A CD54-targeted magneto-responsive nanotheranostic for precision treatment of viral pneumonia via CTSB-mediated PANoptosis inhibition

Abstract

Broad-spectrum therapeutics that simultaneously halt viral replication and mitigate cytokine storms are urgently needed for respiratory viral pandemics. In this study, a magneto-active nanotheranostic platform (IO@CA/P) was developed by conjugating the Cathepsin B (CTSB) inhibitor CA-074 and a CD54-targeting cyclic peptide to superparamagnetic iron oxide nanoparticles. IO@CA/P exhibited broad-spectrum antiviral activity against SARS-CoV-2 (Omicron, Delta), influenza (H1N1, H3N2), and respiratory syncytial virus by blocking viral entry. CD54 targeting combined with magnetic navigation achieved a 16.3-fold increase in pulmonary drug accumulation compared to free CA-074. The SPIO core acted as an antioxidant nanozyme to scavenge reactive oxygen species, synergizing with CA-074 to stabilize lysosomal membranes and block CTSB-mediated PANoptosis. Non-invasive MRI established a quantitative correlation between T1-signal intensity and pneumonia severity, enabling real-time monitoring of drug biodistribution. In murine models of IAV and SARS-CoV-2 infection, IO@CA/P significantly reduced viral loads and pulmonary inflammation via the CTSB-PANoptosis axis. Furthermore, IO@CA/P independently attenuated LPS-induced sterile inflammation and subsequent fibrotic/thrombotic sequelae, highlighting its broader anti-inflammatory applicability beyond direct antiviral mechanisms. Therefore, this host-directed nanotheranostic strategy has potential for clinical translation to combat emerging respiratory viruses and mitigating life-threatening immunopathology.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04531-x.

Introduction

The persistent threat of respiratory viral pandemics, from seasonal influenza to the continuous evolution of SARS-CoV-2 variants, constitutes a formidable global health challenge [1, 2]. Although direct-acting antivirals have saved countless lives, their efficacy is often compromised by rapid viral mutations and the development of drug resistance. Furthermore, mortality in severe respiratory infections is frequently driven not by viral load alone, but by an aberrant host immune response known as a "cytokine storm", characterized by extensive tissue damage and acute respiratory distress syndrome (ARDS) [3, 4]. Consequently, therapeutic strategies are increasingly shifting towards host-directed therapies (HDT), which target conserved host factors critical for viral pathogenesis or the ensuing inflammatory cascade [5]. It offers a promising avenue for developing broad-spectrum, mutation-resilient treatments.

Host antiviral defense and the subsequent inflammatory response are highly dependent on lysosomal protease activities [6, 7]. Specifically, lysosomal Cathepsin B (CTSB) has been identified as a critical host factor that governed viral infection and the initiation of virus-induced inflammatory cell death [8, 9]. For several viruses, including SARS-CoV-2 and influenza virus, CTSB can proteolytically prime their envelope glycoproteins within endosomal compartments, thereby facilitating viral entry and replication [10, 11]. Subsequently, upon infection and lysosomal stress, leaked CTSB can act as a key upstream activator of NLRP3 and Bcl-2, triggering PANoptosis (a coordinated inflammatory cell death pathway involving features of apoptosis, necroptosis, and pyroptosis) and the release of potent pro-inflammatory cytokines like TNF-α, IL-1β and IL-18, thereby fueling the cytokine storm [10–14]. Therefore, inhibiting the CTSB-mediated signaling axis presents a promising dual-action approach to simultaneously suppress viral propagation and mitigate lethal lung injury. However, the systemic administration of potent CTSB inhibitors, such as CA-074 (CA), faces significant hurdles, including poor solubility, limited lung bioavailability, and potential off-target toxicities.

Recently, nanomedicine provides a sophisticated platform to integrate HDT with targeted delivery and non-invasive monitoring. Superparamagnetic iron oxide (SPIO) nanoparticles act as a novel magnetic resonance contrast agent have been widely used for imaging lymph node metastases, brain tumors, atherosclerosis, lung injury and osteoarthritis in clinics [15–20]. Ferumoxtran-10 (Advanced Magnetics, Boston, USA) is currently undergoing clinical trials and has exhibited remarkable efficacy in imaging lymph node metastases [21]. Particularly, SPIO, with a particle size of less than 50 nm, partially evades phagocytosis by the liver and spleen, resulting in an extended in vivo half-life exceeding 6 h and enabling the imaging of other tissue sites [22]. Moreover, SPIO has emerged as an ideal core material for pulmonary applications due to their excellent biocompatibility, intrinsic catalase/peroxidase-like nanozyme activity for scavenging reactive oxygen species (ROS), and unique magnetic properties that enable both magnetic resonance imaging (MRI) and magnetic targeting [23, 24]. However, the efficacy of these nanoparticles in addressing acute systemic infections such as COVID-19 and influenza virus and the underlying biological mechanisms influencing their effects remain unclear. By functionalizing SPIO with specific ligands, it could achieve "theranostic" capabilities, combining precise lung-targeted delivery with real-time monitoring of drug biodistribution. It had been demonstrated that intercellular adhesion molecule-1 (ICAM-1/CD54) was dramatically upregulated on the surface of virus-infected lung epithelial cells and activated vascular endothelium in response to inflammatory cytokines [25, 26]. This specific upregulation made CD54 an excellent “zip code” for targeting the inflamed pulmonary microenvironment, enabling the selective delivery of therapeutic agents to the primary site of infection and injury.

In this study, we developed a magneto-active, dual-targeted theranostic nanoplatform (termed IO@CA/P) designed for the broad-spectrum treatment of severe viral pneumonia. The specific CTSB inhibitor CA-074 (CA) and a high-affinity cyclic peptide targeting CD54 (P) were covalently functionalized via dopamine linkers and then conjugated to the SPIO core through catechol-Fe³⁺ coordination. We hypothesized that this design creates a synergistic “multi-lock” mechanism: 1) the CD54-targeting peptide directed the nanoparticles to the inflamed pulmonary vasculature and infected epithelium, 2) magnetic navigation (MN) further enhanced their retention in the lungs, allowing for MRI-guided delivery, 3) the SPION core acted as an antioxidant nanozyme to mitigate oxidative stress and prevent upstream lysosomal damage, and 4) the released CA-074 directly inhibited CTSB activity. This concerted action was designed to halt both viral propagation and the CTSB-driven PANoptosis cascade. Utilizing both SARS-CoV-2 and influenza infection models, we provided compelling evidence that IO@CA/P achieved significant lung accumulation, potently inhibited viral replication, and rescued lung epithelial cells from PANoptosis-mediated injury, offering a promising, mutation-resistant strategy for combating severe viral pneumonia and secondary cytokine storm.

Materials and methods

Reagents and chemicals

CA-074 (CA) (Purity: ≥99.0%) and dopamine (Purity: ≥99.0%) were purchased from Shanghai Macklin Inc., China. SPIO nanoparticles (Purity: ≥97%; 20-100 nm particle size) was obtained from Xingye Metal Materials Ltd., Hebei, China. Remdesivir (RDV), ribavirin (RBV), dexamethasone (Dex), lipopolysaccharide (LPS), N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide (EDC) and N-hydroxysuccinimide (NHS) were obtained from Yuchuyuan Biochemistry Co., Ltd. (Hangzhou, China). CCK-8 was purchased from APExBIO Technology (Houston, USA). Dulbecco’s modified Eagle’s medium (DMEM) was purchased from vivacell Biotechnology GmbH (Denzlingen, Germany). Fetal bovine serum (FBS) was purchased from Vazyme (Nanjing, China). Penicillin Streptomycin (5000 U/mL) and trypsin (0.25%) were purchased from Baisha Biotech Co., Ltd. (Chengdu, China). Human/mouse TNF-α, human/mouse IL-1β, human IFN-α ELISA kits were purchased from Boster Biological Technology Co., Ltd. (Wuhan, China).

Synthesis of IO@CA/P nanoparticles

To enable surface conjugation, dopamine was employed as an anchor ligand. Dopamine hydrochloride was first per-acetylated using acetic anhydride in anhydrous pyridine to protect both the amine and catechol hydroxyl groups. Selective N‑deprotection was achieved by mild hydrolysis with K₂CO₃ in MeOH at 0°C, yielding O‑acetyl-protected dopamine with a free amine.

Then, CA was activated with EDC and NHS in anhydrous DMF for 2 h at room temperature. The O‑acetyl-protected dopamine was then coupled to the activated CA via amide bond formation. After reaction completion, the acetyl protecting groups on the catechol were removed under gentle conditions (K₂CO₃, MeOH, 0°C) to yield the CA-dopamine conjugate (CA-DA), while preserving the integrity of the epoxysuccinyl pharmacophore.

According to the methods previously reported [27, 28], the CD54‑targeting cyclic peptide (P; sequence: Ac-Cys-Asp-Pro-Arg-Tyr-Lys(PEG4)-Gly-Gly-Gly-Glu-Cys-OH) was synthesized and identified by LC-MS analysis by Shanghai Ziyao Chemical Co., China, and then separately functionalized with dopamine. The peptide was synthesized with an additional C‑terminal cysteine residue, then conjugated to maleimide‑dopamine via thiol‑maleimide Michael addition, producing the peptide‑dopamine conjugate (P-DA). A scrambled control peptide (scrP; sequence: Ac-Cys-Gly-Arg-Tyr-Asp-Ser-Lys(PEG₄)-Gly-Gly-Thp-Cys-OH) with no predicted affinity for CD54 was similarly synthesized and cyclized. The crude CA-DA and P-DA products were purified by preparative high-performance liquid chromatography (HPLC) using a C18 column with a gradient elution of acetonitrile/water (containing 0.1% trifluoroacetic acid). The collected fractions were lyophilized and characterized by LC-MS to confirm purity (>95%).

For nanoparticle assembly, the as‑prepared IO nanoparticles (1 mg/mL) were dispersed in HEPES buffer (10 mM, pH 7.4). CA-DA and P-DA were dissolved in DMSO and added sequentially to the IO dispersion at a mass ratio of CA-DA : P-DA : Fe₃O₄ = 1 : 0.5 : 10. The mixture was sonicated for 5 min and stirred at room temperature for 4 h under nitrogen to prevent catechol oxidation. Surface conjugation occurred via robust bidentate chelation between the catechol groups and Fe³⁺ on the IO surface. The IO@CA/P nanoparticles were separated from unreacted small molecules (e.g., free CA-DA, P-DA, and dopamine derivatives) using magnetic separation. Specifically, the reaction mixture was placed on a magnetic stand for 30 min, and the supernatant containing unbound molecules was discarded. The pellet was resuspended in HEPES buffer (10 mM, pH 7.4) and subjected to three cycles of magnetic separation and washing to ensure removal of non-conjugated impurities. Finally, the purified nanoparticles were dispersed in PBS (pH 7.4) and stored at 4°C until use.

Throughout this manuscript, CA/P referred to the physical mixture of CA-DA conjugate and P-DA conjugate at a mass ratio corresponding to that present in IO@CA/P (CA:P = 2:1, w/w). IO@CA/scrP referred to nanoparticles functionalized with scrP instead of the CD54-targeting peptide P, while maintaining the same CA-074 loading. All dose comparisons (e.g., CA/P vs. IO@CA/P) were performed on an equivalent CA-074 molar basis unless otherwise stated.

Characterization of IO@CA/P

The morphologies of SPIO and the prepared IO@CA/P were assessed using a transmission electron microscope (TEM; JEM-2100plus, HITACHI, Tokyo, Japan) and scanning electron microscope (SEM; JSM-IT800, JEOL, Tokyo, Japan). The Fourier transform infrared (FTIR) spectra of CA, SPIO, and IO@CA/P were determined using a Spectrum One spectrometer (PerkinElmer, USA). The nanoparticle sizes of SPIO and IO@CA/P were measured using a molecular charge analyzer (Malvern Instruments Ltd., Malvern, UK). The magnetic hysteresis loops of SPIO and IO@CA/P were determined using a vibrating sample magnetometer (LakeShore 8600, USA).

In vitro drug release and stability evaluation

The in vitro release profiles of CA, CA-DA, iron, and IO@CA/P were investigated using a dialysis method. Briefly, samples containing equivalent amounts of CA were sealed in dialysis bags (MWCO 3.5 kDa) and immersed in 30 mL of release media: phosphate-buffered saline (PBS, pH 7.4) or lysosomal acidic medium (LAM, pH 5.0). The system was incubated at 37 °C with constant shaking (100 rpm). At predetermined time intervals, 1 mL of external medium was withdrawn for quantification and replaced with an equal volume of fresh medium. The released CA and iron contents were determined by high-performance liquid chromatography (HPLC) and inductively coupled plasma mass spectrometry (ICP-MS), respectively.

The colloidal stability of IO@CA/P was assessed by monitoring the hydrodynamic size and polydispersity index (PDI) over 72 h. Nanoparticles were dispersed in PBS (pH 7.4), LAM (pH 5.0), or rat plasma and incubated at 37°C.

In vitro characterization of P

The Biacore X100 system (GE group, USA) was employed to determine the binding affinity of P and its scrambled control (scrP) to recombinant human CD54 protein (Sino Biological, Beijing, China) by surface plasmon resonance (SPR) as previously reported [29]. Briefly, purified protein was coupled with NTA sensor chip (GE group, USA). The PBS buffers (containing 0.625–5 μM of P) were loaded to detect the response values by using a Biacore 8K detector (Cytiva, Switzerland). The execution time was 200 s, and the velocity of flow was 30 μL/min. The affinity was determined as K_D value according to the instrument’s operating instructions.

FITC-labeled P or scrP was incubated with CD54 ^high-expressing A549 cells (stimulated with TNF-α/IFN-γ), and membrane binding was visualized by confocal microscopy and quantified by flow cytometry. The CD54 expressions on normal and virus-infected human alveolar type II epithelial cells (hAECs) and THP-1 macrophages were analyzed by flow cytometry using an anti-CD54 antibody. Cellular uptake of FITC-labeled peptides into hAECs or THP-1 cells was observed under fluorescence microscopy and quantitatively assessed by flow cytometry. Since high expression of CD54 in alveolar epithelial cells and pulmonary microvascular endothelial cells triggers the recruitment of circulating neutrophils and macrophages, inhibiting this marker can reduce the infiltration of pro-inflammatory cells and thereby ameliorate inflammatory injury [30]. The inhibitory effect of P on neutrophil recruitment to infected hAECs was further evaluated using a Transwell migration assay.

Targeting efficiency analysis

Accumulation of IO@CA/P in the lung tissues using magnetic resonance imaging (MRI) analysis

The male 6-week-old K18-hACE2 (human angiotensin-converting enzyme 2) transgenic mice (background: C57BL/6J) were obtained from Zhejiang Laboratory Animal Center (Hangzhou, China). The National Research Council’s Guide for the Care and Use of Laboratory Animals was implemented in the study. All the antiviral experiments in the study (No. 2023R01011) were approved by the Ethics Committee of Hangzhou Institute of Medicine, Chinese Academy of Sciences and performed in ABSL-3 laboratory of Hangzhou Institute of Medicine, Chinese Academy of Sciences.

Five mice as the control were challenged intranasally with 50 μL of saline, while another five mice were challenged with 50 μL of saline containing 10⁶ copies of SARS-CoV-2 Omicron variant to induce severe viral infection [31]. Similarly, 5 mice were challenged with 10⁶ copies of influenza A virus (IAV). Three days after infection, all animals were fasted for 10 h, and the mice in the IO@CA/P-treated group were intravenously injected with IO@CA/P (10, 25, and 50 mg/kg). Subsequently, MN was applied by attaching a neodymium magnet (15 mm × 10 mm × 2 mm, 3600 Gauss surface field) to the dorsal chest wall immediately after intravenous injection. The magnet was maintained in place for 2 h post-injection. For groups labeled without MN, mice received the same intravenous injection but no magnet was attached. All the mice were monitored using a non-invasive in vivo 7.0T MRI system (PharmaScan7016, Bruker, Germany) under isoflurane anesthesia delivered via a nose cone.

Distribution of IO@CA/P in vivo

After MRI analysis, blood was collected from the mouse jaw, and each mouse was perfused with phosphate-buffered saline (PBS) containing sodium citrate. One milligram of splanchnic tissues or 0.1 mL of blood sample was digested with 0.5 mL of 68% nitric acid and 0.1 mL of 70% hydrochloric acid, heated at 90°C for approximately 1.5 h, and then mixed with 0.1 mL of Triton X-100 for 1 h to ensure complete digestion of the sample. The nonheme iron content in different tissues or blood was determined using ICP-MS system (Agilent 7800, USA) as previously reported [32].

Antiviral activity of IO@CA/P in SARS-CoV-2/IAV-infected hAECs

To evaluate the cytotoxicity of IO@CA/P, hACE2-overexpresing hAECs were respectively treated with SPIO, CA and IO@CA/P at 25-1000 μg/mL for 48 h. Furthermore, to assess the antiviral activities of CA, SPIO, and IO@CA/P in vitro, the cells were pre-exposed with SARS-CoV-2 Omicron/Delta variants, IAV H1N1/H3N2, or respiratory syncytial virus (RSV) (MOI = 2) for 24 h and then respectively treated with CA, SPIO, and IO@CA/P for 48 h. The cell viability of each sample was assessed using CCK8 assay. The 50% inhibitory concentration (IC₅₀) was estimated using OriginPro 2024b software (MA, USA).

The antiviral activities of IO@CA/P on viral infection were assessed as previously described [33–35]. (1) Direct virucidal activity: the virus (100TCID₅₀) was pretreated with IO@CA/P (IC₉₀) at 37°C for 2 h, mixed with 100 μL of cell suspension (containing 5000 cells), and then incubated for 24 h at 37°C. (2) Inhibition of viral adsorption: the cells were cotreated with the virus (100TCID₅₀) and IO@CA/P (IC₉₀) at 4°C for 2 h. Then, the cells were washed with DMEM medium to remove unabsorbed viruses and drug, followed by incubation for 24 h at 37°C. (3) Inhibition of viral replication: precooled cells were exposed to the virus (100TCID₅₀) for 2 h at 37°C, washed, and subsequently treated with IO@CA/P (IC₉₀) at 37°C for 24 h.

Since hAECs highly express TMPRSS2, which cleaves the IAV HA and S protein, TMPRSS2-negative Vero E6 cells were used to avoid interference with the results [36, 37]. The viral loads and genome replication in Vero E6 cells were quantified using fluorescent quantitative PCR (FQ-PCR) analysis. Additionally, after 48 h of incubation, the cleavage of IAV hemagglutinin (HA) and SARS-CoV-2 spike (S) protein was detected by western blot to investigate the inhibition of IO@CA/P on viral entry.

SOD activity and CAT activity assays

SOD activity was assayed using the xanthine oxidase method (hydroxylamine method). Briefly, samples were mixed with xanthine and xanthine oxidase, and then incubated at 37°C for 20 min. Superoxide anions generated during the reaction reacted with hydroxylamine to form nitrite. Upon addition of chromogenic agent, nitrite produced a purple-red complex, and the absorbance was measured at 550 nm. SOD activity was calculated based on a standard curve.

CAT activity was determined by ammonium molybdate colorimetry. Samples were incubated with 65 mmol/L H₂O₂ substrate at 37°C for 1 min. The reaction was terminated by adding ammonium molybdate, forming a stable yellow complex. Absorbance was recorded at 405 nm, and CAT activity was quantified using a standard curve. All procedures were performed in accordance with the manufacturer’s instructions, and results were expressed as U/mg.

CTSB activity and CTSB leakage assays

Cathepsin B activity was measured fluorometrically. The hAECs were lysed in RIPA buffer, and protein concentrations were determined by BCA assay. Equal amounts of protein (20 μg) were incubated with Z-Arg-Arg-AMC substrate (20 μmol/L) in reaction buffer (pH 6.0, containing 2.5 mmol/L DTT) at 37°C for 30 min in the dark. Fluorescence intensity was recorded at Ex/Em = 360/460 nm. A standard curve was generated using free AMC, and enzyme activity of untreated group was used as the control.

The hAECs were seeded on confocal dishes and treated as indicated. After treatment, cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked with 5% BSA. Immunostaining was performed using primary antibodies against CTSB (1:200, Abcam) and LAMP1 (1:200, Abcam), followed by Cy5-labeled anti-mouse IgG (1:500, Invitrogen) and FITC- labeled anti-mouse IgG (1:500, Invitrogen). Nuclei were counterstained with DAPI. Images were captured using a Zeiss LSM 880 confocal microscope. Colocalization of CTSB with LAMP1 was analyzed using ZEN 3.0 software (Zeiss) and quantified by Pearson’s correlation coefficient. A higher R value indicated less CTSB leakage from lysosomes.

Transcriptomic analysis

Total RNA was extracted from hAECs using TRIzol reagent. RNA-seq libraries were constructed and sequenced on an Illumina NovaSeq 6000 platform by Hangzhou LC Biotech Co., China. Clean reads were aligned to the reference genome using Hisat2. Gene expression levels were quantified as FPKM, and differential expression analysis was performed using DESeq2 (|log₂FC| > 1, adjusted P < 0.05). Gene Set Enrichment Analysis (GSEA) was conducted using GSEA software. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment was analyzed using the clusterProfiler R package (P < 0.05).

FQ-PCR analysis

To investigate the viral genome replication of SARS-CoV-2 and IAV nucleoprotein (NP) RNA species, total RNA was extracted using TRIzol. Strand-specific reverse transcription was performed: for vRNA detection, reverse primer specific to IAV NP (5’-TGCAAAAACATCTTCAAGTCTCTG-3’) or SARS-CoV-2 N (5’-CAGACATTTTGCTCTCAAGCTG-3’) was used with SuperScript IV; for mRNA detection, oligo(dT)20 was employed. cDNA was quantified by TaqMan qPCR. For IAV NP (vRNA): forward 5’-AGATGAGTCTTCTAACCGAGGTCG-3’, reverse 5’-TGCAAAAACATCTT CAAGTCTCTG-3’, probe 5’-FAM-TCTGCATTGTCTCCGAAGAAATAAGATCA-TAMRA-3’. For IAV NP mRNA (splice-specific): forward 5’-GGCGATTTAGCTTTGGTACG-3’, reverse 5’-CCATTCTCATTACTCTTCTCG-3’, probe same as above. For SARS-CoV-2 nucleocapsid (N) protein (vRNA): forward 5’-GGGGAACTTCTCCTGCTAGAAT-3’, reverse 5’-CAGACATTTT GCTCTCAAGCTG-3’, probe 5’-FAM-TTGCTGCTGCTTGACAGATT-TAMRA-3’. Reactions were performed using TaqMan Fast Virus 1-Step Master Mix (for vRNA) or Brilliant III Ultra-Fast SYBR Green QRT-PCR (for mRNA) on a CFX96™ Dx System (Bio-Rad, USA) under the following conditions: 50°C for 5 min, 95°C for 20 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. Relative RNA levels were calculated using the 2^−ΔΔCt method with GAPDH as internal control.

Preparation of viral pneumonia model and IO@CA/P intervention

Male K18-hACE2 transgenic mice (22–25 g body weight) were divided randomly into 8 groups: normal control (Ctrl), model control, positive control (25 mg/kg of RDV), SPIO-treated (32 mg/kg, matching the SPIO content in IO@CA/P 50 mg/kg), CA/P-treated (18 mg/kg, equimolar CA-074 to IO@CA/P 50 mg/kg), and IO@CA/P-treated (10, 25, and 50 mg/kg). Each group had 10 mice. Seventy mice were intranasally injected with 30 μL of saline containing 1×10⁶ copies of the SARS-CoV-2 Omicron variant, while mice in Ctrl group were treated with saline. All in vivo comparisons were performed on an equivalent CA-074 molar basis unless otherwise stated. This is now clearly noted in the table legend. The doses of CA/P and RDV were used as previously reported [9, 31]. CA/P, SPIO, RDV or IO@CA/P was intravenously injected once a day for 5 days, while the control and infected mice were injected with saline daily. Approximately 10 min before drug intervention, a neodymium magnet (15 mm × 10 mm × 2 mm, 3600 Gauss) was attached to the back of each mouse in the SPIO-, CA/P-, and IO@CA/P-treated groups and kept in place for 2 h after intravenous injection [38]; other groups underwent the same injection but without magnet attachment. After 5 days of viral infection, all mice were sacrificed using isoflurane. The body weight and wet lung weight of each mouse were weighed, and the blood was collected in EDTA-pretreated tubes.

To investigate the broad-spectrum antiviral and anti-inflammatory effects of IO@CA/P, separate groups of mice were intratracheally injected with IAV (1×10⁶ copies) or LPS (10 mg/kg) to induce viral pneumonia or acute inflammatory lung injury (cytokine storm), respectively [39, 40]. RBV and Dex served as positive controls for IAV infection and LPS-induced inflammation, respectively. Moreover, to further assess the protective effect of IO@CA/P against post-infection lung sequelae (lung fibrosis and microthrombosis), infected mice from each model were treated with the corresponding positive control drugs for 5 days and then treated with IO@CA/P daily for other 7 days [41, 42]. The experimental groups and intervention methods for these fibrosis/thrombosis studies were identical to those described above for the respective acute infection/inflammation models. After 7 days of IO@CA/P treatment, lungs were collected, homogenized in saline, and centrifuged; the supernatants were harvested for subsequent analysis. The levels of TNF-α, IL-1β and IFN-α in media or supernatants were detected using ELISA according to the protocols (detailed in supplement) mentioned in the respective kits (Saipei Biotech Co. Ltd., Wuhan, China). The standard curves were run in duplicate for each plate, with R² > 0.99 required for acceptance.

Lung pathological evaluation

The left lung of each mouse was washed twice with saline, fixed in 4% paraformaldehyde for 72 h, embedded in paraffin, and sectioned into 4-5 μm slices. The lung tissues were stained with hematoxylin and eosin (HE) and then visualized using the Pannoramic Slide Scanners (3D HISTECH, Hungary).

Western blot analysis

Proteins extracted from the cells or lungs were treated with radioimmunoprecipitation assay lysate, and the concentration of each protein sample was determined using the bicinchoninic acid method. Subsequently, 40 μg of protein sample was used for vertical electrophoresis and western blotting. After transfer, the membrane was blocked with 5% bovine serum albumin (BSA) in Tris-buffered saline with Tween 20 (TBST) for 1 h at room temperature. The membrane was treated with primary antibodies (1:1000) (Proteintech, CHI, USA) at 4°C for 10 h, washed twice with TBST, and then treated with the secondary antibody (1:2000) for 1 h at 20°C. Enhanced chemiluminescence (ECL) solution was added to visualize the signals, and the chemiluminescent signals were captured using a chemiluminescence imaging system (Bio-OI, Guangzhou, China) with an exposure time of 30 s to 2 min (adjusted according to signal intensity). The relative expression of each protein was quantified using the imaging system’s software.

Safety evaluation in vivo

The acute toxicity, subacute toxicity and hemolytic assays were detailed in the supplement.

Statistical analysis

Results were presented as the mean ± standard deviation. OriginPro 2024b and GraphPad Prism 9.0 software were employed for plotting and data analysis. For comparisons involving a single time point across multiple groups, one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was used. For longitudinal data (e.g., body weight changes, time-dependent MRI signal intensity, pharmacokinetic profiles), two-way ANOVA with repeated measures was performed, followed by Bonferroni’s post hoc test to correct for multiple comparisons. The statistical significance was denoted as *P < 0.05 and *P < 0.01.

Results

Preparation and characterization of IO@CA/P

As shown in Fig. 1A, IO@CA/P nanoparticles were assembled via catechol-mediated anchoring of dopamine-functionalized CA and CD54-targeting cyclic peptide onto SPIO cores. In contrast to the uncoated square SPIO (nano size: 45.4 ± 4.5 nm; zeta potential: 2.5 mV), the surface of IO@CA/P was coated with CA and P, resulting in nearly spherical particles with an average size of 56.9 ± 4.1 nm and a zeta potential of −18.1 mV (Fig. 1B–1E). The near-zero zeta potential (absolute value ~2.5 mV) indicated poor electrostatic stabilization, which is typical for uncoated SPIOs at neutral pH and explains their tendency to aggregate as previously reported [43, 44]. It further highlighted the importance of surface modification for achieving colloidal stability in our experiments. The FTIR spectrum of IO@CA/P exhibited the characteristic Fe–O stretching vibration of the magnetite core at 597 cm^-1. Successful dopamine anchoring was evidenced by the appearance of Fe–O–C bonds at 1120 cm^-1 and the attenuation of free phenolic O–H stretches. Pronounced amide I bands, along with COO⁻ symmetric stretching at 1610 cm^-1, verified the co-immobilization of CA and P. These spectral features collectively demonstrated the successful layer-by-layer assembly of IO@CA/P (Fig. 1F). SPIO and IO@CA/P presented saturation magnetizations of 82.59 and 71.07 emu/g at 20 kG, respectively. The S-shaped hysteresis loops confirmed their superparamagnetic properties (Fig. 1G). Moreover, TEM dark-field imaging results demonstrated that the sulfur-containing CA was uniformly modified on the surface of SPIO (Fig. 1H).

Fig. 1.

Characterization of IO@CA/P nanoparticles. (A) Schematic illustration of IO@CA/P synthesis. CA, CA-074. P, CD54-targeting cyclic peptide. (B) Morphology assessment of SPIO and IO@CA/P using scanning electron microscopy (SEM) and (C) transmission electron microscopy (TEM). TEM images on the top right showed the amplified appearances of nanoparticles. (D) The sizes of SPIO and IO@CA/P nanoparticles. (E) Zeta potentials of SPIO, CA and IO@CA/P. (F) FTIR spectra of SPIO and IO@CA/P nanoparticles. (G) The magnetic saturation of SPIO and IO@CA/P. (H) The distribution for the elemental mapping of iron (blue), oxygen (yellow), sulfur (orange) and nitrogen (green). (I) In vitro drug release profiles of CA, CA-DA, iron, and IO@CA/P were respectively determined in PBS (pH 7.4) and lysosomal acidic media (LAM; pH 5.0). DA, dopamine. (J) The stability of IO@CA/P in LAM, PBS and plasma observed over 72 h. All data were showed as mean ± SD (n = 5). * P<0.05, ** P<0.01

As shown in the XPS data (Figure S1), in the survey spectra, pure SPIO contained only Fe, O, and adventitious carbon, with the absence of N element. The CA-DA showed no Fe element, but exhibited C, O, and N. For IO@CA/P, a weak Fe 2p signal was retained, while the C and N signals were significantly enhanced. In the high‑resolution C 1s spectra, both the CA-DA and IO@CA/P groups displayed the amide peaks at 288.5 eV, which was absent for pure SPIO. In the N 1s high‑resolution spectra, the main peaks of both samples were located at 400.0 eV. These three features collectively confirmed the successful covalent surface modification.

The in vitro drug release profiles of CA, CA-DA, SPIO, and IO@CA/P were evaluated in both PBS (pH 7.4) and lysosomal acidic media (LAM; pH 5.0) (Fig. 1I). IO@CA/P exhibited robust pH-responsive release behavior, characterized by sustained release at pH 7.4 but accelerated release at pH 5.0, resulting in a 3-fold increase in cumulative release within the acidic environment. In stark contrast, free CA and CA-DA showed rapid and uncontrolled release regardless of the pH. Similarly, iron ions from IO@CA/P and SPIO were rapidly released at pH 5.0, but remained nearly undetectable at pH 7.4. Stability assays further demonstrated that no significant changes in particle size of IO@CA/P were observed in either PBS or plasma over 72 h. In contrast, in LAM, IO@CA/P underwent gradual decomposition due to dissolution, leading to a progressive reduction in particle size (Fig. 1J). These results confirmed that IO@CA/P could effectively prevent premature drug leakage in systemic circulation while enabling precise, stimuli-triggered cargo release within the acidic compartments of target cells.

In vitro characterization of P

The binding specificity and functional activity of P were evaluated in vitro. The results of SPR analysis revealed that P bound directly to recombinant CD54 protein in a concentration-dependent manner, but scrP could not (Fig. 2A). Fluorescence microscopy further demonstrated that FITC-labeled P specifically adhered to the membrane of CD54^high-expressing A549 cells, while scrP could not (Fig. 2B). Notably, the expression of CD54 was markedly upregulated in infected hAECs (>3-fold increase), but only moderately increased (~50%) in infected THP-1 macrophages compared to uninfected controls (Fig. 2C). Accordingly, the uptake of IO@CA/P nanoparticles functionalized with P was substantially enhanced in infected hAECs (14-fold over non-infected hAECs), whereas infected THP-1 macrophages exhibited only 13.6% of the uptake observed in hAECs, as visualized by optical microscopy (Fig. 2D), and quantified by flow cytometry (Fig. 2E). Moreover, P and IO@CA/P effectively inhibited neutrophil recruitment to infected hAECs compared to scrP (Fig. 2F). These findings confirmed that P-functionalized nanoparticles selectively targeted CD54-expressing hAECs and effectively blocked neutrophil adhesion, highlighting the potential for precision delivery to infected alveolar epithelium.

Fig. 2.

The characteristics of CD54-targeting cyclic peptide. (P) in vitro (A) Binding response profiles of P and scrambled control peptide (scrP) to CD54 protein. (B) FITC-labeled P and scrP bound onto the membrane surface of CD54^high-expressing A549 cells. (C) The expressions of CD54 in normal control and infected human alveolar epithelial cells (hAECs) (and macrophages THP-1). (D) Targeting uptake of P and scrP into the hAECs (or THP-1) observed by optical microscopy and (E) quantitatively detected by flow cytometry. Scale bar, 20 μm. (F) Inhibitory effect of P on neutrophil recruitment to infected hAECs. All data were showed as mean ± SD (n = 5). Compared with the control group, *P < 0.05, **P < 0.01; ns, no significance; ND, not detected

Biodistribution of IO@CA/P in the tissues of infected mice

The biodistribution of IO@CA/P was systematically evaluated in virus-infected mouse models. T1-weighted MRI revealed significant accumulation of IO@CA/P in the lungs of IAV (H1N1)-infected mice, as evidenced by enhanced coronal and transverse signals (Fig. 3A-3B). Similarly, robust pulmonary accumulation was observed in SARS-CoV-2 Omicron-infected hACE2-transgenic mice. Quantitative analysis demonstrated a strong positive correlation between the transverse T1-weighted signal intensity and both the administered IO@CA/P dose (R² = 0.992) and the severity of pneumonia (R² = 0.946), indicating that nanoparticle accumulation was proportional to disease severity (Fig. 3C-3D).

Fig. 3.

Biodistribution of IO@CA/P in the tissues of virus-infected mice. (A) Coronal T1-weighted MRI images for assessing IO@CA/P accumulation in the whole body of IAV (H1N1)-infected mice. (B) Transverse T1-weighted MRI images for assessing IO@CA/P accumulation in the lung tissues of IAV (H1N1)-infected mice and SARS-CoV-2 Omicron-infected hACE2-expressing mice. (C) The correlation between transverse T1-weighted signal and IO@CA/P dose. The pink area indicated 95% confidence interval of dataset. (D) The correlation between transverse T1-weighted signal and pneumonia severity. (E) Pharmacokinetic profiles after intravenous injection of CA, IO@CA/scrP and IO@CA/P at 50 mg/kg in the serum and lung tissues of infected mice. The contents of CA were quantified by HPLC analysis. scrP, scrambled control peptide. (F) The fluorescence changes of main organs in infected mice at different time after intravenous injection of Cy5-labeled CA and Cy5-labeled IO@CA/P under magnetic navigation (MN). (G) The contents of CA in feces or urine after 24 h of IO@CA/P injection for investigating the excretion of IO@CA/P. (H) The distribution of IO@CA/P in the organ tissues of infected mice, which was assessed after 1 h of drug administration under MN. (I) TEM images showed the cellular uptake of IO@CA/P in the lung tissues of infected mice after 1 h of treatment. The right images amplified the contents in the yellow frames. Yellow arrows indicated the presence of IO@CA/P. Scale bar, 1 μm. All data were showed as mean ± SD (n = 5). Compared with CA-treated group, * P<0.05, ** P<0.01

Leveraging the magnetic characteristics of SPIO, the effects of MN on the accumulation of IO@CA/P in the lung tissues of infected mice were examined. Pharmacokinetic profiles determined by HPLC analysis showed that IO@CA/P achieved significantly higher and more sustained retention in lung tissue compared to free CA or IO@CA/scrP, while maintaining comparable serum clearance profiles. Notably, treatment with IO@CA/P alone substantially elevated CA levels by 7.9-fold compared to free CA administration, whereas combination with MN resulted in a 16.3-fold increase (Fig. 3E).

Ex vivo fluorescence imaging of major organs under MN further confirmed preferential and prolonged lung accumulation of Cy5-labeled IO@CA/P, whereas free CA exhibited rapid systemic clearance (Fig. 3F). Excretion studies revealed that approximately 78% of the administered CA was eliminated via feces and 20% via urine within 24 h (Fig. 3G). Quantitative tissue distribution analysis confirmed that one-hour post-administration, regardless of the use of MN, IO@CA/P predominantly accumulated in the lung, liver, and kidneys, with minimal accumulation in the heart and spleen (Fig. 3H). Finally, TEM images revealed distinct electron-dense aggregates (characteristic of iron oxide) within the lysosomes of macrophages and alveolar epithelial cells in the lung tissues of IO@CA/P-treated infected animals. In contrast, such structures were rarely observed in the control group, confirming efficient cellular uptake at the target sites (Fig. 3I). Collectively, these results demonstrated that IO@CA/P nanoparticles not only achieved targeted accumulation in virus-infected pulmonary lesions but also enabled non-invasive, real-time monitoring of pneumonia severity by establishing a quantitative correlation between T1-MRI signal intensity and disease progression. Coupled with its therapeutic efficacy in mitigating acute lung injury, IO@CA/P demonstrated a significant theranostic advantage, offering an innovative platform that integrates diagnostic precision with therapeutic potency for precision medicine in respiratory viral infections.

IO@CA/P inhibited viral entry

To evaluate the potential of IO@CA/P as a biocompatible, broad-spectrum antiviral nano-formulation, the antiviral activity of IO@CA/P was evaluated against various respiratory viruses. CCK8 assays showed that administration with CA, SPIO and IO@CA/P at the concentrations of 25–500 μg/mL did not show any significant cytotoxicity in hAECs, indicating their low toxicities (Fig. 4A). IO@CA/P more effectively inhibited cell death induced by SARS-CoV-2 Omicron, Delta, IAV H1N1, H3N2 and RSV compared to the free CA groups, when calculated as the same molar mass of CA (Fig. 4B). In contrast, free SPIO alone showed weak antiviral activity, whose half-maximal inhibitory concentrations (IC₅₀) were higher than 300 μg/mL. Synergistic analysis revealed that the combination index (CI) values ranged from 0.54 to 0.72, indicating that the combination of SPIO and CA exerted a synergistic antiviral effect against all tested strains.

Fig. 4.

Protective effects of IO@CA against various respiratory viruses-induced cell death via inhibiting viral entry. (A) Cytotoxicity of SPIO and IO@CA/P in hAECs. (B) Inhibitory activities of SPIO, CA and IO@CA/P against viruses-induced cell death in hACE2-expressing hAECs. (C) The antiviral actions of SPIO, IO@CA/P and P on the different stages of viral infection in infected hACE2-expressing hAECs. (D) Inhibition of SPIO and IO@CA/P on viral protein synthesis in Vero E6 cells, which were infected with S protein- or HA-pseudotyped pseudoviruses. S protein, SARS-CoV-2 spike protein; HA, influenza A virus (IAV) hemagglutinin. (E) Inhibition of SPIO and IO@CA/P on viral genome replication in Vero E6 cells infected with SARS-CoV-2 Omicron or IAV. (F) Inhibition of CA and IO@CA/P on the cleavage of HA and S protein. All data were showed as mean ± SD (n = 5). Compared with the control (or model) group, * P<0.05, ** P<0.01

To further elucidate the antiviral mechanism of IO@CA/P, time-of-addition experiments were performed in hACE2-expressing hAECs. IO@CA/P exhibited the strongest antiviral effect when administered during the viral entry stage, whereas IO@CA/P lacked direct virucidal efficacy against these 5 respiratory viruses and minimal inhibition was observed during viral adsorption (Fig. 4C).

It was further verified using pseudoviruses coated with S protein or HA, where CA and IO@CA/P significantly inhibited virus-mediated protein synthesis (Fig. 4D). Consequently, the nanoparticles effectively blocked the genomic replication of SARS-CoV-2 Omicron and IAV within host cells (Fig. 4E). In-depth biochemical analysis revealed that the antiviral mechanism of IO@CA/P lay in its ability to specifically inhibit the proteolytic cleavage of HA and S proteins (Fig. 4F), a critical step required for viral fusion, thereby efficiently thwarting the viral invasion and replication. Collectively, these findings demonstrated that IO@CA/P exerted broad-spectrum antiviral activity by specifically inhibiting viral entry via interference with viral glycoprotein cleavage, rather than through direct virucidal effects or interference with viral adsorption.

Nanozyme activity and stabilization of lysosomal integrity by IO@CA/P

Given that viral infections often trigger oxidative stress and lysosomal dysfunction, the mimetic enzyme activities of IO@CA/P were evaluated. As shown in Fig. 5A, IO@CA/P and SPIO exhibited obvious SOD-like and CAT-like activities in vitro, whereas free CA showed minimal enzyme-mimetic effects. However, these nanozyme activities could be reversed by sodium diethyldithiocarbamate (DDC, SOD inhibitor) and 3-amino-1,2,4-triazole (3-AT, CAT inhibitor) (Fig. 5B). Consistently, in infected hAECs, IO@CA/P and SPIO treatment significantly enhanced cellular SOD and CAT activities compared to untreated infected controls. And, IO@CA/P significantly suppressed the surge of ROS induced by SARS-CoV-2 Omicron and IAV (H1N1) (Fig. 5C), compared with the standalone SPIO group. Importantly, these antioxidant capabilities were primarily attributed to the SPIO core, which functioned as a high-efficiency nanozyme to maintain redox homeostasis.

Fig. 5.

IO@CA/P inhibited the production of ROS and the release of CTSB from lysosomes into the cytoplasm. (A) SOD-like and CAT-like activities of CA, SPIO and IO@CA/P in vitro and (B) in infected hAECs. DDC, sodium diethyldithiocarbamate (SOD inhibitor); 3-AT, 3-amino-1,2,4-triazole (CAT inhibitor). (C) Inhibitory effects of SPIO and IO@CA/P on viruses-induced ROS in infected hAECs. (D) Inhibitory effects of IO@CA/P on the activity of CTSB in infected hAECs, which was detected by using Z-Arg-Arg-AMC as the fluorogenic substrate of cathepsin B on fluorophotometer. (E) Inhibitory effects of CA, SPIO and IO@CA/P on the leakage of CTSB from lysosomes into the cytoplasm. The co-localization of CTSB and LAMP1 was assayed by Pearson’s coefficient analysis. A higher R value indicated less leakage of CTSB. Nucleus, DAPI (blue); CTSB, Cy5 (red); LAMP1 (lysosomal marker), FITC (green). Scale bar, 10 μm. All data were showed as mean ± SD (n = 5). Compared with the model group, *P < 0.05, **P < 0.01; ns, no significance

Given the critical role of CTSB in viral entry and inflammation [10–14], the effect of IO@CA/P on CTSB activity was assessed. Using a fluorogenic substrate Z-Arg-Arg-AMC, IO@CA/P was found to significantly inhibit CTSB enzymatic activity in vitro (Fig. 5D). Importantly, virus infection triggered the leakage of CTSB from lysosomes into the cytoplasm, as indicated by reduced colocalization between CTSB and the lysosomal marker LAMP1. However, treatment with free CA or IO@CA/P, but not SPIO alone, effectively preserved lysosomal integrity, maintaining high colocalization of CTSB with LAMP1 (Pearson’s R value > 0.6) (Fig. 5E). These results demonstrated that IO@CA/P exerted a dual-protective effect by scavenging ROS via its nanozyme activity and preventing CTSB-mediated cytosolic damage, thereby stabilizing the intracellular homeostasis of infected airway epithelial cells.

IO@CA/P mitigates viral infection by inhibiting CTSB-mediated PANoptosis

To systematically explore the protective mechanisms of IO@CA/P, the comprehensive transcriptomic profiling of virus-infected hAECs was performed. Differential expression gene (DEG) analysis revealed distinct transcriptional landscapes between the virus-infected and IO@CA/P-treated groups (Fig. 6A). Compared with IAV-infected group, 867 genes were up-regulated and 724 genes were down-regulated in IO@CA/P-treated group; compared with Omicron-infected group, 908 were up-regulated and 605 were down-regulated in IO@CA/P-treated group. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment and circular heatmap analyses identified that these DEGs were predominantly involved in inflammatory, host pattern recognition receptors and programmed cell death pathways, including MAPK signaling pathway, NF-kappa B signaling pathway, TNF signaling pathway, Toll-like receptor signaling pathway, NOD-like receptor signaling pathway, p53 signaling pathway, Apoptosis and mTOR signaling pathway (Fig. 6B-6C).

Fig. 6.

IO@CA/P inhibited virus-induced PANoptosis in hACE2-expressing hAECs. (A) The profiles of differently expressed genes (DEGs) between virus-infected group and IO@CA/P-treated group. (B) The DEGs involved in the Top20 pathways by Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis and (C) their transcriptome characteristics by circular heatmap analysis. (D) The common DEGs involved in the regulation of IO@CA/P in virus-infected hAECs. (E) The hub genes in those common DEGs and the signaling pathways they were involved in, which were performed by chordal plot of KEGG analysis. (F) The common DEGs were enriched in PANoptosis-related signaling pathways (all P<0.05). (G) The protective effects of IO@CA/P against elevated inflammatory cytokines in infected hAECs. (H) The profiles of PANoptosis-related markers in infected hAECs treated with IO@CA/P, which were reversed by over-expression of CTSB (OE). Semi-quantitative results were presented in supplement Figure S3. The data were showed as mean ± SD (n = 3). Compared with the model group, *P < 0.05, **P < 0.01

By using Venn diagram, 108 common DEGs were found in two datasets (Fig. 6D). Then by performing hub gene analysis through chordal plots and GSEA analysis, we mapped the regulatory network of IO@CA/P, which pointed toward a significant modulation of interferon signaling and the PANoptosis-related signaling axis (an integrated pyroptosis-apoptosis-necroptosis pathway) (Fig. 6F). Consistently, IO@CA/P treatment robustly suppressed the secretion of elevated pro-inflammatory cytokines (IFN-α, TNF-α and IL-1β) induced by viral infection, suggesting a potent anti-inflammatory response (Fig. 6G).

To further confirm whether the observed protection was specifically mediated via the CTSB/PANoptosis axis, we examined the expression of key PANoptotic markers. As shown in Fig. 6H and supplement Figure S2, viral infection triggered the dysregulation of markers associated with apoptosis (Bax, Bcl-2, Clv-caspase-3, and Clv-caspase-8), necroptosis (p-MLKL and p-RIP3), pyroptosis (NLRP3, p-GSDMD, Clv-caspase-1), and Panoptosis sensors (ZBP1, ASC and TAK1). However, these changes were significantly reversed by IO@CA/P and CA. Notably, the inhibitory effects of IO@CA/P on these PANoptotic signatures could be largely reversed by the overexpression of CTSB (OE). These results established a causal link between lysosomal CTSB leakage and the induction of PANoptosis, confirming that IO@CA/P exerted its therapeutic efficacy by stabilizing lysosomes and blocking the CTSB-mediated cell death cascade in pulmonary epithelial cells.

Protective effects of IO@CA/P against IAV, SARS-CoV-2, or LPS-induced lung injury

To evaluate the therapeutic potential of IO@CA/P against diverse pathogenic challenges, we established mouse models of acute lung injury (ALI) induced by IAV, SARS-CoV-2, and LPS. As shown in Fig. 7A, the challenged mice exhibited severe diffuse alveolar damage, characterized by profound inflammatory infiltration and structural collapse. Also, the inhibitory effects of the IO@CA/scrP and IO@CA/P groups on IAV, SARS-CoV-2, or LPS-induced inflammatory damage were all lower than those of the IO@CA/P+MN group (supplement Figure S3). However, intervention with IO@CA/P significantly attenuated these pathological features, which exhibited greater protective efficacy than CA/P and SPIO, demonstrating the advantage of combination. Moreover, IO@CA/P treatment effectively reduced lung indexes, viral loads, and pathological scores in the challenged mice (Fig. 7B). Notably, the inhibitory effect of IO@CA/P on LPS-induced lung injury, which did not involve viral entry or CTSB-mediated glycoprotein cleavage, demonstrated its capacity to mitigate sterile inflammation. While this model does not directly interrogate the CTSB-PANoptosis axis established in viral infection models, the observed reduction in pro-inflammatory cytokines (TNF-α, IL-1β, IFN-α) and pathological scores supports the broader anti-inflammatory and tissue-protective utility of IO@CA/P, likely mediated by the SPIO core’s nanozyme activity and the general anti-inflammatory properties of CA-074. Given that an uncontrolled systemic inflammatory cascade was a primary driver of mortality in severe viral infections like IAV and SARS-CoV-2 [45–47], the ability of IO@CA/P to suppress serum pro-inflammatory cytokines provided a robust defense against immune-mediated organ failure. Furthermore, in mice recovered from the initial challenge, IO@CA/P significantly downregulated lung fibrotic markers, including TGF-β1 and PC-III, as well as the microthrombosis indicator D-dimer (Fig. 7C). Taken together, these results demonstrated that IO@CA/P not only mitigated acute inflammation by curbing the cytokine storm but also provides long-term protection against secondary fibrotic remodeling and microvascular thrombosis, offering a multi-faceted therapeutic strategy for viral pneumonia.

Fig. 7.

Protective effects of IO@CA/P against IAV, SARS-CoV-2, or LPS-induced inflammatory damage in mice. (A) Histopathologic profiles in the lung tissues of challenged mice. Scale bar, 50 μm. (B) Inhibition of IO@CA/P on lung indexes, viral loads, pathological scores, and the serum cytokines in challenged mice. (C) Protective effects of IO@CA/P against lung fibrotic indexes [transforming growth factor-beta 1 (TGF-β1) and procollagen type III (PC-III)] and lung microthrombosis index (D-dimer) in mice recovered from IAV infection, SARS-CoV-2 Omicron infection, or LPS challenge. TGF-β1 and PC-III were measured in lung tissue homogenates and normalized to total protein content (pg/mg protein), while D-dimer in plasma was expressed as ng/mL. ND, not detected; LPS, lipopolysaccharide; RBV, ribavirin; RDV, remdesivir; Dex, dexamethasone. The data were showed as mean ± SD (n =10). Compared with the model group, * P<0.05, ** P<0.01

Safety evaluation of IO@CA/P

The acute toxicity of IO@CA/P was initially evaluated by administering 500 mg/kg to mice, a dosage ten times greater than that utilized in the in vivo antiviral study. Moreover, no histological and serum biochemical changes were observed in either the control or IO@CA/P-treated group (Fig. 8A-8B). In the long toxicity assessment, after prussian blue staining, a slight accumulation of SPIO and iron ion was detected in the liver, spleen, and lung tissues of IO@CA/P-treated mice (Fig. 8C). Although the serum levels of ALT alone in IO@CA/P-treated group during the treatment phase were significantly increased comparable to those in control mice, it would turn normal as the control after 2 weeks of recovery (Fig. 8D). Nonetheless, treatment with IO@CA/P demonstrated minimal hemolysis in vitro, with a hemolytic rate below 5% (Supplement Figure S4). Therefore, the synthesized macromolecular prodrug IO@CA/P displayed favorable biocompatibility, and minimal toxicity.

Fig. 8.

The acute toxicity and chronic toxicity of IO@CA/P in mice. (A) Histological assessment of liver, spleen, lung and kidney in mice which were once treated with 500 mg/kg of IO@CA/P. (B) The serum levels of ALT, AST, Cre, BUN and TG in mice for acute toxicity analysis. (C) Histological assessment of liver, spleen, lung, and kidney in mice which were treated daily with 500 mg/kg of IO@CA/P for 30 days and then treated with saline for 15 days. Scale bar, 50 μm. (D) The levels of serum ALT, AST, Cre, BUN and TG in mice for repeat-dose toxicity analysis. The data were shown as mean ± SD (n =10). Red arrows indicated the present of IO@CA/P. Compared with the IO@CA/P-treated group, * P<0.05, ** P<0.01; ns, no significance

Discussion

The management of lethal viral pneumonia remains a formidable clinical challenge, primarily due to the rapid mutational escape of respiratory pathogens and the catastrophic host immune failure characterized by hyperinflammation. While traditional interventions focus predominantly on viral clearance, our findings underscored the clinical imperatives of HDT. By engineering IO@CA/P as a CTSB-targeted nanotheranostic, we successfully integrated precise immunomodulation with real-time monitoring capabilities. This platform departed from conventional single-target approaches by simultaneously addressing the pathogenic triggers and the subsequent inflammatory cascade across various injury models, including IAV, SARS-CoV-2, and LPS challenges. The significant attenuation of pulmonary damage and cytokine release observed in our study suggested that targeting the lysosomal protease CTSB provided a robust, mutation-resistant axis for therapeutic intervention, effectively bridging the gap between acute phase viral inhibition and long-term tissue preservation.

The rationale for targeting CD54 (ICAM-1) stemmed from its dramatic upregulation on virus-infected lung epithelial cells and activated vascular endothelium in response to inflammatory cytokines [48, 49]. Our results confirmed that CD54 expression was markedly increased (>3-fold) in infected hAECs, rendering them as the primary target for CD54-directed delivery. This was corroborated by the 14-fold enhanced uptake of IO@CA/P in infected hAECs compared to non-infected controls. These findings aligned with previous reports demonstrating that CD54 served as an effective "zip code" for inflamed pulmonary tissues [49], and underscored the advantage of targeting the primary site of viral replication and injury. Although free CA-074 exhibited efficient release at both neutral and acidic pH, its clinical utility was severely limited by poor lung targeting, rapid systemic clearance, and lack of synergistic functions. But IO@CA/P integrated CD54‑mediated active targeting, MN for enhanced pulmonary accumulation, prolonged lung retention, and SPIO core‑derived nanozyme activity. Furthermore, leveraging the superparamagnetic properties of the SPIO core, MN achieved a 16.3-fold increase in pulmonary CA accumulation compared to free CA administration. Notably, this enhanced lung retention was accompanied by a strong positive correlation between MRI signal intensity and both nanoparticle dose (R² = 0.992) and pneumonia severity (R² = 0.946), establishing IO@CA/P as a true theranostic platform capable of non-invasively monitoring disease progression and drug biodistribution.

Viral infection induced robust oxidative stress, characterized by excessive ROS generation, which in turn promoted lysosomal membrane permeabilization and subsequent CTSB release into the cytoplasm; on the other hand, eliminating ROS also protected against virus-induced Panoptosis and viral replication [12, 50–55]. Once translocated to the cytosol, CTSB functioned as a key upstream activator of multiple pro-inflammatory and pro-death signaling cascades, culminating in PANoptosis [56–65]. In this study, IO@CA/P exerted potent protective effects through a dual mechanism. The SPIO core exhibited intrinsic SOD-like and CAT-like nanozyme activities, effectively scavenging virus-induced ROS, whereas CA specifically inhibited CTSB enzymatic activity. By targeting both the upstream trigger (ROS) and the central mediator (CTSB), IO@CA/P preserved lysosomal integrity, as evidenced by sustained colocalization of CTSB with LAMP1, and effectively blocked PANoptosis. This was demonstrated by suppression of key PANoptotic markers, including pyroptosis (NLRP3, cleaved GSDMD, cleaved caspase-1), apoptosis (cleaved caspase-3, Bax/Bcl-2 ratio), and necroptosis (p-MLKL, p-RIP3), as well as PANoptosis sensors (ZBP1, ASC, TAK1). The causal role of CTSB was confirmed by CTSB overexpression, which abrogated the protective effects of IO@CA/P and restored PANoptotic signaling.

The therapeutic potential of IO@CA/P was evaluated in three complementary in vivo models: IAV infection, SARS-CoV-2 Omicron infection, and LPS-induced acute lung injury. The inclusion of the LPS model was intended to evaluate the therapeutic potential of IO@CA/P in a virus-independent inflammation, as it recapitulated the sterile inflammation and cytokine storm syndrome frequently observed in severe COVID-19 and influenza patients, where mortality was often driven by uncontrolled host immune responses rather than direct viral cytopathy [66–68]. In this clinically relevant context, IO@CA/P treatment significantly attenuated pulmonary inflammation, reduced histopathological scores, and suppressed serum levels of pro-inflammatory cytokines (TNF-α, IL-1β, IFN-α), showing better efficacy than free CA, SPIO alone, as well as several commercial agents including remdesivir, ribavirin, and dexamethasone. These results underscored the advantage of the integrated nanoplatform in mitigating the immunopathological cascade that underpinned life-threatening respiratory failure. Moreover, beyond acute intervention, IO@CA/P demonstrated marked efficacy in addressing post-infection sequelae, which represented a major unmet clinical need. Patients recovering from severe viral pneumonia frequently experienced persistent symptoms such as cough, chest tightness, and exercise intolerance, often attributed to pulmonary microthrombosis and fibrotic remodeling that remained refractory to conventional antiviral therapies [69–72]. Notably, in mice recovering from viral or LPS challenge, IO@CA/P treatment significantly reduced fibrotic markers (TGF-β1, PC-III) and thrombotic markers (D-dimer), indicating its potential to alleviate long-term complications and promote functional recovery. Unlike IAV or SARS-CoV-2 infection, LPS challenge does not involve viral glycoprotein cleavage or CTSB-mediated viral entry. Accordingly, the protective effects observed in this model, including reduced pulmonary inflammation, decreased histopathological scores, and suppression of pro-inflammatory cytokines, likely reflect the multifunctional properties of IO@CA/P beyond direct CTSB inhibition. Specifically, the SPIO core’s intrinsic SOD-like and CAT-like nanozyme activities contribute to ROS scavenging and attenuation of oxidative stress, while the released CA-074 may exert anti-inflammatory effects through inhibition of CTSB activity in immune cells such as macrophages, where CTSB has been implicated in NLRP3 inflammasome activation [59]. However, CTSB activity/localization and PANoptosis markers were not directly examined under LPS challenge, and CTSB-specific intervention controls were not included. Therefore, the protective effects observed in this model, such as reducing pulmonary inflammation, lowering histopathological scores, and suppressing cytokine release, should be interpreted as evidence of the broad anti-inflammatory and anti-fibrotic applicability of IO@CA/P, rather than as direct confirmation of the CTSB-PANoptosis mechanism. Future studies employing CTSB-knockout mice or selective chemical probes in sterile inflammation models would be required to fully delineate the relative contributions of CTSB inhibition versus nanozyme activity in this context. These findings highlighted the clinical potential of IO@CA/P for managing cytokine storm-driven lung injury, regardless of the initiating insult.

The safety profile of IO@CA/P was rigorously evaluated through acute and subacute toxicity studies. Even at a dose ten times higher than the therapeutic dose (500 mg/kg), IO@CA/P did not induce significant histological or serum biochemical changes. Minimal hemolysis (<5%) was observed in vitro, and although transient ALT elevation was noted during treatment, levels normalized after a 2-week recovery period. These results were consistent with previous reports demonstrating the biocompatibility of SPIO nanoparticles [73–75] and supported the clinical translatability of IO@CA/P.

Despite promising therapeutic efficacy and MRI theranostic capability, several translational hurdles should be acknowledged. Firstly, although the CD54-targeting strategy effectively enhanced pulmonary accumulation, a portion of IO@CA/P was also distributed to the liver and kidneys, reflecting physiological clearance pathways of systemically administered nanoparticles. This off-target accumulation may pose long-term safety concerns, although no overt histopathological changes were noted in these organs. Also, a transient elevation in serum ALT was observed during the treatment phase, which normalized after a 2-week recovery period. It highlighted the need for more detailed hepatotoxicity profiling in future preclinical studies, including assessment of bile duct function and inflammatory infiltrates. Secondly, while we demonstrated efficacy against multiple viral strains, the rapidly evolving nature of respiratory viruses necessitates continuous evaluation against emerging variants. Thirdly, the absence of genetic ablation models (e.g., CTSB-knockout) represented a limitation of the current study. Future investigations would prioritize the use of CTSB-deficient mice or orthogonal chemical probes to definitively exclude any residual off-target contributions. Finally, the LPS model, while informative for assessing anti-inflammatory effects, does not directly interrogate the CTSB-PANoptosis mechanism established in viral infection models. Future studies should investigate the long-term fate of IO@CA/P in the body, explore combination strategies with other host-directed or direct-acting antivirals, and conduct direct mechanistic studies in sterile inflammation models to fully delineate the contributions of CTSB inhibition versus nanozyme activity.

Conclusion

In conclusion, this study developed a CD54-targeted, magneto-active nanotheranostic platform (IO@CA/P) that integrates CTSB-specific inhibitor CA-074 with SPIO nanoparticles for precision therapy of viral pneumonia. IO@CA/P achieved 16.3-fold enhanced pulmonary accumulation under magnetic navigation, enabling MRI-monitored drug delivery and establishing a quantitative correlation between imaging signals and pneumonia severity. Mechanistically, IO@CA/P exerted broad-spectrum antiviral activity by inhibiting viral glycoprotein cleavage, while its SPIO core scavenged ROS and synergized with CA-074 to stabilize lysosomal membranes, thereby blocking CTSB-mediated PANoptosis and inflammatory cytokine release. In murine models of IAV and SARS-CoV-2 infection, IO@CA/P significantly reduced viral loads and pulmonary inflammation by blocking CTSB-mediated PANoptosis. Additionally, in an LPS-induced sterile lung injury model, IO@CA/P attenuated acute inflammation and post-injury fibrotic/thrombotic sequelae, demonstrating its broad therapeutic potential for managing cytokine storm-driven lung injury irrespective of the initial insult. This host-directed nanotheranostic strategy represented potential for combating emerging respiratory viruses and mitigating life-threatening immunopathology.

Author contributions

Jing Chen: Writing – review & editing, Writing – original draft, Methodology, Investigation, Data curation, and Funding acquisition. Song-Lin Jiang and Ying-Jun Lou: Methodology, Investigation, Data curation and Visualization. Zhi-Min Rao, Chun-Li Cai, Qian-Yun Zhang, Yan-Hui Qin, Fang Wu, Karim Malik, and Arman Chowdhury: Methodology, Investigation, and Data curation. Hua-Zhong Ying: Investigation, and Methodology. Chen-Huan Yu: Writing – review & editing, Supervision, Project administration, Data curation, Conceptualization, and Funding acquisition.

Funding

This research was supported by Zhejiang Provincial Natural Science Foundation (No. ZCLMS25H2703), Huadong Medicine Joint Funds of the Zhejiang Provincial Natural Science Foundation (No. LHDMZ22H300009) and National Natural Science Foundation of China (No. 82174272).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.