A transformative perspective on aggregation-induced emission bioimaging: illuminating the complex pathways of metal nanomaterial toxicology

Abstract

The ever-expanding universe of engineered metal nanomaterials (MNMs)—spanning therapeutics, diagnostics, energy, and consumer products—presents a dual imperative: to harness their transformative potential while rigorously ensuring their biological and environmental safety. Contemporary toxicological paradigms, reliant on conventional in vitro assays and post-mortem histology, often yield fragmented, static data that inadequately capture the dynamic, multi-scale journey of MNMs within living systems. This perspective articulates a visionary framework for integrating aggregation-induced emission (AIE)-based bioimaging as a cornerstone technology in nanotoxicology. We posit that the AIE methodology represents not merely an incremental improvement in imaging but a paradigm-shifting toolkit capable of rendering the invisible lifecycle of MNMs visible, quantifiable, and mechanistically interpretable. By transcending the limitations of traditional fluorophores, the AIE-based bioimaging method offers a unique synergy with MNMs, enabling real-time, spatiotemporally resolved, and multifunctional interrogation of nanomaterial fate, transformation, and biological consequences.

---

Environmental significance

The widespread application of engineered metal nanomaterials presents a critical challenge: closing the environmental risk gap between their transformative potential and their long-term biological impacts. This perspective advocates for the integration of aggregation-induced emission bioimaging as a cornerstone technology in environmental nanotoxicology. By enabling the direct, real-time, and mechanistic visualization of MNM fate, transformation, and bio-interactions within living systems and ecologically relevant models, AIE transcends the limitations of conventional, snapshot toxicology. This paradigm shift is essential for moving from reactive hazard assessment to a predictive, mechanism-based understanding. The synergy of AIE with artificial intelligence is poised to decode complex nano–bio interactions, accelerating the development of robust environmental safety guidelines and truly sustainable, safe-by-design nanomaterials.

Introduction

The widespread application of engineered metal nanomaterials (MNMs, with at least one dimension of 100 nm or less) necessitates a robust understanding of their environmental fate and biological impact.^1,2 However, their toxicology is dictated by a dynamic interplay of physicochemical properties—such as size, shape, and dissolution kinetics—that evolve in biological settings.^3–7 Traditional toxicological endpoints and standard analytical techniques, including bulk metal analysis and electron microscopy, fail to capture these critical dynamics.⁸ They lack the spatiotemporal resolution to visualize key processes like uptake, intracellular trafficking, and in situ transformation, creating a fundamental disconnect between an MNM's physical location and its biological effect. This gap hinders mechanistic understanding and predictive risk assessment, creating an urgent need for a method capable of visualizing MNM behavior in real-time within living systems.

While synchrotron-based X-ray fluorescence microscopy (XRF) offers exceptional elemental sensitivity and speciation capacity without exogenous labelling, it requires specialized facilities and provides limited temporal resolution for dynamic processes. Similarly, laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) enables quantitative mapping of elemental distributions with sub-micrometer resolution but is inherently destructive and incompatible with live specimens. Single-particle ICP-MS (sp-ICP-MS), despite its remarkable sensitivity for detecting individual NPs, cannot provide spatial information within intact biological structures. AIE-based imaging uniquely bridges these gaps by enabling real-time, longitudinal visualization in living systems, though we acknowledge that optimal mechanistic understanding will emerge from integrating these complementary approaches rather than relying on any single technique (Table 1). It directly counters the limitation of conventional fluorophores, which suffer from aggregation-caused quenching (ACQ) upon interaction with MNM surfaces.⁹ In contrast, AIE luminogens (AIEgens) exhibit a unique “light-up” fluorescence upon aggregation, a process driven by the restriction of intramolecular motion (RIM).^9,10 Since the discovery of AIE by Tang's group in 2001,¹⁰ the field has expanded rapidly with contributions from numerous research groups worldwide. For instance, Liu's group has pioneered responsive probes for apoptosis and enzymatic activity,¹¹ while Wang's laboratory has advanced environmental applications in aquatic organisms.^12–15 This collective effort has established AIE as a versatile platform spanning materials chemistry, biomedical imaging, and environmental toxicology. This inverted photophysics provides strategic advantages for MNM tracking: (1) a high signal-to-noise ratio via activation specifically at the nano–bio interface, (2) superior photostability and biocompatibility for longitudinal studies, and (3) minimal perturbation of the MNM's surface properties during labeling. Consequently, AIE transforms MNMs into bright, stable optical beacons, enabling the direct, high-fidelity visualization required to decipher their complex environmental toxicology.^16–18 The integration of AIE-based imaging transcends qualitative observation, evolving into a quantitative, multi-parametric platform that directly interrogates the fundamental mechanisms of MNM toxicity. Its applications form a cohesive investigative pipeline, each layer providing unprecedented insight into the nano–bio interface. This perspective examines the current state of bioimaging applications in nanotoxicology, with a particular focus on MNMs, and explores future directions for research and technological development.

Table 1 Comparative analysis of analytical methods in environmental nanotoxicology

Method	Spatial resolution	Temporal resolution	Detection limit	Live specimen compatible	Strengths	Limitations
AIE-based imaging	Diffraction-limited (200–300 nm)	Seconds to minutes	ng mL^−1–μg mL^−1	✓✓✓ (real-time)	Real-time dynamics; multiplexing; live imaging; cost-effective	Indirect mass quantification; photobleaching in extended studies
ICP-MS (bulk)	None	Minutes (endpoint only)	pg L^−1–ng L^−1	✗	Gold-standard quantification; multi-element	No spatial information; destructive; no speciation
sp-ICP-MS	None	Milliseconds (per event)	10–50 nm size detection	✗	Particle-by-particle analysis; size distribution	No spatial context; matrix interferences
LA-ICP-MS	5–100 μm	Minutes–hours	μg g^−1 (tissue sections)	✗	Elemental mapping; quantitative imaging	Destructive; limited sensitivity for light elements
XRF	50 nm–10 μm	Minutes–hours	μg g^−1–mg g^−1	✗ (fixed only)	Elemental speciation; non-destructive	Beamline access limited; radiation damage
Transmission electron microscopy	<1 nm (atomic)	Static (endpoint)	Single atom detection	✗	Ultra-high resolution; structural information	No live imaging; sampling bias; artifact-prone
Confocal	200–300 nm	Seconds–minutes	μg mL^−1	✓✓	Established method; multiplexing	ACQ interference; photobleaching

---

Decoding the dynamic lifecycle with molecular cinematography

AIE labelling transforms static snapshots into a dynamic molecular movie of the MNM journey.¹⁷ Beyond merely confirming uptake, it elucidates the precise kinetics and pathways governing cellular entry. By employing time-lapse confocal microscopy, we can distinguish between energy-dependent endocytosis (e.g., clathrin-mediated, caveolae-dependent) and passive diffusion, a distinction critical for predicting cell-type-specific vulnerability.¹⁹ Following internalization, the exceptional photostability of AIE-MNM conjugates allows for continuous tracking through the endolysosomal cascade, revealing trafficking kinetics, vesicular fusion/fission events, and the critical moment of endosomal escape—a key trigger for cytotoxicity (Fig. 1). Furthermore, long-term imaging uncovers ultimate fate: persistent lysosomal sequestration leading to frustrated phagocytosis and inflammation, exocytosis pathways contributing to transcellular transport, or rare but consequential nuclear envelope association. This temporal and spatial resolution provides a direct visual link between a material's physical journey and its kinetic toxicological profile.

Fig. 1 Schematic of the cellular processing and toxicity mechanism of MNMs. Internalized MNMs undergo lysosomal degradation, resulting in the release of metal ions. These ions can then translocate to critical cellular compartments, leading to adverse effects and toxicity.¹⁹ Adapted with permission from ref. 19, copyright (2021) Royal Society of Chemistry.

In situ direct visualization of biotransformation MNMs: resolving the speciation dilemma

This represents the most transformative application of AIE for MNM toxicology. The central debate—particle-specific effects versus dissolved ion toxicity—requires observing speciation dilemma changes where they happen (the fundamental uncertainty in nanotoxicology regarding whether observed toxicity arises from the nanoparticle itself, from released ions, or from synergistic interactions between both forms—a distinction critical for mechanism-based risk assessment but difficult to resolve with conventional methods).^20,21 Recently, a novel AIEgen was developed, which can specifically convert silver ions (Ag⁺) into silver nanoparticles (AgNPs, also named as AACSNs), with AIEgens coated on the surface of the NPs (Fig. 2a).⁹ By coupling with the AIE-based Ag⁺ detection probe TEZ-TPE-1, a dual sensing system was developed, enabling simultaneous monitoring of AgNPs and their released Ag⁺ (Fig. 2b).²² This allows for the real-time, spatial mapping of dissolution kinetics within a single cell.²² One can visually correlate the zone of intense ion release with immediate local damage, such as mitochondrial fragmentation or glutathione depletion, while areas with intact particle fluorescence may show distinct, physical-stress phenotypes.¹⁹ This capability moves the field beyond bulk measurements of ion concentration, enabling the direct testing of the “Trojan horse” hypothesis (the mechanism whereby nanoparticles serve as efficient delivery vehicles that transport toxic metal ions into cells, bypassing normal cellular defenses and releasing the ionic cargo intracellularly following particle dissolution) and other speciation-dependent mechanisms with spatial precision.

Fig. 2 Direct visualization of biotransformation MNMs. (a) The redox-active AIEgen features an electron donor–acceptor (D–A) structure. AACSNs were synthesized via a reduction reaction followed by AIEgen self-assembly. Fluorescence imaging of 4T1 tumor-bearing mice after intratumoral injection of AACSNs.⁹ (b) AIE-based Ag⁺ detection probe TEZ-TPE-1. Imaging of AgNPs and Ag⁺ distribution in algal cell organelles.²² Adapted with permission from ref. 9 and 22, copyright (2018 and 2025) American Chemical Society.

The molecular design strategy for ion-responsive AIE probes extends well beyond silver, with recent advances enabling detection of a broad spectrum of environmentally and toxicologically relevant metal species. The general design principle involves incorporating metal-coordinating moieties into AIEgen scaffolds, where ion binding triggers conformational restriction and fluorescence activation. Representative examples include: (i) copper ions (Cu²⁺/Cu⁺)—probes based on picolinate or thiosemicarbazide recognition units enable visualization of copper nanoparticle dissolution and Cu-induced neurotoxicity;¹² (ii) zinc ions (Zn²⁺)—dipicolylamine-functionalized AIEgens allow real-time tracking of ZnO nanoparticle dissolution and intracellular zinc homeostasis disruption;⁴ (iii) mercury ions (Hg²⁺)—thioether- or thymine-rich coordination sites enable ultrasensitive detection of mercury nanoparticle transformation and methylmercury formation.²³

Quantitative analysis of the in vivo biodistribution and barrier translocation of MNPs

The development of AIEgens with emission in the near-infrared (NIR-I, 700–900 nm) and second NIR (NIR-II, 1000–1700 nm) windows has been pivotal for advancing whole-organism imaging in nanotoxicology. These wavelengths significantly reduce light scattering, tissue absorption, and autofluorescence, conferring deep-tissue penetration and high spatial resolution for longitudinal studies in intact living models (such as zebrafish).^24,25 The versatility of the AIE platform is further demonstrated by the design of multifunctional probes, including AIE–gadolinium complexes for dual-modal fluorescence/MRI imaging of vascular integrity and AIE–gold hybrid nanostructures that provide correlative fluorescence and computed tomography (CT) signals for sensitive tracking.^9,26 Consequently, AIE-based in vivo imaging delivers non-invasive, real-time data on biodistribution, transcending traditional terminal analyses (Fig. 3a). This capability enables the continuous quantification of MNM accumulation in primary clearance organs (e.g., liver, spleen, and kidneys), the direct visualization of translocation across critical biological barriers (including the blood–brain and placental barriers), and the dynamic assessment of clearance pathways (Fig. 3b).²⁵ It is worth noting that quantitative assessment of blood–brain barrier (BBB) translocation requires rigorous calibration and normalization protocols. Fluorescence intensity measurements in brain tissue must account for: (i) wavelength-dependent light attenuation through cranial structures, (ii) background autofluorescence subtraction using region-of-interest analysis in control animals, and (iii) depth-dependent signal attenuation correction via standardized phantom-based calibration curves. For absolute quantification, we recommend orthogonal validation using: (a) ex vivo brain homogenate analysis by ICP-MS following imaging to correlate fluorescence with mass concentration, (b) tissue clearing techniques that enable volumetric fluorescence quantification while maintaining spatial integrity, and (c) standardized reference materials with known fluorophore concentrations implanted at varying brain depths to establish attenuation correction factors. The integration of machine learning algorithms for automated segmentation of vascular versus parenchymal compartments further enables precise discrimination between true barrier translocation and signals originating from within cerebral blood vessels. The resulting longitudinal datasets yield essential pharmacokinetic parameters—such as tissue residence times and accumulation kinetics—within an ecologically relevant context, thereby effectively bridging the mechanistic gap between in vitro cellular responses and integrated systemic organismal toxicity.

Fig. 3 Quantitative analysis of in vivo biodistribution and barrier translocation of MNPs using AIE-based imaging. (a) Schematic illustration of NIR-II AIEgen design for deep-tissue imaging, enabling real-time visualization of MNM distribution in intact organisms. (b) Representative longitudinal images of MNM accumulation in primary clearance organs (liver and spleen) and translocation across the blood–brain barrier in mouse models. (c) Quantification of fluorescence signals in brain parenchyma versus cerebral vasculature, demonstrating barrier penetration kinetics. (d) Correlation of in vivo fluorescence intensity with ex vivo ICP-MS validation (r² = 0.94), establishing quantitative relationship between the optical signal and MNM mass.²⁵ Adapted with permission from ref. 25, copyright (2022) American Chemical Society.

The versatility of AIE-based imaging has been demonstrated across a phylogenetically diverse range of ecologically relevant organisms. Beyond zebrafish, successful applications now include: (i) terrestrial model organisms such as Caenorhabditis elegans (nematodes) for soil ecotoxicity assessment, enabling real-time visualization of MNM translocation through the intestinal barrier and accumulation in germline tissues;²⁷ (ii) aquatic invertebrates including Daphnia magna (water fleas), where AIE labeling has revealed trophic transfer dynamics and maternal deposition in developing embryos;^16,18 (iii) marine bivalves (oysters), facilitating investigation of hemocyte-mediated nanoparticle trafficking and tissue-specific accumulation patterns relevant to benthic ecosystem health.²⁸

Multiplexed mechanistic interrogation: sstablishing spatial causality

The modular design of AIEgens enables the highest level of mechanistic inquiry: multiplexed, correlative imaging. By conjugating different responsive units to AIE cores, a single probe or a suite of probes can simultaneously report on multiple, interrelated parameters. For example, an AIE-MNM conjugate can be co-localized with microenvironment sensors: probes for lysosomal pH (which often alkalinizes upon dysfunction), or specific reactive oxygen/nitrogen species.²⁹ Activity-based probes: sensors for enzymatic activity, such as caspase-3/7 (apoptosis), cathepsin B (lysosomal membrane permeabilization), or LC3-II (autophagic flux).³⁰ This multi-parametric approach allows researchers to establish direct, spatially resolved cause-and-effect relationships. This transformation enables subjective correlation to be converted into visual evidence of causality, such as demonstrating that MNM-induced ROS bursts preceded and are spatially coincident with subsequent mitochondrial depolarization and the initiation of apoptosis in the same subcellular region. This capability is paramount for constructing detailed, mechanistic pathways of nanotoxicity.

Current challenges and methodological considerations

Despite its transformative potential, the integration of AIE bioimaging into standardized toxicology requires addressing several key challenges that span technical, environmental, and data-related domains. Quantitative calibration and AI-ready data: translating fluorescence intensity into quantitative measures of MNM mass remains complex and necessitates orthogonal validation. Furthermore, to harness the power of artificial intelligence and machine learning, data must be generated and curated in standardized, annotated formats with appropriate metadata. This requires deliberate planning of imaging experiments with quantitative validation embedded in the experimental design rather than as an afterthought. Probe design for environmental relevance: many existing AIEgens have been optimized for biomedical applications, with performance characteristics suited to physiological conditions (pH 7.4, 37 °C, low ionic strength). Their stability and sensitivity must be systematically re-engineered for diverse environmental conditions—including variable pH, salinity, natural organic matter, and sunlight exposure—and validated across a broader spectrum of ecotoxicologically relevant model organisms spanning multiple trophic levels and environmental compartments. Standardization and reproducibility: robust, universally applicable protocols for labeling diverse MNM types are urgently needed to enable reliable cross-study comparisons and facilitate the construction of large, shareable datasets suitable for AI-driven meta-analyses. The field would benefit from community-agreed guidelines for probe characterization, imaging parameters, and data reporting.

Environmental matrix effects on AIE probe performance

The translation of AIE-based imaging from controlled laboratory conditions to complex environmental media presents significant challenges that require systematic investigation. The physicochemical complexity of natural systems can profoundly influence probe behavior, MNM fate, and the interactions between them. Key considerations include: (i) natural organic matter (NOM) interference. Humic and fulvic acids, ubiquitous in aquatic systems, can interact with AIE probes through hydrophobic partitioning or electrostatic interactions. These interactions may lead to: (a) false-positive fluorescence activation via NOM-induced probe aggregation; (b) competitive binding that reduces probe availability for target analyte detection; or (c) fluorescence quenching through electron transfer mechanisms. Strategies to mitigate NOM interference include developing probes with emission in the NIR-II region (>1000 nm) where NOM autofluorescence is minimal, and incorporating shielding moieties (e.g., polyethylene glycol) that reduce non-specific binding. (ii) Colloidal and particulate interactions. Soil colloids, suspended sediments, and mineral particles can significantly alter probe behavior through multiple mechanisms: (a) adsorption of AIE probes, altering their effective concentration and bioavailability; (b) induction of probe aggregation through surface-mediated confinement, potentially generating false signals; and (c) competitive binding with MNMs, modifying their surface chemistry and subsequent AIE labeling efficiency. Pre-incubation controls and centrifugation-based separation protocols are essential to distinguish particles associated from freely dissolved probe populations. (iii) Ionic strength and pH effects. Environmental variability in salinity (ranging from freshwater to marine systems) and pH (spanning acid mine drainage to alkaline soils) dramatically influences multiple interrelated parameters: (a) probe solubility and aggregation propensity; (b) metal ion speciation and probe-metal binding affinities; and (c) MNM dissolution kinetics and surface charge, which in turn affect AIE labeling stability. Probe design must therefore incorporate pH-insensitive fluorophores (e.g., tetraphenylethylene derivatives with pK_a-extended recognition units) and ionic strength-tolerant architectures. (iv) Sediment pore water complexity. The unique chemical environment of sediment pore waters—characterized by low oxygen, high sulfide, and elevated metal concentrations—poses specific challenges including probe degradation via reductive pathways, competitive metal sulfide formation that sequesters target ions, and biofilm interactions that may internalize or degrade probes. (v) Photostability under environmental illumination. Natural sunlight exposure can accelerate probe photobleaching or induce phototransformation, generating unknown degradation products with potentially different properties and toxicological profiles. Accelerated aging studies under simulated solar radiation are essential to establish probe half-lives under environmentally relevant conditions and to characterize phototransformation products. (vi) Tiered validation approach. Addressing these matrix effects requires a systematic, hierarchical validation strategy: (1) controlled laboratory characterization of probe performance in simplified matrices to establish baseline behavior; (2) systematic addition of individual matrix components (NOM, specific colloids, and salts) to identify dominant interference mechanisms; (3) validation in model environmental media (standardized natural water and reference soils) that approximate real-world complexity; and (4) field validation with spike-recovery experiments in authentic environmental samples to confirm performance under true environmental conditions.

Quantitative calibration strategies

Establishing robust quantitative relationships between fluorescence intensity and MNM mass requires multi-tiered calibration approaches that account for the inherent complexities of biological and environmental samples. (i) In vitro calibration standards. Prepare well-characterized AIE-labeled MNM suspensions at precisely known concentrations (validated by ICP-MS). Image these standards under identical conditions to experimental samples (same objective, laser power, gain settings, pinhole aperture) to generate standard curves relating integrated fluorescence density to mass concentration. Account for optical path differences and instrument variability using standardized fluorescent beads as intensity references that enable normalization across imaging sessions and instruments. (ii) Orthogonal validation methods. Multiple complementary techniques can provide independent validation and cross-calibration: radiolabeled MNMs enable direct correlation of radioactivity (mass) with fluorescence in the same specimens, providing calibration curves that inherently account for tissue-specific optical properties and depth-dependent attenuation. sp-ICP-MS analysis of parallel samples quantifies particle number concentration and size distribution, enabling fluorescence-to-particle conversion factors and validation that signal intensity scales linearly with particle number. Correlative light and electron microscopy (CLEM) on identical regions validates that fluorescence signals correspond to identified nanoparticle clusters, confirming specificity and ruling out artefactual signals. Laser ablation ICP-MS on tissue sections adjacent to those used for imaging provides elemental maps that can be registered to fluorescence images for pixel-by-pixel correlation, enabling spatially resolved validation of fluorescence–mass relationships. (iii) Normalization strategies. To correct for sample-to-sample variability in optical properties: implement ratiometric imaging using reference fluorophores with known concentrations (e.g., co-injected fluorescent dextrans and tissue-autofluorescence channels) to normalize analyte signals against internal standards. Use tissue transparency measurements (optical coherence tomography, two-photon penetration depth analysis) to apply depth-dependent attenuation corrections. Develop machine learning models trained on paired fluorescence-ICP-MS datasets to predict MNM mass from multiparametric fluorescence features (intensity, spectral shape, lifetime) across diverse tissue types and experimental conditions.

Standardization and reproducibility: a minimum reporting checklist

To enable cross-study comparability and facilitate data sharing for AI/ML applications, we propose a minimum reporting checklist for AIE-based nanotoxicology studies. Adoption of these guidelines by the research community would substantially accelerate progress toward predictive, mechanism-based toxicology. Probe characterization: chemical structure and purity (HPLC/MS verification, ≥95% purity recommended). Photophysical properties: absorption/emission maxima, quantum yield (in both solution and aggregated states), and molar extinction coefficient. Aggregation behavior: critical aggregation concentration and hydrodynamic diameter in relevant media. Stability: photostability profile (half-life under imaging conditions), pH stability range, thermal stability, and storage conditions. Cytotoxicity: IC₅₀ in relevant cell lines with probe-only controls. Labeling efficiency: number of AIEgen molecules per MNM (quantified by UV-vis or fluorometric assay) and labeling stability over time in relevant media. MNM characterization: core composition, primary size (TEM), hydrodynamic size (DLS), and surface charge (zeta potential) before and after labeling. Surface chemistry: coating material and density, conformation. Dissolution kinetics in relevant media (with/without probe labeling to assess potential modification). Colloidal stability in exposure media (agglomeration state over experimental timeframe). Imaging parameters: microscope specifications: type (confocal, widefield, light sheet), objective (magnification, numerical aperture), and detectors (PMT, GaAsP, CCD, sCMOS). Acquisition settings: excitation wavelength/power, emission filters, pixel dimensions, bit depth, scan speed, and line/frame averaging. Image processing: all processing steps (background subtraction, deconvolution, filtering) with software and parameters specified. Calibration: details of any intensity calibration (bead standards, reference materials, flat-field correction). Experimental conditions: composition of exposure media (pH, ionic strength, NOM content if relevant). Exposure concentration(s) (validated by ICP-MS), duration, and regimen (acute/chronic). Model system: species, strain, developmental stage, culture conditions, and sample size (with justification). Replicates: biological and technical replicates clearly distinguished, with statistical methods specified. Data availability: raw images deposited in public repositories (e.g., BioImage Archive, Zenodo) with standardized metadata following community guidelines (e.g., REMBI). Analysis scripts/code made available for reproducibility (e.g., GitHub, Code Ocean).

Probe safety and interference assessment

A critical yet often overlooked consideration in AIE-based nanotoxicology is the potential toxicological contribution of the probes themselves and their degradation products. Rigorous safety assessment must precede or accompany any imaging study to ensure that observed biological effects reflect MNM toxicity rather than probe artefacts. Required safety assessments: (i) cytotoxicity evaluation: systematic evaluation of probe cytotoxicity across relevant cell lines and organisms using standardized viability assays (MTT, WST-1, LDH release, apoptosis markers) at concentrations equivalent to those used in imaging studies. Both acute (24–48 h) and chronic exposure should be assessed where relevant. (ii) Degradation kinetics and products: investigation of probe degradation kinetics under relevant physiological and environmental conditions (pH gradients, enzymatic activity, photostability) with identification of degradation products via LC-MS/MS. This is particularly important for probes containing ester, amide, or other hydrolytically labile linkages. (iii) Fragment toxicity assessment: evaluation of whether probe fragments or released ions (e.g., tetraphenylethylene derivatives, gadolinium from MRI-capable probes, recognition moieties) independently induce toxicity or modulate MNM behavior through synergistic or antagonistic interactions. (iv) Probe-only controls: essential control experiments using probe-only exposures (identical concentration, duration, and conditions) to distinguish MNM-specific effects from potential probe artefacts. These controls should be included in every experimental design. (v) Metal homeostasis interference: assessment of probe interference with normal metal homeostasis—for example, metal-chelating moieties designed for ion detection may transiently alter bioavailable metal concentrations, potentially confounding interpretation of metal-specific toxicity. (vi) Long-term fate: long-term monitoring in chronic exposure scenarios to assess bioaccumulation potential of intact probes or their persistent degradation products, particularly for probes with high lipophilicity or slow clearance.

Current evidence and recommendations

Recent studies have demonstrated that well-designed AIEgens exhibit minimal toxicity at imaging-relevant concentrations, with LD₅₀ values typically exceeding 100 mg kg⁻¹ in vertebrate models and rapid clearance primarily through hepatobiliary and renal pathways. Nevertheless, these findings cannot be generalized across all probe designs. We therefore advocate for standardized reporting of probe characterization data—including purity, stability, cytotoxicity, and degradation product profiles—as essential components of any AIE-based nanotoxicology study.

pH-dependent stability of AIE labeling

The intracellular journey of MNMs exposes AIE probes to dramatic pH variations—from the neutral extracellular environment (pH 7.4) to early endosomes (pH 6.5–6.0), late endosomes (pH 5.5–5.0), and lysosomes (pH 4.5–4.0). These pH gradients can significantly impact probe performance through multiple mechanisms, and understanding these effects is essential for accurate interpretation of imaging data. Mechanisms of pH-dependent effects: (i) protonation/deprotonation effects. Ionizable functional groups (carboxylates, amines, phenols) commonly incorporated into AIEgens for metal recognition undergo pH-dependent charge changes that alter probe solubility, aggregation propensity, and fluorescence quantum yield. For example, carboxylate-containing probes (negatively charged at neutral pH) may exhibit protonation-mediated charge neutralization at lysosomal pH, leading to reduced electrostatic repulsion, enhanced aggregation, and potentially increased fluorescence—independent of target analyte binding. (ii) Hydrolytic stability. Ester- or amide-linked probe-MNM conjugates may undergo acid-catalyzed hydrolysis in lysosomal compartments, potentially releasing the fluorophore and compromising tracking fidelity. This can create the false impression of MNM dissociation or redistribution when in fact the probe has been cleaved while the MNM remains localized. Control experiments using pH-insensitive linkers (e.g., triazole from click chemistry, polyethylene glycol spacers, and carbamate linkages) are essential to distinguish true MNM trafficking from free probe redistribution. (iii) pH-responsive spectral shifts. Some AIEgens exhibit pH-dependent emission maxima or intensity, which can be exploited for ratiometric pH sensing but also confound interpretation if not properly controlled. Characterizing probe photophysics across the relevant pH range (4.0–7.4) is essential experimental groundwork before biological application. (iv) Acidic organelle accumulation (ion trapping). Weakly basic AIEgens (pK_a > 7) may exhibit ion trapping in acidic compartments, where protonation traps the molecule within lysosomes independent of MNM localization. This can lead to apparent co-localization that reflects probe accumulation rather than true MNM trafficking. Co-localization studies with lysosomal trackers and careful quenching controls (e.g., trypan blue quenching of extracellular signals) are necessary to validate signal specificity. Mitigation strategies: pre-screen AIEgen candidates for pH stability across the physiological range before committing to biological studies. Employ pH-insensitive AIE scaffolds (e.g., tetraphenylethylene without ionizable groups) for tracking applications where pH varies. Implement dual-labeling strategies (AIE + pH-insensitive fluorophore on the same MNM) to distinguish true particle signals from pH artefacts. Conduct control experiments with the free probe (no MNM) at identical concentrations to assess pH-dependent redistribution patterns. Use ratiometric imaging with pH-sensing AIEgens to simultaneously track the particle location and local pH environment, enabling correlation of location with chemical conditions.

Future perspectives and concluding remarks

To fully realize the potential of AIE bioimaging in nanotoxicology, future endeavors should focus on four interconnected priorities: 1. Developing environmentally tuned AIEgens. Designing probes with enhanced stability and selectivity for conditions in soil, aquatic, and sediment systems, moving beyond the biomedical focus that has dominated the field to date. This includes systematic optimization for variable pH, salinity, natural organic matter, and sunlight exposure, as well as validation across diverse ecotoxicological model organisms. 2. Integration with multimodal and AI platforms. Correlating AIE data with complementary techniques—synchrotron-based XRF for elemental mapping, TEM for ultrastructural context, LA-ICP-MS for quantitative distribution—and feeding these combined datasets into AI-driven analysis pipelines to fuse dynamic functional imaging with elemental and structural data. This integration will enable a more complete understanding of MNM fate and effects than any single technique or traditional methods can provide (Fig. 4). 3. AI-driven high-content screening. Implementing AIE-based imaging in automated systems coupled with computer vision algorithms to rapidly profile the bio-interaction kinetics of large MNM libraries, accelerating the safe-by-design cycle. The combination of AIE's molecular specificity with AI's pattern recognition capabilities offers unprecedented throughput for nanotoxicological screening. 4. Building predictive in silico models. Using curated AIE-derived datasets to train machine learning models that can predict the environmental fate and biological pathways of MNMs based on their core physicochemical properties, moving toward a genuinely predictive computational toxicology framework.

Fig. 4 Paradigm shift in nanotoxicology assessment: Traditional versus AIE-enabled approaches. (a) Conventional workflow relies on endpoint measurements (bulk ICP-MS, histology, viability assays) that provide static snapshots of MNM distribution and effects, with temporal gaps that obscure dynamic processes and mechanistic linkages. (b) AIE-enabled approach enables continuous, real-time visualization of the complete MNM lifecycle—from uptake and trafficking to biotransformation and clearance—while simultaneously reporting on multiple toxicity pathways through multiplexed responsive probes. This dynamic dataset supports mechanistic adverse outcome pathway (AOP) construction and provides training data for AI-driven predictive models, fundamentally transforming risk assessment from descriptive to predictive science.

The AIE-AI synergy in practice

The convergence of AIE imaging with artificial intelligence creates transformative opportunities for quantitative analysis and predictive modeling. Specific applications include: (i) automated image segmentation and feature extraction. Deep learning architectures (e.g., U-Net and Mask R-CNN) trained on AIE imaging datasets enable automated identification and quantification of subcellular structures (lysosomes, mitochondria, nuclei) and their association with MNM signals. This transforms subjective visual assessment into objective, high-throughput morphometric analysis. (ii) Spatiotemporal correlation analysis. Recurrent neural networks (RNNs) and long short-term memory (LSTM) networks can analyze time-lapse AIE image sequences to identify temporal patterns linking particle trafficking kinetics (e.g., endosomal escape timing) with subsequent cellular responses (apoptosis initiation), extracting predictive features that precede observable toxicity. (iii) Multiplexed signal deconvolution. When multiple AIE probes with overlapping emission spectra are used, unsupervised machine learning algorithms (e.g., non-negative matrix factorization and independent component analysis) can computationally separate mixed signals, enabling higher-order multiplexing beyond spectral limitations. (iv) Toxicity pathway prediction. Random forest and gradient boosting models trained on AIE-derived features (intracellular distribution patterns, dissolution rates, organelle colocalization) can predict toxicological outcomes with high accuracy. For example, recent work has demonstrated that AIE-based imaging features predict mitochondrial membrane potential disruption with >90% accuracy, enabling early toxicity screening without additional staining. (v) Generative modeling for safe-by-design. Variational autoencoders trained on structure–property–activity relationships derived from AIE imaging data can generate candidate MNM designs with predicted favorable safety profiles, accelerating rational materials development. (vi) Quantitative structure–property relationship (QSPR) enhancement. AIE-derived dynamic parameters (e.g., uptake rate constants, dissolution half-lives, and intracellular residence times) serve as quantitative descriptors that substantially improve QSPR models for predicting environmental fate and toxicity across diverse MNM libraries.

Addressing environmentally relevant exposure scenarios

Moving beyond acute, high-dose laboratory studies requires deliberate optimization of AIE-based approaches for environmentally realistic conditions: (i) sensitivity enhancement for low-dose tracking. Signal amplification through enzymatic cascade systems that multiply fluorescence output per binding event, enabling detection of MNM concentrations in the pg mL⁻¹ range. Development of AIEgens with ultra-high quantum yields (>50% in aggregated state) through rational molecular design incorporating multiple rotor units and heavy atom effects. Implementation of advanced optical configurations including light-sheet microscopy for reduced photobleaching and increased signal integration times, and two-photon excitation for deeper tissue penetration with reduced background. Integration of plasmonic enhancement using gold nanoparticle substrates that amplify AIE fluorescence through localized surface plasmon resonance (LSPR) effects. (ii) Long-term tracking protocols. Establishing probe photostability profiles under chronic illumination conditions, with systematic evaluation of photobleaching kinetics and development of photostable AIEgens (e.g., triphenylamine-based structures with extended π-conjugation). Implementing ratiometric imaging strategies where analyte-responsive signals are normalized to stable internal references, enabling quantitative comparisons across extended time courses. Developing pulse-chase experimental designs where labeled MNMs are introduced and tracked over weeks to months in appropriate model systems (e.g., long-lived organisms and mesocosms). (iii) Capturing aged MNM dynamics. Surface chemistry-independent labeling strategies that maintain fluorescence despite particle aging—recent advances include encapsulation approaches where AIEgens are embedded within protective silica shells that remain associated even as the core MNM undergoes transformation. Multiparametric probes that simultaneously report particle identity (core fluorescence) and aging state (surface-responsive emission), enabling correlation of transformation status with biological outcomes. Time-resolved imaging coupled with machine learning classification to identify characteristic aging signatures (e.g., fluorescence polarization changes indicating aggregation and spectral shifts reflecting surface oxidation). (iv) Chronic toxicity correlation. Longitudinal imaging in organismal models with extended lifespans (e.g., zebrafish aging studies and C. elegans lifespan assays) to correlate cumulative MNM body burden with progressive pathology. Integration with transcriptomic and metabolomic profiling at multiple time points to establish temporal relationships between particle distribution and molecular responses. Development of ex vivo imaging protocols for terminal time points that provide high-resolution spatial context complementary to longitudinal in vivo data.

Regulatory applications and environmental risk assessment

AIE-based approaches hold significant potential to address critical regulatory gaps in environmental MNM risk assessment. Current regulatory frameworks (e.g., OECD testing guidelines) rely primarily on standardized ecotoxicity tests that provide endpoint-based assessments but offer limited mechanistic insight. AIE imaging can directly inform regulatory decision-making through: (i) supporting adverse outcome pathway (AOP) development. By establishing direct visual links between molecular initiating events (e.g., particle dissolution and ROS generation) and adverse outcomes at organismal and population levels, AIE data provide essential mechanistic evidence for AOP construction—a priority under the OECD AOP development program. (ii) Refining species sensitivity distribution models. AIE-enabled comparative toxicokinetics across phylogenetically diverse species (as detailed above) generates quantitative data on species-specific accumulation and clearance parameters, enabling more robust extrapolation in ecological risk assessment. (iii) Validating alternative testing strategies. AIE imaging in high-throughput in vitro systems (e.g., gill cell lines, gut organoids, and zebrafish embryos) can validate these alternatives against in vivo outcomes, supporting the 3R principles and reducing animal testing requirements in regulatory submissions. (iv) Informing safe-by-design guidelines. By revealing structure–activity relationships governing MNM fate and effects (e.g., size-dependent barrier translocation, shape-specific intracellular trafficking, and surface chemistry-dependent dissolution), AIE-derived mechanistic insights enable evidence-based recommendations for designing inherently safer nanomaterials. (v) Environmental monitoring applications. Portable AIE-based sensing platforms (e.g., paper-based AIE probes, smartphone-integrated detection, and field-deployable fluorometers) are under development for real-time monitoring of MNM contamination in water, soil, and biota samples, potentially supporting compliance monitoring under environmental quality standards.

Integrated multimethod approach

Rather than positioning AIE as a replacement for established techniques, we advocate for an integrated multimethod approach that leverages the unique strengths of each technology. For example, an optimal experimental workflow might combine: (i) AIE-based live imaging to identify critical time points and dynamic processes; (ii) ICP-MS for absolute quantification of total metal content; (iii) sp-ICP-MS for particle size distribution analysis; (iv) synchrotron XRF/XANES for high-resolution elemental mapping and speciation confirmation in fixed tissues; and (v) TEM for ultrastructural context. This integrated strategy enables cross-validation while providing complementary information that no single technique can deliver independently.

Implementation pathways for environmental laboratories

For environmental toxicology laboratories considering adoption of AIE-based approaches, we propose tiered implementation pathways that accommodate varying levels of resources and expertise. Entry-level implementation (minimal specialized equipment): utilize commercially available AIE probes (increasingly available from major suppliers such as Sigma-Aldrich, Thermo Fisher, and specialized AIE-focused vendors) for targeted applications. Adapt existing fluorescence microscopes (widefield or basic confocal) with appropriate filter sets matched to AIEgen emission maxima. Begin with simple end-point imaging in fixed specimens before advancing to live-cell studies. Collaborate with AIE-specialized research groups for probe synthesis and characterization while building in-house capacity. Intermediate implementation (moderate investment): establish in-house probe characterization capabilities (spectrofluorometer, dynamic light scattering, HPLC). Develop standardized labeling protocols for commonly studied MNMs (Ag, Au, TiO₂, ZnO, CuO). Implement live-cell imaging with environmental control for dynamic studies. Integrate orthogonal validation (ICP-MS) into experimental workflows. Participate in interlaboratory comparisons to validate reproducibility. Advanced implementation (comprehensive capabilities): invest in state-of-the-art imaging platforms. Develop custom AIEgen libraries for specialized applications. Establish high-content screening infrastructure with automated image analysis. Implement AI/ML pipelines for quantitative feature extraction and predictive modeling. Contribute to public databases and standardization efforts. Practical considerations for environmental researchers: start with well-characterized, commercially available AIEgens to establish protocols before advancing to custom synthesis. Validate probe performance in relevant environmental matrices before proceeding to organismal studies. Include rigorous controls: probe-only exposures, unlabeled MNM comparisons, quenching tests, and autofluorescence controls. Collaborate with imaging core facilities at academic institutions for access to specialized equipment and expertise. Leverage online resources and training workshops. Engage with standardization initiatives and contribute to method development. This tiered approach enables laboratories at all resource levels to incorporate AIE-based methods into their toxicology pipelines, with clear pathways for capability advancement as expertise and funding grow.

Conclusion

AIE bioimaging represents a profound methodological leap forward for the toxicology of metal nanomaterials. By providing direct, dynamic, and mechanistically informative visual data, it bridges critical gaps between observed biological effects and underlying nano–bio interactions—gaps that conventional endpoint assays cannot address. The ability to visualize uptake kinetics, intracellular trafficking, biotransformation, and elimination in real time transforms our understanding of how physicochemical properties translate to biological outcomes. However, the ultimate impact of this technology hinges on our ability to intelligently integrate its rich data outputs. The convergence of AIE with artificial intelligence marks the next frontier, creating a powerful cycle of observation, prediction, and validation. AIE generates high-dimensional, spatiotemporally resolved data that capture the complexity of nano–bio interactions; AI extracts patterns, builds predictive models, and guides experimental design. This synergy transforms imaging from descriptive observation to predictive science. As this interdisciplinary field matures—spanning materials chemistry, optical imaging, molecular biology, data science, and environmental toxicology—the AIE-AI synergy is poised to become a cornerstone technology for nanosafety assessment. It will be instrumental in building a predictive, mechanism-based understanding of nanomaterial fate and effects across diverse organisms and environmental conditions, ultimately informing robust, evidence-based environmental risk assessment and regulatory policy. The path forward requires continued innovation in probe design, standardization of methods, open data sharing, and cross-disciplinary collaboration, but the potential rewards, in terms of both scientific understanding and societal benefit, are immense.

Conflicts of interest

There are no conflicts to declare.

Data availability

This work is a perspective article and does not report new primary data. No datasets were generated or analyzed specifically for this manuscript. The discussions and frameworks presented are based on and refer to data available in the cited published literature. For the original data underlying the referenced studies, readers are directed to the respective publications listed in the references, where applicable data availability statements can be found.

Acknowledgements

We thank the reviewers for their helpful and constructive comments. This study was supported by grants from the Natural Science Foundation of China (No. 42477250 and No. 42207319), and from the National Key Research and Development Program of China under Grant No. 2024YFC3714600, and the Natural Science Foundation of Hubei Province (2025AFA048).

References

1. A. Nel, T. Xia, L. Mädler and N. Li, Toxic potential of materials at the nanolevel, Science, 2006, 311(5761), 622–627.
2. G. Oberdörster, E. Oberdörster and J. Oberdörster, Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles, Environ. Health Perspect., 2005, 113(7), 823–839.
3. Z. Wang, L. Zhang, J. Zhao and B. Xing, Environmental processes and toxicity of metallic nanoparticles in aquatic systems as affected by natural organic matter, Environ. Sci.: Nano, 2016, 3(2), 240–255.
4. B. Molleman and T. Hiemstra, Time, pH, and size dependency of silver nanoparticle dissolution: the road to equilibrium, Environ. Sci.: Nano, 2017, 4(6), 1314–1327.
5. J. Zhang, W. Guo, Q. Li, Z. Wang and S. Liu, The effects and the potential mechanism of environmental transformation of metal nanoparticles on their toxicity in organisms, Environ. Sci.: Nano, 2018, 5(11), 2482–2499.
6. T. Y. Wong, N. Yan, K. K. L. Kwan, Y. Pan, J. Liu, Y. Xiao, L. Wu and H. Lam, Comparative proteomic analysis reveals the different hepatotoxic mechanisms of human hepatocytes exposed to silver nanoparticles, J. Hazard. Mater., 2022, 130599.
7. T. Y. Wong, Y. Wang, K. K. L. Kwan, Y. Pan, A. K. L. Lai, S. Chen, Y. Xiao, K. Zhou, L. Wu, S. Huo, N. Yan and H. Lam, Exploring the Potentials of Silver Nanoparticles in Overcoming Cisplatin Resistance in Lung Adenocarcinoma: Insights from Proteomic and Xenograft Mice Studies, ACS Nano, 2025, 19(39), 34708–34723.
8. N. Yan, A. Parodi, F. Dang and J. Shi, Shedding light on the invisible: aggregation-induced emission-based bioimaging for assessing the environmental impact of emerging nanoscale pollutants, New Contaminants, 2025, 1, e019.
9. X. He, Z. Zhao, L.-H. Xiong, P. F. Gao, C. Peng, R. S. Li, Y. Xiong, Z. Li, H. H. Y. Sung, I. D. Williams, R. T. K. Kwok, J. W. Y. Lam, C. Z. Huang, N. Ma and B. Z. Tang, Redox-Active AIEgen-Derived Plasmonic and Fluorescent Core@Shell Nanoparticles for Multimodality Bioimaging, J. Am. Chem. Soc., 2018, 140(22), 6904–6911.
10. J. Luo, Z. Xie, J. W. Lam, L. Cheng, H. Chen, C. Qiu, H. S. Kwok, X. Zhan, Y. Liu, D. Zhu and B. Z. Tang, Aggregation-induced emission of 1-methyl-1,2,3,4,5-pentaphenylsilole, Chem. Commun., 2001, 1740–1741.
11. X. Shi, N. Yan, G. Niu, S. H. Sung, Z. Liu, J. Liu, R. T. Kwok, J. W. Lam, W.-X. Wang and H. H.-Y. Sung, In vivo monitoring of tissue regeneration using a ratiometric lysosomal AIE probe, Chem. Sci., 2020, 11(12), 3152–3163.
12. N. Yan, Y. Wang, T. Y. Wong, Z. Wu, X. Wang, M. Xie, E. Kutumova, A. Parodi, F. Kolpakov, W.-X. Wang and J. Shi, Differential Mapping of Intracellular Metallic Nanoparticles and Ions and Dynamic Modeling Prediction, ACS Nano, 2025, 19(23), 21644–21659.
13. N. Yan, Y. Wang, T. Y. Wong, Z. Wu, X. Wang, M. Xie, A. Parodi, W.-X. Wang and J. Shi, Spatiotemporal Mapping of the Evolution of Silver Nanoparticles in Living Cells, ACS Nano, 2024, 18(51), 35013–35028.
14. Y. Wang, Y. Hu, J. Mo, T. Yan Wong, J. Liu, P. Alessandro, B. Zhong Tang, W.-X. Wang and N. Yan, Bioprospecting of Chlamydomonas reinhardtii for boosting Biofuel-Related product production based on novel Aggregation-Induced emission active extracellular polymeric substances nanoprobes, Bioresour. Technol., 2024, 399, 130636.
15. Y. Li, N. Yan, T. Y. Wong, W.-X. Wang and H. Liu, Interaction of antibacterial silver nanoparticles and microbiota-dependent holobionts revealed by metatranscriptomic analysis, Environ. Sci.: Nano, 2019, 6(11), 3242–3255.
16. N. Yan and W.-X. Wang, Novel imaging of silver nanoparticle uptake by a unicellular alga and trophic transfer to Daphnia magna, Environ. Sci. Technol., 2021, 55(8), 5143–5151.
17. N. Yan, X. He, B. Z. Tang and W.-X. Wang, Differentiating silver nanoparticles and ions in medaka larvae by coupling two aggregation-induced emission fluorophores, Environ. Sci. Technol., 2019, 53(10), 5895–5905.
18. N. Yan, S. M. J. Tsim, X. He, B. Z. Tang and W.-X. Wang, Direct visualization and quantification of maternal transfer of silver nanoparticles in zooplankton, Environ. Sci. Technol., 2020, 54(17), 10763–10771.
19. N. Yan, B. Z. Tang and W.-X. Wang, Intracellular trafficking of silver nanoparticles and silver ions determined their specific mitotoxicity to the zebrafish cell line, Environ. Sci.: Nano, 2021, 8(5), 1364–1375.
20. M. M. Rohde, C. M. Snyder, J. Sloop, S. R. Solst, G. L. Donati, D. R. Spitz, C. M. Furdui and R. Singh, The mechanism of cell death induced by silver nanoparticles is distinct from silver cations, Part. Fibre Toxicol., 2021, 18(1), 37.
21. C. Levard, S. Mitra, T. Yang, A. D. Jew, A. R. Badireddy, G. V. Lowry and G. E. Brown Jr., Effect of chloride on the dissolution rate of silver nanoparticles and toxicity to E. coli, Environ. Sci. Technol., 2013, 47(11), 5738–5745.
22. N. Yan, Y. Wang, T. Y. Wong, Z. Wu, X. Wang, M. Xie, E. Kutumova, A. Parodi, F. Kolpakov, W.-X. Wang and J. Shi, Differential Mapping of Intracellular Metallic Nanoparticles and Ions and Dynamic Modeling Prediction, ACS Nano, 2025, 19(23), 21644–21659.
23. N. Yan, F. Li, D. Han, L. Yang, Y. Yin, L. Hu, J. Shi and G. Jiang, In Vivo Bioimaging of Mercury Sulfide Nanoparticles Dissolution in the Gut Environment of Zooplankton, ACS Nano, 2025, 20(1), 1639–1651.
24. L. Yang, B. Z. Tang and W.-X. Wang, Near-infrared-II in vivo visualization and quantitative tracking of micro/nanoplastics in fish, ACS Nano, 2023, 17(19), 19410–19420.
25. Y. Qin, X. Chen, Y. Gui, H. Wang, B. Z. Tang and D. Wang, Self-Assembled Metallacage with Second Near-Infrared Aggregation-Induced Emission for Enhanced Multimodal Theranostics, J. Am. Chem. Soc., 2022, 144(28), 12825–12833.
26. Y. Chen, M. Li, Y. Hong, J. W. Y. Lam, Q. Zheng and B. Z. Tang, Dual-Modal MRI Contrast Agent with Aggregation-Induced Emission Characteristic for Liver Specific Imaging with Long Circulation Lifetime, ACS Appl. Mater. Interfaces, 2014, 6(13), 10783–10791.
27. J.-X. Gong, X.-L. Wang, C.-X. Lin, X.-Y. Li, J. Wu, Q.-G. Tan, L. Yang and A.-J. Miao, Gut microbiota mitigate the reproductive toxicity of silver nanoparticles through thiamine-derived metabolites, Nat. Commun., 2025, 16(1), 7294.
28. L. Zhang, H. Jiang and W.-X. Wang, Subcellular imaging of localization and transformation of silver nanoparticles in the oyster larvae, Environ. Sci. Technol., 2020, 54(18), 11434–11442.
29. Y. Luo, X. He and W.-X. Wang, Intracellular accumulation, dissolution, and distribution of AIEgen-coated silver nanoparticles in hemocyte subpopulations of oysters, J. Hazard. Mater., 2022, 440, 129849.
30. H. Shi, R. T. Kwok, J. Liu, B. Xing, B. Z. Tang and B. Liu, Real-time monitoring of cell apoptosis and drug screening using fluorescent light-up probe with aggregation-induced emission characteristics, J. Am. Chem. Soc., 2012, 134(43), 17972–17981.

---

Footnote

† Yanping Li and Sitong Huo contribute equally.

---