Abstract
Tumor metastasis, a leading cause of cancer-related mortality, presents a significant challenge due to its complex, multi-step nature and tumor cells’ adaptability. Current targeted therapies often face limitations as tumor cells activate compensatory signaling pathways, leading to drug resistance. Here, we introduce a novel molecular drug pair of sorafenib (Sora) and curcumin (Cur) self-assembled into Sora-Cur@diploid nanoparticles (diNPs), which were transformed into label-free diNP@pharmaceutical dots (diNP@PDs) with enhanced fluorescence through a specialized formulation process, eliminating the need for chemical modification. diNP@PDs exhibit photoluminescence capability, simultaneously promoting cross-talk among lysosomes, mitochondria, and nuclei. This lysosomes-mitochondria-nuclei interaction induces perinuclear clustering of organelles, triggering nuclear reactive oxygen species (nuROS) generation. The nuROS epigenetically suppresses HDAC2 activity, inhibiting epithelial-mesenchymal transition (EMT) and tumor metastasis. By integrating multi-organelle targeting with molecular therapy, diNP@PDs offer an innovative strategy for both imaging and inhibiting tumor metastasis, highlighting the potential of interorganelle approaches in oncology.
Keywords: Tumor metastasis, Pharmaceutical dots, Lysosomes-mitochondria-nuclei interaction, Nuclear reactive oxygen species, HDAC2
Graphical abstract
1. Introduction
Tumor metastasis, as the leading cause of mortality in cancer patients, remains a critical challenge in oncology [1,2]. While extensive research has been dedicated to elucidating the mechanisms underlying metastasis and developing effective antimetastatic therapies, clinical outcomes have remained suboptimal. This is largely attributable to the intricate, multi-step nature of metastasis, which involves dynamic interactions among genetic, epigenetic, microenvironmental, and systemic regulatory factors [3]. Importantly, despite the potent suppression of specific cancer-related proteins or signaling pathways achieved by targeted therapies, tumor cells often exhibit remarkable adaptability by activating compensatory signaling cascades [4,5]. These compensatory pathways not only counteract therapeutic effects but also drive drug resistance and metastatic progression, thereby highlighting the limitations of current molecularly targeted approaches. To address these challenges, emerging strategies focus on multi-target therapeutic approaches capable of modulating multiple proteins or signaling pathways simultaneously [6,7]. By enhancing crosstalk among these targets, such strategies aim to achieve network pathway targeting, thereby amplifying therapeutic efficacy and overcoming the inherent limitations of single-target interventions [8]. This integrative approach holds significant promise for advancing the development of more effective antimetastatic therapies.
The subcellular localization of these targets within tumor cells underscores the importance of molecular signaling dynamics within sub-organelles, as effective communication among diverse targets is critical for optimizing therapeutic outcomes [9]. Notably, subcellular organelles such as lysosomes and mitochondria play pivotal roles in sustaining tumor cell viability. A defining feature of oncogenesis is the reprogramming of energy metabolism, wherein lysosomes and mitochondria act as both contributors and mediators [10]. Lysosomes actively participate in various physiological functions and diseases, including autophagy, metabolism and immune responses, which promote gene mutation, drug resistance, immune escape, metastasis and cancer recurrence after treatment [11]. Compared to normal cells, tumor cells exhibit enhanced lysosomal activity and elevated autophagy levels. Mitochondria, as the cellular powerhouse, are equally critical due to their bioenergetic and biosynthetic roles, particularly in rapidly metabolizing and proliferating tumors. The interdependence of tumor cells on lysosomes and mitochondria creates a synergistic vulnerability. Lysosomal stress disrupts mitochondrial morphology and function, triggering apoptosis through reactive oxygen species (ROS) signaling. Conversely, mitochondrial autophagy may also induce lysosomal damage, creating a self-amplifying feedback loop that ultimately results in tumor cell death [12]. This lysosome-mitochondrion interaction thus emerges as a promising new therapeutic target. Similarly, the functional interplay between mitochondrion and the nucleus influences tumorigenesis and metastasis [13], indicating that targeting their coordinated activities could represent another viable anti-tumor strategy. However, the therapeutic potential of targeting multi-organelle interactions (e.g., lysosomes-mitochondria-nuclei) remains largely unexplored.
Sorafenib (Sora) as a multikinase targeted inhibitor can block receptor tyrosine kinase (RTKs) and downstream RAS/RAF/MEK/ERK signaling, alleviating tumor growth via anti-proliferative, anti-angiogenic and pro-apoptotic effects [[14], [15], [16]]. Unsurprisingly, cancer patients generally become resistant to Sora after long-term treatment. The organelles, including lysosomes, mitochondria and nuclei, have been explored to play extraordinary roles in Sora resistance and anti-tumour activity [[17], [18], [19]]. It has been reported that mitochondrial biogenesis, ROS homeostasis and lysosome-mediated autophagy largely facilitated Sora resistance in cancer cells, while lysosomal compartment dysregulation robustly improved anti-cancer effect of Sora, indicating synergetic combination of organelles-targeting and molecular-targeting agents may open a new avenue for cancer therapy [20,21]. Alternatively, curcumin (Cur), a polyphenolic compound isolated from turmeric rhizome, possesses excellent anti-cancer potency by inhibiting tumor growth, overcoming drug resistance and triggering tumor immunity [[22], [23], [24]].
Interestingly, our work firstly constructed the self-assembled nanoparticles comprising the molecular drug pair Sora and Cur (Sora-Cur@diploid nanoparticles, diNPs), which exerted potent and consistent pharmaceutical dot-induced luminescence. Through a simple nanofabrication process, we achieved breakthrough fluorescence intensity enhancement, effectively converting the drug molecules into label-free luminescent probes (pharmaceutical dots) without introducing harmful residues from chemical synthesis [25,26]. This innovative approach transforms bioactive molecules into functional fluorescent probes, enabling applications in molecular, organelle, and cellular imaging [[27], [28], [29]]. More importantly, label-free diNP@pharmaceutical dots (diNP@PDs) exhibit an exceptional capacity in promoting crosstalk among lysosomes, mitochondria, and the nucleus. Mechanistically, diNP@PDs induce the perinuclear clustering of lysosomes and mitochondria, facilitating coordinated interactions among these organelles and enhancing their communication with the nucleus. This tripartite organelle coordination was preliminarily demonstrated to enable the regulation of epigenetic pathways, thereby improving therapeutic efficacy and suppressing tumor metastasis.
In this study, we therefore self-assembled Sora and Cur into diNP@PDs to establish a multi-organelle regulatory strategy for metastatic tumor therapy. This interorganelle interaction triggered nuclear reactive oxygen species (nuROS) generation, which epigenetically suppressed histone deacetylase-2 (HDAC2) activity to inhibit epithelial-mesenchymal transition (EMT) and tumor metastasis. Further experiments confirmed that blocking ROS production enhanced HDAC2 activity and consequently restored metastatic phenotypes. Conversely, pharmacological inhibition of HDAC2 using vorinostat (SAHA) effectively suppressed metastasis. These findings collectively establish the nuROS-HDAC2-EMT axis as the mechanistic underpinning of interorganelle synergy. Therefore, the new-developed diNP@PDs, in which Sora was rationally paired with Cur (molecular drug pair), not only enabled pharmaceutical dot-induced luminescence but also evoked extraordinary synergistic anti-metastatic efficacy via a “interorganelle contacts” strategy (Scheme 1).
Scheme 1.
A simple yet powerful label-free strategy was developed to directly convert drug molecules into fluorescent probes, enabling cellular imaging and orchestrating a lysosome-mitochondria-nucleus interplay that elicits nuclear ROS (nuROS) generation. This, in turn, activates an uncharacterized nuROS-HDAC2-EMT axis, thereby abrogating metastasis.
2. Experimental section
2.1. Materials and reagents
Dimethyl sulfoxide, methanol, ethanol and tetrahydrofuran were purchased from Chengdu Kelong Chemical Co., Ltd. Sora (cat. no. 284461-73-0) was purchased from Aladdin Biochemical Technology. Cur (cat. no. 458-37-7) was procured from Meilun Biotechnology. Phosphotungstic acid (cat. no. R128447) was obtained from Rhawn. N-Acetylcysteine (cas. no. 616-91-1) and vorinostat (SAHA, cas. no. 149647-78-9) were purchased from Aladdin Biochemical Technology.
The following antibodies were used for immunohistochemistry (IHC), immunofluorescence (IF), western blot (WB) or flow cytometer (FCM): anti-LAMP1 (Abcam, ab300550), anti-TOMM20 (Abcam, ab309083), anti-Lamin B1 (Abcam, ab133741), VEGFR2 (Immunoway, YT5845), phosphate-VEGFR2 antibody (Immunoway, YP0112), DHODH (Abcam; ab174288), GPX4 (Proteintech, 67763-1-Ig), CD107a (Immunoway, YT0716), HDAC2 (Proteintech, 12922-3-AP), E-cadherin (Immunoway, YT1454), Vimentin (Immunoway, YT5796).
The following dyes and chemicals were used: FluoLyso™ Green (50 nM, L4085S) and MitoTracker® Green (50 nM, 40742ES50) from UElandy Inc. and Yeasen biotechnology Co., Ltd., respectively. LysoBrite red (50 nM, A66439) and PK Mito Deep Red (100 nM, PKMDR-2) from Invitrogen and Genvivo biotechnology, respectively. 4′, 6-diamidino-2-phenylindole (DAPI, D4080) from Bioscience technology Co., Ltd.
2.2. diNP@PDs formation and morphological studies
The diNP@PDs were synthesized using a reprecipitation protocol and subsequently dissolved in a water/DMSO mixture (90/10, v/v). In brief, Sora (5 mg) and Cur (1 mg) were initially dissolved in dimethyl sulfoxide (DMSO, 0.5 mL). The solution was then added dropwise to deionized water (4.5 mL) contained within a glass vial at room temperature. Prior to use, the resulting solution underwent dialysis against deionized water to eliminate DMSO. The concentrations of Sora and Cur were quantified using analytical high-performance liquid chromatography (HPLC). The hydrodynamic diameter, polydispersity index and zeta potential of diNP@PDs were measured via dynamic light scattering utilizing an Anton Paar instrument (Litesizer 500). To examine the morphology of diNP@PDs, TEM images were performed at an operating voltage of 80 kV. Specifically, sample solutions were drop-coated onto carbon-coated copper grids (200 mesh). After allowing for 2 min of adsorption, excess solution was removed with filter water followed by staining with phosphotungstic acid for an additional 2 min.
2.3. Spectral characteristics of diNP@PDs
A 10 μL aliquot of diNP@PDs (1 mg/mL, Sora equivalent) was deposited onto a carbon coated grid, followed by immediate negative staining with phosphotungstic acid for 1 min. EDS-mapping was conducted using a JEM-1400plus microscope operated at 120 kV. Infrared spectra were obtained in the range of 4000 to 400 cm−1 utilizing a FT-IR spectrometer and employing the potassium bromide method. XRD analysis was performed with a Bruker D8 Advance diffractometer, collecting data at a scan rate of 5° per minute over a range of 4° to 40°. XPS spectra were acquired using a Thermo Kalpha analyzer equipped with a monochromatic AlKα source operating at an energy level of 1486.6 eV, focusing on the surface composition of solid samples specifically regarding the elements C, N, O, F, and Cl; the analysis included both spectrograms and quantitative results. The samples underwent triplicate testing after being suspended in deionized water at a concentration of 1 mg/mL. Subsequently, UV-visible spectroscopy was employed to scan within the wavelength range of 200 to 700 nm for acquiring UV spectra. NOESY 2D NMR spectra were recorded on an Avance Neo spectrometer operating at 600 MHz. A comparative analysis was conducted between solutions and lyophilized powders for Sora, Cur, Sora + Cur, diNP@PDs, respectively.
2.4. Molecular dynamics (MD)
MD simulation was conducted to elucidate the mechanism underlying the self-assembly of diNP@PDs. The structures of the small molecules Sora and Cur were optimized using B3LYP/6-31G* method via the Gaussian09 package. Subsequently, electrostatic potential calculations were performed utilizing the HF/6-31G* method in conjunction with Multiwfn software. A topology file for the ligand was generated by Acpype, which was then employed to compute the restricted ESP2 charges. For the Sora-Cur system, twelve Sora molecules, three Cur molecules and one hundred DMSO molecules were initially packed randomly into a cubic box with a length of 60 Å using PACKMOL. This mixture system was subsequently neutralized within a cuboid box containing TIP3P water molecules, ensuring that there was a solvent layer of 10 Å between the edges of the box and the surface of the solute. All MD simulations were executed using GROMACS version 2020.3 with application of the AMBER 99SB force field. For each solvated system, two steps of energy minimization were carried out prior to heating. The first step involved 5000 cycles of minimization employing steepest descent methods at an energy threshold of 300 KJ/(mol/nm). This was followed by another set comprising 5000 cycles utilizing conjugate gradient minimization at an energy threshold of 100 KJ/(mol/nm). Subsequently, the entire system underwent heating from 0 K to 300 K over a duration of 100 ps before being equilibrated for an additional period of 400 ps under constant pressure conditions at 1 atm. In total production phase, a simulation lasting for 50 ns was performed.
2.5. Optical properties of diNP@PDs
The fluorescence spectra were acquired using a SpectraMax iD5 multi-mode microplate reader, covering a wavelength range of 290-750 nm. Briefly, diNP@PDs (1 mg/mL, Sora equivalent) were dissolved in water/DMSO mixtures with varying water fractions ranging from 0 to 95%. Emission spectra of the solutions were recorded at a designed excitation wavelength.
2.6. Cell culture
Murine mammary carcinoma 4T1 cells and human liver cancer MHCC-97H were cultured in DMEM medium. The culture media contained 10% FBS and 1% penicillin/streptomycin. All cell lines were maintained at 37°C in a humidified atmosphere with 5% CO2.
2.7. Cell viability assay
The cells were seeded in 96-well plates at a density of 0.8 to 1 × 104 cells per well and cultured at 37°C for 24 h. Subsequently, the cells were incubated with medium containing Sora, Cur, Sora + Cur and diNP@PDs at varying concentrations for an additional 48 h. Following this incubation period, 100 μL of the thiazolam solution (0.5 mg/mL) was added to each well and allowed to react for 2 h. Afterward, 100 μL of DMSO was introduced into each well to dissolve the resulting formazan crystals. All samples were then measured at a wavelength of 490 nm.
2.8. Western blot
4T1 or MHCC-97H cells were seeded into 12-well plates at a density of 1 × 105 cells per well and allowed to adhere overnight to ensure optimal cellular attachment. Subsequently, the cells were treated with PBS, Cur (2.5 μM), Sora (10 μM), Sora + Cur (10 μM, Sora equivalent) and diNP@PDs (10 μM, Sora equivalent) for a duration of 8 h. Post-treatment, the protein levels were evaluated using a western blot assay. Specifically, the tumor cells were placed on ice and lysed in radio immunoprecipitation assay (RIPA) buffer supplemented with phenylmethylsulfonyl fluoride (PMSF) and phosphatase inhibitors sourced from Beyotime Biotechnology (China) to prevent proteolytic degradation. The resultant suspension was centrifuged, and the protein concentration was quantified using the BCA assay. Subsequently, equal quantities of cell lysates (40 μg) were loaded onto a 10% SDS-PAGE gel and then transferred onto a polyvinylidene fluoride (PVDF) membrane supplied by Millipore (Germany). The membranes were incubated in a blocking solution comprising 5% non-fat milk in TBST for 1 h, followed by overnight incubation with primary antibodies. Thereafter, the membranes were incubated with secondary antibodies at room temperature for an additional 1 h to form the antibody-antigen complex. The immunoreactive products were visualized using an enhanced chemiluminescence detection system. Normalization of all loading samples was conducted based on Tubulin or GAPDH staining.
2.9. Wound healing assay
For the wound healing assay, 4T1 cells were seeded into 6-well plates at a density of 8 × 105 per well and incubated overnight to establish a confluent monolayer. A straight-line scratch was created in the cell monolayer using a scratcher to induce vertical wounds. Following this, the 4T1 cells were treated with free Cur (1.25 μM), free Sora (5 μM), Sora + Cur (5 μM, Sora equivalent) or diNP@PDs (5 μM, Sora equivalent) for durations of 12 and 24 h. Untreated cells served as the control group. Five reference points were randomly selected from each well at different time intervals, and the percentage of wound closure was quantified using ImageJ software.
2.10. Cell invasion assay
The cell invasion assay was performed using BioCoat Matrigel invasion chambers with an 8.0 μm PET membrane sourced from Corning. 4T1 cells were seeded at a density of 5 × 103 per well in the upper chamber filled with serum-free DMEM. Simultaneously, 200 μL of media containing various drugs at a concentration of 5 μM (Sora equivalent) was added to the upper chamber. The lower chamber contained 750 μL of DMEM supplemented with 10% serum, serving as a chemoattractant. The cells were then incubated in a humidified incubator at 37°C for a duration of 48 h. After incubation, the cells were washed three times with 1× PBS and fixed in a 4% paraformaldehyde solution for 15 min. Non-invaded cells residual in the upper chamber were gently removed using cotton swads. The invaded cells that penetrated through the Matrigel layer were subsequently stained with 0.05% crystal violet for 20 min. Images encompassing the entire membrane were captured in segments and seamlessly stitched together using a cell imaging system (10× magnification; XDS200, China). Image analysis was conducted using ImageJ software to quantify the percentage of cell invasion, based on crystal violet intensity relative to the membrane area.
2.11. Fluorescence intensity of lysosomes and mitochondria following various treatments
4T1 cells were seeded into 12-well plates at a density of 1 × 105 cells per well and allowed to adhere overnight to promote optimal cell attachment. Following this incubation, the cells were treated with Cur (1.25 μM), Sora (5 μM) and diNP@PDs (5 μM, Sora equivalent) for a duration of 48 h. Subsequently, the cells were rinsed three times with 1× PBS, after which commercial LysoTracker Green (50 nM, L4085S) and MitoTracker Green (50 nM, 40742ES50) were added for an additional incubation period of 1 h and 30 min, respectively. The cells were then washed three times with 1× PBS and fixed using a solution of 4% paraformaldehyde for 15 min, followed by DAPI staining for another 5 min. After two more washes with 1× PBS, the coverslips were sealed using a fluorescent anti-quenching agent. Finally, fluorescence imaging was conducted utilizing a confocal microscope equipped with a 60× objective lens. Signal collection from CLSM was performed in the FITC channel for both LysoTracker Green and MitoTracker Green. Mean intensity values were captured using ImageJ software.
2.12. Transmission electron microscopy
In vitro studies were conducted using 4T1 or MHCC-97H cells, which were seeded in a 12-well plate at a density of 1 × 105 cells per well. The cells were treated with diNP@PDs (5 μM) for a duration of 48 h. Following treatment, the cells were digested with trypsin and the resulting cell suspensions were centrifuged at 1000 rpm for 5 min. The harvested cells were initially fixed with 0.5% glutaraldehyde at 4°C for 5 min, after which they underwent centrifugation at 12000 rpm for an addition 10 min. Subsequently, the obtained cells were fixed again with 3% glutaraldehyde for a period of 10 h, followed by dehydration using acetone and embedding in resin. Ultrathin sections (60-80 nm) were prepared utilizing an ultramicrotome (UC7rt, Leica). These sections were then stained with lead citrate (EMS) and uranyl acetate (SPI) to facilitate observation under transmission electron microscopy (JEM-1400FLASH, JEOL). The distances between the lysosomes, mitochondria, and nuclei were measured using ImageJ software.
2.13. Lysosome-mitochondrion-nucleus contacts identified through HIS-SIM
4T1 cells were cultured in a confocal dish (Cellvis) at a density of 2 × 105 cells per dish until reached approximately 80% confluence. The cells were then treated with either culture media or diNP@PDs (5 μM) for a duration of 12 h. Based on the experimental requirements, organelles were stained as follows: lysosomes were labeled with LysoBrite Red (Invitrogen, 50 nM) for 10 min, mitochondria were stained with PK Mito Deep Red (Genvivo, 100 nM) for 10 min and the nuclei were stained with Hoechst 33342 (Bioscience) for an additional 10 min. Confocal images were acquired using a High Sensitivity Structured Illumination Microscope (HIS-SIM) from CSR biotechnology.
2.14. Sub-organelle interaction observed by CLSM
4T1 cells were seeded into 12-well plates at a density of 1 × 105 cells per well and allowed to adhere overnight to facilitate optimal cell attachment. Following this incubation period, the cells were treated with PBS, Cur (6.25 μM), Sora (25 μM) and diNP@PDs (25 μM, Sora equivalent) for durations of 24 and 48 h. In accordance with the experimental requirements, organelles were stained as follows: lysosomes were labeled with LAMP1 (Abcam, 5 μg/mL) for 1 h, mitochondria were stained with TOMM20 (Abcam, 5 μg/mL) for 1 h and the nuclei were stained with DAPI (UElandy) for an additional 5 min. Confocal images were acquired using an Olympus CLSM (FV-OSR). Mean intensity values were captured using ImageJ software.
2.15. Proteomics
4T1 cells underwent treatment with PBS or diNP@PDs (5 μM, Sora equivalent) for a duration of 48 h. Subsequently, from each treatment group, three distinct samples containing 3 × 106 cells per sample were collected for protein extraction. The protein concentration was determined using a BCA kit in accordance with the manufacturer's instructions. The samples were fractionated into fractions by high pH reverse-phase HPLC using an Agilent 300 Extend C18 column (5 μm particles, 4.6 mm ID, 250 mm length). Peptide separations were achieved with a gradient ranging from 2% to 60% acetonitrile in 10 mM ammonium bicarbonate (pH 10) over 80 min, resulting in 80 fractions. These fractions were subsequently consolidated into 9 pools and desiccated using vacuum centrifugation. The isolated peptides were analyzed using a Q Exactive™ HF-X mass spectrometer (ThermoFisher Scientific), equipped with a nano-electrospray ion source. This was followed by capillary source analysis on the timsTOF Pro mass spectrometry (Bruker Daltonics). The resulting MS/MS data were processed using the MaxQuant search engine (version 1.6.15.0). Gene Ontology (GO) annotations for the proteome were sourced from the UniProt-GOA database (http://www.ebi.ac.uk/GOA/). GO terms with a corrected p-value of less than 0.05 were considered statistically significant. To identify enriched pathways, the Encyclopedia of Genes and Genomes (KEGG) database was utilized, employing a two-tailed Fisher's exact test to assess the enrichment of differentially expressed proteins against the total identified proteins. Pathways with a corrected p-value of less than 0.05 were considered significant. Additionally, all differentially expressed protein accessions or sequences were cross-referenced against version 11.0 of the STRING database to investigate potential protein-protein interactions.
2.16. Quantitation of nuclear reactive oxygen species (nuROS)
The fluorescent probe 2′, 7′-dichlorofluorescin diacetate (DCFH-DA) was diluted in a serum-free medium to a concentration of 10 μM for the detection of nuROS generation. 4T1 cells were seeded in a 60 mm diameter cell-culture dish at a density of 1 × 106 cells per dish and cultured overnight to ensure adequate adherence. Subsequently, PBS, Cur (6.25 μM), Sora (25 μM) and diNP@PDs (25 μM, Sora equivalent) were added and incubated for 24 h. Following treatment, nuclei were isolated using a Nuclear Extraction Kit (Solarbio) and then incubated with DCFH-DA for 30 min. Fluorescence signals were immediately recorded using flow cytometer (BD FACSVerse) at an excitation wavelength of 488 nm.
2.17. HDAC2 expression levels observed by CLSM
4T1 and MHCC-97H cells were seeded into 12-well plates at a density of 1 × 105 cells per well and allowed to adhere overnight to promote optimal cell attachment. Following this incubation, the cells were treated with PBS, Cur, Sora and diNP@PDs at a concentration of 25 μM (4T1) or 30 μM (MHCC-97H) for a duration of 12 h. Specifically, the tumor cells were placed on ice and fixed in 4% paraformaldehyde buffer for 10 min at room temperature before being incubated in a blocking solution containing 5% goat serum in 1× PBS. Subsequently, the cells were incubated with a commercial HDAC2 antibody (IF dilution: 1: 2000; Proteintech, 12922-3-AP) for an additional period of 1.5 h. Afterward, the cells underwent three rinses prior to incubation with a secondary antibody for another duration of 1.5 h. The cells were then washed three times with 1× PBS and fixed using a solution of 4% paraformaldehyde for 15 min, followed by DAPI staining for another 5 min. The coverslips were sealed using a fluorescent anti-quenching agent after two further washes with 1× PBS. Finally, fluorescence imaging was conducted utilizing a confocal microscope (ZEISS). Signal collection was performed in the FITC channel and mean intensity values were captured using ImageJ software.
2.18. ROS generation following N-acetylcysteine (NAC) treatment
4T1 cells were seeded into 6-well plates at a density of 1 × 106 per well and incubated overnight to establish a confluent monolayer. The experiment was divided into six groups: medium (Control), 2 μM NAC, 4 μM NAC, 25 μM diNP@PDs, 2 μM NAC + 25 μM diNP@PDs and 4 μM NAC + 25 μM diNP@PDs. Following this treatment, the cells were incubated with DCFH-DA for 30 min. Subsequently, fluorescence was immediately measured using flow cytometer (FACSCanto II) with an excitation wavelength set at 488 nm.
2.19. HDAC2 expression levels following NAC treatment
4T1 cells were seeded into 12-well plates at a density of 1 × 105 per well and incubated overnight to establish a confluent monolayer. The experiment was divided into six groups: medium (Control), 2 μM NAC, 4 μM NAC, 25 μM diNP@PDs, 2 μM NAC + 25 μM diNP@PDs and 4 μM NAC + 25 μM diNP@PDs. The cells were subsequently incubated with an HDAC2 antibody. Confocal images were acquired using an Olympus CLSM (FV-OSR) and mean intensity values were analyzed using ImageJ software.
2.20. Wound healing assay following SAHA treatment
For the wound healing assay, 4T1 cells were seeded into 6-well plates at a density of 8 × 105 per well and incubated overnight to establish a confluent monolayer. A straight-line scratch was created in the cell monolayer using a scratcher to induce vertical wounds. Subsequently, the 4T1 cells were treated with diNP@PDs (Sora equivalent) or SAHA at a concentration of 1, 5, 10, 25 and 50 μM for a duration of 24 h. Untreated cells served as the control group. The percentage of wound closure was quantified using ImageJ software.
2.21. In vivo antitumor metastasis activity of test formulations
The female Balb/c mice were obtained from SiPeiFu production. The mice were injected intravenously subcutaneously with 4T1-luc cells at a concentration of 1 × 106 cells per mouse. Three days post-injection, the tumor-bearing mice were randomly divided into six groups (n = 5) and received various treatments as follows: normal group, PBS, PTX (4 mg/kg), Cur (2 mg/kg), Sora (10 mg/kg) and diNP@PDs (10 mg/kg). Treatments were administered every two days for a total of five doses. Bioluminescence image of tumor-bearing mice was performed at specified time points using a Spectrum IVIS system (PerkinElmer). Each mouse was individually marked and monitored throughout the experiment. Mouse body weights were recorded using an electronic balance every two days over a period of 15 consecutive days.
At desired treatment period, the mice were euthanized, and their lungs were isolated and weighed. Additionally, tissue samples were collected, fixed in 4% paraformaldehyde, and subsequently embedded in paraffin. The excised lung tissues were then sectioned for a comprehensive histopathological examination, that included H&E staining, immunohistochemical analysis, and immunofluorescence assays. In the immunohistochemical analysis, the sections were incubated with 3% BSA for 30 min before being exposed to antibodies against HDAC2 (1:2000; Proteintech, 12922-3-AP), E-cadherin (1:300; Immunoway, YT1454) or Vimentin (1:300; Immunoway, YT5796). These prepared sections were analyzed using a digital slice scanning system (Hamamatsu). For the immunofluorescence analysis, organelles were stained as follows: lysosomes with LAMP1 (1:50; Abcam, AB300550), mitochondria with TOMM20 (1:50; Abcam, AB309083) and the nuclear membranes with Lamin B1 (1:50; Abcam, AB133741). Confocal images were acquired using Pannoramic MIDI (3DHISTECH).
2.22. Statistical analysis
All experimental data are presented as the mean ± standard deviation (s.d.) derived from a minimum of three independent measurements. Statistical analyses were performed using GraphPad Prism 9.0 software. Comparisons between two groups were conducted via Student's t-test, while one-way analysis of variance (ANOVA) was employed for multiple group comparisons, as detailed in each figure legend. A p-value of less than 0.05 was deemed statistically significant, whereas a p-value of less than 0.01 was considered highly statistically significant.
3. Results and discussion
3.1. Preparation and characterization of diNP@PDs
The diNP@PDs were synthesized via a straightforward yet versatile nanoprecipitation protocol [30], utilizing only the multikinase inhibitor Sora and the phytochemical Cur without any additional components. Specifically, Sora and Cur dissolved in DMSO were rapidly injected into water. As the proportion of poor solvent (water) increased, the molecules spontaneously self-assembled into water-soluble nanomaterials. The size distribution of diNP@PDs in water was characterized by dynamic light scattering (DLS), accompanied by a photograph showcasing the Tyndall effect observed in the diNP@PDs solution, ultimately yielding a particle size of 286.3 ± 9.7 nm and a polydispersity index (PDI) of 0.177 ± 0.021 (Fig. 1a). Transmission electron microscope (TEM) and energy-dispersive X-ray spectroscopy (EDS) mapping preliminarily verified that diNP@PDs exhibit a core-shell structure. TEM results demonstrated that diNP@PDs particles exhibited a spherical morphology with clearly distinguishable inner and outer layers, providing preliminary evidence of core-shell formation (Fig. 1b and c). Significantly, EDS-mapping distinctly illustrated the distribution of carbon (C), nitrogen (N), oxygen (O), fluorine (F) and chlorine (Cl) in diNP@PDs. The mapping images indicated that F and Cl were enriched within the core of diNP@PDs, thereby confirming the successful co-assembly of Sora and Cur while validating its core-shell structure (Fig. 1d). As depicted in Fig. 1e, the zeta potential values for diNP@PDs, Sora and Cur were −23.04 mV, −22.86 mV, and −20.70 mV, respectively. The relatively higher zeta potential values indicate a generally favorable level of biocompatibility [31].
Fig. 1.
Characteristic and spectroscopic properties of diNP@PDs. (a) Hydrodynamic diameter distribution of diNP@PDs. The inset photographs display the Tyndall effect in the solution. (b) Representative TEM image of diNP@PDs. Scale bar, 200 nm. (c) Element distribution mapping of diNP@PDs, including Carbon (C), Nitrogen (N), Oxygen (O), Fluorine (F), and Chlorine (Cl). (d) Self-assembling schematic diagram of diNP@PDs formulation. (e) Zeta potential of Sora, Cur, and diNP@PDs (n = 3, independent experiments). (f) XPS survey of diNP@PDs. (g-h) Encapsulation efficiency (EE%) (g) and loading efficiency (LE%) (h) of Sora and Cur in diNP@PDs (n = 4, independent experiments). (i-k) UV-vis absorption spectra (i), FT-IR spectra (j) and XRD spectra (k) of Sora, Cur, Sora + Cur and diNP@PDs. (l) NOESY 2D NMR correlation analysis of diNP@PDs for exploring the potential spatial proximity between Sora and Cur. (m-n) Schematic representation of interactions including notable π-π stacking between pyridine ring of Sora and benzene ring of Cur (m) and spatial interactions (n). (o) Snapshots of the formation of diNP@PDs upon molecular dynamics simulation (0-50 ns). (p) A diagram of diNP@PDs in equilibrium at various angles. (q) Intermolecular forces (upper) and hydrogen bonding (lower) in diNP@PDs.
X-ray photoelectron spectroscopy (XPS) analysis provided insights into the binding energies and elemental compositions of the primary surface elements of diNP@PDs (Fig. 1f). As detailed in Table S1, the surface composition of diNP@PDs was composed of carbon (65.84%), nitrogen (9.55%), oxygen (13.14%), fluorine (9.02%) and chlorine (2.45%). Furthermore, Fig. S1 presents the binding energies and chemical states obtained through curve fitting of high-resolution spectra for the coating material. Consistent with expectations, this coating was composed of carbon, nitrogen, oxygen, fluorine, chlorine—the five elements constituting the nanoparticles. The XPS analysis for C1s indicated the presence of C=C at 284.41 eV, C-C/C-H at 285.00 eV, C-N/C-O at 285.80 eV, C=O at 288.21 eV, CF at 290.65 eV, and CF2 at 292.36 eV on surface of diNP@PDs. The N1s analysis revealed peaks corresponding to -N = at 398.88 eV and C-N at 399.74 eV. Similarly, the O1s results exhibited signals for C=O at 531.24 eV and C-O at 533.05 eV. The F1s spectrum confirmed the presence of C-F at 687.9 eV. The Cl2p results identified chloride ions at 197.68 eV and organic-Cl species at 200.39 eV on the nano-surface. These findings suggested that carbon and oxygen were predominantly located on the nanoparticles' surfaces, while characteristic elements such as nitrogen, fluorine, and chlorine were present in lower concentrations. Preliminary speculation indicates that Sora might be primarily concentrated within the nanoparticles’ core. The encapsulation efficiency (EE) for Sora and Cur within diNP@PDs was found to be 98.23 ± 0.89% and 98.64 ± 0.56%, respectively, with corresponding drug loadings (DL) of 78.34 ± 1.67% and 14.17 ± 2.46% (Fig. 1g and h). Traditional nanocarriers often exhibit limited drug affinity, resulting in drug loadings that fall below 10% [32]. In contrast, self-assembled nanosystems, which do not rely on carrier materials, inherently demonstrated superior encapsulation rates and drug loadings capacities [33,34]. Furthermore, stability tests in serum showed that the particle size of diNP@PDs remained unchanged over an extended period, with no attenuation in fluorescence intensity (Fig. S2), indicating their excellent stability during in vivo applications. These self-assembled nanoparticles offer significant advantages [35], such as enhanced solubility and rapid dissolution rates, thereby improving the bioavailability of insoluble drugs to a certain extent.
3.2. Self-assembly mechanism of diNP@PDs
Next, the intermolecular forces that influence the self-assembly and stability of diNP@PDs were systematically examined. As illustrated in Fig. 1i, a comparative analysis of the UV-vis absorption spectra of Sora and Cur solutions revealed that the diNP@PDs spectrum exhibited two additional tailing peaks within the wavelength ranges of 300-390 nm and 500-590 nm. This observation suggested the presence of π-π interactions between Sora and the aromatic skeletons of Cur. Fourier transform infrared spectroscopy (FT-IR) provided valuable insights into the nature of interactions among bioactive compounds following encapsulation (Fig. 1j). As expected, the FT-IR spectrum of diNP@PDs, which comprises both Sora and Cur, displayed characteristic peaks corresponding to each component. For Cur, the characteristic peaks at 3517.58 cm−1 (stretching vibrations of -OH), 1627.65 cm−1 (α-β-unsaturated ketone C=O), 1602.58 cm−1 (stretching vibrations of C=C), 1510.01 cm−1 (aromatic in plane bending vibrations of C=C), and 962.32 cm−1 (aromatic out-of-plane bending vibrations of = C-H) were observed [36]. Meanwhile, for Sora, the characteristic peaks at 3380.66 cm−1, 3338.23 cm−1, and 3299.66 cm−1 corresponded to the stretching vibrations of -NC=O. Notably, in the spectrum of diNP@PDs, the absence of -OH peaks at 3517.58 cm−1 from Cur and -NC=O at 3380.66 cm−1, 3338.23 cm−1, and 3299.66 cm−1 from Sora indicated that hydrogen bonding plays a significant role in the formation of diNP@PDs nanoparticles.
The structural properties and crystallinity characteristics of these formulations were analyzed using X-ray diffraction (XRD). Specifically, XRD patterns were recorded over a range spanning a 2θ value of 4° to 40° (Fig. 1k). The XRD pattern revealed that pure Sora and Cur exhibited crystalline structures; however, upon nanoparticle formation into diNP@PDs, no distinct single peaks were observed, indicative of an amorphous state resulting from self-assembly processes. This amorphous state subsequently enhanced the solubility and bioavailability of both Sora and Cur. Table S2 summarized the notable cross-peaks observed in the NOESY 2D NMR spectra of diNP@PDs. The corresponding peaks demonstrated strong spatial interactions between NH-8 of Sora and H-4/H-4’ of Cur, characterized by chemical shifts of 9.07 and 7.54 ppm (Fig. 1l), respectively. Structurally, this electron-rich donor system consists of two ortho-methoxylated phenols and the enol tautomer of Cur. These findings suggested that self-assembly of diNP@PDs involves multiple structural contributions, including aromatic rings, pyridine rings, amide groups, ketonic groups, and the formation of a conjugated system. Notably, hydrogen bonds are likely to form between the ketonic group of Cur and the nitrogen atoms in Sora. Furthermore, π-π stacking interactions between aromatic and pyridine rings play a crucial role in stabilizing the assembly (Fig. 1m and n).
To further investigate the self-assembly mechanism underlying the formation of diNP@PDs, molecular dynamics (MD) simulations were conducted (Fig. S3). Fig. 1o illustrated the structural evolution of the diNP@PDs system across different simulation time points, revealing significant aggregation events involving both Sora and Cur throughout the process. Initially, the system was characterized by clustering among individual Sora molecules, which subsequently underwent spontaneous aggregation involving multiple units, each comprising both Sora and Cur monomers. The simulation trajectory revealed that certain Cur molecules integrated into the aggregates formed by Sora, while others remained at the periphery. The final stable complex of diNP@PDs, with Sora molecules rendered in surface view and Cur molecules depicted using a stick model, was illustrated from various angles in Fig. 1p. A stable structure composed of multiple Sora molecules was centrally located within the system, with Cur molecules embedded within the voids of this Sora-formed architecture. We subsequently analyzed the noncovalent interactions involved in the self-assembly process (Fig. 1q). Hydrogen bonds were established between Sora and Cur molecules; specifically, the carboxyl group (-C=O) of Cur acted as a hydrogen bond acceptor and interacted with the nitrogen atom (-NC=O) of Sora. Additionally, the negatively charged oxygen atom in the carboxyl group of Cur engaged in electrostatic interactions with the positively charged nitrogen atom present in the pyridine ring of Sora, thereby contributing to the internal stability of their alignment. Furthermore, π-π stacking interactions were observed between the pyridine ring of Sora and the aromatic ring of Cur. Collectively, these multifaceted interactions played a vital role in ensuring the stability of the nanoparticles. These theoretical findings were corroborated by experimental results obtained through various spectroscopic observations.
3.3. Photophysical properties of label-free diNP@PDs
In our previous studies, we successfully formulated various drug molecules into pharmaceutical dots through pharmaceutical approaches, which exhibited significant fluorescence enhancement suitable for bioimaging [25,26]. This straightforward and practical method seamlessly converts drug molecules into label-free probes, retaining their therapeutic efficacy while circumventing the cytotoxicity associated with traditional chemical probe synthesis. To identify the optimal diNP@PDs formulation for fluorescence performance, we investigated the optical properties of Sora, Cur and diNP@PDs (Fig. 2a–c). Free Cur exhibited relatively weak luminescence in solution, yet its emission was significantly weakened or even quenched in the aggregated state, demonstrating a typical aggregation-caused quenching (ACQ) effect (Fig. 2b). Intriguingly, diNP@PDs was non-emissive in pure DMSO but exhibited a significant enhancement in luminescence as the water fraction increased within a water/DMSO mixture (Fig. 2c), akin to the aggregation-induced emission (AIE) effect [37]. Inspired by this observation, we subsequently measured the fluorescence intensities of diNP@PDs and free Cur (Fig. S4). As expected, diNP@PDs displayed significantly higher fluorescence intensity than free Cur (p < 0.0001). Under identical conditions, the fluorescence intensities of Sora and Cur individually were virtually negligible, illustrating an optical property characterized by “0 + 0>2” (Fig. 2d and e). Here, “0” denotes the negligible optical behavior of Sora and Cur individually, while “2” represents the enhanced optical behavior of diNP@PDs upon their combination. These unique optical properties were characterized using fluorescence spectroscopies. However, this fluorescence phenomenon in drug molecules is not entirely analogous to AIE. The fluorescence intensity of diNP@PDs remains stable in aqueous solution in the absence of DMSO. Moreover, although Sora and Cur are individually non-emissive in the aggregated state, a pronounced luminescent phenomenon emerges upon their co-assembly into stable pharmaceutical dots. This observation distinguishes the phenomenon from conventional aggregation- or crystallization-based emission mechanisms. Specifically, AIE activates fluorescence through aggregation-induced restriction of intramolecular motion (RIM), whereas aggregation-enhanced emission (AEE) enhances pre-existing emission through multiple synergistic mechanisms, including RIM and π–π stacking. Crystallization-induced emission (CIE) refers specifically to fluorescence enhancement arising from crystalline packing. In contrast, the phenomenon observed here originates from supramolecular assembly of intrinsically non-emissive drugs and excipients into nanoscale pharmaceutical dots, without any covalent chemical modification. We therefore term this phenomenon “pharmaceutical dots-induced emission”, emphasizing the intrinsic emissive properties acquired by pharmaceutical molecules themselves.
Fig. 2.
Optical characterization of diNP@PDs. (a-c) Fluorescence spectra of Sora, Cur and diNP@PDs were acquired in water/DMSO mixtures with increasing water fractions (60-95%, v/v). (d) Visual fluorescence observations under 365 nm UV irradiation, demonstrating formulation behavior across the water fraction gradient (0-95%, v/v). (e) Fluorescence emission profiles (λex: 460 nm; λem: 550 nm) comparing the intensity of Sora, Cur and diNP@PDs in aqueous media. (f-h) The excitation and emission spectra of Sora (f), Cur (g) and diNP@PDs (h). (i-j) Quantum yield (QY) and fluorescence lifetimes (τ) of diNP@PDs in solution (i) and solid powder (j) state. (k) Solvent-dependent photophysical properties of diNP@PDs. The fluorescence characteristics were investigated in organic solvents of varying polarity (in descending order): DMSO > MeOH > EtOH > THF. All measurements were conducted at an excitation wavelength of 420 nm. (l) Excitation-independent fluorescence behavior of diNP@PDs in aqueous solution. (m) The hydrodynamic diameter of diNP@PDs exhibited significant solvent-dependent variations: 95% water (283.52 nm); 80% water (376.84 nm); 50% water (553.63 nm); 25% water (817.11 nm); Pure DMSO (832.53 nm).
To further clarify the optical properties of diNP@PDs, we systematically characterized the excitation and emission spectra in both solution and solid powder states. As illustrated in Fig. 2f and g, free Sora presented a distinct excitation peak at 260 nm and an emission peak at 420 nm, while Cur showed a typical excitation maximum at 440 nm and emission at 610 nm. Differently, diNP@PDs possessed a primary excitation peak at 440 nm, accompanied by a notable red-shifted emission peak at 570 nm (Fig. 2h). It is well acknowledged that a smaller Stokes shift generally causes severe spectral overlap, which further triggers light backscattering and fluorescence self-quenching effects. In comparison, diNP@PDs achieved a large Stokes shift of 130 nm, which greatly reduced self-absorption interference and resisted background biological autofluorescence. To uncover the intrinsic intramolecular interaction mechanisms of diNP@PDs, we further determined its quantum yield (QY) and fluorescence lifetimes (τ) in liquid and solid states to clarify the radiative and non-radiative decay processes. Quantitative results showed that diNP@PDs in aqueous solution exhibited a QY value of 1.81% and a fluorescence lifetime of 0.5716 ns (Fig. 2i), whereas the powder sample delivered a slightly decreased QY (1.48%) and a prolonged lifetime of 0.7606 ns (Fig. 2j). Collectively, these steady-state and transient fluorescence parameters comprehensively reflected the unique luminescence decay behavior of diNP@PDs at different states.
Solvent polarity is a critical factor for evaluating the optical performance, as polar microenvironment changes can modulate the energy levels of internal chromophores and further alter the optical absorption and fluorescence responses of nanoparticles. In this work, diNP@PDs exhibited obvious solvent-dependent fluorescence behaviors, with its fluorescence intensity gradually changing along with the increase of solvent polarity and reaching the maximum value in DMSO (Fig. 2k). We speculated that the surface passivation structure of diNP@PDs was the key factor regulating its solvent-responsive emission performance. Intriguingly, distinct from most conventional AIE luminogens, diNP@PDs maintained fixed emission positions under variable excitation wavelengths, showing excitation-independent fluorescence features (Fig. 2l). This unique optical property effectively avoids signal deviation caused by excitation fluctuation and eliminates biological autofluorescence interference, endowing diNP@PDs with great application potential in high-precision bioimaging and specific cell labeling. In addition, DLS was applied to verify the correlation between morphological evolution and photoluminescence properties. DLS results confirmed the successful formation of uniform pharmaceutical dots with the increasing water fraction (Fig. 2m), which further explained the structural basis for the excellent fluorescence performance of diNP@PDs in aqueous environments.
4. diNP@PDs induced L-M-N contacts in 4T1 cells model
4.1. In vitro migration inhibition efficacy of diNP@PDs
These label-free diNP@PDs exhibited pronounced intracellular fluorescence, whereas free Cur and Sora yielded negligible fluorescent signals (Fig. 3a and b), substantiating their utility in tracking self-assembled diNP@PDs vehicles within living cells. To further compare the cytotoxic effects of Sora, Cur, Sora + Cur and diNP@PDs, 4T1 cells were further incubated for 48 h at designated concentrations of Sora and subsequently assessed using the MTT assay. The results indicated that tumor cell cytotoxicity was significantly inhibited in a dose-dependent manner (Fig. 3c), with diNP@PDs exhibiting slightly superior antitumor effects among the tested groups. Following the cell cytotoxic assay, we unexpectedly observed a notable increase in trypsin digestion time for cells treated with diNP@PDs (Fig. 3d; Video S1). Interestingly, we observed that cells treated with Sora and diNP@PDs for an equivalent duration of digestion exhibited fewer morphological alterations and enhanced cell adhesion properties compared to those in the control and Cur groups. Cell adhesion is a crucial factor in tumor metastasis, with the initial step of tumor invasion involving the detachment and dissociation of individual tumor cells from the primary tumor due primarily to the loss of intercellular adhesion molecules [38].
Fig. 3.
Assessing the cellular imaging capabilities and anti-migratory efficacy of diNP@PDs in vitro. (a) Representative confocal images illustrating the augmented cellular imaging capability of diNP@PDs in contrast to free Cur and Sora. Scale bar, 40 μm. (b) Flow cytometric quantification of cellular internalization efficiency in 4T1 cells treated with free Cur and diNP@PDs (n = 3). (c)In vitro cytotoxicity of test formulations against 4T1 cells determined by MTT assay after 48 h treatment. IC50 for Cur, 32.94 ± 2.47 μM; IC50 for Sora, 1.73 ± 0.45 μM; IC50 for Sora + Cur, 1.51 ± 0.44 μM; IC50 for diNP@PDs, 1.06 ± 0.26 μM. (d) Representative images showing trypsin digestion differences after 48 h treatment with test formulations at a concentration of 10 μM (equivalent to Sora dose). Compared with other groups, diNP@PDs significantly prolonged the trypsin digestion time, indicating a marked enhancement in cell adhesion following treatment. This suggests that the impact of diNP@PDs on cell proliferation is less pronounced than its effect on cell migration. Scale bar, 100 μm. (e-f) Representative phase-contrast images (e) and quantitative analysis of wound closure (f) after various treatment. Scale bar, 100 μm. (g-i) Representative images (g) and quantitative analysis (h,i) of cell migration and invasion assessing the effects of different treatments via transwell chamber assay. Scale bar, 100 μm. (j) Representative confocal images of EMT markers E-cadherin and Vimentin expression following various treatment. Scale bar, 40 μm. (k) Immunoblotting analysis of EMT markers E-cadherin and Vimentin expression relative to Tubulin following various treatment. Statistical significance of p values in b was calculated via two-tailed Student's t-test, and f,h,i was calculated via one-way ANOVA with a Tukey post-hoc test.
Supplementary data related to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103681
The following are the Supplementary data related to this article:
Inspired by this intriguing phenomenon, we conducted wound healing assays and trans-well chamber assays to evaluate the efficacy of these drugs in inhibiting cellular invasion and migration capabilities. In the wound healing assay, nearly complete closure was achieved within 24 h post-injury in the control group (Fig. 3e and f). Cells treated with free Sora, Sora + Cur and diNP@PDs displayed significantly reduced migratory capabilities, with recorded wound-healing rates of approximately 52.39%, 34.19%, and 11.53%, respectively. These treated cells exhibited diminished migratory capacity, as evidenced by their slower wound closure compared to untreated controls. Under identical conditions, diNP@PDs demonstrated a remarkable ability to effectively suppress both migration and invasion of tumor cells (Fig. 3g–i). To further validate the effect of diNP@PDs on reducing tumor migration, immunofluorescence and western blot analysis was employed to quantify the protein expression levels of EMT markers like E-cadherin and Vimentin (Fig. 3j and k). The downregulation of E-cadherin coupled with the overexpression of Vimentin in cancer was strongly correlated with enhanced tumor invasion [39]. Notably, in the diNP@PDs group, there was a significant increase in E-cadherin expression levels alongside a marked reduction in Vimentin expression levels.
4.2. Induction of L-M-N contacts by diNP@PDs
Given the inhibitory effect of diNP@PDs on tumor migration and invasion, we further explored its underlying molecular mechanism. First, we investigated the cellular microstructure in breast cancer and liver cancer cell lines using electron microscopy to scrutinize cells treated with diNP@PDs (Fig. 4a–S5-6). Notably, subcellular organelle interactions among lysosomes, mitochondria and the nuclei drew our attention. We hypothesized that diNP@PDs may modulate the contacts among these organelles. To further test this hypothesis, we identified the contact sites between lysosomes, mitochondria and nuclei (Fig. 4b). In diNP@PDs-treated 4T1 cells, the average distance between mitochondria-mitochondria (M-M) contacts was determined to be 16.66 ± 8.88 nm, which aligned with the mitochondria-lysosomes (M-L) contact distance of 21.52 ± 9.92 nm and the mitochondria-nuclei (M − N) contact distance of 44.89 ± 18.77 nm. These distances were significantly shorter than those observed in untreated cells, where M-M contacts averaged 242.02 ± 134.23 nm, M-L contacts averaged 235.54 ± 106.77 nm, and M − N contacts averaged 225.28 ± 172.37 nm. Similar findings were also observed in MHCC-97H cells (Fig. S7). Subsequently, we investigated the interactions among mitochondria, lysosomes, and the nuclei in 4T1 cells using structured illumination microscopy (SIM). We also confirmed that lysosomes labeled with LysoBrite Red and mitochondria labeled with PK Mito Deep Red could simultaneously establish contact with the nucleus after treatment with diNP@PDs (Fig. 4c).
Fig. 4.
Lysosome-mitochondrion-nucleus interactions induced by diNP@PDs. (a) Representative TEM images showing lysosome (L) - mitochondrion (M) - nucleus (N) contact in 4T1 cells after 48 h treatment. Scale bar, 2 μm. (b) Quantified distance distributions between organelles (M-M, M-L, M − N). The additional images were shown in Fig. S5–6. These results revealed that diNP@PDs treatment significantly increased perinuclear clustering of M-L populations, suggesting potential modulation of nuclear function through organelle-nuclear crosstalk. (c) Representative structured illumination microscopy (SIM) images of L-M-N contacts in 4T1 cells after 12 h treatment with diNP@PDs, with lysosomes labeled using LysoBrite Red (50 nM) and mitochondria labeled using PK Mito Deep Red (100 nM). (d-e) Representative confocal images (d) and semi-quantitative analysis (e) showing the colocalization of lysosomes with nuclei. in 4T1 cells following 48 h treatment. Scale bar, 20 μm. (f-g) Representative confocal images (f) and semi-quantitative analysis (g) showing the colocalization of mitochondria with nuclei in 4T1 cells following 48 h treatment. Scale bar, 20 μm. (h) Representative confocal images of L-M-N contacts in 4T1 cells stained for LAMP1 (lysosome), TOMM20 (mitochondrion), and DAPI (nucleus) following 48 h treatment. (i-j) Mean intensity values of lysosome (i) and mitochondrion (j) surrounding the nucleus corresponding to (h) showing L-M-N contacts. (k,l) Immunoblotting analysis of VEGFR signaling pathway components (VEGFR, pVEGFR, ERK, pERK) (k) and subcellular localization markers (lysosomal: CD107a; mitochondrial: DHODH, GPX4) (l). (m) Schematic representation of L-M-N contacts. Data are shown as mean ± s.d. Statistical significance of p values in e,g,i,j was calculated via one-way ANOVA with a Tukey post-hoc test.
Recently, the intricate contacts between mitochondria and lysosomes have emerged as a pivotal nexus in tumor metabolism, cell death signaling, and drug resistance [40]. Targeting this complex interplay holds significant promise for the development of novel anticancer therapies [41]. Concurrently, the nucleus serves as a central regulatory hub in tumorigenesis and tumor progression. For instance, intranuclear genetic mutations, such as the inactivation of TP53, or epigenetic alterations can lead to uncontrolled cell cycle progression, inhibition of apoptosis, and activation of oncogenes, thereby promoting tumor proliferation [42]. Moreover, oncogenic proteins within the nucleus, such as ANLN, interact with RNA polymerase II and form transcriptional condensates through phase separation, thereby enhancing the expression of oncogenes [43]. Additionally, the abnormal activation of nuclear receptor signaling, such as that of ERα and AR, promotes tumor growth [44]. These findings underscore the potential of targeting nuclear abnormalities to develop novel therapeutic strategies to improve treatment efficacy. However, the intricate contacts among mitochondria, lysosomes, and the nucleus remain largely unexplored.
To investigate more precisely how diNP@PDs affects sub-organelle function, cells underwent various treatments for 48 h at a lower concentration of drugs (5 μM). As shown in Fig. 4d–g, the administration of diNP@PDs resulted in a notable enhancement in the communication dynamics between the cell nucleus and organelles, such as lysosomes and mitochondria, compared to free drugs. Furthermore, diNP@PDs induced a significant accumulation of lysosomes and mitochondria around the cell nucleus, and this substantial aggregation led to noticeable changes in the morphology of the nucleus. As evidenced by the fluorescence intensity calculations, the average fluorescence intensity of lysosomes and mitochondria localized around the nucleus was significantly higher in tumor cells treated with diNP@PDs than in those treated with free Cur and Sora (Fig. 4e and g). These findings indicate that diNP@PDs modulate lysosomes-mitochondria-nuclei contacts interorganelle communication, thereby potentially attenuating tumor cell metastasis.
The confocal images of LAMP1 on late endosomal-lysosomal membranes and TOMM20 on outer mitochondrial membranes further demonstrated that treatment with diNP@PDs for 24 and 48 h resulted in a higher frequent of contacts among lysosomes, mitochondria and the nuclei (Fig. 4h and S8). This was evidenced by a mean fluorescence intensity that surpassed that observed in free monomer and untreated cells. As shown in Fig. 4i and j, this conclusion is consistent with the fluorescence colocalization spectra of mitochondria and lysosomes surrounding the nuclei.
Sora demonstrated a remarkable ability to not only block the extracellular regulated kinases (ERK)-mediated cell signaling pathway, thereby directly suppressing cell proliferation, but also effectively inhibit the activity of VEGFR. This dual action indirectly impeded tumor neovascularization. Consequently, Sora served to restrain both the growth and dissemination of tumor cells [45]. Elucidating the mechanism of action underlying the formation of diNP@PDs with Sora and Cur is highly warranted. Western blot analysis was employed to assess the biological function of diNP@PDs revealing the involvement of the VEGFR signaling pathway (Fig. 4k). In both the Sora + Cur and diNP@PDs groups, concurrent treatment significantly decreased the expression levels of pVEGFR and pERK, implying a synergistic effect; however, the expression levels of total VEGFR and ERK remained relatively unchanged. These results indicate that diNP@PDs not only retain the original function of Sora in inducing the phosphorylation of VEGFR but also augment the efficacy of Sora.
Furthermore, in addition to the aforementioned functions, the effects of diNP@PDs on lysosomes, mitochondria, and the nuclei remain to be fully unraveled. CD107a, a lysosomal-associated membrane protein (LAMP), is characterized by highly glycosylated transmembrane proteins that serve as lysosomal markers [46]. The inactivation of dihydroorotate dehydrogenase (DHODH) suppresses extensive mitochondrial ferroptosis through the inhibition of glutathione peroxidase 4 (GPX4); both processes occur within the mitochondria. Therefore, to verify the novel functions of diNP@PDs, key proteins in lysosomes and mitochondria were analyzed via western blotting. As shown in Fig. 4l, the results revealed a reduction in CD107a expression, indicating that tumor inhibition may occur through enhanced autophagy. Subsequently, we examined the effects of diNP@PDs on GPX4 and DHODH expression, revealing that diNP@PDs exert a significant sensitizing effect on mitochondrial ferroptosis in 4T1 cells. Collectively, these findings demonstrated that diNP@PDs not only preserved the functionality of Sora and modulated VEGFR phosphorylation, but also extended the functional capabilities of Sora to induce interorganelle contacts among mitochondria, lysosomes, and the nuclei (L-M-N contacts) (Fig. 4m).
4.3. Exploring the mechanism of L-M-N contacts through proteomics analysis
Given the ability of diNP@PDs to facilitate communication among lysosomes, mitochondria, and the nucleus, we next investigated whether this interorganelle coordination contributes to their antimetastatic activity. As shown in Fig. 4, treatment with diNP@PDs markedly promoted the perinuclear clustering of lysosomes and mitochondria, indicating enhanced spatial proximity and potential interorganelle interactions. These findings prompted us to hypothesize that diNP@PDs suppress tumor metastasis, at least in part, by promoting interorganelle contacts and thereby modulating downstream signaling pathways involved in metastatic progression.
To validate this hypothesis, we conducted a comprehensive proteomics analysis to explore the underlying mechanisms through which diNP@PDs influences tumor metastasis. A suite of bioinformatics analyses including Gene Ontology (GO) analysis, Kyoto Encylopedia of Genes and Genomes (KEGG) pathways, chord diagram analysis, and Gene Set Enrichment analysis (GSEA), was performed to analyze the significantly differentially expressed proteins (DEPs) between the diNP@PDs and control groups. Initially, we visualized the distribution of DEPs using a volcano plot (Fig. 5a). A total of 2244 proteins were identified as differentially expressed between the diNP@PDs and control groups. Specifically, 1102 proteins exhibited elevated expression levels (diNP@PDs/control Ratio ≥ 1.502, p < 0.05), while 1142 proteins showed reduced expression levels (diNP@PDs/control Ratio ≤ 0.666, p < 0.05) in response to diNP@PDs treatment. Through GO enrichment analysis of these proteins, the cellular component (CC) results indicated that DEPs were predominantly localized within the nucleus (28.35%), cytoplasm (24.57%), and mitochondrion (20.29%) (Fig. 5b). This distribution underscores the significant roles of diNP@PDs in the cytoplasm, mitochondrion, and nucleus. Employing high-throughput and high-precision parallel reaction monitoring (PRM)-based targeted proteomics techniques, we conducted an independent analysis that revealed that regulated expression of proteins associated with lysosomes, mitochondria, and the nuclei, such as Ctsa, Mpc1, Top2a, and Hmgb1, in the diNP@PDs-treated group. This finding underscores the multi-targeting effects of diNP@PDs on these organelles and highlights their role in modulating the expression of metastasis-related proteins, thereby reinforcing the inhibitory effects of diNP@PDs on cell metastasis (Fig. 5c).
Fig. 5.
The supratargeting mechanism through proteomics analysis. (a) Volcano plots of differentially expressed proteins (DEPs) in 4T1 cells treated with Control or diNP@PDs (25 μM). Proteins are represented by dots, with upregulated proteins highlighted in red and downregulated proteins in blue. (b) Proteomics analysis showing DEPs in distinct subcellular compartments of 4T1 cells treated with Control or diNP@PDs. The subcellular structure annotations for DEPs were classified based on Gene Ontology (GO) analysis. (c) PRM-based proteomic profiling elucidates multi-organelle targeting mechanism of diNP@PDs. Through high-resolution parallel reaction monitoring (PRM) targeted proteomics. (d-e) GO enrichment analysis of up-regulated (d) and down-regulated (e) genes in diNP@PDs compared to control groups. (f-g) GSEA enrichment analysis (f) and KEGG pathway (g) of lysosome-related genes. (h,i) Circos plots of upregulated (h) or downregulated (i) DEPs and their associated enriched pathways. (j) Venn diagram of the most differential survival genes from GEPIA2, DEPs and proteins relative to L-M-N. (k) Relative protein levels of 27 proteins located in lysosomes, mitochondria, and nuclei identified via Venn diagram. (l) Survival analysis plots of nuclear HDAC2 in breast invasive carcinoma (BRCA) patients using the GEPIA database (http://gepia.cancer-pku.cn/). (m) A Gene Set Enrichment Analysis (GSEA) for pathway enrichment of Reactive Oxygen Species based on identified DEPs, represented by normalized enrichment score (NES) and nominal p value.
Given the substantial alterations in the number of DEPs observed following diNP@PDs treatment compared to untreated controls, we proceeded to investigate the potential pathways enriched within the treated cohort. GO analysis showed that the up-regulated proteins were predominantly enriched in molecular functions (MF) and biological processes (BP) related to mitochondria, including “oxidoreductase activity”, “NADH dehydrogenase activity”, “mitochondrial gene expression”, “oxidative phosphorylation”, and “nucleoside phosphate metabolic process” (Fig. 5d). Mitochondria, as essential organelles in most metabolic pathways, played a critical role in energy production and biomass generation. They function as metabolic sensors, regulate cancer cell apoptosis, and initiate signaling pathways associated with cancer cell migration, invasion, metastasis, and resistance to therapeutic interventions [47]. Moreover, the rapid metabolism and active proliferation of tumor cells heighten their reliance on lysosomes and mitochondria [11]. Lysosomal stress can disrupt mitochondrial morphology and function, thereby triggering apoptosis via ROS signal transduction [48]. This suggests that diNP@PDs may induce lysosome-mitochondrion interactions to promote ROS generation. Conversely, the down-regulated proteins were predominantly associated with nuclear functions, including “protein phosphatase regulator activity”, “translation initiation factor activity”, “regulation of protein depolymerization”, “mitotic cell cycle phase transition”, and “DNA replication” (Fig. 5e). Collectively, these findings indicated that diNP@PDs may inhibit cancer metastasis by mediating the biological interactions among lysosomes (Fig. 5f and g), mitochondria, and the nuclei, thereby modulating key metabolic and nuclear processes.
In addition, the chord diagram illustrated that up-expressed proteins were significantly associated with several representative pathways, including “Lysosome”, “PPAR signaling pathway”, and “Tryptophan metabolism” (Fig. 5h). These pathways collectively mediated lysosomal, mitochondrial and nuclear functions, highlighting the intricate interplay between these organelles. These enrichment analysis results further suggested that redox metabolism might play a pivotal role in inhibiting tumor metastasis. Conversely, down-regulated proteins were associated with pathways related to the “Pentose phosphate pathway”, “DNA replication” and “Cell cycle” (Fig. 5i). The “Pentose phosphate pathway” is associated with purine synthesis, while the “DNA replication” pathway influences DNA acetylation and methylation processes. These findings suggested that diNP@PDs’ inhibition of tumor metastasis is closely linked to epigenetic alterations within tumors [49,50].
To further identify targeted proteins through which diNP@PDs impacts tumor epigenetics, a Venn diagram was utilized to intersect breast invasive carcinoma (BRCA)-related survival genes from GEPIA2 with DEPs derived from proteomics analysis (Fig. 5j). Among the 48 genes/proteins identified by GEPIA2 and DEPs, 27 proteins were found to be associated with lysosomes, mitochondria, and nuclei (Fig. 5k). Given that epigenetic regulation primarily occurs within the nucleus, we identified eight proteins closely associated with the nucleus among the aforementioned 27 proteins. Analysis of the most differentially expressed survival genes alongside clinical data indicated that HDAC2 in the nucleus may serve as a potential therapeutic target for addressing tumor metastasis (Fig. 5l and S9). HDAC2 overexpression is correlated with metastasis and progression, positioning it as both a potential therapeutic target and a prognostic indicator for select cancer patients [51,52]. Clinical analyses revealed that the elevated levels of HDAC2 expression are associated with poor overall survival outcomes in human breast cancer patients [53]. That is to say, diNP@PDs may inhibit tumor metastasis by modulating the expression of HDAC2. As previously discussed, both redox metabolism and HDAC2 were implicated in tumor metastasis, necessitating an exploration of their relationship. Under pathological conditions, redox dysregulation results in excessive ROS accumulation, causing oxidative stress and cellular damage [54]. Our GSEA analysis revealed significant alterations in pathways related to “reactive oxygen species” between the diNP@PDs and control groups (Fig. 5m).
4.4. Inhibiting tumor metastasis via HDAC2 sensing nuclear ROS
Mitochondria are the primary site of ROS generation, while lysosomes can regulate mitochondrial ROS production through various mechanisms [55]. Thus, M-L contacts facilitated increased ROS production (Figs. 5m, 6a, and S10). Furthermore, ROS generation exhibited concentration- and time-dependent characteristics (Fig. 6b and c). Upon treating tumor cells with diNP@PDs, the formation of L-M-N contacts leads to the aggregation of mitochondria and lysosomes around the nucleus, potentially promoting the translocation of mitochondrial ROS to the nucleus (Fig. 6d and e). Following nuclear isolation, GAPDH was undetectable in the nuclear fraction, and the nuclei exhibited intact morphology, confirming that the isolated nuclei were of adequate purity for subsequent analyses (Fig. S11). Additionally, we confirmed that the fluorescence signal originating from diNP@PDs did not interfere with ROS probe detection (Fig. 6f). Therefore, we hypothesized that this organelle aggregation around the nucleus is associated with heightened nuclear ROS (nuROS) sensing, which in turn downregulates HDAC2 function and subsequently inhibits cell migration activity (Fig. 6g–i). Moreover, these findings are consistent with the observed downregulation of HDAC2 expression in MHCC-97H cells (Fig. S12).
Fig. 6.
Inhibiting tumor metastasis via HDAC2 sensing nuclear ROS. (a) Representative confocal images of intracellular ROS generation after the indicated treatment. (b) Dose-response relationship of ROS production in 4T1 cells after 24 h treatment with diNP@PDs (5-50 μM). (c) Time-course analysis of ROS generation in 4T1 cells treated with 20 μM diNP@PDs. (d-f) Nuclear ROS (nuROS) in 4T1 cells post-treatment. Quantitative analysis of nuROS levels by flow cytometry (d-e). Purple bars: nuROS levels after various treatment, measured via DCFH-DA staining; Blue bars: drug fluorescence in the FITC channel after cellular uptake (f). Nuclear fractions were isolated using a Nuclear Extraction Kit (Solarbio®), following manufacturer's protocol. (g-h) HDAC2 immunofluorescence in 4T1 cells treated with Cur (6.25 μM), Sora (25 μM) and diNP@PDs (25 μM) for 12 h. (i) Immunoblotting analysis of HDAC2 expression after the indicated treatment. (j) Flow cytometry analysis of cellular ROS levels after indicated treatments with the ROS inhibitor NAC or diNP@PDs. (k,l) Representative confocal images (k) and mean intensity values (l) of cellular immunofluorescence of HDAC2 after indicated treatments with the ROS inhibitor NAC or diNP@PDs. Scale bar, 10 μm. (m-n) Representative images (m) and quantitative analysis (n) of wound healing assay to measure the migration ability of the HDAC2 inhibitor SAHA and diNP@PDs. (o) Schematic illustration of the mechanism underlying tumor metastasis inhibition via the nuROS-HDAC2-EMT pathway. Data are shown as mean ± s.d. Statistical significance of p values in b,c,e,f,h,j,l was calculated via one-way ANOVA with a Tukey post-hoc test, and this in n was calculated by a two-tailed Student's t-test.
To further elucidate the impact of nuROS on HDAC2 expression, we employed the ROS scavenger N-acetylcysteine (NAC) and the HDAC2 inhibitor SAHA. Treatment with NAC at concentrations of 2 or 4 μM resulted in a reduction in ROS levels (Fig. 6j) and concomitant upregulation of HDAC2 expression (Fig. 6k and l). Conversely, diNP@PDs reversed the effects of NAC, increasing ROS levels and downregulating HDAC2 expression. Additionally, treatment with the SAHA significantly reduced cell migration rates (Fig. 6m and n), confirming that HDAC2 promotes tumor metastasis and that diNP@PDs can inhibit metastasis through HDAC2 modulation. These findings strongly suggested that HDAC2 may sense and respond to changes in nuclear ROS levels, ultimately inhibiting tumor metastasis (Fig. 6o).
5. Therapeutic efficacy of diNP@PDs in the 4T1 lung metastasis mice model
5.1. In vivo suppression of tumor metastasis by diNP@PDs
To evaluate the in vivo imaging capability of diNP@PDs, we compared the fluorescence intensities of diNP@PDs and the mixture of Sora + Cur in 4T1 tumor-bearing mice. The results showed that the mixture exhibited negligible fluorescence signals in the tumor-bearing lungs, whereas diNP@PDs displayed strong detectable signals (Fig. S13), indicating that diNP@PDs can be used to trace the in vivo biodistribution of the self-assembled pharmaceutical dots.
Given the capacity of diNP@PDs to suppress the metastatic potential of 4T1 cells, we first established a metastasis-inhibition model using 4T1 cells pretreated with diNP@PDs and subsequently implanted these cells into Balb/c mice. Specifically, 4T1 cells were pretreated with diNP@PDs at a concentration of 10 μM for 12 h before being intravenously injected into mice (Fig. 7a). We observed that diNP@PDs exhibited a significantly greater inhibitory effect on pulmonary metastasis in the 4T1 lung metastasis model compared with Cur or Sora alone (Fig. 7b and c). The mice were then monitored over a 15-day period post-injection, during which their body weights were meticulously recorded (Fig. 7d). Histopathological examination of H&E-stained lung sections and measurement of lung weight revealed a significant reduction in breast cancer lung metastasis in the group of cells pretreated with diNP@PDs, as compared to both the control group and the monomer group.
Fig. 7.
In vivo antitumor metastasis activity of diNP@PDs. (a) Representative images of lungs on day 15 after inoculation with diNP@PD-treated 4T1 cells (n = 5). (b) H&E-stained lung sections showing breast tumor metastasis. Scale bar, 2.5 mm. (c) Lung weights measured at day 15 post-treatment. (d) Body weight changes in tumor-bearing mice over time. (e) Schematic illustration of administration regimen for assessing the impact of diNP@PDs on 4T1 cell metastasis in vivo. First, 4T1 cells were incubated with diNP@PDs for 12 h to obtain metastasis-inhibited cells. These cells were then inoculated into Balb/c mice. Three days later, the test formulations (10 mg/kg, Sora equivalent) were administered every two days for a total of five doses. (f) Representative images (upper) and H&E sections (lower) of the lungs collected from different groups. (g) Tumor-bearing lung weight after indicated treatment on day 15 (n = 5). (h) Quantification of metastatic lesions after various treatment. (i) Body weight changes in tumor-bearing mice. (j) Schematic illustration of administration regimen for diNP@PDs i.v. injection to 4T1-luc bearing Babl/c mice. 4T1-luc cells were administered intravenously. Subsequently, the test formulations (PTX: 4 mg/kg; Cur: 2 mg/kg; Sora or diNP@PDs: 10 mg/kg) were delivered every two days, culminating in a total of five doses. Bioluminescence imaging was employed on days 0, 2, 5, 8, and 12 to assess tumor metastasis. On day 15, mouse lungs were harvested for further investigation. (k) Representative images of metastatic lungs after indicated treatments on day 15 (n = 5). (l) Tumor-bearing lung weight of 4T1-luc tumors after indicated treatment on day 15 (n = 5). (m-o) H&E staining of lung sections (m), quantification of metastatic lesions (n), enumeration of metastatic lung nodules (o) to assess the inhibitory effects on metastasis after various treatment (n = 5). Statistical significance of p values in c,h,l,n,o was calculated via one-way ANOVA with a Tukey post-hoc test.
Based on these findings, we aimed to evaluate the sustained effects of diNP@PDs on breast tumor lung metastasis following routine treatment, which involved intravenous injection of drug-pretreated cells along with continued administration over time (Fig. 7e). As illustrated in Fig. 7f, gross lung morphology and H&E staining revealed that the therapeutic efficacy of diNP@PDs was superior to that of Sora, Cur, and Sora + Cur, as evidenced by the lung weight of tumor-bearing mice (Fig. 7g) and the area of tumor nodules (Fig. 7h). Throughout the experimental period, body weights within the diNP@PDs group remained stable (Fig. 7i), suggesting minimal toxicity associated with routine administration of the treatment. These findings indicated that intravenous injection of drug-pretreated cells combined with regular administration can effectively suppress breast cancer lung metastasis.
Building on these promising results, we further evaluated the therapeutic efficacy of diNP@PDs in a lung metastasis model using 4T1-luc-bearing mice. Paclitaxel (PTX), a first-line chemotherapeutic agent for breast cancer, has been reported to exacerbate metastasis in mouse models of the disease [56]. Consequently, PTX was employed as the reference group to investigate the effects of various treatments on metastasis. Following intravenous injection of 4T1-luc cells for 3 days, tumor-bearing mice were randomly divided into five groups and treated with PBS, PTX (4 mg/kg), Cur (2 mg/kg), Sora (10 mg/kg), diNP@PDs (10 mg/kg) five times every two days (Fig. 7j). Fifteen days after the initial treatment, harvested lungs from the mice (Fig. 7k) showed substantial inhibition of metastasis by diNP@PDs, as assessed by lung weight measurement (Fig. 7l) and H&E staining (Fig. 7m). Quantification using ImageJ software further confirmed these findings, with significant reductions in both tumor area (Fig. 7n) and the number of metastatic lung nodules (Fig. 7o) in the diNP@PDs group. Serial bioluminescence imaging further provided direct evidence that diNP@PDs significantly suppressed breast cancer lung metastasis compared to the control, PTX, Cur, and Sora groups (Fig. 8).
Fig. 8.
diNP@PDs-induced L-M-N contacts in vivo regulate HDAC2 signaling and attenuate metastasis. (a) Schematic illustration of lung metastasis inhibition after being treated with diNP@PDs. (b) Representative bioluminescence images of tumor-bearing mice at specified time points after indicated treatments (n = 3). (c-d) Immunohistochemical staining (c) and quantitative analysis (d) of EMT markers after indicated treatment. Scale bar, 25 μm. (e-f) HDAC2 expression in metastatic lungs after indicated treatment. Scale bar, 100 μm. (g) Representative confocal images of L-M-N contacts in metastasis-bearing lungs stained for LAMP1 (lysosome), TOMM20 (mitochondrion), and Lamin B1 (nucleus). Scale bar, 10 μm. Data are shown as mean ± s.d.
5.2. In vivo suppression mechanism of tumor metastasis
Building upon our earlier findings of diNP@PDs-mediated metastasis inhibition in 4T1 tumors (Fig. 8a and b), we proceeded to examine the expression of canonical EMT markers, namely E-cadherin and Vimentin. Immunohistochemical staining was performed on lung metastatic tissues to quantitatively assess the protein levels of these two EMT markers (Fig. 8c). The results revealed that the diNP@PDs treatment markedly upregulated E-cadherin expression while significantly downregulating Vimentin levels (Fig. 8d). Meanwhile, immunohistochemical detection of HDAC2 protein further verified that diNP@PDs exerted the strongest suppressive effect on HDAC2 expression among all experimental groups (Fig. 8e and f).
To further explore the subcellular structural alterations triggered by diNP@PDs, we conducted co-localization staining with three specific organelle markers: the lysosomal marker LAMP1, the mitochondrial marker TOMM20, and the nuclear membrane marker Lamin B1. In the control, PTX, Cur, and Sora groups, lysosomes and mitochondria were sparsely dispersed in tumor cells. In comparison, tumor cells from diNP@PDs treatment group displayed extensive co-localization of lysosomes and mitochondria, which aggregated into dense clusters surrounding the nucleus (Fig. 8g). Collectively, these results evidence that diNP@PDs facilitate the formation of lysosome-mitochondrion-nucleus contact sites in tumor tissues, which is highly consistent with our in vitro cellular observations. Mechanistically, diNP@PDs inhibit tumor metastasis by triggering interorganelle contact formation, which further modulates nuclear HDAC2 expression and thereby restrains the EMT process.
To preliminarily assess the in vivo biocompatibility of diNP@PDs and other formulations, histological H&E staining of major mouse organs and routine hematological analyses were conducted (Fig. S14). H&E histological examination demonstrated no obvious pathological damage in major organs across all treatment groups, confirming the good tissue compatibility of diNP@PDs. Furthermore, serum biochemical indicators related to hepatic and renal function, including aspartate aminotransferase (AST), alanine aminotransferase (ALT) and UREA, remained within physiological normal ranges after diNP@PDs administration. These findings collectively validate the favorable in vivo biosafety of this co-delivery nanoplatform.
6. Conclusion
To facilitate effective tracking or in vivo imaging of drugs or drug carriers, fluorescent probe labeling is commonly employed. However, this chemical modification often introduces significant cytotoxicity and adverse effects. Sora and Cur molecules, in their native solution states, exhibit either negligible luminescence or low fluorescence intensity. Moreover, Cur undergoes ACQ in aggregated states, which further limits its imaging capabilities. To enhance their imaging potential, we utilized supramolecular self-assembly to fabricate this molecule drug pair into diNP@PDs without any additional modifications, thereby achieving a non-fluorescent-to-fluorescent transformation (0 + 0>2 fluorescence effect).
In vitro assays revealed that diNP@PDs suppress the migration and invasion of tumor cells. Surprisingly, TEM analysis uncovered a novel mechanism whereby diNP@PDs induce cross-talk among lysosomes, mitochondria, and nuclei in 4T1 cells. These contacts induced the aggregation of lysosomes and mitochondria around the nuclei, facilitating the generation of nuROS and creating a high-concentration ROS environment within the nucleus, which modulated the epigenetic landscape, inhibiting HDAC2 activity and thereby disrupting the EMT process (nuROS-HDAC2-EMT axis), ultimately suppressing the metastasis of breast cancer.
CRediT authorship contribution statement
Rujing Wang: Writing – original draft, Investigation, Conceptualization. Lan Zou: Validation, Investigation. Dandan Mi: Visualization. Chuan Wang: Validation, Methodology. Zijian Song: Investigation. Tiantian Liu: Data curation. Mengnan Zhao: Writing – original draft, Methodology. Dong Wang: Project administration, Methodology. Sanjun Shi: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This study was financially supported by the National Natural Science Foundation of China (82405161), the Natural Science Foundation Project of Sichuan Province (2026NSFSC0609), the China Postdoctoral Science Foundation (GZC20240170 and 2024M760307), the Sichuan Province Joint Fund for Science and Education (2024NSFSC2124), the Sichuan Province Administration of Traditional Chinese Medicine (25MSZX536), the Joint Innovation Fund of Health Commission of Chengdu and Chengdu University of Traditional Chinese Medicine (WXLH202403272). We also wish to thank Huan Xu, Lu Yang and Yili Wang from Chengdu University of TCM for their helps in providing flow cytometer and multi-mode microplate reader facilities.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103681.
Contributor Information
Mengnan Zhao, Email: zhaomengnan@cdutcm.edu.cn.
Dong Wang, Email: dwang@cdutcm.edu.cn.
Sanjun Shi, Email: shisanjuns@cdutcm.edu.cn, shisanjuns@163.com.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Data availability
Data will be made available on request.
References
- 1.Gao Y., Xie J., Yang Z., et al. Functional tumor-derived exosomes in NSCLC progression and clinical implications. Front. Pharmacol. 2025;16 doi: 10.3389/fphar.2025.1485661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Sandhbor P., John G., Bhat S., et al. Immune response recalibration using immune therapy and biomimetic nano-therapy against high-grade gliomas and brain metastases. Asian J. Pharm. Sci. 2025;20 doi: 10.1016/j.ajps.2025.101021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Huang Y., Hong W., Wei X. The molecular mechanisms and therapeutic strategies of EMT in tumor progression and metastasis. J. Hematol. Oncol. 2022;15:129. doi: 10.1186/s13045-022-01347-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bergholz J.S., Zhao J.J. How compensatory mechanisms and adaptive rewiring have shaped our understanding of therapeutic resistance in cancer. Cancer Res. 2021;81:6074–6077. doi: 10.1158/0008-5472.CAN-21-3605. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Labrie M., Brugge J.S., Mills G.B., et al. Therapy resistance: opportunities created by adaptive responses to targeted therapies in cancer. Nat. Rev. Cancer. 2022;22:323–339. doi: 10.1038/s41568-022-00454-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Doostmohammadi A., Jooya H., Ghorbanian K., et al. Potentials and future perspectives of multi-target drugs in cancer treatment: the next generation anti-cancer agents. Cell Commun. Signal. 2024;22:228. doi: 10.1186/s12964-024-01607-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Haki M., Bayat R. Innovative approaches for molecular targeted therapy of breast cancer: interfering with various pathway signaling. Int J Mol Cell Med. 2025;14:533–551. doi: 10.22088/IJMCM.BUMS.14.1.533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Jiang Q., Li M., Li H., et al. Entrectinib, a new multi-target inhibitor for cancer therapy. Biomed. Pharmacother. 2022;150 doi: 10.1016/j.biopha.2022.112974. [DOI] [PubMed] [Google Scholar]
- 9.Cheng Y., Qu Z., Jiang Q., et al. Functional materials for subcellular targeting strategies in cancer therapy: progress and prospects. Adv. Mater. 2023 doi: 10.1002/adma.202305095. [DOI] [PubMed] [Google Scholar]
- 10.Liu Z., Liu Y., Kang X., et al. Subcellular organelle targeting as a novel approach to combat tumor metastasis. Pharmaceutics. 2025;17:198. doi: 10.3390/pharmaceutics17020198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Tang T., Yang Z.Y., Wang D., et al. The role of lysosomes in cancer development and progression. Cell Biosci. 2020;10:131. doi: 10.1186/s13578-020-00489-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hao T., Yu J., Wu Z., et al. Hypoxia-reprogramed megamitochondrion contacts and engulfs lysosome to mediate mitochondrial self-digestion. Nat. Commun. 2023;14:4105. doi: 10.1038/s41467-023-39811-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yang J., Li Q., Zhou M., et al. Concurrent impairment of nucleus and mitochondria for synergistic inhibition of cancer metastasis. Int. J. Pharm. 2021;608 doi: 10.1016/j.ijpharm.2021.121077. [DOI] [PubMed] [Google Scholar]
- 14.Chen L., Zhang Y., Zhang Y.X., et al. Pretreatment and analysis techniques development of TKIs in biological samples for pharmacokinetic studies and therapeutic drug monitoring. J. Pharm. Anal. 2024;14 doi: 10.1016/j.jpha.2023.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang H.L., Ma X., Guan X.Y., et al. Potential synthetic lethality for breast cancer: a selective sirtuin 2 inhibitor combined with a multiple kinase inhibitor sorafenib. Pharmacol. Res. 2022;177 doi: 10.1016/j.phrs.2021.106050. [DOI] [PubMed] [Google Scholar]
- 16.Huang X., Kuerban K., Fan J., et al. IL-2-loaded liposomes modified with sorafenib derivative exert a synergistic anti-melanoma effect via improving tumor immune microenvironment and enhancing antiangiogenic activity. Asian J. Pharm. Sci. 2025;20 doi: 10.1016/j.ajps.2025.101020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Colombo F., Trombetta E., Cetrangolo P., et al. Giant lysosomes as a chemotherapy resistance mechanism in hepatocellular carcinoma cells. PLoS One. 2014;9 doi: 10.1371/journal.pone.0114787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wu H., Wang T., Liu Y., et al. Mitophagy promotes sorafenib resistance through hypoxia-inducible ATAD3A dependent axis. J. Exp. Clin. Cancer Res. 2020;39:274. doi: 10.1186/s13046-020-01768-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yang X.D., Kong F.E., Qi L., et al. PARP inhibitor olaparib overcomes sorafenib resistance through reshaping the pluripotent transcriptome in hepatocellular carcinoma. Mol. Cancer. 2021;20:20. doi: 10.1186/s12943-021-01315-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Gavini J., Dommann N., Jakob M.O., et al. Verteporfin-induced lysosomal compartment dysregulation potentiates the effect of sorafenib in hepatocellular carcinoma. Cell Death Dis. 2019;10:749. doi: 10.1038/s41419-019-1989-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zhuang J., Li N., Zhang Y., et al. Esterase-activated theranostic prodrug for dual organelles-targeted imaging and synergetic chemo-photodynamic cancer therapy. CCS Chem. 2022;4:1028–1043. [Google Scholar]
- 22.Wang Y., Zeng Y., Yang W., et al. Targeting CD8+ T cells with natural products for tumor therapy: revealing insights into the mechanisms. Phytomedicine. 2024;129 doi: 10.1016/j.phymed.2024.155608. [DOI] [PubMed] [Google Scholar]
- 23.Zhao C., Zhou X., Cao Z., et al. Curcumin and analogues against head and neck cancer: from drug delivery to molecular mechanisms. Phytomedicine. 2023;119 doi: 10.1016/j.phymed.2023.154986. [DOI] [PubMed] [Google Scholar]
- 24.Tang C., Liu Y., Liu S., et al. Curcumin and its analogs as potential epigenetic modulators: prevention of diabetes and its complications. Pharmacology. 2022;107:1–13. doi: 10.1159/000520311. [DOI] [PubMed] [Google Scholar]
- 25.Wang R.J., Zou L., Yi Z.W., et al. PLGA nanoparticles loaded with curcumin produced luminescence for cell bioimaging. Int. J. Pharm. 2023;639 doi: 10.1016/j.ijpharm.2023.122944. [DOI] [PubMed] [Google Scholar]
- 26.Zou L., Wang R., Zhao M., et al. PLGA confers upon conventional nonfluorescent molecules luminescent properties to trigger 1O2-induced pyroptosis and immune response in tumors. J. Nanobiotechnol.Nanobiotechnol. 2025;23:35. doi: 10.1186/s12951-025-03094-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Luo G.H., Xu T.Z., Li X., et al. Cellular organelle-targeted smart AIEgens in tumor detection, imaging and therapeutics. Coord. Chem. Rev. 2022;462 [Google Scholar]
- 28.Liu J., Cabral H., Mi P. Nanocarriers address intracellular barriers for efficient drug delivery, overcoming drug resistance, subcellular targeting and controlled release. Adv. Drug Deliv. Rev. 2024;207 doi: 10.1016/j.addr.2024.115239. [DOI] [PubMed] [Google Scholar]
- 29.Mi P., Cabral H., Kataoka K. Ligand-installed nanocarriers toward precision therapy. Adv. Mater. 2020;32 doi: 10.1002/adma.201902604. [DOI] [PubMed] [Google Scholar]
- 30.Chen X., Hu Z., Zhou L., et al. Self-assembling a natural small molecular inhibitor that shows aggregation-induced emission and potentiates antitumor efficacy. Nanoscale Horiz. 2021;6:33–42. doi: 10.1039/d0nh00469c. [DOI] [PubMed] [Google Scholar]
- 31.Chatterjee S., Das A., Das A., et al. Synthesis, evaluation, and biocompatibility study of magnetite nano particles in normal cells and cancer cells for health care application. Nano-Struct. Nano-Objects. 2024;39 [Google Scholar]
- 32.Liu Y., Yang G.Z., S J., et al. Development of high‐drug‐loading nanoparticles. ChemPlusChem. 2020;85:2143–2157. doi: 10.1002/cplu.202000496. [DOI] [PubMed] [Google Scholar]
- 33.Wu D., Zhou B., Liu Y., et al. Tailoring carrier-free nanoparticles based on natural small molecule assembly for synergistic anti-tumor efficacy. Asian J. Pharm. Sci. 2025;20 doi: 10.1016/j.ajps.2024.100992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Tang Y., Liu T.T., Wang C., et al. Janus microneedles enabling orderly vitiligo treatment through spatiotemporal sequential promoting melanogenesis to transfer via PAR2 activation. Asian J. Pharm. Sci. 2026;21 doi: 10.1016/j.ajps.2026.101184. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Yang T., Liu Z., Fu Z., et al. Albumin nanoassembly bi-directionally manipulated ferroptosis in tumor and CD8+ T cells for triple-negative breast cancer therapy. Asian J. Pharm. Sci. 2025;20 doi: 10.1016/j.ajps.2024.100970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Manimaran S., SambathKumar K., Gayathri R., et al. Medicinal plant using ground state stabilization of natural antioxidant curcumin by keto-enol tautomerisation. Nat. Prod. Bioprospect. 2018;8:369–390. doi: 10.1007/s13659-018-0170-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Shen L., Zhang Q.L., Yao Y., et al. Alkyl chain length-regulated in situ intelligent nano-assemblies with AIE-active photosensitizers for photodynamic cancer therapy. Asian J. Pharm. Sci. 2024;19 doi: 10.1016/j.ajps.2024.100967. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Hamidi H., Ivaska J. Every step of the way: integrins in cancer progression and metastasis. Nat. Rev. Cancer. 2018;18:533–548. doi: 10.1038/s41568-018-0038-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Xu S., Huang H., Tang D., et al. Diallyl disulfide attenuates ionizing radiation-induced migration and invasion by suppressing Nrf2 signaling in non-small-cell lung cancer. Dose Response. 2021;19 doi: 10.1177/15593258211033114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wong Y.C., Ysselstein D., Krainc D. Mitochondria-lysosome contacts regulate mitochondrial fission via RAB7 GTP hydrolysis. Nature. 2018;554:382–386. doi: 10.1038/nature25486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Li Z., Zou J.H., Chen X.Y. Response to precision medicine: current subcellular targeting strategies for cancer therapy. Adv. Mater. 2023;35 doi: 10.1002/adma.202209529. [DOI] [PubMed] [Google Scholar]
- 42.Sun X.Q., Klingbeil O., Lu B., et al. BRD8 maintains glioblastoma by epigenetic reprogramming of the p53 network. Nature. 2023;613:195–202. doi: 10.1038/s41586-022-05551-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Cao Y.F., Wang H., Sun Y., et al. Nuclear ANLN regulates transcription initiation related Pol II clustering and target gene expression. Nat. Commun. 2025;16:1271. doi: 10.1038/s41467-025-56645-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Zhao L.J., Zhou S.T., Gustafsson J.A. Nuclear receptors: recent drug discovery for cancer therapies. Endocr. Rev. 2019;40:1207–1249. doi: 10.1210/er.2018-00222. [DOI] [PubMed] [Google Scholar]
- 45.Shen Y.Y., Wang H.L., Ma Z.Y., et al. Sorafenib promotes treg cell differentiation to compromise its efficacy via VEGFR/AKT/Foxo1 signaling in hepatocellular carcinoma. Cell. Mol. Gastroenterol. Hepatol. 2025;19 doi: 10.1016/j.jcmgh.2024.101454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Xia Y., Xu F., Xiong M., et al. Repurposing of antipsychotic trifluoperazine for treating brain metastasis, lung metastasis and bone metastasis of melanoma by disrupting autophagy flux. Pharmacol. Res. 2021;163 doi: 10.1016/j.phrs.2020.105295. [DOI] [PubMed] [Google Scholar]
- 47.Chen F.L., Xue Y.C., Zhang W.K., et al. The role of mitochondria in tumor metastasis and advances in mitochondria-targeted cancer therapy. Cancer Metastasis Rev. 2024;43:1419–1443. doi: 10.1007/s10555-024-10211-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Boya P., Andreau K., Poncet D., et al. Lysosomal membrane permeabilization induces cell death in a mitochondrion-dependent fashion. J. Exp. Med. 2003;197:1323–1334. doi: 10.1084/jem.20021952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Ping M., Li G., Li Q., et al. The NRF2-CARM1 axis links glucose sensing to transcriptional and epigenetic regulation of the pentose phosphate pathway in gastric cancer. Cell Death Dis. 2024;15:670. doi: 10.1038/s41419-024-07052-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Li X., Wang S., Xie Y., et al. Deacetylation induced nuclear condensation of HP1γ promotes multiple myeloma drug resistance. Nat. Commun. 2023;14:1290. doi: 10.1038/s41467-023-37013-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Parveen R., Harihar D., Chatterji B.P. Recent histone deacetylase inhibitors in cancer therapy. Cancer. 2023;129:3372–3380. doi: 10.1002/cncr.34974. [DOI] [PubMed] [Google Scholar]
- 52.Wu X., Wang J., Liang Q., et al. Recent progress on FAK inhibitors with dual targeting capabilities for cancer treatment. Biomed. Pharmacother. 2022;151 doi: 10.1016/j.biopha.2022.113116. [DOI] [PubMed] [Google Scholar]
- 53.Chen H., Yu S., Ma R., et al. Hypoxia-activated XBP1s recruits HDAC2-EZH2 to engage epigenetic suppression of ΔNp63α expression and promote breast cancer metastasis independent of HIF1α. Cell Death Differ. 2024;31:447–459. doi: 10.1038/s41418-024-01271-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Zheng X.L., Yang J.J., Wang Y.Y., et al. RIP1 promotes proliferation through G2/M checkpoint progression and mediates cisplatin-induced apoptosis and necroptosis in human ovarian cancer cells. Acta Pharmacol. Sin. 2020;41:1223–1233. doi: 10.1038/s41401-019-0340-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Zhao M., Antunes F., Eaton J.W., et al. Lysosomal enzymes promote mitochondrial oxidant production, cytochrome c release and apoptosis. Eur. J. Biochem. 2003;270:3778–3786. doi: 10.1046/j.1432-1033.2003.03765.x. [DOI] [PubMed] [Google Scholar]
- 56.Chang Y.S., Jalgaonkar S.P., Middleton J.D., et al. Stress-inducible gene Atf3 in the noncancer host cells contributes to chemotherapy-exacerbated breast cancer metastasis. Proc. Natl. Acad. Sci. U. S. A. 2017;114:E7159–E7168. doi: 10.1073/pnas.1700455114. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data will be made available on request.