mLumiOpto Is a Mitochondrial-Targeted Gene Therapy for Treating Cancer

Mitochondria, the cell’s powerhouses, are vital signaling organelles that regulate key cellular processes essential for cell growth and function (1). Mitochondrial genetics and biochemical metabolisms are implicated in various aspects of the cancer cell metastatic cascade, including motility, invasion, microenvironment modulation, plasticity, and colonization (2). Given their crucial role, mitochondria have emerged as a promising target for cancer treatment (3). Numerous mitochondrial-targeted anticancer therapies, such as mitocans (4), mitochondriotropics (5), and mitochondriotoxics (6), have been developed. However, these therapies typically target specific signaling pathways or proteins, such as hexokinase (7), Bcl2 family proteins (8), thiol redox (9), and VDAC/ANT (10), which may undergo unpredictive mutations during long-term treatment, leading to drug resistance and reduced efficacy (11). Consequently, the clinical translation of mitochondrial-targeted therapies has not yet succeeded.

The mitochondrion consists of two membranes, a relatively permeable outer membrane and a highly folded and impermeable inner mitochondrial membrane (IMM). Proper mitochondrial function relies on maintaining the electrical potential gradient across the IMM, known as ΔΨ_m. A profound and sustained dissipation of ΔΨ_m is a crucial trigger for cell death (12), making it a potential strategy for cancer treatment. Chemical uncouplers (e.g., FCCP and CCCP; refs. 13, 14) or permeability transition pore (mPTP) activators (e.g., Atr and ployP; refs. 15, 16) have been used to depolarize ΔΨ_m. However, lack of cancer specificity limits their in vivo utility, as mitochondrial activity is critical for all cells. Genetic methods can modulate mitochondrial function in specific tissues but do not directly target ΔΨ_m and often cause irreversible side effects. Thus, there is currently a dearth of approaches that can directly and specifically disrupt cancer cell ΔΨ_m. Recently, we developed mitochondrial optogenetics (mOpto) by expressing heterologous light-gated channelrhodopsin 2 (ChR2) in the IMM with a mitochondrial leading sequence (MLS; ref. 17). Sustained blue light illumination led to irreversible ΔΨ_m depolarization and substantial cell death in cells expressing mitochondrial ChR2. Despite its capability to induce cytotoxicity, mOpto requires external light that is difficult to penetrate deep tissues, limiting its in vivo utility and clinical translation for cancer treatment.

To achieve in vivo targeting of mitochondria, we developed a new generation optogenetic tool called mitochondrial luminoptogenetics (mLumiOpto). Specifically, we coexpressed cationic channelrhodopsin (CoChR), a blue light–gated channelrhodopsin from Chloromonas oogama (18) in the IMM and nanoluciferase (NLuc), an emission spectrum–matched luciferase from the deep-sea shrimp Oplophorus gracilirostris (19), in the cytosol of the same cells. We used a cancer-enhanced promoter (cfos) to maximize selective expression of mLumiOpto genes in tumor cells. Additionally, we designed a mAb-tagged exosome-associated adeno-associated virus (mAb-Exo-AAV) vehicle to deliver mLumiOpto genes to tumors in vivo. We hypothesized that mLumiOpto effectively induces cancer mitochondrial depolarization and cytotoxicity with the synthesized endogenous bioluminescence. Furthermore, we hypothesized that mAb-Exo-AAV facilitates cancer-specific gene delivery and functional expression of mLumiOpto, allowing targeted elimination of cancer cells with minimal side effects via synergizing mitochondrial depolarization–mediated cell death and mAb or AAV-mediated in situ immunity within the tumor microenvironment (TME).

To test these hypotheses, we examined the ability of mLumiOpto to induce mitochondrial depolarization and cytotoxicity across different cancer cell types, including glioblastoma (GBM) and triple-negative breast cancer (TNBC). We then assessed the cancer-specific surface binding, internalization, transduction efficiency, biodistribution, and tumor-specific expression of mLumiOpto. The therapeutic efficacy of mLumiOpto, delivered via AAV or mAb-Exo-AAV, was evaluated in preclinical mouse models with GBM or TNBC xenograft. Our results demonstrated that mLumiOpto effectively induces cancer cell death and significantly reduces tumor burden without impairing normal organs or tissues.

The animal studies were conducted according to the Institutional Animal Care and Use Committee (IACUC) Protocols IACUC-2022A00000035 and IACUC-2022A00000029, which were approved by the Institutional Biosafety Committee at the Ohio State University.

Viral Production Cells 2.0 (Gibco, cat. #A49784, RRID: RRID: CVCL_0045) were maintained in a chemically defined viral production medium supplemented with 4 mmol/L GlutaMAX in shaker flasks on an orbital shaker at 135 rpm. The human cervical cancer cell line HeLa (CLS, cat. #300194/p772_HeLa, RRID: CVCL_0030) was cultured in DMEM supplemented with 10% (v/v) FBS and 2 mmol/L L-glutamine. The TNBC cell lines MDA-MB-231 (ATCC, cat. #HTB-26, RRID: CVCL_0062), MDA-MB-231-FLuc (GenTarget, cat. #SC059-Puro, RRID: CVCL_YZ80), MDA-MB-468 (ATCC, cat. #HTB-132, RRID: CVCL_0419), and BT-20 (ATCC, cat. #HTB-19. RRID: CVCL_0178), human GBM LN-229 cells (ATCC, cat. #CRL-2611, RRID: CVCL_0393), and mouse GMB GL261 cells (Creative Bioarray, cat. #NCL-2108P28, RRID: CVCL_Y003) were cultured in DMEM/F12 medium supplemented with 10% FBS, 4 g/L glucose, 4 mmol/L L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. The mouse TNBC cell line 4T1-FLuc (ATCC, cat. #CRL-2539-LUC2, RRID: CVCL_5I85) was cultivated in RPMI-1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin. The human GBM cell lines U87 (ATCC, cat. #HTB-14, RRID: CVCL_0022) and U87-FLuc (ATCC, cat. #HTB-14-LUC2, RRID: CVCL_UR33) were maintained in Eagle Minimum Essential Medium with 10% FBS and 8 µg/mL blasticidin. Human GBM U251 (Millipore Sigma, cat. #09063001, RRID: CVCL_0021) and drug-resistant U251-TMZ (in-house developed) cells were maintained in Eagle Minimum Essential Medium with immune suppression markers IL1β, IL17A, and IL232 mmol/L L-glutamine, 1% nonessential amino acids, 1 mmol/L sodium pyruvate, and 10% FBS. All cell lines were incubated at 37°C and 5% or 8% CO₂ in a humidified incubator (Eppendorf). All media, supplements, and bioreagents used in this study were purchased from Thermo Fisher Scientific unless otherwise specified. All cell lines or patient-derived xenograft (PDX) lines, except specified, were commercially purchased, authenticated via polymorphic short tandem repeat analysis at University Genomics Core, and confirmed Mycoplasma-free with in-house PCR targeting 16S rRNA genes. The latest test date for all cell banks with stock vials of 30 to 100 was March 21, 2024. The time between cell thaw and use in our experiments was 2 to 3 weeks.

The ABCB, CoChR, and eYFP gene fragments were amplified from CAG-ABCB-ChR2-eYFP (17), AAV-Syn-CoChR-GFP (RRID: Addgene_59070; ref. 20), and pcDNA3.1-PsChR2-eYFP (RRID: Addgene_69057; ref. 21), respectively. The PCR primers are ABCB-forward, ABCB-reverse, CoChR-forward, CoChR-reverse, eYFP_1-forward, and eYFP_1-reverse (Supplementary Table S1). These gene fragments were cloned into pcDNA3.1-PsChR2-eYFP backbone vector using the HiFi Assembly Kit (New England Biolabs).

The ABCB-CoChR gene fragments were amplified from CMV-ABCB-CoChR-eYFP and cloned into pcDNA3.0-Magneto2.0-p2A-mCherry (RRID: Addgene_74308) backbone vector. The primers ABCB-CoChR_1-forward and ABCB-CoChR_1-reverse are listed in Supplementary Table S1.

The NLuc, 2A, and ABCB-CoChR-mCherry gene fragments were PCR amplified from pNL-CMV-NLuc (Promega, #N1091), pcDNA3.0-Magneto2.0-p2A-mCherry (RRID: Addgene_74308), and CMV-ABCB-CoChR-mCherry, respectively. The amplified genes were cloned into the CMV-ABCB-CoChR-mCherry vector. The PCR primers are NLuc_1-forward, NLuc_1-reverse, 2A-forward, 2A-reserve, ABCB-CoChR_2-forward, and ABCB-CoChR_2-reserve (Supplementary Table S1).

The NLuc, GFP, and 2A-ABCB-CoChR-mCherry gene fragments were PCR amplified from pNL-CMV-NLuc, CMV-myc-mito-GFP (RRID: Addgene_71542), and CMV-NLuc-2A-ABCB-CoChR-mCherry, respectively. The amplified genes were cloned into CMV-NLuc-2A-ABCB-CoChR-mCherry vector. The primers used are NLuc_2-forward, NLuc_2-reserve, GFP-forward, GFP-reserve, 2A-ABCB-CoChR-forward, and 2A-ABCB-CoChR-reserve (Supplementary Table S1).

The NLuc-2A-ABCB-CoChR gene fragment was PCR amplified from CMV-NLuc-2A-ABCB-CoChR-mCherry and cloned into the pAAV-DJ/8 expression vector (Cell Biolabs) following the manufacture’s instruction. The primers are NLuc-CoChR_2-forward and NLuc-CoChR_2-reverse (Supplementary Table S1).

The biosimilar of cetuximab, anti-EGFR mAb (Bio X Cell), was tagged to the surface of Exo-AAV via a mPEG-DSPE linker to generate the TNBC-targeting mAb-Exo-AAV. Following the procedure developed in our previous study (22, 23), Exo-AAV was labeled with fluorescent dye Cy5.5 PE [only for In Vivo Imaging System (IVIS) imaging] and modified with mPEG-DSPE at a molar ratio of 1:10,000:6,000,000 (Exo-AAV:Cy5.5:mPEG-DSPE). The Exo-AAV-PEG-Cy5.5 was then conjugated with anti-EGFR mAb via DSPE-PEG-NHS linker with a molar ratio of 1:2,680:13,000.

The purified AAV was digested with DNAse I to extract ssDNA and titrated using RT-PCR with primers: forward: 5′-ATT​GTC​CTG​AGC​GGT​GAA​A-3′, reverse: 5′-CAC​AGG​GTA​CAC​CAC​CTT​AAA-3′. The size distribution, morphology, biomarkers, and purity of mAb-Exo-AAV were characterized using NanoSight, transmission electron microscope (TEM), and Western blotting. The AAV packed in each exosome was titrated using RT-PCR with the same primers, and Exo-AAV was titrated using NanoSight to calculate AAV packing rate in Exo (i.e., AAV copy per exosome particle). The surface binging of mAb-Exo-AAV to TNBC was determined using flow cytometry, as previously described (24).

Cells were stained with mitochondrial membrane potential fluorescent dye TMRM (100 nmol/L) or MitoView 633 (25 nmol/L), as previously described (25). The fluorescence of TMRM and MitoView 633 was imaged with the 543- and 635-nm laser, respectively, using the Olympus FV1000 confocal microscope (Olympus). Images were analyzed offline using ImageJ software (RRID: SCR_003070, NIH).

Cells cultivated on a 15-mm glass-bottom dish were transfected with CMV-ABCB-CoChR-eYFP plasmid. Forty-eight hours after transfection, cells were loaded with MitoTracker Deep Red (250 nmol/L) for 30 minutes. The CoChR-YFP and MitoTracker were simultaneously imaged with confocal microscopy. For NLuc and CoChR co-expression analysis, cells were transfected with CMV-NLuc-GFP-2A-ABCB-CoChR-mCherry plasmid. Forty-eight hours later, the expression of NLuc-GFP and CoChR-mCherry was simultaneously imaged with confocal microscopy. Images were processed offline using ImageJ software for colocalization analysis.

mAb-Exo-AAV and AAV carrying mLumiOpto genes were labeled with sulfo-Cyanine5.5 and sulfo-Cyanine3 (red) from Lumiprobe, respectively. TNBC MDA-MB-468 cells were infected with BacMam GFP Transduction Control (green, Thermo Fisher Scientific) for 24 hours. Then mAb-Exo-AAV-Cy5.5 was incubated with TNBC MDA-MB-468 cells at 37°C for 2 hours. The fluorescence of GFP and Cy5.5 was imaged with the 510 and 694 nm laser, respectively, to monitor the internalization of mAb-Exo-AAV. AAV-Cy3 was incubated with MDA-MB-468 cells that were stained with DAPI for 20 minutes. The fluorescence of DAPI and Cy5.5 was imaged with the 461- and 694-nm laser, respectively.

Cells cultivated on glass coverslips or tissue sections were fixed in 4% formaldehyde in PBS and treated with PBS containing 10% goat serum and 0.3% Triton X-100 to block nonspecific staining. Samples were then incubated overnight at 4°C with anti-TOMM20 (1:200 dilution; Abcam, cat. #ab205486, RRID: AB_2943509) and anti–CYCS (1:200 dilution; Cell Signaling Technology, cat. #11940, RRID: AB_2637071) primary antibodies, Thereafter, samples were washed with 0.1% BSA in PBS, blocked again with blocking buffer (30 minutes at room temperature), and incubated with 1:200 diluted secondary antibodies labeled with AF488 and AF647 in 1% BSA, 1% goat serum, and 0.3% Triton X-100 in PBS. Images were acquired by confocal microscopy.

Cells were seeded onto 96-well plates at a density of 5 × 10⁴ cells/mL and transfected with mLumiOpto plasmid (DNA:cells = 1.2 µg:1 × 10⁶ cells). Forty-eight hours later, ViviRen (0–60 µmol/L), an engineered luciferin purchased from Promega (P1232), was added to the culture. The mock-transfected cells subjected to the same treatment were used as control. Two days later, cell viability was measured using the MTT assay (Thermo Fisher Scientific) following the manufacturer’s instruction.

To delineate the mechanistic pathway underlying mLumiOpto-mediated cytotoxicity, cells were treated with mLumiOpto for 48 hours with or without the presence of a pan-caspase inhibitor Z-VAD-FMK (20 µmol/L), a necroptosis inhibitor 7-Cl-O-Nec-1 (100 µmol/L), a caspase-9–specific inhibitor Z-DEVD-FMK (100 µmol/L), a caspase-3 (CASP3)–specific inhibitor Z-LEHD-FMK (20 µmol/L), a caspase-8–specific inhibitor Z-IETD-FMK (20 µmol/L), an mPTP opening inhibitor cyclosporin A (10 µmol/L), or a mitochondrial-specific antioxidant MitoQ (300 nmol/L). Cell viability was measured using a TC20 automated cell counter (Bio-Rad) or the MTT assay.

Cell lysate or AAV samples were subjected to SDS-PAGE in NuPAGE 4% to 12% Bis-Tris gels. After electrophoresis, proteins were electro-transferred to a polyvinylidene difluoride membrane (Thermo Fisher Scientific) and subjected to the immunoblot assay by primary antibodies followed by secondary antibodies. Antibodies of cleaved CASP3 (Cell Signaling Technology, cat. #9661, RRID: AB_2341188), cleaved PARP (Cell Signaling Technology, cat. #9148, RRID: AB_10827981), LC3B (Cell Signaling Technology, cat. #8899, RRID: AB_2797680), CYCS (Cell Signaling Technology, cat. #11940, RRID: AB_2637071), and GAPDH (Cell Signaling Technology, cat. #5174, RRID: AB_10622025) were purchased from Cell Signaling Technology. TOMM20 (Abcam, cat. #ab205486, RRID: AB_2943509) was purchased from Abcam. VP1, VP2, and VP3 (ARP American Research Products, cat. #03-61057, RRID: AB_1540382) were obtained from ARP American Research Products. Quantification analysis of blots was performed using ImageJ software. Targeted bands were normalized to GAPDH.

Six-week-old nude (J:NU HOM Homozygous for Foxn1<nu>) mice (RRID: IMSR_JAX:007850), with an equal number of males and females, were stereotactically injected with human GBM cells as previously described (22). Briefly, 0.5 × 10⁶ U87 cells (ATCC, cat. #HTB-14, RRID: CVCL_0022), a widely used human GMB cell line (26, 27), were suspended in 3-µL growth medium and implanted into the frontal region of the cerebral cortex at 0.4 µL/minutes using Stoelting Just for Mouse Stereotaxic Instrument (Stoelting). The burr hole in the skull was closed with sterile bone wax, and 5 mg/kg carprofen was administrated immediately before surgery and every 12 to 24 hours for 48 hours post-surgery. The intracranially xenografted mice were monitored daily for 7 days and used for in vivo studies.

The GBM PDX line was provided by Dr. Jann Sarkaria at Mayo Clinic and maintained at low passages (2–4) in NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice (RRID: IMSR_JAX:005557) as previously described (28). To establish the PDX model, 0.5 × 10⁶ PDX cells were intracranially implanted into nude mice following the same process as U87 implantation. When PDX tumor reached ∼1 mm³, detected with MRI in ∼week 6, the xenografted mice were treated with saline (control) or mLumiOpto. GBM PDX tumor growth was monitored using MRI until week 14 or tumor volume reached >3.5 mm³.

Five million MDA-MB-231-FLuc or MDA-MB-231 cells were orthotopically injected into the mammary fat pad of 6-week-old NSG (RRID: IMSR_JAX:005557) female mice. When tumor volume reached ∼75 to 100 mm³, mice were intravenously administrated anti-EGFR mAb-Exo-AAV (2 × 10¹³ ptc/kg-BW), free AAV (60 × 10¹³ vg/kg-BW), ViviRen (2 mg/kg-BW), or saline in 50 µL via tail vein injection (n = 6/group). Four days later, mice in the mAb-Exo-AAV group received daily ViviRen injections (2 mg/kg-BW) for three consecutive days. Tumor volumes were measured using an external caliper every 2 days. At the end of the treatment, tumor tissues and major organs were harvested for paraffin sectioning, hematoxylin and eosin (H&E) staining, and biochemical analysis.

Three million mouse 4T1-Fluc cells (ATCC, Cat. # CRL-2539-LUC2, RRID: CVCL_5I85) were subcutaneously injected into the mammary fat pad of 6-week-old BALB/cJ female mice (RRID: IMSR_JAX:000651). When tumor volume reached ∼75 to 100 mm³, mice were randomly divided into two groups (n = 6/group) and received saline or anti-EGFR mAb-Exo-AAV (2 × 10¹³ ptc/kg-BW) on day 0 (i.v. injection), followed by ViviRen administration (2 mg/kg-BW) via tail vein injection on days 4 to 6. Tumor volumes were measured every 2 days using an external caliper, and the wet weight of terminal tumors was recorded at the end of the study. Tumor tissues were used for flow cytometry analysis to detect immune cells infiltration in the TME.

Cells were seeded in clear-bottom black well plates and transfected with Renilla luciferase (RLuc) or NLuc plasmid. After 48 hours, the medium was replaced with a colorless medium containing ViviRen (0–30 µmol/L). Bioluminescence was detected at 0, 2, 4, 6, and 20 hours using IVIS Lumina Series III (PerkinElmer) at 470 nm.

TNBC tumor-bearing mice received a single dose of mLumiOpto mAb-Exo-AAV (1 × 10¹³ ptc/kg-BW, i.v. injection) and ViviRen (2 mg/kg-BW, i.v. injection). Twenty-four hours later, mice were imaged with IVIS to measure NLuc luminescence. To evaluate tumor-specific targeting and biodistribution, TNBC MDA-MB-231-FLuc xenograft mice were injected with mAb-Exo-AAV (1 × 10¹³ ptc/kg-BW, i.v. injection). Mice were imaged after 24 hours with an exposure time of 10 seconds. Tumor-specific targeting was determined by (i) analyzing the overlay of NLuc luminescence and tumor (24); (ii) ex vivo IVIS imaging of the harvested tumor and major organs; and (iii) analyzing the transcripts of tumor and major organs.

MRI was performed using the BioSpect 94/30USR system (Bruker BioSpin) at the Small Animal Imaging Core Facility of the Ohio State University. T2-weighted scans were acquired with the following parameters: TR/TE: 2500/33 (ms), FA: 180 (degree), NEX: 2, FOV: 20 mm × 15.313, matrix: 256 × 196, slice thickness: 1 mm, slice distance: 1 mm, and slices: 18. Mice were anesthetized, and a 0.2 mmol/kg gadolinium-based contrast agent was administrated intraperitoneally before imaging. The mice were then secured on an animal bed and placed in the MRI scanner. A rectal thermometer was used to measure body temperature, and respiration and heart rate were monitored using the Small Animal Monitoring System (Small Animals Instruments) during the imaging session.

The tissues were dehydrated in ethanol, cleared in xylene, embedded in paraffin, sectioned at 5 μm, and mounted on frosted microscope slides. The paraffin-sectioned slides were dewaxed with xylene and gradient hydrated with 100% to 50% ETOH and dH₂O. The hydrated slides were stained with hematoxylin, rinsed with deionized water, dipped in 1% HCl in 70% ETOH, immersed in 1% NH₄OH for blue color development overnight, and stained with eosin for 30 seconds. The stained slides were dehydrated in 95% and 100% ethanol and cleared in xylene.

Apoptotic cells were assessed using the DeadEnd Fluorometric TUNEL System (Promega) according to the manufacturer’s instruction. Briefly, the slides were mounted with Permount mounting medium, with the nuclei counterstained with DAPI. The slides were then dried and imaged with a fluorescent microscope. Data were analyzed using ImageJ software.

Total RNA was extracted and purified from frozen tissues using the RNeasy mini kit (Qiagen). cDNA was synthesized from 500 ng of RNA using QuantiTect Reverse Transcription Kit (Qiagen). qRT-PCR was performed using Select Master Mix (Thermo Fisher Scientific) in a Bio-Rad IQ5 detection system (Bio-Rad). The transcript levels of the CoChR and NLuc genes were normalized to the average levels of Gapdh and Rpl32. PCR primers are listed in Supplementary Table S2, with specificity confirmed by 1% agarose gel electrophoresis and melt curves. Fold differences in mRNA expression were calculated using 2^−ΔΔCt.

AAV (3.3 × 10¹² vg/kg-BW) was intracerebroventricularly (i.c.v.) injected to ∼2 mm GBM U251 xenografted nude mice (n = 4). EGFR mAb-Exo-AAV (2 × 10¹³ ptc/kg-BW) was i.v. injected to 100 mm³ TNBC MDA-MB-231 xenografted NSG mice (n = 4). Seven days post-infection, tumor samples were harvested and dissociated with a Tissue Dissociation Kit (101Bio). The 1 × 10⁶ dissociated tumor cells were stained with 1 µg of anti-Ki67 antibody labeled with AF488 (Cell Signaling Technology, cat. #9129, RRID: AB_2687446) and 1 µg of NLuc antibody (Promega, cat. #N7000, RRID: AB_3095534) labeled with AF647 (Thermo Fisher Scientific) at 37°C for 30 minutes. The average fluorescence intensity was determined using the FACSCalibur flow cytometer (Becton-Dickinson), and data were analyzed using FlowJo 7.6.1 software (RRID: SCR_008520, TreeStar).

Freshly isolated tumor tissues were dissociated with a Tissue Dissociation Kit following the manufacturer’s instructions. The dissociated tumor cells were stained with AF488 anti-CD8 antibody (BioLegend, cat. #100723, RRID: AB_389304) and APC anti-CD11c antibody (BioLegend, cat. #100723, RRID: AB_389304) and analyzed by flow cytometry to assess infiltrated immune cells.

Chemokines and cytokines within the TME were measured using a Luminex-based multiplexing assay kit (Luminex Corporate). A preconfigured 26-plex chemocytokine assay kit (EPX070-20835-901) was purchased from Thermo Fisher Scientific. All assay reagents were prepared following manufacturer’s instructions. The Luminex assay was performed in 96-well plates provided with the kit, and the raw data of mean fluorescence intensity were read using the Luminex MAGPIX with xPONENT software.

The blood samples were drawn from the heart for blood cell count using HemaVet 950FS (Drew Scientific). The erythrocytes (red blood cell and hemoglobin) and leukocytes (lymphocyte, monocyte, white blood cell, and neutrophil) were titrated to analyze the general peripheral immune response.

The experimental data were presented as mean ± SEM. Statistical comparisons among groups were performed using the two-way ANOVA Tukey multiple comparisons test or one-way ANOVA Holm–Sidak multiple comparisons test. P < 0.05 was considered statistically significant. The distribution of data was tested using the Shapiro–Wilk normality test.

All raw data generated in this study are available upon request from the corresponding author.

We constructed an expression vector to coexpress light-gated rhodopsin in the IMM and an emission spectrum–matched luciferase in the cytoplasm of cancer cells. Specifically, we synthesized the mLumiOpto (NLuc-2A-ABCB10-CoChR) by cloning CoChR (peak λ_ex = 470 nm) and NLuc (peak λ_em = 460 nm), which were fused through a cleavable 2A linker, into a pcDNA3.0 expression vector (Fig. 1A). We chose CoChR over the more commonly used ChR2 due to its higher (∼10-fold) photocurrent (18) and greater efficiency in inducing mOpto-mediated ΔΨ_m depolarization (Fig. 1B). NLuc was selected for its much brighter bioluminescence than other discovered blue light–emitting luciferases such as RLuc (29) when paired with ViviRen, an engineered luciferin (Fig. 1C). We fused the ABCB10 MLS to the N-terminal of CoChR to ensure mitochondrial expression. Consistent with previous studies (17), ABCB10 MLS led to high-level and mitochondrial-specific CoChR expression across various tumor cell lines, including HeLa and TNBC MDA-MB-231 (Fig. 1D), as indicated by the strong overlap between eYFP (fused with CoChR) and MitoTracker (a mitochondrial indicator). Confocal microscopy confirmed coexpression of NLuc (fused with eGFP) and CoChR (fused with mCherry) in mLumiOpto-transfected MDA-MB-231 cells (Fig. 1E).

To validate the functionality and efficiency of mLumiOpto in mediating cancer cell mitochondrial depolarization, MDA-MB-231 cells were transfected with the NLuc-2A-ABCB10-CoChR plasmid and treated with varying doses (0–100 μmol/L) of ViviRen. ViviRen elicited intracellular NLuc luminescence (Fig. 1F) and ΔΨ_m depolarization (Fig. 1G) in a dose-dependent manner, confirming the functional expression of NLuc and CoChR proteins and the capability of mLumiOpto to depolarize mitochondria. Prolonged exposure to ViviRen (48 hours) caused a dose-dependent reduction in cell viability in NLuc-2A-ABCB10-CoChR–transfected MDA-MB-231 cells (Fig. 1H). This mLumiOpto-mediated cytotoxicity was also observed in multiple human GBM (U251 and U87) and TNBC (BT-20 and MDA-MB-468) cell lines, exhibiting substantial cell death with ViviRen induction (Fig. 1I). Importantly, neither mLumiOpto plasmid transfection nor ViviRen induction alone significantly affected cancer cell ΔΨ_m (Supplementary Fig. S1) and viability (Fig. 1H and I).

We next investigated the mechanisms underlying mLumiOpto-mediated cancer cell death. We examined the effect of apoptosis inhibitor Z-VAD-FMK and necroptosis inhibitor ecrostatin-1 (Nec-1) on the viability of mLumiOpto-expressing MDA-MB-231 cells treated with ViviRen. Z-VAD-FMK significantly reduced cell death, whereas Nec-1 had no noticeable effect (Fig. 2A), suggesting activation of caspase-dependent apoptosis. To determine whether the apoptotic pathway was intrinsic or extrinsic, we co-treated cells with ViviRen and caspase-specific inhibitors. Both caspase-9 inhibitor Z-DEVD-FMK and Casp-3 inhibitor Z-LEHD-FMK, but not caspase-8 inhibitor Z-IETD-FMK, effectively attenuated mLumiOpto-induced cytotoxicity (Fig. 2A), indicating activation of the intrinsic apoptotic pathway. Apoptosis activation was confirmed by increased expression of apoptosis markers cleaved Casp-3 and PARP (Fig. 2B), Casp-3 activity (Fig. 2C), and terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling–positive cells (Fig. 2D) in mLumiOpto-treated cells compared with controls (i.e., mock-transfected). We also observed increased cytochrome C release in mLumiOpto-treated cells compared with controls, revealed by punctuated cytochrome C staining disparate from TOMM20 (Fig. 2E) and confirmed by elevated cytosolic cytochrome C expression (Supplementary Fig. S2A). The expression of the autophagy marker LC3B also increased in treated TNBC cells (Supplementary Fig. S2B), whereas the necrosis marker HMGB remained unchanged (Supplementary Fig. S2C). Similar results were observed in GBM U251 cells, where the apoptosis inhibitor attenuated mLumOpto-induced cytotoxicity, but necrosis inhibition was trivial (Supplementary Fig. S2D). Consistently, the expression of cleaved caspase-9, cleaved Casp-3, and LC3B, but not the necrosis marker HMGB1, significantly increased following mLumiOpto treatment, revealed by intracellular flow cytometry (Supplementary Fig. S2E).

Further investigations revealed nuclear condensation and fragmentation in treated TNBC cells, as observed through Syto24 staining (Fig. 2F). Western blot analysis confirmed DNA damage in mLumiOpto-treated cells, indicated by significantly increased γH2AX expression (Fig. 2G). We also examined the effects of mPTP inhibitor cyclosporin A and mitochondrial-specific antioxidant MitoQ on mLumiOpto-mediated cytotoxicity, finding that neither significantly influenced cell death (Fig. 2H). These data indicate that mLumiOpto induces cancer cell cytotoxicity primarily through mitochondrial-mediated intrinsic apoptosis and DNA damage, independent of canonical mPTP opening and excessive oxidative stress.

To deliver synthesized mLumOpto genes to cancer cells in vivo, we constructed an AAV expression vector using the commercial hybrid serotype AAV-DJ/8 with a heparin-binding domain mutation, which shows high infection efficiency in vivo (30). Additionally, we utilized the cfos promoter (31) to enhance cancer-selective gene expression. AAV was produced in a stirred-tank bioreactor and purified using ion-exchange liquid chromatography as previously described (32). The size (∼20 nm) and morphology of purified AAV DJ/8 were verified using TEM (Fig. 3A). Western blotting confirmed the expression of AAV capsid proteins VP1, VP2, and VP3 (Fig. 3B). The cfos promoter mediated remarkably higher GFP expression in U87 and MDA-MB-231 compared with noncancerous normal human astrocytes and mammary epithelial cells 184B5 (Supplementary Fig. S3A), confirming its high cancer selectivity. ViviRen triggered robust luminescence (Fig. 3C) and substantial mitochondrial depolarization (Fig. 3D) in AAV-transduced U87 cells, demonstrating functional expression of mLumiOpto proteins. Along with observed mitochondrial collapse, ViviRen induction caused dramatic cell death in various AAV-transduced GMB cell lines, including the drug-resistant U251-TMZ cells compared with the controls (saline, AAV only, and ViviRen only; Fig. 3E). Importantly, ViviRen induction did not significantly affect the viability of mLumiOpto AAV cocultured normal human astrocytes (Supplementary Fig. S3B) and 184B5 cells (Supplementary Fig. S3C).

To assess the in vivo gene delivery efficiency of AAV, we compared i.v. and i.c.v. injections in the GBM model. qRT-PCR analysis revealed that i.c.v. injection achieved GBM tumor-specific mLumiOpto gene delivery, with remarkably higher levels of NLuc (Fig. 3F) and CoChR (Supplementary Fig. S4) expression than i.v. injection. Live animal IVIS imaging (Fig. 3G) and ex vivo imaging of the isolated organs (Fig. 3H) confirmed functional mLumiOpto expression in AAV-transduced (via i.c.v. injection) GBM xenografts but not the normal organs. Flow cytometry revealed 80% to 90% of NLuc⁺ GBM and TNBC cells in the xenografts (Supplementary Fig. S5), indicating high in vivo infection efficiency. Consequently, direct intracranial administration was used for subsequent anti-GBM efficacy studies.

To evaluate mLumiOpto’s antitumor efficacy, U87 xenografted mice were randomly divided into four groups (n = 8–10/group) and received i.c.v. injection of saline (control), AAV only (3.3 × 10¹² vg/kg-BW), mLumiOpto 1 (AAV dose: 1.6 × 10¹² vg/kg-BW), and mLumiOpto 2 (AAV dose: 3.3 × 10¹² vg/kg-BW), respectively. Mice in the mLumiOpto groups received ViviRen (2 mg/kg-BW) through tail vein injection daily for three consecutive days following AAV administration. Survival of mice in mLumiOpto groups was significantly prolonged compared with controls (Fig. 4A), with similar body weight profiles across all groups (Fig. 4B). H&E staining showed a dramatic reduction in tumor burden with mLumiOpto treatment (Fig. 4C). IHC staining revealed increased cleaved CASP3 and Ki67 expression in mLumiOpto tumors (Fig. 4D), indicating apoptosis induction and proliferation inhibition. The immunofluorescence assay revealed evident cytochrome C release in the treated group, implying mitochondrial depolarization and injury (Supplementary Fig. S6). No damage to normal organs (brain, heart, lung, liver, spleen, and kidney) was detected (Fig. 4E), and no behavior changes were observed in the treated mice. IVIS imaging on day 28 (i.e., 10 days after the last ViviRen administration) revealed a dramatic reduction in GBM tumor volume in mLumiOpto groups (Fig. 4F; Supplementary Fig. S7). Endpoint MRI on day 44 confirmed reduced tumor burden in both mLumiOpto groups (Fig. 4G). It is worth noting that antitumor efficacy was comparable between the two mLumiOpto groups, suggesting further AAV dose optimization is needed.

We further investigated mLumiOpto’s anticancer efficacy using a GBM PDX mouse model. Mice received i.c.v. AAV injections (3.3 × 10¹² vg/kg-BW) in weeks 6, 8, and 10, followed by ViviRen induction (2 mg/kg-BW). MRI at the endpoint showed remarkable tumor burden reduction in the mLumiOpto-treated group compared with the saline group (Fig. 5A). H&E staining confirmed reduced tumor cell density (Fig. 5B). mLumiOpto significantly extended survival compared with controls (Fig. 5C), without affecting mouse body weights (Fig. 5D). IHC staining with cleaved CASP3 and Ki67 antibodies indicated mLumiOpto-induced apoptosis and proliferation inhibition (Fig. 5E). Similar to U87 xenograft models, immunofluorescence staining detected obvious cytochrome C release from mitochondria to the cytoplasm, suggesting mitochondrial depolarization following mLumiOpto treatment (Supplementary Fig. S8). H&E staining revealed no injury in major organs including the brain, heart, lung, liver, spleen, and kidney (Fig. 5F). These evaluations demonstrated the great potential of AAV-delivered mLumiOpto for managing GBM through depolarizing mitochondria.

To achieve highly efficient and targeted delivery of mLumiOpto genes to cancer cells in vivo, we developed a mAb-Exo-AAV delivery vehicle. We first produced high-quality and high-yield Exo-AAVs using viral production cells 2.0 in a 2-L stirred-tank bioreactor. We then surface-tagged the purified Exo-AAV with an EGFR mAb, cetuximab, via mPEG-DSPE linker, creating TNBC-targeting mAb-Exo-AAV (Fig. 6A). NanoSight analysis showed a mAb-Exo-AAV size distribution of 133.4 ± 74 nm (Fig. 6B), and TEM confirmed the morphology and particle size (Fig. 6C). The packed AAV exhibited high AAV packing rates of 20 to 50 vg/ptc of exosomes, with a mean of ∼30 vg/ptc. Alexa Fluor 488 dye was used to label mAb and detect the tagging ratio, which revealed 20 to 100 copies of mAb on each Exo-AAV particle. Flow cytometry showed strong surface binding of anti-EGFR mAb-Exo-AAV to EGFR⁺ TNBC MDA-MB-231 (>60%) and MDA-MB-468 (>95%) cells (Fig. 6D). Confocal microscopy confirmed that Cy5.5-labeled mAb-Exo-AAV bound to the surface of MDA-MB-468 cells within 20 minutes of incubation (Fig. 6E) and internalized within 2 hours. We also observed accumulation of Cy5.5-labeled AAV in >95% of cells 30 minutes post-transduction (Fig. 6F), indicating high transduction efficiency. ViviRen triggered strong NLuc bioluminescence in mAb-Exo-AAV–transduced MDA-MB-231 cells, demonstrating functional NLuc expression (Fig. 6G).

To evaluate the potential immune response of mAb-Exo-AAV in the TME and its effect on general immunity, we administered mAb-Exo-AAV, free AAV, and saline (control) to healthy BALB/cJ mice via tail vein injection. Two weeks later, whole blood samples were collected and analyzed. mAb-Exo-AAV had no significant effect on blood cell counts except for lymphocytes, which were approximately 30% higher than control (Fig. 6H) but within the reference range (0.9–9.3 K/µL). In contrast, free AAV significantly increased white blood cells and neutrophils while reducing monocytes (Fig. 6H). These results suggest that mAb-Exo-AAV causes less peripheral immunity than free AAV, but this observation requires further investigation.

We next conducted a comprehensive evaluation of mAb-Exo-AAV–delivered mLumiOpto for tumor treatment, including dosage tolerance, tumor targeting, biodistribution, and antitumor efficacy. To investigate the tolerated dosage and potential toxicity, various doses of mAb-Exo-AAV were i.v. injected into C57BL/6J mice, followed by administration of ViviRen 3 days later to induce intracellular bioluminescence. The mice maintained normal body weight, indicating no major toxicity at the tested dosages (Supplementary Fig. S9A). Whole blood analysis reported normal counts of erythrocytes, leukocytes, and thrombocytes (Supplementary Fig. S9B–S9D). Furthermore, H&E staining of major organs revealed no apparent inflammation, apoptosis, or necrosis (Supplementary Fig. S9E). Consistent with histology, echocardiograph showed normal cardiac function (Supplementary Fig. S9F). No liver and kidney damage were observed in mAb-Exo-AAV-treated BALB/cJ mice, indicated by similar serum levels of alanine transaminase, aspartate aminotransferase, blood urea nitrogen, and creatinine between the two groups (Supplementary Fig. S9G). These findings suggest that mAb-Exo-AAV is a safe vehicle for delivering mLumiOpto genes and mLumiOpto technology has minimal toxicity in healthy animals.

To assess tumor-specific targeting and biodistribution, TNBC MDA-MB-231 xenograft NSG female mice were i.v. administrated anti-EGFR mAb-Exo-AAV. Live animal IVIS imaging demonstrated strong NLuc luminescence in the tumors (Fig. 7A), indicating that mAb-Exo-AAV specifically targeted tumors. Consistent with in vivo imaging, ex vivo imaging detected bright Cy7 fluorescence in tumors, not in normal organs (Fig. 7B). Further biodistribution analysis in MDA-MB-231 xenograft mice revealed significant CoChR (Fig. 7C) and NLuc (Supplementary Fig. S10) expression in tumor tissue of mAb-Exo-AAV mice but not in that of controls. Importantly, CoChR expression in normal organs (e.g., heart, brain, lung, spleen, kidney, intestine, pancreas, stomach, colon, and liver) of mAb-Exo-AAV mice was undetectable (Fig. 7D). Collectively, these data demonstrated that anti-EGFR mAb-Exo-AAV specifically targets tumors and delivers mLumiOpto genes to EGFR⁺ TNBC in vivo.

To assess anticancer efficacy, anti-EGFR mAb-Exo-AAV was i.v. injected into MDA-MB-231 xenografts when tumor volume reached 25 to 50 mm³. Tumor-bearing mice given saline, free AAV, or ViviRen only served as controls. Remarkably, TNBC tumor stopped growing shortly after ViviRen induction, with tumor volume eventually shrinking in the treatment (i.e., mAb-Exo-AAV+ViviRen) group. In contrast, tumors grew rapidly in all control groups (Fig. 7E). H&E staining of paraffin-sectioned TNBC tumors, harvested 4 to 5 days after the last ViviRen induction, revealed severe cell death and reduced tumor cell density in mLumiOpto-treated mice compared with controls (Fig. 7F). Importantly, no apparent damage or injury was observed in normal organs of mLumiOpto-treated mice (Fig. 7G). Together, these results indicate that mAb-Exo-AAV–delivered mLumiOpto has high antitumor efficacy and minimal off-target toxicity.

The anticancer effectiveness of mLumiOpto was further assessed in immunocompetent mouse models. EGFR⁺ mouse TNBC 4T1 xenografted BALB/cJ female mice were administrated either saline or anti-EGFR mAb-Exo-AAV on day 0, followed by daily ViviRen injections on days 4 to 6. IVIS imaging after the first ViviRen induction showed robust luminescent response in 4T1 xenografts (Fig. 8A), confirming tumor-specific and functional NLuc expression. Similar to the MDA-MB-231 xenografts, 4T1 tumor growth was significantly inhibited in mAb-Exo-AAV mice compared with controls (Fig. 8B). Notably, the terminal tumor wet weight in mAb-Exo-AAV–treated mice was only 6% to 8% of that in the controls (Fig. 8C). Intriguingly, we observed significant infiltration of CD11c⁺ dendritic cells (DC) and CD8⁺ T cells in the TME of mLumiOpto-treated mice compared with the controls (Fig. 8D), suggesting enhanced tumoral immunity. To investigate the immune regulation effect of mAb-Exo-AAV carrying mLumiOpto, a Luminex multiplex assay was performed to measure chemokines and cytokines in the TME. There was significant upregulation of TFNγ, IL2, IL4, IL13, and IL12p70, whereas modulation of IL1β, IL17A, and IL23 was minimal (Fig. 8E). Additional analysis of immunosuppressive markers CD33 and CD39 revealed slight downregulation of Tregs (14.2%–16.9% in the saline group vs. 7.9%–15.1% in the mAb-Exo-AAV group) and minimal change in myeloid-derived suppressor cells (7.3%–11.6% in the saline group vs. 8.39%–11.3% in the mAb-Exo-AAV treatment group; Supplementary Fig. S11).

Mitochondria have been considered a promising therapeutic target for cancer treatment, but translating this concept into clinical practice has proven challenging. Although certain drugs targeting mitochondria have demonstrated promise in preclinical studies, their effectiveness in clinical trials has been limited due to the low efficacy, drug resistance, and side effect from lack of specificity. To overcome these limitations, we introduce mLumiOpto, a novel therapeutic strategy that specifically and directly destroys cancer mitochondria to induce cancer cell death. We use AAV and mAb-Exo-AAV platforms to deliver mLumiOpto genes to cancer cells in vivo, maximizing potency and minimizing off-target effects. Our preclinical mouse xenograft models demonstrate that AAV (i.c.v. administration) and mAb-Exo-AAV (i.v. administration)–delivered mLumiOpto effectively kill cancer cells and inhibit tumor growth without causing noticeable side effects, highlighting its potential for targeted cancer therapy.

Optogenetics is widely used for the precise manipulation of cell membrane excitability, but its application to intracellular organelles, particularly mitochondria, has been limited. In this study, we expanded optogenetics to target and damage cancer mitochondria, aiming to eliminate cancer cells. Specifically, we introduced mitochondrial-targeted luminoptogenetics, a novel approach enabling dynamic mitochondrial manipulation both in vitro and in vivo. This strategy integrates endogenous bioluminescence from the NLuc-ViviRen pair into our mOpto system, creating the advanced, external light-independent mLumiOpto technology. Our design leverages the principle that luciferases emit light at specific wavelengths when paired with substrates like luciferin (33). For instance, RLuc emits blue light (λ_Peak ∼ 470 nm) with coelenterazine (34). Exploiting the spectral overlap between luciferase emission and rhodopsin absorption, bioluminescence has proven successful in activating rhodopsin channels and controlling neuronal activity in live animals (35). We identified NLuc as the ideal luciferase for dynamic and efficient mitochondrial control in vivo due to its monomeric structure, stability, high and sustained luminescence at low substrate doses, and nontoxicity to cells. NLuc emits bright, sustained bioluminescence at low ViviRen concentrations, is small (encoded by a 513-bp gene), ATP-independent, and uniformly distributed intracellularly (19). Our in vitro studies show that NLuc-ViviRen pair-generated bioluminescence effectively activates the mitochondrial CoChR channel, leading to ViviRen dose-dependent ΔΨ_m depolarization without external light. By harnessing endogenous bioluminescence, mLumiOpto overcomes the challenges of delivering external light to deep tissues and minimizes potential side effects on surrounding healthy tissues. This ability to manipulate mitochondrial function in freely moving animals renders mLumiOpto a powerful tool for in vivo investigations, from studying mitochondrial dysfunction mechanisms to developing mitochondrial-targeted therapies.

The capability of mLumiOpto in inducing cancer cell death was validated using in vitro cell lines and in vivo tumor xenograft mouse models. Consistent with our previous observations that mOpto mediates light intensity–dependent cytotoxicity, mLumiOpto induced cancer cell death in a ViviRen dose-dependent manner and was cytotoxic to all treated cancer cell types. Dose dependence is crucial for optimizing treatment efficacy while minimizing side effects due to varying sensitivities among cancer cells. Our animal studies further demonstrated mLumiOpto’s ability to eliminate tumor cells in vivo, effectively inhibiting tumor growth in various xenograft mouse models, including multiple TNBC subtypes and intracranial GBM. This is particularly significant given the lack of effective treatment options for those aggressive and recurrent cancers. Notably, neither ViviRen nor mLumiOpto expression alone exhibited any deleterious effects on cancer cell mitochondria and viability. These findings highlight mLumiOpto as a robust and versatile approach for targeted cancer cell death in both preclinical and translational settings. Its ability to effectively combat tumor growth across diverse cancer types positions mLumiOpto as a promising therapeutic strategy for various malignancies.

Our studies provide valuable insights into the mechanisms of mLumiOpto-mediated cancer cell death. We found that this cell death primarily occurs through the intrinsic apoptotic pathway, accompanied by DNA damage and autophagy, and is independent of mtROS and mPTP opening. These findings highlight mLumiOpto’s ability to eliminate cancer cells without relying on cancer type–associated proteins or signaling pathways. It is worth noting that we also detected activation of mitophagy in mLumiOpto-treated cancer cells. Although mitophagy does not induce apoptotic cell death through cytochrome C release as a default pathway, under conditions of extreme stress, where mitochondrial damage is too extensive, it can lead to cell death, either through the release of cytochrome C and subsequent apoptosis or through autophagy-dependent mechanisms. The extent to which mitophagy contributes to mLumiOpto-mediated cell death requires further investigations.

A major challenge in current cancer treatment is the development of drug resistance, which comprises the efficacy of conventional therapies. Proposed mechanisms include cancer heterogeneity, the TME, cancer stem cells, inhibition of cell death pathways, and mutations in therapeutic targets (36, 37). mLumiOpto employs heterologous CoChR genes that do not target or rely on cancer-associated proteins or pathways such as EGFR, VEGF, ALK, PI3K/AKT/mTOR, or p53, which are often impaired during long-term treatment. Consequently, mLumiOpto-based approaches may be less prone to developing drug resistance and more efficient in eliminating heterogeneous cancer cells compared with conventional therapies. These characteristics make mLumiOpto a promising strategy for overcoming drug resistance, offering the potential for improved therapeutic outcomes in treating diverse cancer types, although further evaluation is needed in future studies.

AAV vectors have gained considerable attention as a promising approach for delivering therapeutic genes (38). The FDA has approved AAV-delivered Luxturna and Zolgensma for treating rare inherited blindness and spinal muscular atrophy, respectively, and numerous clinical studies are exploring AAV-based gene therapy for various diseases (39–41). However, free AAV administrated through i.v. injection faces challenges in cancer treatment, including lack of specific targeting, relatively low infection efficiency, and preexisting AAV neutralizing antibodies in a significant portion of the patients (42–45). Exosomes, extracellular nanovesicles secreted by cells, have emerged as an alternative vehicle for delivering therapeutic genes, including AAV vectors, due to their low antigenicity and toxicity. Compared with free AAV, Exo-AAV protects genetic material from degradation and increases circulation stability. Additionally, mAbs can bind to specific surface receptors on cancer cells, directing exosomes carrying therapeutic genes to target cells. This combination (mAb-Exo-AAV) allows for targeted gene delivery, enhancing the efficiency and specificity of mLumiOpto therapy. Several preclinical studies have demonstrated the potential of Exo-AAV in treating conditions such as hemophilia A and B (46) and liver disease (42). In this study, we established a novel mAb-Exo-AAV platform by surface tagging an anti-EGFR mAb (targeting 52%–89% of TNBC) to Exo-AAV using a DMPE-PEG-NHS linker. We also integrated mPEG-DSPE into the exosome membrane to improve circulation stability. Our animal results demonstrated that mAb-Exo-AAV effectively targets tumors, achieving high-level and functional mLumiOpto expression specifically in tumor tissues with undetectable nonspecific distribution in normal organs. Importantly, our strategy allows for facile surface conjugation of various mAbs, including dual mAbs, to Exo-AAV, expanding the applicability of this approach to a broader range of patients and various cancer types or subtypes.

Remarkably, our study not only demonstrated the highly efficient and cancer-specific gene delivery capability of anti-EGFR mAb-Exo-AAV but also revealed its ability to induce an antitumor immune response in immunocompetent TNBC xenograft mouse models. The mechanisms underlying this enhanced tumoral immunity are multifaceted. First, EGFR activation is known to stimulate cancer proliferation, DNA damage repair, and metastasis (47). Anti-EGFR mAbs, such as cetuximab or panitumumab, mediate antibody-dependent cell cytotoxicity within tumors and inhibit DNA repair via BRCA (48). Our mAb-Exo-AAV, with a high mAb surface tagging rate, facilitates immune responses within the TME. Second, the immune response triggered by AAV capsid immunity can reactivate memory CD8⁺ T lymphocytes through histocompatibility (MHC) class I presentation, leading to tumor destruction (49). After specifically targeting tumor cells, mAb-Exo-AAV is internalized to release AAV intracellularly. The AAV enters the nucleus, whereas its capsid undergoes proteasomal degradation, enhancing MHC surface expression, CD8⁺ T-cell activation, and adaptive immunity within the TME. Third, our flow cytometry analysis revealed an increase in DCs in freshly harvested tumor tissues, aligning with the “cross-priming” mechanism (50). Apoptotic cancer cells release tumor antigens into the TME, which are subsequently captured by antigen-presenting DC cells, facilitating CD8⁺ T-cell activation. These activated CD8⁺ T cells selectively target cancer cells, creating a more favorable TME for immune cell infiltration and further boosting tumoral immunity. Future studies will use multiple immunocompetent and humanized mouse models to fully investigate the immune effects in the TME, decipher the underlying mechanisms, and examine the potential synergistic anti-cancer efficacy of mLumiOpto technology and the mAb-Exo-AAV gene delivery vehicle.

In summary, our study introduces mLumiOpto, an innovative luminoptogenetic approach targeting cancer mitochondria to trigger cytotoxicity. Our findings highlight the potential of mAb-Exo-AAV–delivered mLumiOpto as a promising strategy for inducing targeted cancer cell death and activating immune response activation in the TME. These advancements provide valuable insights for developing novel therapeutic strategies that address major cancer treatment challenges, including reduced drug resistance and enhanced efficacy. Further evaluations using more clinically relevant models, such as metastatic and PDX humanized mouse models, are essential to explore the translational potential of this promising therapeutic strategy. Comprehensive assessments of IND-directed toxicology, biodistribution, pharmacokinetics, and pharmacodynamics are crucial for future clinical trials. Importantly, mLumiOpto technology has great potential for treating other challenging cancers, such as recurrent GBM and non–small cell lung cancers, by substituting the cancer-targeting mAb on the surface of Exo-AAV. Finally, mitochondria play an essential role in tumorigenesis, metastasis, and stemness, making mLumiOpto a powerful tool in mechanistic studies due to its ability to dynamically modulate mitochondria.

No disclosures were reported.

K. Chen: Data curation, formal analysis, visualization, methodology. P. Ernst: Data curation, formal analysis, investigation, methodology. A. Sarkar: Data curation, formal analysis. S. Kim: Data curation, visualization. Y. Si: Data curation, formal analysis. T. Varadkar: Data curation, formal analysis. M.D. Ringel: Methodology, writing–review and editing. X.M. Liu: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing. L. Zhou: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing.

This work was supported by DoD BCRP W81XWH2110066/67 (X.M. Liu and L. Zhou), NIH NCI R01CA262028 (X.M. Liu and L. Zhou), and NIH R01HL156581 (L. Zhou). The authors thank the Comparative Pathology, Small Animal Imaging Shared Resource, Flow Cytometry Shared Resource, and Center of Electron Microscopy and Analysis at the Ohio State University for the assistance with tissue sectioning, IVIS imaging, flow cytometry, nanoparticle tracking assay, echocardiography, and transmission electron microscopy.