plasmablasts, CD4⁺ T, CD8⁺ T, NK, macrophage, DC, and neutrophils in the LNP-M/pSpike group compared to the control. The percentage of MDSCs was reduced in the LNP-M/pSpike group, consistent with FACS results (Fig. 4C). The levels of DEGs for specific cell markers were shown in Fig. S4d. KEGG and GO enrichment analyses highlighted significant antigen presentation by B cells, macrophages and DCs, immune receptor reactions by NK, and activation of T cells in the LNP-M/pSpike group (Fig. S4E and S4F).

Fig. 4

Single-cell analyses of splenocytes from mice immunized with LNP-M/pSpike. Mice were immunized three times using LNP/DNA, LNP/mRNA, or EP/DNA at 2-week intervals, and splenocytes were taken two weeks after the last dose for FACS and scRNA-seq analyses. A FACS of immune cell subtypes. B UMAP of CD45⁺ immune cells. C Bar graphs showing the percentages of immune cell subtypes. D Scatter plots compared the outgoing and incoming interaction strengths in a 2D space between the LNP-M/pSpike group and other groups. E Analysis of total TCR clonotype abundance by sample and type using the abundance contig function. F Assessment of CDR3 peptide length by sample using the length contig function. G Clonal homeostatic space representations, showing the relative proportional space occupied by specific clonotypes of TCR amino acid across different vaccine groups. H Box plot showing clonotype diversity of VDJC genes. Clonotype diversity was calculated as the Shannon index, inverse Simpson index, or Chao index within T cells. I BCR clonotype abundance in various vaccine groups. J CDR3 peptide length. K Relative proportional space occupied by specific clonotypes of BCR amino acids. L clonotype diversity of IG genes. Data were represented as means ± SD. Statistical significance was set at *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; ns, not significant

To identify the signaling pathways contributing to the dramatic signaling changes in immune cell activation, we analyzed the differential outgoing and incoming interaction strengths of each signaling pathway in B2_B, CD4_T, CD8_T, NK, macrophage, and DC. In the LNP-M/pSpike group, incoming signaling of L1CAM in B2_B, IL-4 in CD4_T, IFN-II and PECAM1in CD8_T, NKG2D and CD86 in NK, IL-1in macrophage emerged as the most predominantly increased signals. In contrast, TIGIT signal in immune cells predominantly increased in the control (Fig. 4D). These largest incoming interaction strengths indicated increased induction of activated immune cells and decreased induction of immune inhibitory cells by LNP-M/pSpike. Compared to LNP-M/Spike-mRNA and EP/pSpike, the LNP-M/pSpike vaccine group exhibited increased signaling changes in the incoming signaling of CD80 and IFN-II in immune cells and decreased signaling changes in the incoming signaling of TIGIT.

The T cell receptors (TCRs) recognize peptide antigens presented by major histocompatibility complexes (MHCs) from APC to promote T cell activation [33,34,35]. We first analyzed TCR sequences from scRNA-seq, which revealed a significant increase in the number and abundance of TCRs and complementary-determining region 3 (CDR3s) in the LNP-M/pSpike group compared to the control (Fig. 4E and F). The space-occupied area and clonal diversity of TCR clonotypes also increased (Fig. 4G and H). A substantial increase in the number and abundance of B cell receptor (BCR) clones was observed in the LNP-M/pSpike group compared to the control group (Fig. 4I), as was the number or proportion of various CDR3 lengths and space occupied (Fig. 4J and K). The clonal diversity of VDJ genes was also higher in the LNP-M/pSpike group (Fig. 4L). Gene set variation analysis (GSVA) showed that LNP-M/pSpike immunization induced the strongest TCR signaling pathway, and LNP-M/Spike-mRNA induced the strongest BCR signaling pathway among the four groups (Fig. S5A and S5B). These TCR and BCR sequencing data revealed that the LNP-M/pSpike vaccine activates both humoral and cellular immunity, with a preference for T cell response.

LNP-M-delivered vaccines and antibodies elicit potent therapeutic effects in various tumor models

We investigated the therapeutic benefits of LNP-M-delivered DNA-encoded biologics, including vaccines and antibodies. We targeted the tumor surface antigen PD-L1 and mutant p53-R172H using LNP-M to deliver DNA-encoded mouse PD-L1 and mouse p53^R172H with treatments administered intramuscularly on days 7 and 17, and intratumorally on day 24 after tumor inoculation. Treatment with the LNP-M/pPD-L1 vaccine in the A20 subcutaneous tumor model induced PD-L1-specific antibodies (Fig. 5A) and significantly suppressed tumor growth (Fig. 5B and C). Similarly, treatment with the LNP-M/pR172H vaccine in the 344SQ subcutaneous tumor model significantly induced anti-R172H antibodies in mouse sera (Fig. 5D) and suppressed tumor growth (Fig. 5E and F).

Fig. 5

Therapeutic effects of LNP-M/DNA against PD-L1, p53^R172H, PD1, or p53^R282W in several syngeneic tumor models. A-C LNP-M/DNA expressing PD-L1 elicits anti-PD-L1 antibodies and anti-tumor activities. BALB/c mice (n = 5 per group) were immunized with three doses of LNP-M/DNA expressing mouse PD-L1 at days 7, 17, and 24. A20 tumor cells were inoculated on day 0. The first twos doses were delivered intramuscularly and the last dose intratumorally. Two weeks after the last dose, sera were taken to measure the titers of anti-PD-L1 IgG antibodies (A), and tumor volumes (B) and inhibition (C) were assessed. D-F LNP-M/DNA expressing p53-R172H elicits anti-p53 antibodies and anti-tumor activities. 129 Sv/E mice (n = 5 per group) were immunized with three doses of LNP-M/DNA expressing mouse p53-R172H at days 7, 17, and 24. 344SQ tumor cells were inoculated on day 0. The first two doses were delivered intramuscularly and the last dose intratumorally. Two weeks after the last dose, sera were taken to measure the titers of anti-p53^R172H IgG antibodies (D), and tumor volumes (E) and inhibition (F) were assessed. G-I LNP-M/DNA expressing the antibody against mouse PD1 elicits anti-tumor activities. MC38 tumor cells were inoculated into C57BL/6J mice, and LNP-M/pPD1-mAb was injected intratumorally on days 14 and 21 post-tumor inoculation (n = 5 mice per group). Five days after the first dose, the expression of PD1 on CD8⁺ T cells from the tumors was detected using FACS. Tumor volumes (H) and inhibition (I) were assessed. J-N LNP-M/DNA expressing the antibody against human p53^R282W elicited anti-tumor activities in MC38 syngeneic models. MC38-p53^KO/R282W cells were injected subcutaneously into mice and given intratumoral injections of LNP-M/pR282W-mAb on days 14 and 21 post-tumor inoculation (n = 5 mice per group). The percentages of Cd45⁺ and human Fc⁺ cells within tumors were determined using FACS (J). Tumor volumes were measured (K), and tumor inhibition was calculated (L). In the metastatic model, MC38-p53^KO/R282W cells were inoculated intravenously (n = 10 mice per group), and mice were treated with LNP-M/pR282W-mAb; representative images of metastatic nodules on the lung surface and animal survival were shown in (M). MC38-p53^KO/R282W cells were inoculated intraperitoneally before treatment (n = 10 mice per group). Representative images of rectal MC38-p53^KO/R282W tumors and animal survival were shown in (N). O-Q LNP-M/pR282W-mAb inhibits HupT3 tumors. NSG mice were inoculated with HupT3 cells with an endogenous p53 R282W mutation on day 0, given PBMCs (1 × 10⁷/mouse) intravenously on day 10, and treated by LNP-M/pR282W-mAb on days 14, 19, and 24 (n = 5 mice). FACS detected the percentage of human CD56⁺ NK cells in TILs (O), tumor volumes were measured (P), and tumor inhibition rates were calculated (Q). Data were presented as mean ± SD. Statistical significance was set at **p < 0.01, ***p < 0.001 and ****p < 0.0001; ns, not significant

Next, for LNP-M-delivered DNA-encoded antibody therapy, we examined the effect of the immune checkpoint PD1 and the p53^R282W antibodies. The subcutaneous tumor models were intratumorally injected with LNP-M/pPD1-mAb or LNP-M/pR282W-mAb on days 14 and 21 after tumor inoculation. In the MC38 subcutaneous tumor model, LNP-M/pPD1-mAb treatment disrupted PD1 expression on tumor-infiltrating CD8⁺ T cells (Fig. 5G) and significantly reduced tumor growth (Fig. 5H and I). Similarly, administration of LNP-M/pR282W-mAb in the MC38-p53^KO/R282W tumor model resulted in significant suppression of tumor growth, with 40% achieving a complete response (CR, i.e., tumor rejection; Fig. 5J–L).

We examined the anti-metastatic activity of the LNP-M/pR282W-mAb in an MC38 lung tumor model. MC38-p53^KO/R282W tumor cells were inoculated into the tail vein of animals, and intravenous LNP-M/pR282W-mAb treatments were administered 7 and 12 days after tumor inoculation. LNP-M/pR282W-mAb significantly suppressed lung tumor growth and tumor nodule formation and prolonged animal survival (Fig. 5M). Next, both tumor cells and the treatments were delivered intraperitoneally, and LNP-M/pR282W-mAb significantly suppressed intestinal tumor growth and increased animal survival (Fig. 5N). Finally, we evaluated the activity of LNP-M/pR282W-mAb against human cancer cells. Human pancreatic adenocarcinoma HupT3 cells expressing endogenous p53^R282W were inoculated into NSG mice on day 0, and human PBMCs were injected into tumor-bearing mice on day 10. LNP-M/pR282W-mAb was delivered intratumorally five days after tumor cell inoculation. We found that treatment with LNP-M/pR282W-mAb increased the number of tumor-infiltrating NK cells (Fig. 5O), leading to tumor cell destruction and tumor growth suppression (Fig. 5P), thereby significantly improving animal survival (Fig. 5Q). These results indicate that LNP-M-delivered cancer vaccines and antibodies effectively inhibit tumor development and metastasis against several solid tumors.

The therapeutic function of LNP-M/pR282W-mAb requires CD8⁺ T and NK cell immune responses

We investigated the underlying mechanisms of action of LNP-M/pR282W-mAb. Antibody-induced NK cells play a crucial role in cancer immunotherapy through antibody-dependent cellular cytotoxicity (ADCC) [36, 37]. After intratumoral administration of LNP-M/pR282W-mAb, we observed a marked increase in the percentages of CD107a⁺NK and IFN-γ⁺NK cells within the tumor (Fig. 6A and B). Additionally, the proportion of NKT cells expressing CD107a and IFN-γ also rose significantly (Fig. 6C and D), indicating a robust ADCC effect. APCs play a critical role in delivering tumor antigens to activated T cells, which in turn attack tumor cells [38, 39]. We found a significant increase in the numbers of CD103⁺CD11c⁺ and CD8⁺CD11c⁺ DCs within tumors treated with LNP-M/pR82W-mAb (Fig. 6E and F). There was a higher expression of activation markers, such as CD80, CD86, MHC-II, and MHC-I, on DCs in the LNP-M/pR282W-mAb group than in the control group (Fig. 6G and H). Furthermore, the proportions of tumor-infiltrating multifunctional IFN-γ⁺CD8⁺, IL-2⁺CD8⁺, and TNF-α⁺CD8⁺ T cells were significantly increased (Fig. 6I and J). Blocking antibodies were used to deplete CD8⁺ T or NK cells in treated animals and the therapeutic effects of LNP-M/pR282W-mAb were significantly impeded by each blocking antibody (Fig. 6K and L). However, depleting CD4⁺ T cells did not significantly affect the survival of mice treated with LNP-pR282W-mAb (Fig. 6M). These results suggested that anti-tumor immune responses of CD8⁺ T and NK cells are crucial for the therapeutic benefits of LNP-pR282W-mAb.

Fig. 6

The mechanistic study of anti-tumor effects induced by LNP-M/pR282W-mAb. A and B The levels of CD107a and IFN-γ expression in NK cells within TILs. C and D The expression of CD107a and IFN-γ in NKT cells within TILs. D and E The percentages of CD103⁺CD11c⁺ and CD8⁺CD11c⁺ cells within tumors. G and H The expression of DC activation markers (CD80, CD86, and MHC) on CD11c⁺ cells within tumors. I and J The percentages of multifunctional CD8⁺ T cells expressing IFN-γ, IL-2, and TNF-α in tumors. K-M The impact of blocking CD8⁺ T cells, NK cells, or CD4⁺ T cells on animal survival. Each animal (n = 10 per group) received intraperitoneal injections of 0.5 mg anti-mouse CD4, CD8α or NK1.1 mAb two days before the first dose of LNP-M/pR282W-mAb. The second and third doses were administered on days 5 and 12. N and O Memory immunity in mice treated with LNP-M/pR282W-mAb. Mice treated with LNP-M/pR282W-mAb were rechallenged with MC38-p53^KO/R282W cells on the left flank, with naïve mice serving as the control group. N Tumor growth curves. O Survival rate. P Representative images of FACS and statistical analysis of CD8⁺ T cells labeled by CD44 and CD62L in the splenocytes from both groups. Q Splenic memory T cells from mice treated with LNP-M/pR282W-mAb. Mice carrying subcutaneous MC38-p53^KO/R282W tumors were treated with LNP-M/pR282W-mAb, and the spleen was taken for FACS analyses of CD44^lowCD62L^high (stem cell memory): CD44^highCD62L^high (central memory), and CD44^highCD62L^low (effector memory) CD8.⁺ T cells. Data were presented as means ± SD. Statistical significance was set at **p < 0.01, ***p < 0.001, and ****p < 0.0001

To assess the anti-tumor memory from the LNP-M/pR282W-mAb treatment, we rechallenged mice that had a CR contralaterally with MC38-p53^KO/R282W cells. Tumor growth was substantially suppressed, and survival improved in these animals compared to naive controls (Fig. 6N and O). We analyzed memory CD8⁺ T cells from the spleens of rechallenged mice, which are key to long-lasting anti-tumor immunity. LNP-M/pR282W-mAb treatment elicited a robust increase in effector memory (CD44^highCD62L^low, T_em) and central memory (CD44^highCD62L^high, T_cm) CD8⁺ T cells and a decrease in naïve stem cell-like memory (CD44^lowCD62L^high, T_scm) CD8⁺ T cells (Fig. 6P and Q). These results suggest CD8⁺ T_em and T_cm cells contribute to the enduring protective effects of LNP-M/pR282W-mAb against tumors.

LNP-M/pR282W-mAb promotes the expansion of TCR diversity in CD8⁺ T cell anti-tumor immune responses

To further investigate the functional changes of CD8⁺ T cells within the tumor microenvironment following the LNP-M/pR282W-mAb treatment, we performed scRNA-seq of tumor tissues and defined seven cell subsets: non-immune_Cell1, non-immune_Cell2, T_cells, Mast cells, NK cells, APC1, and APC2 (Fig. S6A and S6B). T cells were enriched in the treated group, in agreement with the FACS data (Fig. S6C). Next, we refined the CD3⁺ cell populations into 12 subsets: Naïve CD8⁺ T (CD8_Tn), exhausted CD8⁺ T (CD8_Tex1 and CD8_Tex2), progenitor exhausted CD8⁺ T (CD8_Tpro), Memory-like CD8⁺ T ( CD8_Tem), Progenitor exhausted CD4⁺ T (CD4_Tpro), Regulatory T ( Treg), T follicular helper ( Tfh), T helper 1 (Th1), T helper 17 (Th17), Double negative T (DNT) cells, and CD3⁺ macrophages (CD3_Mac) (Figs. 7A, B, S6D and S6E). UMAP analysis revealed a remarkable increase in CD8_Tex1, CD8_Tex2, CD8_Tpro, CD4_Tpro, Tfh, DNT and CD3_Mac cells and a decrease of CD8_Tem, Treg, CD8_Tn, Th1, and Th17 cells (Fig. 7C). We applied Cytokine Signaling (CytoSig) [26] to analyze the cellular response to signaling molecules (e.g., cytokines, chemokines, and growth factors) of scRNA-seq data and found stronger signaling from IL-2, Brain-derived neurotrophic factor (BDNF), Epidermal growth factor (EGF), CXCL12, and IFN-γ in CD8_Tex1 cells, BDNF, IL-36, and CXCL12 in CD8_Tex2 cells, IL-1A, EGF, IL-36, and IFN-I in CD8_Tpro cells, and BDNF, EGF, and IFN-I in CD8_Tem cells in the LNP-M/pR282W-mAb group than the control (Fig. 7D). LNP-M/pR282W-mAb decreased the inhibitory signals from cytokines like IL-3, TNF-related weak inducer of apoptosis (TWEAK), IL-4, and TGF-β1 in CD8_Tex1 and CD8_Tex2 cells, IL-3 and TWEAK in CD8_Tpro cells, and IL-22, IL-3, TWEAK, IL-4, and TGFB1 in CD8_Tem cells (Fig. 7D). These results imply that LNP-pR282W-mAb promotes the activation of anti-tumor inflammatory cytokine signals and inhibits the activation of immunosuppressive signals in the CD8⁺ T cell-mediated anti-tumor activity. Efficient clearance of tumor cells relies heavily on MHC-I signaling of APCs that deliver tumor antigens to activated CD8⁺ T cells [40]. We found enhanced MHC-I receptor activity, antigen processing, and presentation in CD8_Tex2, CD8_Tpro, and CD8_Tem cells in the LNP-pR282W-mAb group (Fig. S6F). These results indicated that LNP-M/pR282W-mAb boosts MHC antigen presentation from APCs to T cells.

Fig. 7

Expansion of TCR diversity induced by LNP-M/pR282W-mAb in CD8⁺ T cell anti-tumor immune responses. A UMAP visualization of T cell-associated populations pooled across samples, which clustered into 12 distinct clusters. B Identification expression of representative marker genes. C The percentages of various T cell subtypes in the tumors from each group. D Summary of signaling activities in T cell subsets. The heat map includes cell types with significant differences in signaling activities between LNP-M/pR282W-mAb group and control group. E UMAP visualization overlay identifying the network interaction of clonotypes shared between clusters along the single cell dimension reduction. The relative proportion of clones transitioning from a starting node to a different cluster, visualized by arrows in four CD8⁺ T cell cluster networks. F Analysis of total TCR clonotype abundance by sample and type using the abundance contig function. G Assessment of CDR3 peptide length by sample using the length contig function. H Alluvial plots illustrating the frequencies of TCR clonotypes from each sample, in relationship to the top V(D)J pairing frequencies of expanded clonotypes in each group (right) and contacts (left) among T cell clusters. I The relative proportional space occupied by specific clonotypes of TCR across T cell subsets. J Dynamics of dominant clonotype sequences (amino acids) of TCRs across samples, colored by the types of dominant sequences. K Box plot showing clonotype diversity of 12 T cell subpopulations. Clonotype diversity was calculated as the clonal expansion index, cross-tissue migration index, or state transition index within T cell subpopulations

Subsequently, single-cell TCR mRNA sequencing of the tumor-infiltrating T cells (scVDJseq) was performed. Visualized TCR clonotype network interactions, along with single-cell dimensional reduction, illustrated the relatively increased proportions of CD8_Tex1, CD8_Tex2, CD8_Tpro, or CD8_Tem clones in LNP-M/pR282W-mAb group (Fig. 7E). Overlaying the clonal interaction network with UMAP allowed us to visualize the directionality of the network interactions among the CD8_Tn, CD8_Tex1, CD8_Te2, and CD8_Tpro clusters. The bidirectional movement of clonotype distribution of CD8_Tpex and CD8_Tex showed an increased interconnection between these two clusters through their TCRs in the LNP-M/pR282W-mAb group compared to the control group. The CD8_Tn and CD8_Tem subsets exhibited an efflux of the proportion of clones from the starting node in these clusters out to the CD8_Tex1, CD8_Tex2, and CD8_Tpex clusters, with more unique clones moving from the starting node to the ending node in the LNP-M/pR282W-mAb group than in the control group. Furthermore, a higher proportion of clones originating from CD8_Tex1, CD8_Tex2, and CD8_Tpex moved into the CD8_Tem clusters in the LNP-M/pR282W-mAb group. A significant increase in the number and abundance of TCRs and their complementary-determining region 3 (CDR3s) was observed in the LNP-M/pR282W-mAb group compared to the control (Fig. 7F and G). Next, we used alluvial plots to analyze the relationships between the top V(D)J pairing frequencies of the expanded TCR clonotypes and cell types. Tracking TCR clonotypes based on scVDJseq data revealed an increase in the number and expansion of clonal TCRs in the CD8_Tex1, CD8_Tex2, CD8_Tpex, CD8_Tem subsets following the LNP-M/pR282W-mAb treatment (Fig. 7H). TCR enrichment analysis revealed that T cell receptor V(D)J recombination and TCR complex were enhanced in CD8_Tex2, CD8_Tpro, and CD8_Tem cells in the LNP-M/pR282W-mAb group (Fig. S6G). The clonotypes of the dominant hyperexpanded CDR3 sequences increased in T cells after LNP-M/pR282W-mAb treatment (Fig. 7I and J). TCR clonal diversity was higher among CD8_Tex1, CD8_Tex2, and CD8_Tpro clusters in the LNP-M/pR282W-mAb group than that in the control group (Fig. 7K). These data revealed a visible CD8⁺ T cell immune response in tumors treated with LNP-M/pR282W-mAb.

Discussion

LNP/DNA-based delivery offers a cost-effective solution for gene therapy, vaccines, and antibody-based therapeutics [41, 42]. In this study, we evaluated all three FDA-approved LNP formulations in delivering DNA plasmids and found that LNP-B and LNP-M are effective for reporter gene expression in vitro and in vivo. LNP-M has a lower cost because DMG-PEG 2000 in LNP-M is cheaper than ALC-0159 in LNP-B. We used LNP-M with 3% DMG-PEG 2000 to deliver three vaccine antigens and two mAbs into mice and evaluated their therapeutic activities and underlying mechanisms.

The spike protein is the primary antigen used in vaccines against COVID-19 [43,44,45]. To assess the effectiveness of LNP-M delivering DNA-encoded spike, we immunized mice intramuscularly three times with a dosage of 40 μg/mouse for the DNA vaccine or 5 μg/mouse for the mRNA vaccine. The humoral immune response induced by LNP-M/pSpike was significantly higher than that of the control and comparable to those of LNP-M/Spike-mRNA and EP/Spike (three doses of 40 μg/mouse). LNP-M/pSpike elicited stronger T cell responses than LNP-M/Spike-mRNA and EP/Spike. These immune response profiles suggest that LNP-M is a versatile delivery system for DNA vaccines with broad applications in fighting infectious diseases like COVID-19. Both the DNA and RNA vaccines delivered by LNP-M induced stronger immune cell activation signals with fewer immune-suppressive signals than those delivered by EP/pSpike. This was likely due to the adjuvant activity of LNP-M [46]. At the same nucleic acid quantity, LNP/mRNA generally leads to significantly higher levels of protein expression than LNP/DNA due to the more efficient translation process of mRNA once inside the cell. However, DNA plasmids are inherently more stable, easier to manufacture at scale, and do not require cold-chain logistics, which are critical considerations for global deployment, particularly in low-resource settings.

PD-L1 is a cell surface protein that is frequently overexpressed on many tumors and is associated with the evasion of immune surveillance [47, 48]. PD-L1 binds to the PD1 receptor and inhibits the activation and proliferation of T cells, thereby reducing the anti-tumor immunity [49]. PD-L1 can serve as a target for tumor vaccines by inducing specific immune responses that kill tumor cells [32]. We evaluated the anti-tumor activity of LNP-M/PD-L1 that expressed mouse full-length PD-L1 against A20 tumor cells endogenously expressing the mouse PD-L1. When mice were intramuscularly immunized with LNP-M/PD-L1, antibodies against PD-L1 were induced to inhibit A20 tumorigenesis, presumably by blocking the interaction between PD1 and PD-L1 and restoring the ability of the immune system to kill tumor cells. Given PD-L1 is heavily modified post-translationally [50], in vivo expressed PD-L1 should be a better immunogen than PD-L1 fusion proteins purified from E. coli [51,52,53].

Mutations in the p53 gene, a crucial tumor suppressor, are common in approximately half of all cancers [54, 55]. These mutations often impair p53 function, leading to uncontrolled cell growth and tumor development. We developed a vaccine expressing the mouse p53^R172H protein delivered by LNP-M and assessed its activity against 334SQ tumor cells with an endogenous p53 ^R172H. Mice immunized with this vaccine with two intramuscular doses and one intratumoral dose developed tumors at a lower rate than the control. Previous studies have suggested that therapeutic tumor vaccines mainly suppress tumor growth and metastasis through a CD8⁺ T cell immune response mediated by APC [56, 57]. Further research is required to understand how the LNP-M/DNA vaccines augment the anti-tumor activities of APCs.

mAbs target specific antigens on cancer cells or immune cells while sparing other cells [58]. We evaluated the efficacy of intratumoral LNP-M-delivered mAbs targeting either a membrane protein (immune checkpoint PD1) or an intracellular protein (mutant p53^R282W). R282W is a hotspot p53 mutation with unique phenotypes in protein instability, aggregation, and gain-of-function (p53 R282W remains largely unfolded under physiological conditions) [59, 60]. LNP-M-delivered pPD1-mAb or pR282W-mAb effectively and sustainably suppressed tumor growth and metastasis in syngeneic mouse models and humanized mouse models. The therapeutic effects are primarily driven by the APC-mediated CD8⁺ T cell immune response accompanied by ADCC mediated by NK cells. Our scRNA-seq data supports the idea that LNP-M-delivered pR282W-mAb promotes CD8⁺ T cell maturation from naïve T cells to memory T cells and that its therapeutic efficacy depends on CD8⁺T and NK cells. Tumor-infiltrating B cells and plasma cells (TIL-Bs) play crucial roles in modulating the anti-tumor activity of NK and CD8⁺ T cells, contributing to the suppression of tumor growth and metastasis [61]. TIL-Bs inhibit tumor development via antigen presentation and antibody production to support both T cell responses and innate mechanisms. We believe that LNP-M-delivered pR282W-mAb amplifies the antibody effector functions of TIL-Bs, representing a novel form of intratumoral therapy.

Conclusions

In conclusion, our study demonstrates that LNP-M is an effective and safe delivery tool for DNA-based prophylactic and therapeutic agents. This study had several limitations. First, the precise mechanisms by which the LNP-M-delivered vaccine antigens and antibody treatments induce protective immunity remain unclear. Second, LNPs, particularly those with better cationic lipids, need to be optimized for more efficient delivery of DNA-encoded payloads. Finally, LNP-M-delivered DNA may require other agents to achieve a CR for cancer therapy.