Zinc-coordination-driven mitochondrial nanodisruptors to ablate cancer stemness for potentiating immunotherapy in colorectal malignancies

Abstract

Cancer stemness, sustained by mitochondrial metabolic plasticity, drives immunotherapy resistance in colorectal cancer (CRC). We reengineer zeolitic imidazolate framework-8 (ZIF-8), a zinc-coordination-driven metal-organic framework, from inert nanocarrier to a multifunctional mitochondrial synchronizer that disrupts stemness through tripartite bioenergetic targeting. A hyaluronic acid (HA)-modified nanoarchitecture (HA/ZGA) co-delivers glucose oxidase (GOx) and 5-aminolevulinic acid (5-ALA), leveraging ZIF-8’s intrinsic Zn²⁺ release for dual metabolic blockade: Zn²⁺ inhibits glucose transporters and electron transport chain activity, while GOx depletes glucose to amplify energy stress. Concurrently, 5-ALA enables cristae-confined photodynamic ROS generation, collapsing oxidative phosphorylation and silencing pluripotency regulators. This metabolic triad coordinately eradicates stemness, reverses PD-L1/CD44 co-expression, and triggers cytotoxic T cell infiltration. In translational models, HA/ZGA achieves potent tumor regression, suppresses metastasis, and establishes immunological memory against rechallenge. Our work redefines ZIF-8 as a therapeutic chameleon that dismantles mitochondrial energetics, ablates stemness, and reignites antitumor immunity, exemplifying nanomaterial repurposing to bridge metabolic intervention with immune potentiation.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04184-w.

Introduction

Colorectal cancer (CRC) persists as a leading contributor to global oncologic mortality, with cancer stemness orchestrating therapeutic resistance through metabolic adaptability and immune-editing mechanisms [1–3]. The metabolic duality of CRC stem cells, simultaneously exploiting aerobic glycolysis (Warburg effect) and mitochondrial oxidative phosphorylation, creates a self-reinforcing bioenergetic network that sustains pluripotency and evades immune surveillance [4–6]. Targeting this metabolic axis requires precision strategies capable of synchronously dismantling extracellular nutrient supply and intracellular energy transduction, a therapeutic challenge yet to be adequately addressed [7–8].

Emerging enzymatic approaches using glucose oxidase (GOx) demonstrate potential by depleting intratumoral glucose, thereby inducing ATP crisis and PARP-1-mediated synthetic lethality through oxidative DNA damage [9–11]. However, tumor cells rapidly adapt through GLUT1 upregulation and lactate shuttle activation, necessitating complementary strategies to block compensatory metabolic pathways [12–14]. Parallel advances in photodynamic metabolism modulation employ 5-aminolevulinic acid (5-ALA) to generate mitochondriotropic protoporphyrin IX (PpIX), which upon irradiation produces spatially confined reactive oxygen species (ROS) that selectively disrupt cristae ultrastructure and FoF₁-ATP synthase assembly [15–17]. The integration of these modalities, systemic glucose deprivation and subcellular mitochondrial sabotage, presents a logical yet underexplored approach to eradicate metabolic plasticity in cancer stem cells.

The success of such integrated metabolic therapy hinges on a delivery platform with intrinsic bioactivity to bridge extracellular and intracellular targeting [18–19]. Zeolitic imidazolate framework-8 (ZIF-8), a pH-responsive metal-organic framework classically utilized for drug encapsulation, emerges as an ideal candidate through its inherent zinc ion (Zn²⁺) bioactivity [20–22]. Beyond its cargo-protective function, ZIF-8 dissolution in acidic tumor microenvironments releases Zn²⁺ that actively participate in metabolic reprogramming via three synergistic mechanisms: (1) Structural mimicry of glucose’s hydroxyl configuration enables competitive inhibition of GLUT1 substrate binding, suppressing glucose influx [23]; (2) High-affinity chelation with mitochondrial Complex III’s heme c₁ disrupts cytochrome bc₁ activity, uncoupling electron transport from proton motive force generation [24]; (3) Displacement of zinc ions in β-catenin’s zinc-finger domains impairs its nuclear translocation, silencing Wnt/β-catenin stemness signaling [25]. This functional triad repositions ZIF-8 from a passive nanocarrier to a bioenergetic disruptor with autonomous anti-stemness activity.

To bridge mechanistic insight with therapeutic implementation, a hyaluronic acid-engineered ZIF-8 nanocomposite (HA/ZGA) was developed for co-encapsulation of glucose oxidase (GOx) and 5-aminolevulinic acid (5-ALA) (Scheme 1A). The platform orchestrates sequential metabolic disruption through three harmonized phases: (1) Zn²⁺-mediated GLUT1 blockade establishes a glucose-deprived microenvironment via structural mimicry of substrate-binding domains, priming tumors for enzymatic intervention; (2) GOx catalyzes intratumoral glucose conversion to gluconic acid and hydrogen peroxide, synergizing nutrient deprivation with oxidative stress potentiation; (3) Mitochondrially accumulated protoporphyrin IX from 5-ALA generates cristae-confined reactive oxygen species under photoirradiation, inducing lipid peroxidation of inner membrane components and disassembling electron transport chain supercomplexes (Scheme 1B). This triphasic intervention dismantles cancer stemness through coupled bioenergetic and redox collapse. ATP depletion silences chromatin remodeling complexes essential for pluripotency transcription factor expression, while oxidative modification of stem cell-specific detoxification enzymes cripples stress adaptation mechanisms. Concurrently, the metabolic crisis promotes immunogenic cell death signatures, including surface exposure of damage-associated molecular patterns that facilitate dendritic cell activation and subsequent cytotoxic T lymphocyte recruitment. The coordinated eradication of therapy-resistant stem cell populations and reconfiguration of tumor-immune microenvironments collectively demonstrate the system’s capacity to synergize metabolic targeting with immune potentiation.

Scheme 1.

(A) Schematic synthesis of HA/ZGA and (B) therapeutic action of synergistic immuno-starvation treatment

Materials and methods

Materials

HA (CAS no.9004-61-9, Sigma), GOx (CAS no.9001-37-0, Sigma), Zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O) (CAS no.10196-18-6, Aladdin), 5-Aminolevulinic acid hydrochloride (5-ALA) (CAS no.5451-09-2, Aladdin), 2,4-dimethylimidazole (CAS no.930-62-1, Aladdin), Enhanced BCA Protein Assay Kit (P0010S, Beyotime), EdU-488 cell proliferation detection kit (C0071S, Beyotime), ATP assay kit (S0026, Beyotime). 2-NBDG (CAS no.186689-07-6, MCE). AnnexinV-FITC/PI apoptosis detection kit (40302ES60, Yeasen). Anti-GLUT1 antibody (A6982) was obtained from ABclonal Co., Ltd. anti-SOX2 (ab97959), anti-OCT4 (ab19857), anti-KLF4 (ab215036), anti-Nanog (ab317506), anti-CD3 (ab237721) and anti-CD8α (ab316778) were purchased from Abcam Biotech Co., Ltd.

Preparation of HA/ZGA

HA/ZGA was prepared by sequentially loading GOx and 5-ALA into the ZIF-8 and then funtionalized them with HA (Scheme 1A). Firstly, ZIF-8 was synthesized by the coordination of zinc nitrate and 2-methylimidazole. GOx (1 mg) was dissolved in double-distilled water (ddH₂O, 4.0 mL) as solution A, 4 mg of 5-ALA was dissolved in 1.0 mL of ddH₂O as solution B. The ratio of ZIF-8 to GOx and 5-ALA was 5:1:4. Thus, 5 mg of preheated ZIF-8 (at 60 ℃ drying) was slowly added to solution A with stirring (500 rpm) for 2 h, following the solution B was slowly added into the above mixture, stirred for another 24 h to form ZGA NPs. Finally, the harvest ZGA NPs was dispersed in HA (3 mg/mL) and stirred for 2 h, washing and centrifuged at 12,000 rpm for 10 min, the resulting solid was HA/ZGA NPs.

Coomassie blue staining

Coomassie blue staining was a widely used protein staining technique that visualized proteins by combining bright blue G-250 with proteins to stain them blue in PAGE gels. The synthesized NPs were fully dissolved in ddH₂O (with a pH value of 5.0), the solution containing GOx was mixed with 5× loading buffer and heated at 95℃ for 5 min. The 20 µL samples was taken for electrophoresis with an 10% SDS-PAGE gel. This gel was stained by Coomassie brilliant blue G250 for 4 h and shaker rinsed with ddH₂O until protein strips appeared.

Drug release behavior of HA/ZGA

The release amount of GOx and 5-ALA from HA/ZGA was calculated by a dialysis method conducted at 37℃ under 300 rpm gentle shaking. Briefly, 5 mg of HA/ZGA dissolved in 2 mL of PBS with pH value at 5.0 or 7.4, then take 2 mL HA/ZGA solution into a dialysis bag and immerse it into a centrifuge tube (15 mL) pre-loaded with 5 mL PBS (pH 5.0 or 7.4). Collect 1 mL PBS solution from the centrifuge tube at predetermined time to calculated the amount of GOx or 5-ALA released from HA/ZGA in the dialysis bag, add 1 mL of fresh PBS back into the centrifuge tube to maintain 5 mL volume.

Loading efficiency (LE) of GOx and 5-ALA

The loading efficiency was calculated using the following equations:

Where C_GOx and C_5−ALA were the amount of GOx and 5-ALA in HA/ZGA NPs, respectively. V was the volume of HA/ZGA solution, and M_HA was the weight of HA. The amount of 5-ALA or GOx in HA/ZGA NPs was calculated by the BCA Protein Assay Kit (for GOx) or UV-vis (for 5-ALA) standard curves, respectively.

Evaluation of GOx activity in HA/ZGA

Considering that GOx can efficiently catalyze the oxidation of glucose into H₂O₂ and glucuronic acid, a process that consumes glucose and elevates acidity. Therefore, we examined glucose consumption and pH changes to reflecting the activity of glucose oxidase. Simply, 1 ml of HA/ZGA (1 mg/mL) was mixed with 2 ml of DMEM medium (glucose 4500 mg/L), 24 h later, the absorption peak of the mixture was measured by a UV-vis spectrophotometer (UV2450, Shimadzu). The changes of pH values was detected by pH Meter (PHS-3E, INESA Scientific Instrument Co.,Ltd, Shanghai). GOx (500 µg/ml ) was the positive control.

Cell culture and treatment methods

Mouse CRC cell line CT26 and human CRC cell line RKO were incubated in DMEM supplemented with 10% FBS (Biological Industries) and 1% streptomycin-penicillin (HyClone). Cells were treated with different treatments of normal saline (NC), 5-ALA, ZGA and HA/ZGA, followed by exposed to laser irradiation for 3 min (660 nm laser, 2.0 W/cm^− 2), except NC group. Then, the follow-up experiments were conducted.

Cellular internalization and cytotoxicity of HA/ZGA

Time course of cellular internalization of HA/ZGA was evaluated by fluorescence microscope and flow cytometer, respectively. Cellular uptake of free 5-ALA, ZGA and HA/ZGA into CT26 and RKO cells were determined by fluorescence microscope and flow cytometry.

The MTT assay was used to measure cell toxicity in CT26 or RKO cells, cells were seeded into 96-well plates, following incubated with treatments of normal saline (NC), 5-ALA, ZGA and HA/ZGA for 24 h. Discarding the medicated medium and rinsing them with normal saline, the cells (except NC group) were exposed to 660 nm laser (2.0 W/cm^− 2) irradiation for 3 min and incubated normally for 1d. Adding 0.5 mg/mL MTT solution (20 µL) into each well for incubation 4 h, then carefully removing the MTT solution, 150 µL of DMSO was added and gently shake for 15 min to ensure complete dissolution, and the absorbance at 570 nm was recorded for analysis cell viability.

The Live/Dead Staining Kit (C2015M, Beyotime) was also used to measure cell toxicity of HA/ZGA, cells were treated with different treatments followed by exposed to 660 nm laser irradiation for 3 min (2.0 W/cm^− 2), except NC group. Prepare the staining solution according to the kit instructions, each well was incubated with 100 µL staining solution at 37 °C for 30 min and imaged under fluorescence microscope.

The anti-proliferative efficacy of HA/ZGA

Colony formation and EdU incorporation assays were used to study the effect of anti-proliferation of HA/ZGA. For the colony formation assay, an equal volume of 0.1 mL cell suspension (100–500 cells) was seeded into 12-well plate, cell drug and laser irradiation treatment methods ibid., then cultured for about 5–7 days until cell clones were formed. The cellular colonies were fixed with 4% paraformaldehyde, and stained with crystal violet, taken pictures and analyzed for cloning efficiency.

For the EdU incorporation assay, cells were seeded into 96-well plates at a density of 5 × 10⁴ cells per well and incubated overnight. Cell drug and laser irradiation treatment methods ibid. Then, cells were stained with EdU dye following the the manufacturer’s recommendations (C0071S, Beyotime).

Apoptosis detection by flow cytometry

CT26 and RKO cells were plated into 6-well plates to reach about 70% confluence, cell drug and laser irradiation treatment methods ibid. The treated cells were collected and stained with Annexin V-FITC and PI for 20 min according to the kit’s instructions (40302ES60, Yeasen). Note that both suspension cells and adherent cells are collected. The cell apoptosis were analyzed by flow cytometry.

Detection of intracellular ROS

Intracellular ROS was detected by ROS kit (S0033, Beyotime). Cells were plated into 6-well or 24-well plates, cell drug and laser irradiation treatment methods ibid. Cells were incubated with fluorescent probe DCFH-DA for 20 min following the kit’s instructions, then intracellular ROS were evaluated by fluorescence microscope and flow cytometry.

JC-1 assay

Loss of mitochondrial membrane potential (MMP) serves as a hallmark event of early apoptotic stage. JC-1 kit (C2006, Beyotime) was used to assess the change of MMP. CT26 or RKO cells were plated into 6-well or 24-well plates, cell drug and laser irradiation treatment methods ibid. Cells were stained with JC-1 solution following the JC-1 kit’s instructions. Then, the change of MMP was recorded by fluorescence microscopy or flow cytometry.

Glucose uptake assay

2-NBDG, a fluorescently labeled deoxyglucose analog, was used to measure glucose uptake or transport of CT26 or RKO cells. The cells was incubated with a 2-NBDG solution (200 µM of 2-NBDG) for 30 min at 37℃. After washed with cold PBS and the cells were directly observed by a fluorescence microscope, or collected cells for flow cytometry analysis at the FITC channel.

Western blotting

Total proteins from cells or tissues were collected and quantified by a commercial BCA protein assay kit. The protein samples were subjected to SDS-PAGE gels, then the separated proteins were transferred onto PVDF membranes. After blocked with 5% (w/v) non-fat milk, the PVDF membranes containing proteins were incubated with indicated primary antibodies overnight at 4℃ with gently shaking, the secondary antibodies with HRP labels were applied for 2 h at room temperature, the protein bands were visualized by using an enhanced chemiluminescence kit.

In vitro co-culture experiment

CT26 cells were seeded into 6-well plates, cell drug and laser irradiation treatment methods ibid. Removed the medicated medium and added fresh medium culture for another 24 h, the culture supernatant was collected. Spleen immune cells of CT26 mouse were obtained and co-cultured with the culture supernatant for 48 h, then the co-cultured immune cells were collected and stained with fluorochrome-conjugated antibodies for immune cells surface tags (FITC anti-mouse CD3 antibody and PE anti-mouse CD4 antibody are used to label CD4⁺ T cell; APC anti-mouse CD8α antibody and FITC anti-mouse CD3 antibody are used to label CD8⁺ T cell), and then analyzed by flow cytometry.

Sphere formation assay

After the cells are treated with drug and laser irradiation, removed the medicated medium and added fresh medium culture for another 24 h. The sphere formation assay was performed as follows: The cells suspension were diluted in serum-free medium and seeded into non-coated 6-well plates. The lower layer of the 6-well plate was coated with a mixture 0.7% agarose and serum-free medium, the ratio was 1:1. The cell suspension was mixed 1:1 with 0.7% agarose and spread in the middle layer. After solidifying, added the serum-free medium to the upper layer. Culture for about 7 days, the formation of cell spheres was observed under an optical microscope.

In vivo imaging and bio-distribution of HA/ZGA

Tumors were obtained by subcutaneously injecting 0.1 mL CT26 cells (1.0 × 10⁶ cells) into BALB/c mice. Mice received injections of the same volume of saline, 5-ALA (2 mg/kg), ZGA (4 mg/kg) and HA/ZGA (4 mg/kg) via the tail vein when tumors reached 80 mm³. CT26 tumor-bearing mice were real-time imaged at different time using a live-imaging system and analyzed the bio-distribution of HA/ZGA NPs.

In vivo tumor suppression experiments

Mice were subcutaneously injecting 0.1 mL CT26 cells (1.0 × 10⁶ cells) to construct subcutaneous tumor model. When tumor vulume reached about 80–100 mm³, mice were randomly divided into four groups (n = 5) and intravenously injected with equal amounts of saline, 5-ALA (2 mg/kg), ZGA (4 mg/kg) and HA/ZGA (4 mg/kg). The tumor regions were exposed to 660 nm laser irradiation (2.0 W cm^− 2, 5 min) after tail vein injection for 6 hours. Body weight and tumor volume were detected during treatment period. Tumors and organs (such as heart, liver, spleen, lung, kidney) were harvested for subsequent experiments.

Recurrent tumor models

For recurrent tumor model, the method of tumor inoculation was the same as above. When tumor growth reached about 250 mm³, tumors were excised surgically, recovery 4 days after surgery, mice were intravenously injected with equal amounts of saline, 5-ALA (2 mg/kg), ZGA (4 mg/kg) and HA/ZGA NPs (4 mg/kg) on an alternate day basis for 14 days. Tumor volumes and body weights were monitored every day. Mice were euthanized and spleens and tumors were harvested for subsequent experiments. Single-cell suspensions isolated from recurrence tumor or spleen were stained with fluorochrome-conjugated antibodies to analyze the proportion of helper T cells referred as CD4⁺ T cell (Th, CD3⁺, CD4⁺), cytotoxic T cells referred as CD8⁺ T cell (CTLs, CD3⁺, CD8a⁺), Treg cells (CD4⁺, CD25⁺, FOXP3⁺) and DCs (CD80⁺, CD86⁺, CD11c⁺) were detected by flow cytometry.

Statistical analysis

All data are presented as mean ± SD obtained from at least three independent assays unless otherwise noted. We used GraphPad Prism software (version 9.0) for statistical analysis. Differences among groups were conducted through ANOVA, and P < 0.05 was considered statistically significant.

Result and discussion

Preparation and characterization of HA/ZGA

In this work, as shown in Scheme 1A, the porous feature of ZIF-8 was exploited to load GOx & 5-ALA and then modified hyaluronic acid (HA) on its surface by virtue of electrostatic interaction, which enabling the generation of the tumor-targeting nanoparticles. The transmission electron microscopy (TEM) images of pure ZIF-8 with a well-defined rhombic structures and nanometer scale, while the morphology of ZGA or HA/ZGA NPs was irregular and has fuzzy surface (Fig. 1A-C). Meanwhile, these nanoparticles had obvious Tyndall effect, ZIF-8 as a nanocarrier has an average diameter of 114.7 nm, while the obtained HA/ZGA NPs with an average size about 142.6 nm, the larger size was apparently due to the loading of GOx and 5-ALA by a physical adsorption mechanism (Fig. 1D-F). The size of HA/ZGA NPs could be relatively stable at about 152.8 nm over 7 days in different solutions (Figure S1B, Supporting Information). In addition, as shown in Fig. 1G, the zeta potentials of the obtained ZIF-8, ZGA and HA/ZGA NPs were about at 5.3, -6.6 and − 12.6 mV, respectively. The changes of zeta potential also suggested the successful loading of GOx & 5-ALA and modification of HA.

Fig. 1.

Synthesis and characterization of HA/ZGA. A-C) TEM images of ZIF-8 (A), ZGA (B), and HA/ZGA (C). D-F) Size distribution and Tyndall effect of ZIF-8 (D), ZGA (E), and HA/ZGA (F). G) Zeta potential of ZIF-8, ZAG and HA/ZGA. H) UV-visible absorption spectra of ZIF-8, 5-ALA, GOx, ZAG and HA/ZGA. I) FTIR spectra of ZIF-8, 5-ALA, GOx and ZGA. J) The photographs of BCA solution treated with HA/ZGA NPs for 12 h. K) The residual levels of glucose in DMEM-high glucose (4500 mg/mL) when mixed with GOx (500 µg/mL) or HA/ZGA (1 mg/mL). L) The changes of pH value when DMEM-high glucose (4500 mg/mL) mixed with GOx (500 µg/mL) or HA/ZGA (1 mg/mL)

As the UV-visible spectrophotometry shown in Fig. 1H, it was evident that the absorbance peaks of GOx and 5-ALA were close to each other, which was confirmed by the spectral curves of GOx and 5-ALA. (Figure S1C-D, Supporting Information). The FTIR spectral characteristics of ZIF-8, 5-ALA, GOx and ZGA NPs demonstrated the successful synthesis of nanoparticles (Fig. 1I). Moreover, ZGA NPs retained the similar FTIR pattern as that of pure ZIF-8, revealed that the loading of GOx or 5-ALA had little impact on crystalline structures of carrier ZIF-8. The coomassie blue staining showed that GOx was roughly distributed 70 kD, which also confirmed the successful loading of GOx into the ZIF-8 (Figure S1A, Supporting Information). As shown in Fig. 1J, we further confirmed the existence of GOx in HA/ZGA NPs by adding different concentrations of HA/ZGA into the BCA working solution. GOx is a protein that undergoes a color reaction with the BCA working solution (P0010S, Beyotime), the higher the protein content, the darker the color. Additionally, we examined the enzyme activity of GOx in HA/ZGA NPs by recording the changes of the glucose content or the pH value, as GOx catalyze the oxidation of glucose into glucuronic acid and H2O2 resulting in lower glucose content and the pH value. As shown in Fig. 1K-L, the glucose content and pH value of the DMEM medium were decreased in HA/ZGA, which indicated that the catalytic activity of GOx in the HA/ZGA was not affected.

HA/ZGA NPs release behavior experiment was used to study the pH-responsive delivery performance. As shown in Figure S1F, Supporting Information, only about 40% GOx released from HA/ZGA NPs at pH 7.4, while more than 90% GOx released at pH 5.0. For 5-ALA, the release of 5-ALA was 51.5% and 80.7% at pH 7.4 and pH 5.0, respectively (Figure S1G, Supporting Information). These results demonstrated the pH-sensitive property of HA/ZGA NPs, which will facilitate their rapidly release in acidic tumor micro-environments. The pH-responsive release property stemmed from the collapse of ZIF-8 under acidic condition according to the previous report [24]. Finally, the loading efficiencies (LE) of GOx or 5-ALA in HA/ZGA NPs were calculated to be 14.6% using a BCA protein quantification kit (Figure S1E, Supporting Information) and 19.1% using the UV-vis spectroscopy (Figure S1C, Supporting Information), respectively. Therefore, the above data confirmed that the nanocarrier ZIF-8 has satisfactory loading rate and pH-responsive release behavior while ensuring the catalytic activity of GOx, which providing powerful conditions for effective tumor therapy.

The cellular internalization and in vitro anti-CRC effects of HA/ZGA

The time course of HA/ZGA NPs cellular uptake showed that maximum intracellular accumulation was achieved after 6 h incubation for both CT26 (Fig. 2A) and RKO cells (Figure S2A, Supporting Information). Additionally, HA/ZGA NPs exhibited stronger fluorescence intensity in CT26 (Fig. 2B) and RKO cells (Figure S2B, Supporting Information) than free 5-ALA and ZGA NPs, demonstrating that HA modification enhanced HA/ZGA NPs’ cellular uptake.

Fig. 2.

In vitro cell uptake and cytotoxicity of HA/ZGA. (A) Time course of HA/ZGA NPs cellular uptake in CT26 cells, Scale bar: 100 μm. (B) Cellular uptake of 5-ALA, ZGA and HA/ZGA in CT26 cells. Scale bar: 100 μm. (C) Cell viability of CT26 and RKO cells incubated 24 h with 5-ALA, ZGA and HA/ZGA NPs. (D) Images of live (green) and dead cells (red) stained with Calcein-AM/PI dye. Scale bar: 100 μm. (E) Colony formation images and analysis for CT26 and RKO cells. (F) EdU labeling images and analysis for CT26 cells across different groups. Scale bar: 100 μm. (G) EdU labeling images and analysis for RKO cells across different groups. Scale bar: 100 μm. (H) Apoptosis images and analysis for CT26 and RKO cells detected by flow cytometry (n = 3, ^***P < 0.001)

The results of MTT assay as shown in Fig. 2C, the HA/ZGA NPs exhibited a higher inhibitory effect on cell viability than the free 5-ALA and ZGA, this may be attributed to improved PDT efficiency and tumor targeting of HA/ZGA NPs. 5-ALA triggered PDT depends on its metabolite PpIX (endogenous photosensitizer) generates a large amount of reactive oxygen species (ROS) under laser irradiation. In our experiment, free 5-ALA, even under laser irradiation, exerts minimal cytotoxicity maybe due to its rapid metabolism and the low bio-availability.

In addition, as showed in Fig. 2D, the Live/Dead staining assays also supported that HA/ZGA nanoplatform significantly amplified the phototherapy efficiency of free 5-ALA. Compared to the 5-ALA or ZGA group, more red fluorescent signal (indicate dead cell) and fewer green fluorescence signal (indicate live cell) were observed in the HA/ZGA group, suggesting a stronger anti-CRC effect. The results of colony formation assay (Fig. 2E) and EdU proliferation assay (Fig. 2F, G) demonstrated that HA/ZGA NPs exhibited a better anti-proliferative ability than the other groups. Moreover, as shown in Fig. 2H, HA/ZGA NPs promoted cell apoptosis in CT26 and RKO cells models. Therefore, the above results indicate that HA/ZGA NPs has a good tumor suppression effect in vitro by inhibiting cell proliferation and promoting cell apoptosis.

The mechanism of HA/ZGA-mediated stemness suppression

Glucose is the most abundant nutrient in circulation blood and the most frequent metabolic substrate utilized by malignant cells with high metabolism. The growth, metastasis and recurrence of tumor cells are heavily dependent on a continuing energy supply [7]. Therefore, starvation therapy targeting glucose metabolism specifically, which results in energy metabolic disorders, restricted synthesis, and inhibition of tumor growth, has been developed for cancer therapy.

Here, HA/ZGA NPs was proposed for inhibiting tumor growth and recurrence by targeting energy metabolism of tumor cells, such as cutting off energy sources or reducing ATP production. Given that mitochondria is the main sites for synthesis of ATP in cells, while the loaded 5-ALA as a photosentitizer precursor is converted to protoporphyrin Ⅸ (PpⅨ), the latter produces large amount of reactive oxygen species (ROS) in mitochondria under near-infrared irradiation. Therefore, we firstly detected intracellular ROS production. Compared to the other groups, HA/ZGA NPs showed brighter and more intense green fluorescence, indicating more ROS production in CT26 cells (Fig. 3A, B) and RKO cells (Figure S3A, B, Supporting Information). Since mitochondria are highly sensitive to ROS accumulation produced by 5-ALA mediated photodynamic therapy (PDT), oxidative stress can induce mitochondrial dysfunction, and thus further reduce ATP production [19]. Mitochondrial membrane potential (MMP) is not only an indicator of mitochondrial function, but also an indicator of early apoptosis, thus JC-1 assay was used to study the change of MMP. Compared to the other groups, HA/ZGA NPs showed more JC-1 monomers (Green) and fewer JC-1 aggregates (Red) in CT26 cells (Fig. 3C) and RKO cells (Figure S3C, Supporting information), indicating MMP decline. Flow cytometry results further confirmed that more JC-1 aggregates (Red) were converted to monomers (Green) in CT26 cells (Fig. 3D) and RKO cells (Figure S3D, Supporting information), indicating the loss of MMP and mitochondrial dysfunction. Moreover, HA/ZGA NPs significantly reduced ATP level in CT26 cells (Fig. 3E) and RKO cells (Figure S3E, Supporting information), suggesting mitochondrial dysfunction leads to less ATP production.

Fig. 3.

Anti-tumor mechanism of HA/ZGA NPs in vitro. (A) Intracellular ROS of CT26 cells detected by fluorescence imaging. Scale bar:100 μm. (B) Intracellular ROS of CT26 cells detected by flow cytometry. (C) The change of mitochondrial membrane potential (MMP) in CT26 cells was detected by JC-1 assay, Scale bar:50 μm. (D) The change of MMP in CT26 cells was detected by flow cytometry. (E) Fluorescence imaging for intracellular 2-NBDG uptake of CT26 cells across different groups. Scale bar: 100 μm. (F) Flow cytometry analysis for intracellular 2-NBDG uptake of CT26 cells across different groups. (G) Representative immunofluorescence images for GLUT1 expression in CT26 cells. Scale bar: 50 μm. (H) Representative western blot images for GLUT1 expression in CT26 cells. (I) ATP content in CT26 cells. (J) Sphere-forming abilities of CT26 cells across different groups. Scale bar: 50 μm. K) Western blot showed the effects of different treatments on the expression levels of SOX2, OCT4, Nanog and KLF4 in CT26 cells. L-O) The percentage of immune cells in CT26 cells was analyzed by flow cytometry, including L, N) CD4 + T cells, M, O) CD8 + T cells. (^***P < 0.001, one-way ANOVA)

To further investigate the effect of HA/ZGA NPs on energy metabolism of CRC cells, 2-[N-(7-nitrobenz-2-oxa-1,3-diaxol-4-yl)amino]-2-deoxyglucose, generally referred as 2-NBDG, was employed to monitored the level of glucose uptake by cells [24]. As shown in Fig. 3E and Figure S3E, a weak green fluorescence signal was observed in the HA/ZGA NPs group, suggesting reduced glucose uptake. Flow cytometry results also confirmed a reduction in glucose uptake, as evidenced by the leftward shift of the curve (Fig. 3F and Figure S3F, Supporting information). Given that glucose transporter 1 (GLUT1) is a key rate-limiting protein in hyperactive glucose metabolism and up-regulated in colorectal cancer [26], we determined the protein levels of GLUT1 in CT26 and RKO cells by immunofluorescence and western blot. The results of both immunofluorescence (Fig. 3G and Figure S3G, Supporting information) and western blot (Fig. 3H and Figure S3H, Supporting information) displayed that HA/ZGA NPs reduced the expression of GLUT1, which showed the same trend with the glucose uptake, supporting that Zn²⁺ could effectively inhibit the expression of GLUT1 to decrease energy intake according to the previous report [23]. Obviously, the above results suggested that the PpIX derived from 5-ALA metabolism in mitochondria induced ROS accumulation causing mitochondrial damage and down-regulation of GLUT1 expression by Zn²⁺ released from ZIF-8, interfered with ATP production and energy intake, and thus synergistically inhibited tumor growth and recurrence.

Cancer stem cells (CSCs) represent a small population of tumor cells that are thought to be responsible for therapy resistance, recurrence, and metastasis through their capacity to self-renew and differentiate into heterogeneous downstream lineages of cancer cells [27]. Here, we performed sphere-formation assay to explore whether HA/ZGA NPs involved in regulating tumor stemness. Our data showed that the spherical formation ability and number of CT26 cells and RKO cells in the HA/ZGA group notably decreased (Fig. 3J and Figure S3J, Supporting information), compared with the other groups. Coherently, according to the outcomes of western blot revealed that the expression level of cancer stem cell-related proteins [28] such as SOX2, OCT4, Nanog and KLF4 in CT26 and RKO cells were markedly reduced when the CRC cells were treated with HA/ZGA NPs (Fig. 3K and Figure S3K, Supporting information), suggesting HA/ZGA mediated decrease in energy uptake attenuated their stemness properties including markedly reduced sphere forming ability and expression of stemness markers. In addition, an increasing number of studies have shown that ATP is not only an important source of energy for tumor metastasis and recurrence, but also released ATP by stressed and dying cells can influence diverse immune cell activities to trigger an anticancer immune response [29–30]. We thus designed an in vitro co-culture assay to evaluate the effects of HA/ZGA NPs on immunological cells extracted from BALB/C mouse spleens. As shown in Fig. 3L-O, flow cytometry results revealed that HA/ZGA NPs could obviously increase the proportion of CD4⁺ and CD8a⁺ T cells. Helper T lymphocytes (Th), CD4⁺ T cells, are important components of the immune system and can directly kill tumor cells [31]. Cytotoxic T lymphocytes (CTL), or CD8a⁺ T cells has been proposed as an indicator for the assessment of CD8⁺ T cells recruitment or activity, which is closely related to tumor immune infiltration, immunosuppressive environment formation, cancer progression [32–33]. The results strongly supported the potential of HA/ZGA NPs to trigger antitumor immunity.

In vivo tumor suppression effect of HA/ZGA

CT26 tumor-bearing mouse model was used to evaluated the in vivo therapeutic effect of HA/ZGA NPs. Firstly, the bio-distribution and tumor aggregation of HA/ZGA NPs in CT26 tumor bearing mice were detected by a living imaging system. As shown in Fig. 4A, owing to the presence of hyaluronic acid (HA), HA/ZGA NPs exhibited excellent tumor targeting ability than free 5-ALA and ZGA NPs. Moreover, as shown in Fig. 4B, compared with other organs, brighter and more intense fluorescence signals was observed in dissected tumor of the HA/ZGA group, which further confirmed that HA/ZGA NPs exhibited superior tumor-targeting ability.While HA modification is primarily intended for targeted delivery to CD44-overexpressing tumor sites, it is important to note that baseline expression of CD44 receptors in tissues such as the liver can also contribute to biodistribution and presents a potential pathway for off-target accumulation. Hemolytic test results demonstrated that HA/ZGA NPs within the tested concentrations are safe and feasible for intravenous delivery.

Fig. 4.

In vivo bio-distribution and anti-tumor efficacy evaluation of HA/ZGA. (A) Real-time in vivo imaging of CT26 tumor-bearing mice at different time points following intravenous injection of 5-ALA, ZGA and HA/ZGA NPs. (B) Fluorescence imaging of major organs and tumors. (C) Hemolysis rate and images of HA/ZGA NPs at different concentrations. (D) Schematic diagram of CT26 tumor-bearing mouse model. (E) Tumors images from different groups (n = 5). (F) Tumor volume curves from different groups (n = 5, one-way ANOVA). (G) Body weight curves of different groups (n = 5). (H) Tumor inhibition ratios and tumor weights (n = 5, one-way ANOVA). (I) Representative immunohistochemistry images for SOX2, OCT4, Nanog and KLF4 expressions. Scale bar:50 μm. (J) Western blot analysis for SOX2, OCT4, Nanog and KLF4 expressions

Next, the subcutaneous tumor model was used to evaluate the anti-tumor efficacy of HA/ZGA NPs, as shown in Fig. 4D. HA/ZGA exhibited excellent anti-tumor efficacy, the mass and volume of excised tumors from different groups were consistent with the tumor’s inhibition behavior (Fig. 4E-H). According to the results of immunohistochemistry (IHC) and western blot revealed that the expression level of SOX2, OCT4, Nanog and KLF4 in tumor tissues were markedly reduced when CT26 tumor bearing mice were treated with HA/ZGA NPs (Fig. 4I and J), further suggesting HA/ZGA NPs attenuated tumor stemness property by reduced expressions of stemness markers. Weight changes of CT26 tumor-bearing mice of all treatment groups showed an upward trend and only fluctuated slightly during the treatment period (Fig. 4G), Hematoxylin and Eosin (H&E) staining of major organs showed no obvious abnormalities (Figure S4A, Supporting information), the blood biochemical index of tumor-bearing mice that used to assess liver and renal toxicity, such as aspartate aminotransferase (AST), alanine amino transferase (ALT), blood urea nitrogen (UREA) and creatinine (CRE), also exhibited negligible changes in all treatment groups, indicating the negligible systemic toxicity of HA/ZGA (Figure S4B-E, Supporting information). The results demonstrated the antitumor activities and bio-safety of our HA/ZGA NPs in CRC therapy.

HA/ZGA inhibit tumor recurrence by anti-tumor immune response

Recurrence of colorectal cancer (CRC) is the major reason for poor clinical outcomes and cancer deaths [3]. Recurrence model of CT26 tumor-bearing mice was used to study the effect of HA/ZGA NPs on preventing post-surgical recurrence, as presented in Fig. 5A. Compared to the other groups, HA/ZGA NPs significantly prevented tumor recurrence, and the size of recurrent tumors was also significantly reduced (Fig. 5B, C). The tumor inhibition rate of HA/ZGA group was the highest (Fig. 5D). Moreover, according to the results of immunohistochemistry (IHC) and western blot revealed that the expression level of GLUT1 in recurrence tumors were markedly reduced when CT26 tumor bearing mice were treated with HA/ZGA NPs (Fig. 4E and F), and the expression levels of SOX2, OCT4, Nanog and KLF4 in recurrence tumors were markedly reduced in HA/ZGA groups (Fig. 4G), suggesting HA/ZGA NPs attenuated stemness properties of the recurrence tumors.

Fig. 5.

The efficacy evaluation of HA/ZGA NPs on tumor recurrence in vivo. (A) Schematic diagram of recurrent tumor mouse model. (B) In vivo imaging of mice from different groups at surgery day (Day 0) and day 18th after surgery (Day 18). (C) Recurrence tumors images from different groups (n = 5). (D) Tumor inhibition ratios and recurrence tumor weight (n = 5, one-way ANOVA). (E) Representative IHC images for GLUT1 expression in recurrence tumors. Scale bar: 50 μm. (F) Western blot showed the effects of different treatments on the expression level of GLUT1 in recurrence tumors. (G) Western blot showed the effects of different treatments on the expression levels of SOX2, OCT4, Nanog and KLF4 in recurrence tumors. H-O) The percentage of major immune cells in mouse spleen was analyzed by flow cytometry, including H, L) CD4⁺ T cells; I, M) CD8⁺ T cells; J, N) Tregs; K, O) DCs

Dying or necrosis tumor cells had an important role in triggering anti-tumor immunity through immunogenic cell death (ICD) in vivo. We hypothesize that this mechanisms also exist in this study. A large amounts of tumor antigens or the damage-associated molecular patterns (DAMPs) be released from the dying tumor cells induced by starvation or photodynamic therapy (PDT), which can promote the transition of immature dendritic cells (DCs) into mature DCs [28]. Mature DCs, as the antigen presenting cells (APCs), activating CD8 + T cells referred as cytotoxic T lymphocytes (CTLs) by presenting antigens, then activated CTLs can clear tumor cells. Helper T lymphocytes (Th) referred as CD4 + T cells, works with CD8 + T cells to directly kill tumor cells and virus-infected cells. While regulatory T cells (Tregs), being recognised as prototypical immunosuppressive cells, which not only inhibit the maturation of DCs but also suppress the proliferation and differentiation of CD8 + T cells within the tumor microenvironment [32].

Therefore, flow cytometry was employed to analyze the fluctuations of various immune cells in recurrent tumors and spleens following different treatments, thereby elucidating the potential mechanisms underlying the anti-tumor immunity conferred by HA/ZGA NPs. The results of flow cytometry revealed that DCs (Fig. 5K, O), CD4+ (Fig. 5H, L) and CD8 + T cells (Fig. 5I, M) were obviously increased while Tregs (Fig. 5J, N) was decreased in the spleen of HA/ZGA NPs treated mice. Moreover, the proportion of these immune cells in the tumors of HA/ZGA NPs treated mice showed the same trend (Figure S5C-J, Supporting information). As shown in Figure S5A, IHC detection of CD3 and CD8α showed that there was infiltration of CD8 + T cells in recurrence tumors of HA/ZGA NPs treated mice. Western blot results also confirmed that a significant elevation of CD3 and CD8α proteins in recurrence tumors of HA/ZGA NPs treated mice (Figure S5B). In a word, the aforementioned results indicated that HA/ZGA NPs exerts immunotherapeutic effects by regulates systemic immunological responses and inhibits the development of recurrence tumors.

Conclusion

We engineered a multifunctional nanoplatform (HA/ZGA) through rational repurposing of ZIF-8, a conventional metal-organic framework, into a stimuli-responsive theranostic carrier co-loading glucose oxidase (GOx) and 5-aminolevulinic acid (5-ALA). Surface functionalization with hyaluronic acid bestowed tumor-targeting precision and enhanced systemic stability, enabling efficient CD44-mediated tumor accumulation post-administration. The therapeutic paradigm operates through coordinated bioenergetic sabotage: (1) GOx-mediated glucose deprivation synergized with Zn²⁺-induced GLUT1 blockade to establish extracellular nutrient siege; (2) 5-ALA-derived protoporphyrin IX enabled cristae-targeted photodynamic destruction of mitochondrial ultrastructure, collapsing oxidative phosphorylation; (3) Concerted ROS overproduction and ATP depletion dismantled stemness-maintaining transcriptional networks, erasing chemoresistant cell populations. In translational models, HA/ZGA demonstrated profound tumor growth inhibition and complete suppression of metastatic recurrence, while maintaining an exemplary safety profile. Therefore, our findings redefine ZIF-8 as a therapeutic chameleon capable of carrier-cargo cooperativity, establishing a new blueprint for nanomaterials repurposing in next-generation cancer immunotherapy. By synchronously dismantling cancer stemness infrastructure and reversing immunosuppressive niches, HA/ZGA exemplifies how rational nanoarchitectonics can bridge metabolic intervention with immune potentiation—a paradigm shift in CRC management.

Author contributions

**Y. Long: ** performed data curation, writing, and original draft preparation. **WT. Cheng and HL. Tian: ** methodology, data curation, writing/review and editing, software, data curation, investigation. **CH. Huang, ZQ. Zhang and Yongfeng Jia: ** formal analysis, supervision, funding acquisition, writing/review and editing. **SJ. Zheng: ** resources, project administration, funding acquisition. All authors read and approved the final manuscript. ( **Y. Long** and **WT. Cheng** contributed equally to this work).

Funding

Declaration.

This work was supported by grants from the National Key R&D Program of China (2023YFC3402100), National Natural Science Foundation of China (82560493, 823B2081, 82130082), Sichuan Provincial Department of Science and Technology (2025ZNSFSC0001), the Hainan Province Science and Technology Special Fund (ZDYF2026SHFZ039); Academic Enhance-ment Support Program of Hainan Medical University (XSTS2025133; XSTS2025139); the specific research fund of The Innovation Platform for Academicians of Hainan Province (grant no. YSPTZX202208); Sichuan Science and Technology Program (2025ZNSFSC1042); the Postdoctoral Fellowship Program and China Postdoctoral Science Foundation under Grant BX20250245; a multidisciplinary interdisciplinary innovation team for the protection and utilization of authentic medicinal materials in the Three Gorges Reservoir Area (CXTD202303).

Data availability

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Animal protocols were approved by the Administration Committee of Experimental Animals in Sichuan Province and the Ethics Committee of Sichuan University.

Consent for publication

All authors of this study agreed to publish.

Competing interests

The authors declare no competing interests.