Nanomaterial signatures program biomolecular condensates via triphasic separation for chemoplasticity remodeling

Zheng, Liu-Ting; Yan, Zeng-Shuai; Li, Xin-Yue; Chang, Jia-Jia; Tan, Xiao-Qi; Wang, Yu-Xing; Ding, Hong-Ming; Liu, Qin; Ma, Yu-Qiang; Huo, Da

doi:10.1038/s41467-025-64623-4

Download PDF

Article
Open access
Published: 29 October 2025

Nanomaterial signatures program biomolecular condensates via triphasic separation for chemoplasticity remodeling

Nature Communications volume 16, Article number: 9554 (2025) Cite this article

4793 Accesses
3 Citations
Metrics details

Subjects

Abstract

Membraneless organelles form by phase separation and regulate cell behavior. We show that cholesterol-patterned AuNPs program nanomaterial-induced stress granules (NSGs) by lowering G3BP1 condensation barriers through a solid–liquid–liquid triphasic sequence: nanomaterials recruit hnRNPC, which then engages G3BP1 to nucleate gel-like condensates. We map NSG microenvironments (temperature, polarity, pH, and proteasome activity), uncover dual disassembly—a slow VCP/19S-dependent route and a rapid SUMO/20S-dependent backup—and show that NSGs remodel chemo-plasticity: they mitigate doxorubicin/cisplatin toxicity in normal tissues yet sensitize tumors to nocodazole in vivo. Local induction and selective dissolution of NSGs thus offers a strategy to decouple efficacy from toxicity. Our results establish design rules linking nanomaterial surface chemistry to condensate programming and provide actionable levers to steer therapeutic outcomes.

Introduction

Membrane-less organelles (MLOs) are dynamic cellular structures formed through liquid-liquid phase separation (LLPS) of scaffold proteins and RNAs, playing critical roles in various biological processes¹. Stress granules (SGs), a typical example of MLOs, emerge in eukaryotic cells subjected to stimulation, including oxidative stress, heat-shock, or hyperosmotic treatment^2,3. Under such conditions, the translational activity of mRNA is temporally halted. By taking RNA-binding protein bound to mRNA as the core, followed by the recruitment of client proteins, SGs can be established that serve a pro-survival role^3,4. Among the various types of SGs that differ in origination, G3BP1 is ubiquitous that functions as the central hub owing to its intrinsic interaction with RNA as mediated by the RNA-recognition motif ⁵. In responding to stress relief, SGs undergo programmed disassembly to unleash sequestered mRNAs for resuming translation⁶. As such, the formation of SGs offers the stressed cells a means to temporarily control the energy consumption to balance the damage repair and homeostasis processes.

Emerging evidence indicates a correlation between SGs and the chemo-plasticity of cells^7,8,9. When exposed to chemotherapeutics such as cisplatin, wild-type cancer cells that extensively express SGs demonstrate a significantly higher survival rate compared to their counterparts lacking G3BP1¹⁰. The modulation of chemo-plasticity by SGs can primarily be attributed to the sequestration of pro-apoptotic components within the organelle¹¹. Due to their metastable nature, the influence of SGs on chemo-plasticity can be transient and temporally regulated^8,12, presenting unique advantages over conventional chemotherapy and identifying MLOs as promising targets for cancer treatment. However, several limitations need to be addressed to achieve this goal. Firstly, the complex structure of SGs requires drugs capable of simultaneously interfering with multiple proteins, which is beyond the capability of most small molecules¹³. Additionally, despite the varied origins of different SG subtypes, their contribution to mitigating chemo-toxicity is relatively consistent, and the presence of pro-apoptotic counterparts remains unclear, which is critical for tumor eradication. Furthermore, therapies aimed at modulating SGs must possess spatial precision to confine their influence to specific tissues, thereby mitigating global side effects. Meeting these criteria is essential for the successful development of therapeutics targeting the SGs-chemo-plasticity axis, which remains technically challenging.

The advancement of multifunctional and integrated therapeutic systems through nanotechnology represents a potential approach to address these challenges. Gold nanoparticles (AuNPs) are considered useful platforms in cancer therapy due to their biocompatibility, ease of surface functionalization, and superior stability^14,15. Although current research has observed passive interactions of nanomaterials with G3BP1 and other stress-related counterparts, the ability of rationally designed nano-platforms to remodel SGs remains unclear. While some evidence suggests that the physicochemical features of AuNPs significantly influence nanomaterials-protein interplay¹⁶, yet there lacks a comprehensive study revealing how these signatures impact the interaction between nanomaterials and SGs.

In this study, we constructed a library of AuNPs with varying signatures in terms of surface charge and hydrophobicity. Utilizing chemo-etching-assisted proximity labeling and immunofluorescence techniques, we systematically decoded the network of adsorbed proteins in an unbiased manner and examined their correlation with the surface functionalization features of AuNPs. Surprisingly, among the various candidates, a platform partially enclosed by a cholesterol (CLS)-terminated ligand demonstrated the ability to establish SGs through heterogeneous growth by serving as the seeds. Regarding the unprecedented establishment of SGs in a stress-free manner, we next employed computational simulation to assess the underlying mechanisms. We proposed and validated through truncation assays and proximity assessment that a stepwise phase separation occurs via sequential recruitment of hnRNPC and G3BP1 by CLS-ligand. Considering the physical properties of nanomaterials and adsorbed proteins, this process was termed “solid-liquid-liquid” tri-phase separation, ultimately leading to the formation of non-canonical SGs (c-SGs), namely, nanomaterials-induced SGs (NSGs). Unlike c-SGs, NSGs exhibit a gel-like structure that hinders component exchange with diluted phases, confirmed by the photo-bleach recovery assay. Given this unique feature, the dissociation pathway of NSGs is expected to differ from that of c-SGs, further explored using inhibition assays and transcriptome analysis. The disassembly of NSGs revealed an unprecedented dual-mode pathway dominated by a slow yet robust 19S proteasome-dependent mechanism alongside a 20S-dependent backup pathway functioning upon inactivation of the former. Formation of NSGs was validated to be capable of reshaping the chemo-plasticity of cells, thereby leading to context-dependent therapeutic outcome remodeling. Through spatially controlled introduction of CLS-functionalized AuNPs as the inducer of NSGs, we investigated a set of ideas, including in vivo chemo-toxicity mitigation and sensitization in normal and cancerous tissues, respectively. The role of NSGs in these processes was further validated using mice with liver and tumor tissue, conditional knockout of pivotal NSGs genes. A schematic illustration provided in Fig. 1 summarizes the core concepts and findings of the study, offering insight into the role of NSGs in nanomaterials metabolism and shedding light on the development of artificial, functional MLO for precision disease management.

**Fig. 1: Formation mechanism and key features of NSGs.**

Results

Nanomaterials signature-regulated SGs

In this study, we synthesized a library of gold nanoparticles (AuNPs) with varying sizes from 10 nm to 180 nm (refer to Fig. 2A), employing seed-mediated growth¹⁷. The size of the synthesized AuNPs was characterized using transmission electron microscopy (refer to Supplementary Fig. 1). To investigate the interaction between these nanoparticles and stress granule-associated proteins, we analyzed the impact of differently sized AuNPs on the condensation of G3BP1 protein within HeLa cells. Notably, an inverse correlation was found between the extent of G3BP1 condensation and the size of the AuNPs (refer to Fig. 2B). Based on these observations, we selected AuNPs with a diameter of 70 nm due to their minimal influence on G3BP1 condensation, thereby reducing the confounding effects of nanomaterial size. In subsequent discussions, unless otherwise specified, AuNPs refer to 70-nm gold nanoparticles.

**Fig. 2: Suitably functionalized nanomaterials stimulate NSGs.**

We functionalized AuNPs with polyethylene glycol (PEG) of different molecular weights, as shown in Fig. 2A, and characterized their optical features using UV-vis spectra (Supplementary Fig. 2). We then examined how PEG chain length affected the cellular uptake of AuNPs by HeLa cells and G3BP1 condensation. Longer PEG chains reduced both G3BP1 condensation and AuNP endocytosis (Supplementary Fig. 3). For a balanced performance, we fixed PEG molecular weight at 5 kDa for further analysis. Next, we investigated the role of AuNP surface charge and hydrophilicity in regulating G3BP1 condensation. We used NH₂-/COOH-/OMe-PEG-SH and mixtures of CLS-PEG-SH and mPEG-SH at varying ratios (20%-100%) to study these effects. Specifically, NH₂-modified and 60% CLS-AuNPs significantly stimulated SGs formation (Supplementary Fig. 4, 4.37 and 3.05-fold, respectively, compared to control) and showed 68.5% colocalization with the MLOs (Fig. 2C-E), the highest among all groups. Positively charged materials, like those rendered by the NH₂ group, can either stimulate stress response due to intrinsic toxicity (for instance, gold nanorods¹⁴) or provoke SGs by acting as RNA-binding proteins that aid the G3BP1 assembly^4,16. The mechanism for CLS-modified nanomaterials, excluding the likelihood of seeding SGs through electrostatic interaction, is yet to be defined.

To ascertain if phase separation induced by 60% CLS-AuNPs is a universal phenomenon, we extended our assay to other LLPS-prone proteins such as YTHDF2, TDP43, and SIRT2, using NaAsO₂ (AS) as a classic inducer. We observed that 60% CLS-AuNPs significantly reduced the phase-separation energy barriers for G3BP1 and YTHDF2 proteins in cells (Supplementary Fig. 5A, showing reductions of 1.74-, 1.44-, and 1.47-fold for G3BP1 and 1.08-, 1.01-, and 1.24-fold for YTHDF2 at 15 min, 30 min, and 60 min post-treatment, respectively, relative to AS-treated control groups). As an essential component of SGs, YTHDF2 can form droplets that closely overlap with SGs through coacervation^18,19, which explains the ability of nanomaterials to stimulate YTHDF2 puncta. Given the difference in size between SGs induced by nanomaterials and canonical SGs stimulated by AS, the former were termed NSGs. Conversely, the condensation of SIRT2 and TDP43—proteins forming MLOs only weakly associated with SGs²⁰—was unaffected by 60% CLS-AuNPs (Supplementary Fig. 5A). These findings indicate that 60% CLS-AuNPs preferentially target components associated with SGs.

A common outcome of nanomaterials’ exposure to biological media such as serum is the protein corona effect, where proteins rapidly cover the surface of nano-substances. It is unclear whether proteins adsorbed prior to cellular entry of 60% CLS-AuNPs are involved in nanomaterials-G3BP1 interactions. A simplified assay was conducted to examine the biodistribution of the protein corona after endocytosis. To distinguish extracellular from intracellular proteins, serum proteins conjugated with a click-handle were prepared for cell culture and subsequent formation of the protein corona, as shown in Supplementary Fig. 5B. Tetrazine-expressing proteins—either remaining on nanomaterials (Component I) or detached (Component II)—were isolated using a click-reaction and magnetic beads. Albumin, as the predominant component, was used as a reference for expression analysis in both components (Supplementary Fig. 5C). A significant reduction in protein corona was observed for 60% CLS-AuNPs, consistent with previous reports indicating different fates for protein corona and nanomaterials involving cycling endosomes²¹. Although the precise mechanism behind the disassembly of the protein corona in this context has not been determined, current findings suggest that previously adsorbed proteins did not play a key role in the nanomaterials-G3BP1 interaction.

Mechanism of NSG formation

Endoplasmic reticulum stress (ER-stress) frequently occurs and serves as an early event stimulating the formation of SGs²². We first examined the presence of ER-stress in cells exposed to 60% CLS-AuNPs, indicated by the expression level of CHOP and BiP (Fig. 2F). The absence of ER-stress ruled out the possibility of 60% CLS-AuNPs stimulating NSG through a classic mechanism involving misfolded protein accumulation²³. Next, potential pathways underlying the action of 60% CLS-AuNPs were screened using RNA-sequencing. Cells exposed to 40 kDa mPEG modified AuNPs were used as a negative control, wherein the absence of noticeable NSGs was confirmed. Thus, altered genes (compared to untreated cells) in the mPEG group can be identified as irrelevant to NSGs assembly, reflecting the cellular response to the nanomaterial itself, rather than the signature. Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were conducted with a focus on significantly altered genes that only appeared in the CLS group. Results showed no candidates had a clarified interplay with SGs (Fig. 2G), suggesting that the induction of NSGs proceeds independently of transcriptional level changes.

Considering the tendency of nanomaterials to recruit proteins, especially those with intrinsically disordered regions, we hypothesize that variations in the nanomaterial signature can affect the preference of recruited proteins as well as their conformation, thereby impacting SGs assembly. To thoroughly identify the adsorbed proteins, we developed a chemo-etching-assisted proximity labeling method to capture the fragile adsorbed complex on the nanomaterial surfaces, as illustrated in Fig. 2H, alongside co-immunoprecipitation (Co-IP) techniques to uncover the proteins associated with G3BP1. Analysis of surface-adsorbed proteins revealed that NSGs formed in response to 60% CLS-AuNPs stimulation differ compositionally from canonical SGs induced by AS (Fig. 3A). Notably, NSGs contained a higher proportion of proteins associated with translation processes, including ribosomes and rRNA binding factors. This observation suggests that NSGs might have the capacity to induce transient cell cycle arrest and preserve ATP by recruiting translation-associated components, thus exerting a pro-survival role in stressed cells.

**Fig. 3: Mechanisms of NSG assembly driven by hnRNPC recruitment.**

Subsequently, we compared the proteins adsorbed on the nanomaterial surfaces—captured via photo-proximity labeling—with those that interact closely with G3BP1 in NH₂- and 60% CLS-AuNPs groups. Among the common proteins identified, we observed that nine proteins within the 60% CLS-AuNPs group exhibit potential association with G3BP1 (Fig. 3B). The ability of these components to undergo LLPS was further assessed based on the proportion of disordered regions and physicochemical properties of each protein using PhaSePred²⁴. The calculated scores are as follows: hnRNPU (0.947), SYNCRIP (0.945), hnRNPC (0.920), DHX9 (0.917), hnRNPK (0.914), HSPA9 (0.793), PTBP1 (0.612), ACTB (0.498), and FN1 (0.166). Conversely, only three proteins from the NH₂-group interacted with G3BP1, yielding an average phase separation score of 0.677, which is significantly lower than that of the CLS group (0.746), aligning with their differences in nanomaterials-NSGs colocalization efficiency. We propose that the enhanced phase separation potential of the proteins found in the CLS group may contribute to their increased capability to seed NSGs compared to NH₂-modified materials.

Combining the interaction components of G3BP1 in both experimental groups, we observed the presence of hnRNPC (Fig. 3B). Previous analyses indicated that mPEG-functionalized AuNPs could induce NSGs, albeit to a lesser extent than CLS- or NH₂-modified counterparts. We extended the photo-proximity labeling to the mPEG group, alongside co-IP assays to elucidate the interactome between nanomaterials and G3BP1, respectively (referred to as material and G3BP1 side in Fig. 3C). Venn diagram analysis revealed hnRNPC as the only overlapping component linking NSGs derived from different origins and nanomaterials, showcasing its capacity to undergo LLPS by serving as clients (Fig. 3D).

We subsequently examined the role of hnRNPC in establishing NSGs. Given the bi-directional interaction of hnRNPC with G3BP1 and nanomaterials, we performed a competition assay to evaluate hnRNPC’s preference towards NSGs versus canonical SGs (denoted as f-SGs). As shown in Fig. 3E, simultaneous exposure to 60% CLS-AuNPs and AS resulted in the co-existence of NSGs and f-SGs, distinguishable based on their colocalization with dye-labeled nanomaterials. A stronger association between hnRNPC and nanomaterials compared to G3BP1 was confirmed by the higher protein abundance in NSGs (Figs. 3F, 1.29-fold of f-SGs, p < 0.05). Furthermore, colocalization analysis demonstrated that the introduction of either CLS-AuNPs or AS treatment promoted a stronger association between G3BP1 and hnRNPC (Fig. 3G). Notably, the co-presence of NSGs and f-SGs (60% CLS + AS group) further strengthened the interaction between these two components compared to AS treatment alone. This observation confirmed that within NSGs, hnRNPC interacts more intimately with G3BP1 compared to f-SGs.

In stressed cells, regardless of the type of stimuli, we observed the aberrant translocation of hnRNPC from the nucleus to the cytoplasm, consistent with its role as a client that promotes the formation of SGs (Fig. 3H, I). Notably, the presence of nanomaterials further enhances the cytoplasmic entry of hnRNPC compared to G3BP1 condensation alone (AS group). We hypothesize that the recruitment of hnRNPC by nanomaterials may be a critical event promoting the establishment of NSGs. To test this hypothesis, we performed the same assay in cells lacking hnRNPC (hnRNPC KO, Fig. 3K). The absence of hnRNPC did not affect AS’s ability to induce SGs, likely due to the presence of multiple alternative clients (Fig. 3L). In contrast, hnRNPC KO significantly reduced the capacity of nanomaterials to provoke NSGs, with G3BP1 condensation decreased by 1.38-fold (Fig. 3L). These results clearly demonstrate the role of hnRNPC in facilitating NSG formation. Collectively, our findings suggest a model of NSG formation wherein the translocation of hnRNPC from the nucleus to the surface of CLS-modified nanomaterials initiates the recruitment of G3BP1, thereby driving NSG assembly (Fig. 3M).

An additional question is how the formation of NSGs affects the cellular capacity to form canonical SGs and their interaction. To investigate this, cells were simultaneously exposed to 60% CLS nanomaterials and canonical SG inducers (AS, sorbitol, and heat-shock). As shown in Supplementary Fig. 6A, this co-exposure resulted in the concurrent formation of both NSGs and canonical SGs in all tested conditions, indicating that NSG formation minimally impacts the G3BP1 pool. The experiment was then modified so that exposure to canonical SG inducers occurred after nanomaterial stimulation. It was observed that prior NSG generation substantially reduced the ability of cells to form canonical SGs (Supplementary Fig. 6B). Previous findings have established that hnRNPC plays a key role in mediating interactions between nanomaterials and G3BP1, making nanomaterials a less competitive substrate for G3BP1 phase separation compared to conventional stressors. Therefore, simultaneous exposure did not significantly reduce the cells’ ability to form canonical SGs. In contrast, sequential exposure enabled nanomaterials to recruit G3BP1 more effectively, leading to depletion of the protein pool and a diminished ability of cells to form canonical SGs.

Molecular dynamics simulations of tri-phase separation

To generate the mechanistic hypothesis, we employed all-atom MD simulations to explore the sequential recruitment of hnRNPC and G3BP1 to AuNPs, where 500 ns production simulations were performed following pre-equilibration, and the final 200 ns, corresponding to the equilibrium phase, were used for subsequent analyses (details described in the Methods section). Compared to mPEG, the incorporation of 60% CLS-PEG results in the formation of larger patches on the outer surface of AuNPs due to strong hydrophobic interactions among CLS molecules (Supplementary Fig. 7A–C). These hydrophobic interactions reduce the conformational flexibility of CLS-PEG and extend the residence time of water molecules on the PEG surfaces (Supplementary Fig. 7D–F), thus altering local hydrophilicity. Notably, the surface undulations of 60% CLS resulted in a surface area of 252.98 nm², which is significantly higher than that of the other two groups (163.23 nm² and 184.80 nm² for mPEG and 100% CLS, respectively) (Supplementary Fig. 7G). These features collectively suggest that 60% CLS-AuNPs exhibit distinct adsorption behaviors when interacting with hnRNPC.

Subsequently, we simulated the interaction between hnRNPC and nanomaterials, placing hnRNPC on top of three different types of AuNPs. Successful adsorption was confirmed across all tested groups with negligible differences (Fig. 4A–I and Supplementary Figs. 8 and 9); however, the binding strength and interfaces varied significantly (Fig. 4J, K). Specifically, a much closer interaction between hnRNPC and 60% CLS-AuNPs was observed compared to the other two groups in terms of contact surface area (CSA: 46.22, 30.41, and 41.47 nm² for 60%-CLS, 100%-CLS, and mPEG, respectively) and binding energy (−421.38, −197.98, and −319.08 kcal/mol). These differences were further supported by the greater number of residues at the hnRNPC-nanomaterial interface compared to the other two groups (Fig. 4M–O). Specifically, N-terminal residues of hnRNPC, including K8, T9, D10, P11, R12, M14, K42, R61, and P88 (Fig. 4N), predominantly contributed to the recruitment of hnRNPC to 60% CLS-AuNPs. Additionally, interactions between hydrophobic residues (P145, P148, Y132, Y133, and Y137) and CLS at the hnRNPC-60% CLS-AuNP interface facilitated hnRNPC adsorption (Fig. 4D). In contrast, the total or partial absence of similar interactions or compensatory forces in fully cholesterolized (100% CLS group) or cholesterol-devoid (mPEG group) groups safely explained the weakened adsorption performance of hnRNPC.

**Fig. 4: Simulated landscape for the adsorption of hnRNPC onto AuNPs.**

We further investigated the interaction between hnRNPC, stabilized by nanomaterials, and G3BP1. Using ClusPro²⁵ and AlphaFold3²⁶, we predicted the complex structure to identify the binding region between the NTF2L domain of G3BP1 and hnRNPC. The structures obtained from both approaches are nearly identical (Supplementary Fig. 10A), demonstrating that the binding of the NTF2L pocket with Coil_111-123 residues of hnRNPC is critical for recruitment. To confirm the stability of the complex structure, energy minimization, a short pre-equilibration in NVT and NPT ensembles, and final 500 ns production simulations were conducted. The NTF2L-hnRNPC binding energy was calculated over the equilibrated final 200 ns, indicating a stable complex with a binding energy of −258.24 kcal/mol (Supplementary Fig. 10B). Hydrophobic interactions mediated by residues in Coil_111-123, such as H111, P114, P116, L117, L118, and F122, play a key role in stabilizing the NTF2L-hnRNPC interaction (Fig. 5A, B). This attraction is similar to how the NTF2L domain recruits other G3BP1 clients, such as Caprin-1 or USP10, in the formation of stress granules²⁷.

**Fig. 5: Mechanism underlying hnRNPC-G3BP1 interaction.**

After optimizing models describing the interaction of hnRNPC with either G3BP1 (NTF2L domain) or nanomaterials, a ternary complex including all these components was constructed (Fig. 5C, D). The presence of nanomaterials did not impair hnRNPC’s ability to interact with the NTF2L domain of G3BP1, as indicated by the preservation of Coil_111-123 residues (such as P116, L117, L118, and F122) in the protein-protein interface (inset Fig. 5C). There might be contributions from other bystander proteins in stabilizing the hnRNPC-G3BP1 interaction, which are beyond the capacity of current simulation approaches to reveal. To validate the MD-informed hypothesis that hnRNPC interacts with G3BP1 through the NTF2L domain in cellular environments, an assay was conducted in cells overexpressing truncated G3BP1 (ΔNTF2L). As shown in Fig. 5E, super-resolution imaging was used to analyze the interaction of G3BP1 or its mutant with hnRNPC, measuring the distance between these proteins. The absence of the NTF2L domain significantly weakened the hnRNPC-G3BP1 interaction, as indicated by their enforced segregation (mean distance: 0.53 and 2.85 μm for control and ΔNTF2L groups, respectively). This finding confirmed the hypothesis of previous simulations and validated that the contribution of bystander components to the recruitment of G3BP1 is limited with respect to hnRNPC.

Combining the computational findings and experimental results, the formation mechanism of the nanomaterials-hnRNPC-G3BP1 ternary complex can be described as follows: The modification with CLS-PEG ligands at a density of 60% creates a microenvironment conducive to hnRNPC adsorption, involving both hydrophobic interactions and hydrogen bonds to varying extents. This step is crucial for the further recruitment of G3BP1, as the absence of hnRNPC prevented NSG assembly. Through multivalent interaction of the Coil_111-123 region on the hnRNPC side with the NTF2L domain of G3BP1, G3BP1 condensation around AuNPs is initiated, establishing NSGs. Unlike c-SGs that rely mainly on coacervation, NSG assembly is regulated through sequential client recruitment. In this scenario, hnRNPC (liquid phase) acts as the central hub linking nanomaterials (solid phase) and G3BP1 (liquid phase), shifting the LLPS of proteins from co-condensation to a cascade-like process. This mechanism is termed ‘solid-liquid-liquid’ tri-phase separation to describe the formation of NSGs.

Microenvironment of NSGs

Emerging evidence indicates that the formation of MLOs contributes to creating a metastable lumen in the cytosol, differing significantly between the condensed phase (intra-MLO) and dilute phase (inter-MLO)^11,28,29. This characteristic allows cells to adjust metabolic activities dynamically without changing the entire cytoplasmic environment, balancing various reactions under subtly different conditions. Therefore, understanding the microenvironment of MLOs is essential for predicting their biological implications, which is the focus of this study.

We categorized the microenvironment of NSGs into physical and chemical classes, including temperature, polarity, pH, and proteasome activity. To analyze the local microenvironment of NSGs without affecting the overall cellular environment, we applied a biorthogonal approach using HaloTag labeling³⁰, as illustrated in Fig. 6A. Environment-responsive dyes were conjugated with a chloroalkane ligand recognized by Halo-protein fused to the C-terminal of G3BP1, resulting in dye immobilization within SGs and visualization of the physicochemical microenvironment.

**Fig. 6: Characterization of the microenvironment in NSGs.**

For the physical microenvironment of NSGs, rhodamine B was chosen as the temperature reporter³¹. With increased temperature, energy consumption by rhodamine B via non-radiative dissipation is enhanced, reducing fluorescence intensity. Colocalization analysis (1D, Fig. 6B) supported the labeling conciseness between ligand-functionalized dye and NSGs. Compared to c-SGs stimulated by AS, the calculated fluorescence intensity of NSGs was higher than canonical counterparts (1.69-fold, Fig. 6F and Supplementary Fig. 11), indicating a lower internal temperature. Similarly, coumarin-343 dye estimated polarity³², revealing high polarization in both NSGs and canonical ones (Fig. 6C). According to Ye and colleagues³³, polarized microenvironments facilitate macromolecule influx into certain biomolecular condensates, promoting secondary phase separation, consistent with NSG features established through multiple phase separations.

We proceeded to analyze the chemical properties of NSGs using Protonex- and LLVY-conjugated ligands, which respond to pH and proteasome variations, respectively. The Protonex dye exhibits increased fluorescence intensity with decreasing pH levels. Interestingly, as illustrated by the 1D fluorescence distribution plot (Fig. 6D), G3BP1 translocates from regions of high pH (canonical SGs) to areas of low pH (NSGs). This observation aligns with our previous proximity labeling results, indicating significant differences in the enrichment of proteins closely interacting with G3BP1 between NSGs and those stimulated by oxidative stress. Changes in proximal proteins correspond to expected variations in pH levels. Our subsequent investigations revealed that pH is not the only altered chemical feature of NSGs compared to c-SGs. Utilizing LLVY as a cleavable, fluorogenic substrate for evaluating proteasome activity, we observed that G3BP1 acts as the proteolytic center within NSGs (suggested by its perfect overlap with dye molecules) and the absence of similar findings in the AS group (Supplementary Fig. 12, 5.5‰ of NSGs). Proteasomes, being highly susceptible to damaged proteins, suggest an accumulation of misfolded components in NSGs. As these MLO-associated clients unfold, the packing of these components may intensify, leading to expected fluidity variation. To test this hypothesis, our subsequent analysis will focus on elucidating the physical properties of NSGs.

NSG dissociation mechanism

Enrichment of proteasomes suggests rigorous degradation occurring in certain regions, leading to a much-shortened half-life for NSGs. The timeframe assessment of NSGs revealed sustainable generation lasting for at least 8 h, followed by disassembly during the next 2 h, as indicated by the fold of changes in G3BP1 condensation extent measured at 2, 4, 6, 8, and 10 h (Supplementary Fig. 13: 1.57, 1.57, 1.84, 2.89, and 2.27-fold of control), whereas c-SGs typically undergo dissociation within 4 h post formation. As such, the stability of NSGs is unexpected and suggests concentrated proteasomes were either dysfunctional or acting in ways other than degradation machinery.

Our previous study indicated SGs provoked by nanomaterials might feature reduced fluidity compared to canonical ones stimulated by AS or heat stress¹⁶. It is unknown whether different nanomaterials exhibit similar implications. A comparative analysis of the physical features of NSGs and SGs elicited by canonical inducers (AS and sorbitol) was conducted using fluorescence recovery after photobleaching (FRAP) assays. Figure 7A shows the fluorescence recovery ratio of NSGs progressively reduced with extended durations of stimulation by 60% CLS-AuNPs (maximum recovery ratio: 85.5%, 64.7%, 40.6%, and 21.5% for AS, sorbitol, and treatments at 1 and 4 h with 60% CLS-AuNPs, respectively). This supports the notion that incorporation of CLS-AuNPs contributes to incremental solidification of NSGs, partially explaining the decreased degradative power of enriched proteasomes due to restricted traffic in solidified organelles.

We categorized the generated SGs in cells stressed with CLS-AuNPs into two groups: NSGs and those without nanomaterials’ involvement (f-SGs). We repeated the FRAP assay on these groups to assess differences in fluidity. Under identical conditions, NSGs had a significantly lower recovery ratio compared to f-SGs (Fig. 7B, C, recovery values: 58.8% for f-SGs and 23.7% for NSGs). The preservation of liquid-like properties in f-SGs ruled out a global change in fluidity due to nanomaterial exposure, indicating a local effect in NSGs. Dissolution of MLOs can occur through programmed dissolution within the droplet, which is not possible when exchange between dense and diluted phases is prohibited^34,35, as seen in NSGs. Therefore, a dissociation mechanism differing from its liquid-like counterparts was expected for NSGs.

To understand NSGs degradation, we tested several inhibitors linked to SGs disassembly and nanomaterials-induced counterparts, including myricetin (HSP70 inhibitor), ML346 (HSP70 agonist), Grp94 (HSP90 inhibitor), AMP-PCP (HSP90 agonist), and CB5083 (VCP-p97 inhibitor). After confirming that these inhibitors do not induce SG formation alone (Supplementary Fig. 14A), we co-stimulated them with nanomaterials. Notably, VCP inhibitors notably accelerated NSG degradation (Supplementary Fig. 14B), showing the importance of the 19S proteasome in NSG maintenance. Similar results in HeLa^{G3BP1 OE} cells suggest the anti-degradation role VCP served is dispensable of G3BP1 abundance (Fig. 7D). Tests with other VCP-p97 inhibitors (DBeQ and NMS-859, Fig. 7E, F) supported these findings. As part of the 19S proteasome, VCP is known for degrading heat stress-induced SGs via ubiquitinated G3BP1, opposing our case³⁶. It is thus hypothesized that alternative cytoplasmic degradation pathways might activate upon VCP-p97 inhibition. We next examined G3BP1 condensation under co-stimulation with 60% CLS-AuNPs and deubiquitinating enzyme inhibitor Spautin-1 or E1-activating enzyme inhibitor TAK-243. Neither inhibitor alone caused significant G3BP1 condensation (Supplementary Fig. 15). Co-stimulation with TAK-243 preserved NSGs better (Fig. 7G), while Spautin-1 reduced NSG stability. These results indicate a negative correlation between NSG preservation and ubiquitination extent.

We investigated the ubiquitination level of G3BP1 when the VCP-p97 pathway was inhibited. Comparing the expression of ubiquitinated G3BP1 in NSGs with the overall cytoplasmic protein ubiquitination level before and after CB5083 treatment, we found decreased G3BP1 ubiquitination as NSGs appeared (Fig. 7H). During this time, NSG formation exceeded degradation. Inhibition of VCP increased ubiquitinated G3BP1 levels and caused NSGs to depolymerize, indicating an alternative ubiquitination-related degradation pathway activable upon VCP inhibition. Stable ubiquitination level of cytoplasmic protein suggests the influence of this alternative pathway was limited (Fig. 8A, B). Combined CB5083 and nanomaterials treatment accelerated G3BP1 clearance with respect to nanomaterials alone (Fig. 8C), revealing strong degradation power compared to the VCP-mediated pathway. Thus, it can be concluded that two mechanisms coexist for NSG degradation: the slower VCP-p97-dependent route and a rapid, alternative one activated upon VCP inhibition. RNA-sequencing showed increased mRNA levels related to 20S proteasome activity, especially PSMA, PSMB, and PSME post 19S inhibition (Fig. 8D). Ubiquitinated proteins can be degraded by a 19S-independent mechanism associated with SUMOylation and 20S proteasome activity, often removing unfolded or damaged proteins³⁷. We validated the role of 20S proteasome by stabilizing NSGs under nanomaterial stimulation with PYR-41, an ubiquitination inhibitor (Fig. 8E, F, Supplementary Fig. 16). This stability was largely restored by introducing a SUMOylation inhibitor (2-D08), confirming the contribution of SUMOylation to NSG degradation. Under physiological conditions, the 19S-mediated route is the rate-limiting step, whereas the 20S-dominated degradation path, once inactive, can accelerate NSG clearance. Using proximity analysis, we validated the hypothesis of 20S proteasome-mediated degradation. We analyzed the interaction of PSMB6 (part of the 20S proteasome complex) with nanomaterials before and after inhibition of VCP. Suppressed VCP activity increased the interaction of G3BP1 with the 20S proteasome, as indicated by their reduced distance to each other (Fig. 8G), whereas the absence of nanomaterials abolished this effect. This suggests that the 20S proteasome functions as a backup degradation pathway exclusively for NSGs, which is inhibited by 19S counterparts under normal conditions.

**Fig. 8: SUMOylation-dependent 20S pathway underlies NSGs dissociation.**

We also investigated whether there exist degradation pathways other than the above-mentioned ones. By concurrently inhibiting both sides of proteasomes using CD5083 and Tomatine, we observed NSGs enlargement compared to control (Fig. 8H), which ruled out the presence of compensatory degradation pathways. Inhibition of 20S proteasome activity alone had a negligible influence on NSG stability, confirming the dormant state of this degradation pathway under the modulation of functionally intact VCP (Fig. 8J). Additionally, we explored whether similar dissociation actions can be extended to c-SGs. According to literature, SGs induced by heat stress typically undergo VCP-mediated degradation for resolution³⁶. We thus assessed the effect of 19S proteasome on the condensation of G3BP1 in heat-stressed cells. A rescue effect exerted by CB5083 on SGs was observed, excluding the presence of alternative degradation pathways in this context (Fig. 8I), which supports the specificity of dual-degradation pathways only in cases associated with nanomaterials.

Potential of NSGs in remodeling chemo-plasticity

We further examined the potential biological functions of NSGs. By introducing a VCP inhibitor, we found an effective way for NSGs-specific dissociation to assess their biological implications by using RNA-sequencing. Subsequently, bioinformatic analysis, as shown in Fig. 9A, suggests that NSG inhibition compensatively activates multiple pathways, including MAPK and PI3K-Akt, which feature the highest enrichment factors (Fig. 9B). Given their pro-survival characteristics, it is hypothesized that NSG generation plays a role in maintaining cellular homeostasis under stress conditions induced by nanomaterials. To evaluate this hypothesis, assays were performed in which cells exposed to nanomaterials designed to induce NSGs were treated with pro-apoptotic or pro-inflammatory agents. As shown in Supplementary Fig. 17A, the presence of NSGs was associated with reduced activation of inflammatory and apoptosis pathways, indicated by lower nuclear translocation of NF-κB and reduced expression of cleaved caspase 8. When NSGs were depleted using CB5083, the observed effect of nanomaterials was not present, indicating that the pro-survival outcome was attributable to the presence of NSGs. Additionally, changes in mRNAs associated with microtubules were noted due to NSG dissipation. During SGs' growth, clients are translocated to premature droplets via the microtubule network^3,38. Interplay between NSGs and the cytoskeleton suggests that, despite differences in composition and fluidity, both types of SGs exhibit similarities during their growth stages.

**Fig. 9: NSGs-modulated chemo-plasticity.**

Recent research highlighted that by sequestering biomolecules (such as untranslated mRNA or proteins) and thus affecting their abundance and function, SGs can act as central hubs governing the chemo-plasticity of cells³⁹. This raises the question of whether NSGs have similar implications and if this feature could be used for chemo-plasticity intervention. Therefore, we evaluated the cytotoxic effects of four chemotherapeutics alongside NSGs. Starting with Doxorubicin, a widely used drug that functions as a replication inhibitor, we analyzed Cleaved PARP expression levels as indicators of DNA damage. Upon introducing the VCP inhibitor, NSGs underwent rapid degradation mediated by the 20S proteasome, and a significant increase in Cleaved PARP levels was observed (Fig. 9C, 2.06-fold of control). These findings indicate that NSGs considerably mitigate the cytotoxicity caused by Doxorubicin in cancer cells.

We examined the impact of NSGs on chemo-plasticity in response to microtubule-targeting agents. Nocodazole and Paclitaxel, which function differently, were introduced as presented in Fig. 9D. It was observed that nanomaterials alone intensified the microtubule depolymerization caused by nocodazole at concentrations of 50 nM or higher. Introduction of VCP inhibitor mitigated this effect on the cytoskeleton, as indicated by morphology (Fig. 9E, sphericity values: 0.65, 0.78, and 0.95 for 60% CLS, 1 nM, 50 nM, and 100 nM; Supplementary Fig. 17B, isotropy values: 0.61, 0.51, and 0.37 for 60% CLS, 1 nM, 50 nM, and 100 nM, respectively), highlighting the role of NSGs as nocodazole-sensitizers in chemo-plasticity remodeling. Conversely, when paclitaxel, a stabilizer of polymerized microtubules, was used instead of nocodazole, the presence of NSGs reduced the effectiveness of the chemotherapeutic (Fig. 9E, F). These findings suggest that NSGs can modify the chemo-plasticity of cancer cells, making them more sensitive to microtubule depolymerization. This analysis was replicated in cisplatin-treated cells (Fig. 9G), showing reduced expression of the DNA damage biomarker γ-H2AX (Fig. 9H, expression levels: 0.97, 0.73, and 0.87 for nanomaterials combined with cisplatin at 10 μM, 50 μM, and 250 μM, respectively), which returned to basal level upon NSG dissociation. These results indicate that NSGs’ effects on cancer chemo-plasticity are context-dependent (Fig. 9I). For nocodazole, NSGs may enhance its sensitizing effect for tumor management. For other chemotherapeutics, combination with CLS-AuNPs could potentially reduce unwanted cytotoxicity, especially in healthy tissues like the liver. To test this hypothesis further, our subsequent studies will explore the pro- and anti-chemo-action effects of nanomaterials in cancerous and normal tissues, respectively.

In vivo implications of NSGs

Previous experience showed a correlation between the organ distribution of injected nanomaterials and their hydrophilicity. To identify the tissues for chemo-toxicity mitigation, we assessed nanomaterial enrichment over time post administration. We focused on the liver, spleen, and kidney due to their higher likelihood of enriching nanomaterials. Accumulation of AuNPs in tissues was confirmed shortly after administration, which increased consistently over time. The liver had the highest enrichment (maximal accumulation: 40.9% at 48 h), followed by the spleen (16.7% at 48 h) and kidney (7.2% at 24 h). The presence of nanomaterials is essential for NSGs provocation. However, it is unclear whether nanomaterials enriched in certain tissues preserve such potential and their contribution to chemo-toxicity mitigation. We evaluated tissue responses to Doxorubicin, cisplatin, and paclitaxel. Nanomaterials reduced drug side effects as indicated by the lower levels of damage-related biomarkers. Thus, nanomaterial presence ensured NSG generation with biological implications in vitro.

We proceeded to assess the feasibility of using NSGs-regulated chemo-plasticity for solid tumor management. A xenograft animal model was established as illustrated, and CLS-AuNPs were intratumorally administered once the volume of tumor tissue reached a certain value (Fig. 10C, Trial#1, Supplementary Fig. 18A, B). Subsequently, systematic administration of nocodazole at differing dosages (denoted by 0.5× and 1×) was conducted (I–V as illustrated). Both the growth curves of the solid tumors and the overall survival of mice receiving corresponding treatments were recorded. Results indicated that the presence of NSGs, even when the dosage of chemotherapeutics was halved (group I), exerted enduring yet robust tumoricidal effects, outperforming the outcome of chemotherapy alone at normal dosage (group III). Abrogation of anti-tumoral action was observed when nanomaterials were employed alone for disease management, further supporting that the observed tumor regression should be attributed to the synergistic action between chemo-plasticity remodeling and nocodazole. To assert this conclusion, cell lines devoid of G3BP1 were employed for xenograft establishment. After replicating the above treatment, tumor tissues were collected and dissociated into single-cell suspensions for flow-cytometry analysis (labeling with Annexin V-FITC/PI for gating). In wild-type tissues, the viability of solid tumors receiving varying treatments (I, III, and V as representatives, viability: 45.1%, 69.2%, and 94.7%, respectively) corresponded with their growth rates. Notably, the knockout of G3BP1 substantially weakened the anti-tumoral effect of treatment I (HeLa^{G3BP1 KO}), confirming that NSGs, rather than nanomaterials, underlie the sensitized chemotherapy.

**Fig. 10: In vivo therapeutic outcome.**

Furthermore, the capacity of nanomaterials to protect liver tissue from systemic chemotherapy-associated side effects in tumor-bearing animals was analyzed (Fig. 10D, Trial#2). Cisplatin was selected for the study due to its notable liver toxicity, and necropsy and immunofluorescence imaging were conducted to decipher the tissue damage at the bulk and cellular levels, respectively. Typically, cisplatin-induced liver damage caused the accumulation of apoptotic hepatocytes in lobules, characterized by fragmented nuclei surrounded by vacuoles (inset necropsy, IV group). The proportion of damaged tissues due to cisplatin exposure was quantified, and results demonstrated that the co-presence of nanomaterials markedly mitigated the liver toxicity of the drug (%damaged region: 42% and 10% for chemotherapy alone and with nanomaterials, respectively). Expression levels of DNA damage markers (Cleaved PARP) further confirmed the liver-protective function of nanomaterials. As there existed the possibility that mitigation of liver toxicity with nanomaterials proceeds in an NSG dispensable manner, liver-specific adeno-associated virus serotype 8 (AAV8) carrying TBG (thyroxine-binding globulin) promoter and shRNA targeting G3BP1 and hnRNPC (Fig. 10E, AAV denoted by #1 and #2, respectively), two components integral to the establishment of NSGs, were employed. Recorded serum biomarkers indicated that knockdown of each protein in the liver reduced the protective effect exerted by nanomaterials to a comparable extent, safely attributing the mitigated chemo-toxicity to NSGs.

During the systematic exposure to nanomaterials and chemotherapeutics, serum was collected for immune response assessment. Using an antibody array, the expression levels of pro-inflammatory cytokines in peripheral circulation were analyzed. Dye-labeled blots can be found in Fig. 10F and Supplementary Fig. 18C, with results ruling out the prevalence of an inflammatory response (Fig. 10G). Additionally, it was confirmed that exposure to nanomaterials caused negligible changes in organs beyond the liver, as indicated by intact tissue morphology (Supplementary Fig. 19). These findings collectively ensured the safe usage of CLS-modified nanomaterials for local chemo-plasticity reshaping and thereby modulating the therapeutic outcome.

Discussion

Nanomedicines have been extensively utilized for disease management. Most research efforts focus on understanding the implications of certain nanotherapeutics at transcriptional and translational levels^40,41. Upon exposure to nanomaterials, their outer surfaces are rapidly passivated by biomolecules both intra- and extracellularly, a phenomenon known as the protein corona effect (PCE)⁴². While PCE is widely accepted as resulting from random protein condensation enclosing nanomaterials, recent findings suggest that PCE may be a highly regulated and preferential process⁴³. Flexible proteins with intrinsic disordered regions are more likely to be adsorbed by nanomaterials compared to rigid ones. Once adsorbed, these flexible proteins undergo forcible unfolding to various extents depending on interactions with ligands functionalizing the nanomaterials, which cause their rigidity to reduce⁴⁴. Such a process facilitates multivalency binding with bystander biomolecules, thus creating an environment that benefits the formation of MLOs. Despite these observations, the influence of specific types of MLOs on the action and metabolism of nanomedicine remains unclear. Herein, we provide compelling evidence demonstrating that suitably functionalized nanomaterials can induce the formation of NSGs, which differ compositionally and physicochemically from their canonical counterparts.

For different types of SG subtypes (e.g., heat-shock-induced or AS-induced), although they differ in composition, a mild-to-moderate stress condition is required for their generation. Despite the difference in stimulation, c-SGs generally initiate from the nucleation of RNP-mRNA complex, followed by its growth through recruiting client proteins that establish a liquid-like coacervated droplet³. Less is known about non-canonical SGs that originate under stress-free conditions. Through stepwise all-atom simulations, it was shown that the sequential recruitment of hnRNPC and G3BP1, mediated respectively by multivalency interactions between CLS-hnRNPC and Coil_111-123 residue (hnRNPC) and NTF2L domain (G3BP1), can promote NSGs formation in the absence of stress. Truncation and KO assays confirmed that the sequential phase separation of hnRNPC and G3BP1 is crucial for NSGs formation. While secondary phase separation within already established MLO is observed in FUS, Tau, and TDP43-composed counterparts^45,46, there is no similar observation during the growth stage of SGs, whose client components were previously believed to be recruited in parallel⁵. Intriguingly, both NSGs and dysfunctional MLOs observed in functionally degenerative neurons feature less fluidity with respect to conventional counterparts like SGs, revealing that the sequential phase separation might be crucial for establishing a more closely packed microstructure. Our findings collectively suggest a seeded-growth-like mechanism underlying the growth of NSGs, termed tri-phase separation. Future research is encouraged to investigate the possibility of sequential phase separation occurring during the nucleation stage of other MLOs beyond SGs.

With the advancement of imaging modality, the microstructure of MLOs can be revealed in more detail regarding composition information. For the function executed by MLOs, predictions are made based on a collective consideration of physical, chemical, and biological features, although proper methods to assess the former two are still lacking¹. To address this challenge, a biorthogonal functional imaging approach was developed in this study using HaloTag-rendered, NSGs-specific imaging. By introducing physicochemical microenvironment-responsive, Halo-ligand-conjugated dyes, a set of parameters, including local temperature, polarity (physical parameters), pH, and proteasome activity (chemical parameters), was analyzed. Despite physical similarities between NSGs and canonical counterparts, their sub-SGs distribution of G3BP1 differed significantly based on measured local pH. Additionally, it was found that proteasome enrichment levels in NSGs are higher than those in canonical counterparts. During SG disassembly, proteasome recruitment is typically associated with imminent dissociation via degradation of ubiquitinated scaffold proteins³⁶. Therefore, the preservation of proteasome-collecting NSGs was unexpected. FRAP assays indicated that the gel-like structure of NSGs helps to hinder the interaction with proteasomes, thereby mitigating disassembly.

The disassembly of SGs can occur through ubiquitination or chaperone mediation. For ubiquitination, G3BP1 interacts with VCP-p97, leading to its proteasomal cleavage³⁶. In the chaperone-mediated pathway, HSP70 and HSP90 force refolding of scaffold proteins, destabilizing droplets⁴⁷. Inhibitor assays showed that chaperone pathways do not contribute to NSGs disassembly. Blocking the VCP function accelerates NSGs dissociation. Transcriptome and inhibitor assays confirmed dual modes of NSGs disassembly: VCP-mediated 19S proteasome acts slowly under normal conditions due to the gel-like structure of NSGs. Dysfunction of 19S leads to 20S proteasome compensating via SUMOylated G3BP1 recognition, speeding up NSGs disassembly. This finding aids in understanding non-canonical MLO disassembly and provides a precision inhibitor for modulating NSGs activity.

By manipulating NSGs using VCP inhibitors, their association with chemo-plasticity was confirmed. Subsequently, NSGs were leveraged as the chemosensitivity remodeling hub for in vivo applications. Striking a balance between the cytolytic effect on solid tumors and minimizing side effects on normal tissues has long been a challenging goal in clinical trials, often resulting in compromises in therapeutic outcomes. This study demonstrates that the site-specific provocation of NSGs holds promise for smart disease management. Specifically, for drugs whose efficacy is mitigated by NSGs, nanomaterial delivery stimulates these MLOs in liver tissues normally susceptible to therapeutic damage, thereby mitigating chemo-toxicity. Conversely, for drugs that synergize with NSGs, localized tumor introduction of nanomaterials provides a feasible yet robust method to reduce therapeutic dosage, achieving safe disease control in a biocompatible manner.

Research interests in targeting MLOs are increasing, particularly in the context of neurodegeneration due to the increasingly clarified association between disease progression and dysfunctional protein coacervates. Among various disease-related MLOs, hnRNPC and G3BP1 have been frequently highlighted, either through indirect association with disease progression as MLO scaffolds or direct contribution. As central compartments of NSGs, the potential of their generation in neurons to mitigate the liquid-to-gel transition of certain MLOs and optimize this process for disease management warrants further investigation. Another promising direction involves validating NSGs’ ability to destabilize already solidified components. Existing evidence supports the possible translocation of components from neurodegeneration-related granules to nanomaterials; however, precise control over this process remains unfulfilled. Sequential recruitment of candidate proteins via tailored functionalization of nanomaterials (as discussed herein and future designs beyond) may provide a pathway to targeted abrogation of MLOs through protein relocation. For nanomaterials within the size range examined in this study (30 nm and above), specific surface functionalization is required to enable blood-brain barrier permeability, which may reduce interactions between endocytosed nanomaterials and MLOs. One approach to address this involves utilizing established inter-organ communication pathways for drug delivery⁴⁸.

Nanotoxicity studies have extensively examined the biological effects of nanomaterials on transcription, translation, and metabolism. With more metastable cellular structures as MLOs, identified, understanding their interaction with nanomaterials in terms of nanotoxicity is just beginning. This study proposes a framework for identifying nanomaterial-associated MLOs and decoding their biological functions. Future research should investigate how certain nanomedicines affect the homeostasis of MLO networks in organs beyond tumor tissues, such as the liver, which accumulates most intravenously injected nanomaterials. These findings could provide insights into therapeutic potentials akin to NSGs or inform the design of safer next-generation nanomedicines. Additionally, although cytotoxicity linked to ultrafine nanomaterials has been reported¹⁴, this phenomenon is often attributed to oxidative stress. The extent to which ultrasmall nano-vehicle-induced MLO reprogramming contributes to cytotoxicity remains unclear and requires further investigation.

Our study also presents several limitations. While our all-atom MD simulations provide mechanistic insights into NSG formation, we must acknowledge the inherent limitations of computational modeling at nanomaterial-biointerfaces as an emerging frontier. Current force fields face challenges in accurately parameterizing dynamic ligand–protein interactions, necessitating ongoing methodological refinement. Future studies could enhance physical accuracy by integrating machine learning potentials⁴⁹. Due to reproducibility and biosafety concerns, we utilized noble metal-based nanomedicine, whose cost-effectiveness should be carefully evaluated before clinical translation. Additionally, the impact of NSGs on other therapeutics, particularly recently emerged precision drugs targeting the immune system, should be explored in future studies to expand the applicability of our therapeutic approach. Furthermore, while we focused on the protective effects of nanomaterials on the liver, spleen, and kidneys, evaluating therapeutic benefits in other tissues and disease models beyond cancer that involved an expanded cohort will help establish a more comprehensive understanding of NSG-induced chemo-plasticity reprogramming.

In conclusion, our study provides valuable insights into the MLO-nanomaterial interplay network, systematically evaluating the formation/dissociation mechanisms of NSGs and their microstructure details at multiple levels. Our findings enhance the understanding of nanomedicine, providing guidelines for future design of precision therapeutics. We also hope the study can inspire further exploration of promising methodologies to harness the therapeutic potential of artificial MLOs.

Methods

Ethical statement

This study was approved by the Experimental Animal Care and Use Committee of Affiliated Nanjing Drum Tower Hospital of Nanjing University Medical School (2023AE01016).

Chemicals and antibodies

Gold (III) chloride trihydrate (HAuCl₄·3H₂O, ≥99.9%), ascorbic acid (AA, ≥99.0%), sodium borohydride (NaBH₄, ≥98%), sodium bicarbonate (NaHCO₃, ≥99.5%) hexadecyltrimethylammonium bromide (CTAB, ≥99%), hexadecyltrimethylammonium chloride (CTAC, 25% wt/vol), sodium arsenite (NaAsO₂, ≥90%) Bovine serum albumin (BSA) and lipopolysaccharides (LPS, Escherichia coli O55:B5) were all obtained from Sigma–Aldrich (United States). D-Sorbitol solution, dimethyl sulfoxide (DMSO, anhydrous), Cy5-PEG_2k-SH, NHS-PEG_5k-SH, FA-PEG_1k-SH, 4-[3-(Trifluoromethyl)-3H-diazirin-3-yl] benzylamine hydrochloride (≥98%), potassium iodide (KI, ≥99%), iodine (I₂, ≥99.8%), puromycin dihydrochloride (≥98%), doxorubicin hydrochloride (98%), nocodazole (≥98%) and Streptavidin were all obtained from Aladdin (P.R. China). Paclitaxel (98%), cisplatin (65%), coumarin (98%), Iron (II) chloride tetrahydrate (FeCl₂·4H₂O, ≥98%), Iron (III) chloride (FeCl₃, ≥99%), Ammonium hydroxide solution (28 wt% NH₄OH), (3-Aminopropyl)triethoxysilane (APTES, ≥99%) were purchased from Macklin (P.R. China). OMe-PEG_1k-SH, OMe-PEG_5k-SH, OMe-PEG_10k-SH, OMe-PEG_20k-SH, OMe-PEG_30k-SH, OMe-PEG_40k-SH, COOH-PEG_5k-SH, and NH₂-PEG_5k-SH were purchased from JenKem Technology (P.R. China). CLS-PEG_5k-SH was purchased from YusiYY (P.R. China). Protonex Red 600 was purchased from KeyGEN BioTECH (P.R. China). Rhodamine B (97%), Tetrazine-sulfo-NHS ester, and TCO-Biotin were purchased from Bidepharm (P.R. China). HaloTag ligand (2-(2-(6-chlorohexyloxy) ethoxy) ethanamine hydrochloride, 97%) and LLVY-AMC-N₃ were purchased from Picass (P.R. China). TAK-243, Spautin-1, Myricetin, Grp94, AMP-PCP, CB5083, DBeQ, 2-D08, NMS-859, PYR-41, tomatine, and etoposide were all purchased from MedChemExpress (MCE, United States). All chemicals and solvents are used as received without further purification. Nuclear stain DAPI, Dulbecco’s Modified Eagle Medium (DMEM), phosphate-buffered saline (PBS), penicillin/streptomycin (100×), and fetal bovine serum (FBS) were obtained from Invitrogen (Thermo Fisher Scientific, United States). Deionized (DI) water with a resistivity of 18.2 MΩ·cm was used in all experiments, which was prepared using an ultrapure water system (Arium Mini, Sartorius, Germany). Primary antibodies: Anti-BiP Rabbit mAb (#3177, 1:400, Cell Signaling, United States), anti-CHOP Mouse mAb (#2895, 1:400, Cell Signaling), anti-G3BP1 Rabbit mAb (#61559, 1:400, Cell Signaling), Alexa Fluor 488-conjugated anti-G3BP1 Rabbit mAb (#94496, 1:400, Cell Signaling), anti-G3BP1 Mouse mAb (66486-1-LG, 1:500, Proteintech, P.R. China), anti-hnRNP C1/C2 Rabbit mAb (ab133607, 1:500, Abcam, UK), anti-ubiquitinated protein Mouse mAb (04-263, 1:200, Merck, Germany), anti-Cleaved PARP (Asp214) Rabbit mAb (#5625, 1:500, Cell Signaling), anti-Caspase8 Mouse mAb (#9746, 1:500, Cell Signaling), Alexa Fluor 647-conjugated anti-gamma H2A.X (ab195189, 1: 200, Abcam), Alexa Fluor 488-conjugated anti-Alpha Tubulin Mouse mAb (322588, 1:500, Invitrogen), anti-PSMB6 Rabbit mAb (GTX23331, 1:200, Genetex, United States), anti-Albumin Rabbit mAb (ab207327, 1:2000, Abcam), anti-NF-κB p65 Rabbit mAb (#8242,1:800, Cell Signaling), anti-Cleaved Caspase3 Rabbit mAb (#9664, 1:1000, Cell Signaling). Secondary antibodies: Alexa Fluor-488-labeled goat-anti-rabbit IgG (H+L) cross-adsorbed secondary antibody (A-11008, 1: 500, Invitrogen), Alexa Fluor 555-conjugated goat-anti-rabbit IgG H&L (#4413, 1:1000, Cell Signaling), Alexa Fluor 488-conjugated goat-anti-mouse IgG H&L (#4408, 1:1000, Cell Signaling), anti-Rabbit IgG, HRP-linked Antibody (#7074, 1:3000, Cell Signaling).

Synthesis and functionalization of Au nanospheres

AuNPs were synthesized using the seed-mediated growth method¹⁷. Briefly, gold nanoseeds were prepared by mixing HAuCl₄ (0.5 mM), NaBH₄ (10 mM), and CTAB (200 mM), followed by vortexing for 2 min until the solution turned brown. The mixture was incubated at 37 °C for 3 h to decompose excess NaBH₄. For 10 nm AuNPs, gold nanoseeds were added to a growth solution containing HAuCl₄ (0.5 mM), CTAC (200 mM), and ascorbic acid (100 mM), followed by 30 min incubation with shaking. Larger AuNPs (30–180 nm) were obtained through iterative seed-mediated growth with controlled injection of HAuCl₄ (0.5 mM) at 2 mL/h using a syringe pump. Particle size and morphology were characterized by TEM (JEM-1400, JEOL, Japan) and UV-vis spectroscopy (1800PC, JingHua, P.R. China).

For charged AuNPs, NH2-PEG-SH and COOH-PEG-SH were added into the aqueous solution of freshly obtained nanospheres to reach a final concentration of 10 μg/mL. After 4 h vortexing, FA-PEG-SH and Cy5-PEG-SH were added and incubated overnight. The product was collected by centrifugation at 15,000 × g for 15 min, washed once with water, and re-dispersed in deionized water to OD ~ 5.

For CLS-PEG-SH modification, CLS-PEG-SH and OMe-PEG-SH were mixed at varying ratios (ranging from 20% to 100%) with AuNPs, followed by FA-PEG-SH/Cy5-PEG-SH addition. For photo-crosslinkable AuNPs, diazirine-conjugated-PEG-SH was synthesized by reacting 4-[3-(Trifluoromethyl)-3H-diazirin-3-yl] benzylamine hydrochloride with NHS-PEG-SH in NaHCO₃ (pH~8.0) at 4 °C overnight, then mixed with AuNPs and functionalized as above.

Cell culture and treatments

Human cervical carcinoma cell line (HeLa) was obtained from American Type Culture Collection (ATCC, P.R. China) and cultured in the growth medium of Dulbecco’s Modified Eagle’s Medium supplemented with 10% FBS and 1% penicillin/streptomycin. Cells were cultured at 37 °C under 5% CO₂ in a humidified incubator (MCO-170AICUV-PE, PHC, Japan). Upon reaching ~85% confluency, cells were trypsinized using 0.25% trypsin (KeyGen Biotech, P.R. China) and passaged for further experiments. For stress granule induction, cells were treated with: sodium arsenite (AS, 0.3 M, 30 min), sorbitol (0.4 M, 90 min), or various functionalized AuNPs (OD ~ 5, 1:100 dilution in complete media for 4 h unless otherwise specified). Inhibitors were applied at: myricetin (50 μM, 3 h), ML346 (20 μM, 4 h), Grp94 (4 nM, 4 h), AMP-PCP (50 μM, 4 h), spautin-1 (10 μM, 4 h), TAK-243 (1 μM, 20 min), CB5083 (10 μM, 30 min), DBeQ (10 μM, 4 h), NMS-859 (5 μM, 6 h), Tomatine (5 μM, 4 h), 2-D08 (200 μM, 24 h), PYR-41 (10 μM, 4 h), LPS (5 μg/mL, 24 h) and Etoposide (50 μM, 24 h). Chemotherapeutic agents were administered at: Nocodazole (1, 50, or 100 nM, 2 h), Paclitaxel (1 or 5 μM, 4 h), and Doxorubicin (2 μM, 2 h). For heat stress, HeLa cells in an 18-well plate (ibidi, Germany) were incubated at 42 °C with 5% CO₂ for 1 h.

Transfection

The SLenti-CMV-EGFP-G3BP1-puro lentivirus was purchased from OBiO (TS0517, P.R. China), for sgRNA plasmids (sgRNA sequence: CTCATTAACATACTGAACGA):

5′-ATGGACTATCATATGCTTACCGTAACTTGAAAGTATTTCGATTTCTTGGCTTTATATATCTTGTGGAAAGGACGAAACACCGCTCATTAACATACTGAACGAGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGCTTTTTTG-3′

It was provided by Beyotime (L02615, P.R. China). Truncated G3BP1 plasmids were constructed by Tsingke Biotech (P.R. China). CO., Ltd (Contact Da Huo for reasonable sharing request). In a typical assay, HeLa cells were seeded at 1 × 10⁵ cells per well. After overnight culture at 37 °C with 5% CO₂, cells reaching 30%–50% confluency were transfected with lentivirus at MOI = 10 in fresh medium containing polybrene (final concentration 5 μg/mL). Following 24-h incubation, the viral supernatant was replaced with complete medium. At 48 h post-transfection, EGFP-G3BP1 expression was preliminarily assessed by fluorescence microscopy. When cell density reached 50%–70% confluency, selection was initiated using 1 μM puromycin-containing medium for one week to establish stable transductions.

DL-Stbl3 competent cells (DLC106, Tsingke Biotech) were transformed with the hnRNPC KO plasmid following the standard heat-shock protocol. Positive clones were expanded in LB/ampicillin, and plasmids were purified using a commercial kit (I613KA4995, Sangon Biotech, P.R. China) following the manufacturer’s instructions. For plasmid transfection, HeLa cells were seeded at 5 × 10³ cells/well in chamber slides and allowed to reach 50%–70% confluency. Plasmid DNA (1 μg) was delivered using Lipofectamine LTX (A12621, Thermo Fisher Scientific) according to the manufacturer’s instructions, with transgene expression assessed 24-48 h post-transfection.

AAV8-TBG-vehicle was a kind gift from Dr. Pengxiang Min; the corresponding sequence is as follows:

shG3BP1 sequence: (target: CGGGAATTTGTGAGACAGTAT)

5′- CCGGCGGGAATTTGTGAGACAGTATCTCGAGATACTGTCTCACAAATTCCCGTTTTT -3′

shhnRNPC sequence: (target: GCCTTCGTTCAGTATGTTAAT)

5′- CCGGGCCTTCGTTCAGTATGTTAATCTCGAGATTAACATACTGAACGAAGGCTTTTT -3′

Fluorescence imaging and data processing

Cells were fixed with 4% paraformaldehyde (P0098, Beyotime) for 10 min at room temperature and washed three times with PBS. Subsequently, cells were permeabilized with 0.1% Triton X-100 (v/v, 9036-19-5, Sigma) in PBS for 15 min before blocking with PBS containing 5% goat serum for 1 h at room temperature. Target proteins were immuno-stained with diluted primary antibodies overnight at 4 °C, followed by incubation with appropriate secondary antibodies (except when using directly labeled primary antibodies). Fluorescence imaging was performed using a TIRF microscope (Thunder Imager with Infinity TIRF system, Leica, Germany) equipped with 405 nm, 488 nm, 561 nm, and 638 nm lasers and a 100× plan-Apochromat objective (NA 1.47). Cells grown in 18-well glass-bottom dishes (ibidi) were imaged in epi mode. Acquired images were processed using Leica LAS X software (version 3.2.1) for computational clearing and deconvolution, achieving final resolutions of 136 nm (x-y axis) and 300 nm (z-axis). Three-dimensional reconstruction and quantitative analysis were performed using Imaris software (version 9.9, Oxford Instruments, UK) with Surface, Spot, and Cell modules, applying consistent thresholds. The analysis included measurements of: Total object volume, intensity mean, shortest distance to recognized surface, sphericity, and colocalization. Isotropy of microtubules was quantified using ImageJ software with FibrilTool and OrientationJ plugins.

Fluorescence Recovery After Photobleaching (FRAP) and time-lapse imaging

HeLa^{G3BP1-EGFP OE} cells were seeded into an ibidi 18-well plate at a concentration of 5 × 10³ cells/well and cultured overnight at 37 °C with 5% CO₂ before being treated with either sodium arsenite (AS), sorbitol, or dye-labeled 60%CLS-AuNPs. Following treatment, cells were washed with pre-warmed D-PBS and moved into a stage-top live imaging chamber (ibidi). FRAP was performed using a laser-scanning Leica inverted microscope with a 100× oil immersion objective with a numerical aperture (NA) equal to 1.47 (Apo oil immersion, Vizla, Germany). At least three stress granules (1–2.5 μm diameter) per condition were selected for photobleaching (10 ms pulse, 30 mW) followed by time-lapse imaging at 2-s intervals. Data normalization and analysis were performed using Leica LAS X software (version 3.2.1).

Analysis of intracellular protein corona

To investigate protein corona adsorbed on nanospheres, HeLa cells were co-cultured with diazirine-modified 60%CLS-AuNPs (50 μL, 10¹² particles mL^-1) for 4 h. Following removal of unbound nanospheres, cells were irradiated with 365 nm ultraviolet light (15 mW/cm², 20 min) to facilitate covalent protein-nanosphere crosslinking. Cells were then lysed using RIPA buffer (10 min, PD0013D Beyotime) and centrifuged at 18,000 × g to pellet the nanosphere-protein complexes. The resulting precipitate was subjected to chemical etching with KI/I₂ (500 μL, containing 2.5 mM KI and 0.34 mM I₂) to liberate bound proteins, which were subsequently purified through 3 kDa molecular weight cutoff ultrafiltration. Protein identification was performed by label-free LC–MS analysis as described in the following section.

Analysis of extracellular protein trafficking

Magnetic Fe₃O₄ nanoparticles were synthesized via chemical co-precipitation of Fe²⁺ and Fe³⁺ salts under a nitrogen atmosphere, using ammonium hydroxide as the precipitant. The nanoparticles were washed and collected magnetically. Surface amination was achieved by treating the Fe₃O₄ nanoparticles with (3-aminopropyl) triethoxysilane (APTES) under acidic conditions. The resulting aminated nanoparticles (Fe₃O₄-NH₂) were thoroughly washed and dried. For streptavidin conjugation, streptavidin was first activated in PBS using excess EDC and NHS. The activated streptavidin was then reacted with Fe₃O₄-NH₂ nanoparticles. The conjugated beads (Fe₃O₄-Streptavidin) were magnetically separated and blocked with 1% bovine serum albumin (BSA) to minimize non-specific binding.

To analyze the extracellular protein corona, serum proteins were first labeled with tetrazine-sulfo-NHS ester (targeting lysine residues). These tetrazine-labeled proteins were adsorbed onto nanomaterials to form a corona. The corona-coated nanoparticles were isolated, washed, and incubated with cells in wild-type serum. After photo-ligation and cell lysis, intracellular corona components and total lysate were subjected to click chemistry-mediated immunoprecipitation using a TCO-Biotin linker (RT, 15 min), followed by capture with the prepared Fe₃O₄-Streptavidin magnetic beads. The captured protein corona components were then quantified by Western blotting (WB).

Co-IP experiments and proteomics

The Co-immunoprecipitation (Co-IP) experiment was conducted utilizing the Classic IP Kit (#26146, Thermo Fisher Scientific) to selectively enrich G3BP1 protein along with its interacting partners, in accordance with the manufacturer’s instructions. A total of 18 experimental samples were analyzed: three groups (NH₂-AuNPs, 60% CLS-AuNPs, and mPEG-AuNPs, on nanomaterials vs G3BP1-IP alone) × n = 2 independent biological replicates each. For each replicate, HeLa cells were treated as described in the main text. The proteomic analysis was conducted by Verygenome Technology Co. Ltd. For the sample preparation, cells were lysed in RIPA buffer (10 min, 4 °C), and nanomaterial–protein complexes were isolated by chemo-etching-assisted proximity labeling (see Fig. 2H). After elution, proteins were reduced with 10 mM DTT (56 °C, 30 min), alkylated with 20 mM iodoacetamide (RT, dark, 30 min), and digested overnight at 37 °C with sequencing-grade trypsin (1:50 enzyme:protein, w/w). Peptides were desalted on C18 StageTips and dried under vacuum prior to LC–MS/MS. Peptides were analyzed on an Easy-nLC 1000 UHPLC coupled to a Q Exactive HF Orbitrap mass spectrometer (Thermo Fisher Scientific). Peptides were separated on a 15 cm × 75 µm ID C18 column (2 µm particle size) using a 90 min linear gradient from 5% to 35% solvent B (0.1% formic acid in acetonitrile) at 300 nL/min. MS acquisition was in data-dependent mode: full MS scans at 60,000 resolution (m/z 350–1,600), AGC target 3e6, max IT 50 ms; top 20 precursors fragmented by HCD (NCE 27), MS/MS resolution 15,000, AGC target 1e5, max IT 100 ms, dynamic exclusion 30 s. Raw files were processed in Proteome Discoverer v2.4 (Thermo) using the Sequest HT search engine against the UniProtKB/Swiss-Prot human database (release 2023_02). Parameters: Enzyme (trypsin, max 2 missed cleavages); Fixed modification (carbamidomethyl, C); Variable modifications (oxidation (M), acetyl (protein N-term)); Precursor mass tolerance (10 ppm); Fragment mass tolerance (0.02 Da); Minimum peptide length (7 amino acids); Peptide-spectrum match (PSM) and protein-level FDR ≤ 1% using Percolator; Minimum 1 unique peptide per protein for identification; Score cut-off: q-value < 0.01; PTM localization by ptmRS node (threshold 0.75). Label-free quantification was performed using precursor ion intensities normalized across runs. Statistical analysis of differential protein abundance was conducted in Perseus v1.6.15.0 using log₂-transformed LFQ intensities; missing values were imputed from a normal distribution (width 0.3, downshift 1.8). Significance was assessed by two-sided Student’s t-test with Benjamini–Hochberg FDR correction (FDR 0.05). GO and KEGG enrichment were performed using STRING v11.5 and DAVID v6.8. Visualization was done in GraphPad Prism 9 and Cytoscape v3.9.1.

All-atom molecular dynamics simulation

Given that the AuNPs (~70 nm) in the experiment are much larger than those of the hnRNPC and G3BP1 ( < 10 nm), we modeled the AuNPs using a flat Au(111) substrate with the size of 10.38 × 10.49 nm² ⁵⁰. mPEG and CLS-PEG with 23 repeat units were used to modify the AuNP surfaces. The three surface-modified AuNPs are composed of 100% mPEG (mPEG), 100% CLS-PEG (100% CLS), and a combination of 40% mPEG and 60% CLS-PEG (60% CLS), respectively. Each PEG molecule is covalently bonded to the AuNP surface via Au-S bonds. The structure of hnRNPC was sourced from AlphaFold (https://alphafold.ebi.ac.uk, UniProt: P07910), while the crystal structure of the NTF2L domain of G3BP1 was obtained from the Protein Data Bank (PDB entry: 7SUO⁵¹). For the hnRNPC-NTF2L complex, the protein was initially positioned at the center of a cubic box, ensuring that the minimum distance between the protein and its mirror image exceeds 3 nm. The system was solvated in a 12 × 12 × 12 nm³ cubic box with 53,997 water molecules and 0.15 M NaCl, comprising 169,216 atoms in total. In the case of the AuNPs or the protein-AuNPs complex, the minimum distance between the PEGs or the protein and the mirrored gold substrate is greater than 2 nm. For the protein-AuNPs complex, the protein was initially placed above the AuNPs, with the minimum distance between the protein and the PEGs exceeding 1 nm. Each complex was solvated in water, with NaCl added to neutralize the system at a concentration of 0.15 M. The simulation boxes for the mPEG, 60% CLS, and 100% CLS systems had lateral dimensions of 10.38 × 10.49 nm², with z-dimensions of 12.99, 13.48, and 13.49 nm, respectively. Each system contained 4536 Au atoms, 168 PEG molecules, and one hnRNPC protein, solvated with 32,511, 32,304, and 31,002 water molecules, yielding total atom counts of 136,052, 142,713, and 143,631 for the mPEG, 60% CLS, and 100% CLS systems, respectively.

For the hnRNPC-NTF2L complex, the system was first energy-minimized using the steepest descent method until convergence was achieved. This was followed by 500 ps of pre-equilibration in the NVT ensemble and an additional 500 ps in the NPT ensemble, during which all protein heavy atoms were restrained with harmonic potentials of 1000 kJ·mol⁻¹·nm⁻². Finally, the restraints were removed, and 500 ns production MD simulations in the NPT ensemble were carried out with a 2 fs timestep. For the AuNP and AuNP-protein systems, after energy minimization, the systems also underwent 500 ps of NVT and 500 ps of NPT pre-equilibration, during which all gold atoms, as well as the heavy atoms of the proteins and PEGs, were restrained with harmonic constraints of 1000 kJ·mol⁻¹·nm⁻². Restraints on the proteins and PEGs were subsequently released, followed by 500 ns production simulations in the NPT ensemble with a 2 fs timestep. For each system, we performed three independent simulations to ensure reproducibility. The temperature was maintained at 310 K using the V-rescale thermostat⁵² with a time constant of 0.2 ps, while the pressure was kept at 1 atm using the Parrinello-Rahman barostat with a time constant of 2.0 ps and a compressibility constant of 4.5 × 10⁻⁵ bar⁻¹. Notably, in the case of AuNPs or the protein-AuNPs complex, pressure coupling was applied only in the z-direction. The final 200 ns trajectory with an interval of 1 ns from each run was used for binding energy analysis.

All-atom molecular dynamics (MD) simulations were performed using the GROMACS 2020.6 package⁵³ with the Amber14SB force field⁵⁴ and the TIP3P water model. The General Amber force field⁵⁵ (GAFF) parameter for mPEG and CLS-PEG, including methoxy and cholesterol, was generated using AmberTools23⁵⁶. RESP2 charge⁵⁷ were obtained through B3LYP/def2TZVP quantum calculations of the electrostatic potential, performed with the Gaussian09 program and fitted using Multifwn⁵⁸. Parameters for the gold atom were obtained from previous work^59,60. The Lennard–Jones interaction was computed using a cutoff of 1.2 nm, and long-range electrostatic interactions were treated using the particle-mesh Ewald (PME) method. The bonds involving hydrogen atoms were constrained by the LINCS algorithm⁶¹. Periodic boundary conditions were applied in all three directions.

Binding energies were calculated using the molecular mechanics/generalized Born surface area (MM/GBSA) method⁶². The binding energy is composed of four terms, including the van der Waals interaction (${\Delta {{{\rm{E}}}}}_{{{{\rm{VDW}}}}}$), the electrostatic interaction (${\Delta {{{\rm{E}}}}}_{{{{\rm{ELE}}}}}$), the polar solvation energy (${\Delta {{{\rm{E}}}}}_{{{{\rm{polar}}}}}$), and the nonpolar solvation energy (${\Delta {{{\rm{E}}}}}_{{{{\rm{SA}}}}}$). The contact surface area (CSA) was calculated by

$${{{\rm{CSA}}}}=\left({{\mbox{SASA}}}_{{\mbox{A}}}+{{\mbox{SASA}}}_{{{{\rm{B}}}}}-{{\mbox{SASA}}}_{{\mbox{AB}}}\right)/2$$

where SASA represents the solvent accessible surface area, A and B represent the individual molecules, while AB denotes the complex. Typical snapshots and corresponding key residue interface graphs were visualized using PyMOL⁶³.

NSG microenvironment measurement method

Four distinct HaloTag ligand-dye conjugates were synthesized under light-protected conditions: (1) Protonex Red-HaloTag ligand was prepared by mixing 10 mM Protonex Red and HaloTag ligand solutions in a 1:3-5 molar ratio in PBS buffer at 4 °C overnight (final concentration: 80 μM); (2) LLVY-AMC conjugate was synthesized by combining 10 mM LLVY-AMC, 10 mM HaloTag ligand, and 5 mM acetylene-succinimide in a 2:3:2 molar ratio in PBS buffer with DMSO, followed by 37 °C overnight shaking (final concentration: 1 mM); (3–4) Rhodamine B- and Coumarin-HaloTag ligands were prepared at final concentrations of 14.6 mM and 20 mM, respectively.

For microenvironment imaging, cells were stained with HaloTag ligand-dye conjugates at optimized concentrations: Protonex-HaloTag ligand (0.1 μM), LLVY-HaloTag ligand (1 μM), Rhodamine B-HaloTag ligand (10 nM), and Coumarin-HaloTag ligand (0.1 μM) for 1 h prior to treatment with either sodium arsenite (AS) or 60%CLS-AuNPs, followed by fixation with 4% paraformaldehyde and fluorescence microscopy imaging.

Transcriptome

The transcriptome libraries were prepared using Trizol (R401-01, Vazyme, P.R. China) following the manufacturer’s instructions. Transcriptome analysis was provided by Wuhan Frasergen Genomic Medicine CO., Ltd. Raw reads were processed using trim-galore, and clean reads were mapped to mm 10 from UCSC genome1 using Hisat2 (v2.1.0) with default parameters. Differential analysis was performed using edgeR algorithms with fold change ≥1.2 and p-value ≤ 0.05 as the thresholds. For data analysis visualization, the CNSknowall Online Tool (https://cnsknowall.com) was harnessed.

Bio-distribution

Female BALB/C nude mice (4–6 weeks) were purchased from the Comparative Medicine Centre of Yangzhou University and bred at the Specific Pathogen Free (SPF) Experimental Animal Center of Affiliated Nanjing Drum Tower Hospital of Nanjing University Medical School. Mice were housed under a 12-h light/dark cycle at 22–24 °C and 40%–60% humidity with ad libitum access to food and water. Choice of animal sex was made based on the subsequent study, wherein a cervical tumor xenograft was established. For the sample size, it has been specified in each section. Once reaching a body weight of roughly 25 grams, the mice were randomly divided into several groups (detailed treatment as described in each case) post an intravenous injection of CLS-AuNPs (20 mg/Kg body weight) through the tail vein. At certain time points, mice were sacrificed to harvest tissues including liver, spleen, and kidneys. For those tissues, they were weighed, followed by repeated freeze-thawing to entirely release the nanomaterials within cells. Afterwards, the tissues were dissolved in heated aqua regia. DI water was supplemented into the solution to make a final volume of 5 mL containing 1% nitric acid (vol/vol). The content of gold in each sample was measured by using ICP-MS (Agilent 7500ce, Agilent, United States), and the biodistribution profiles of nanomaterials were calculated based on the weight of tissue measured right after collection.

Chemo-toxicity mitigation and chemo-sensitization

To evaluate the action of NSGs formation related to chemo-toxicity mitigation. BALB/c nude mice received a treatment regimen identical to that of the biodistribution assay. After 48 h, the mice received an i.v. injection of chemotherapeutic agents through the tail vein (dissolved in saline, 10, 15, and 10 mg/Kg body weight for Doxorubicin, cisplatin, and paclitaxel, respectively). For mice that received only nanomaterials or chemotherapeutics, we followed the same dosage. Peripheral blood of mice was collected, and potential tissue toxicity caused by either nanomaterials or chemotherapeutics was reflected by the measured levels of ALT (Alanine Transaminase), GGT (Gamma-Glutamyl Transferase), and Scr (Serum Creatinine). For β2-MG (β-2-Microglobulin), PDGF (Platelet-derived growth factor), and TNF-α (Tumor necrosis factor α), we harnessed ELISA kits for their measurement (β2-MG: #D721167, Sangon Biotech, P.R. China; PDGF and TNF-α, #KTE4053 and # KTE7015, all from Abbkine, P. R. China).

To evaluate the chemo-toxicity mitigation effect in tumor-bearing mice. The xenograft tumor model was first established using wild-type or gene-edited HeLa cells through subcutaneous injection (4 × 10⁵ cells per mouse) suspended in PBS into the flank of the mouse. According to the ARRIVE guide 2.0⁶⁴, mice successfully developed a tumor xenograft were included for subsequent study. After the size of the tumor grew to 100 μm³, mice were randomly grouped into treated and control groups for subsequent analysis. Animals were randomized into treatment arms using a computer-generated list. Cage distribution was balanced (maximum three animals per cage per treatment). Allocation was concealed by coding treatment syringes/vials; dosing and outcome acquisition were performed by technicians blinded to group labels where feasible. The maximal tumor burden permitted by our IACUC (e.g., 1.5 cm maximal diameter or 1500 mm³ volume) was not exceeded in any experiment. For the dosage of chemotherapeutics, we followed the treatment regimen as described above, except for nocodazole (5 mg/Kg body weight). For both nanomaterials and chemotherapeutics, they were administered every four days. Mice were sacrificed once their tumor tissue reached a volume of 1 mm³ or at the end of the therapeutic outcome study. Both the growth curves of the tumor tissue and the overall survival rate of mice in each group were recorded to assess the chem-sensitization action rendered by NSGs.

Protein array assay

Mouse serum was collected and maintained at −80 °C prior to analysis. Antibody array assays (Mouse Inflammation Cytokine Array, Ray Biotech Inc, United States) were carried out according to the manufacturer’s instructions. In brief, the diluted serum was incubated with the microarray glass substrate under gentle shaking, followed by multiple washes to remove unbound cytokines. Afterwards, a cocktail of antibodies and an equivalent amount of dye-conjugated Streptavidin were added stepwise and subjected to repeated washes with PBS. The signals were visualized by using a laser scanner equipped with a filter for the corresponding dye (Axon GenePix). Data was extracted using ScanArray Express.

Flow-cytometry analysis

For the quantification of the viability of cancer cells, mice in different groups were sacrificed at 8 h post-treatment, with their tumor tissues harvested and cleaned rapidly through ice-cold PBS to remove blood and tissue debris. Afterwards, the tumor tissue was minced into pieces with a mean volume of 2~5 mm³ using an eye scissor. These tissue fragments were immediately transferred to DMEM/F12 culture medium (Gibco) supplemented with 30 μg of type IV collagenase (Sigma–Aldrich) and type I DNase (Roche) and incubated for 4 h at 37 °C under gentle shaking. The softened tumor tissue was next filtered through a 400-mesh screen, followed by centrifugation (200 × g for 5 min). Cell debris in the supernatant was discarded, and the cells were resuspended with PBS for labeling with Annexin V-FITC/PI (#KGA108, Keygen Biotech., P.R. China) following the manufacturer’s instructions. Flow-cytometry analysis was conducted on FACS Verse (BD Bioscience, United States) and processed using FlowJo software (version 10.0.7).

Statistical analysis

All experiments were repeated n = 3 independent biological replicates unless stated otherwise. The unit of analysis for statistics was the per-replicate mean to avoid inflating n with technical repeats. Data are shown as mean ± SD (unless box/violin plots are used, for which center, box, and whiskers respectively mean median, IQR, and 5th–95th percentiles; all data points were displayed unless their presentation prevents the plot from being clearly visualized, in this case, the entire information can be found in the source data). Statistical tests are stated in each legend; two-sided tests were used with Tukey correction for multiple comparisons after one-way ANOVA (normality met) or Kruskal–Wallis otherwise. P-values, test statistics (t/F/df), and sample sizes can be found in legends or Source Data. No data were excluded unless pre-specified (see “Reporting summary” section).

No formal statistical power calculation was performed prior to this study because effect sizes for nanomaterial-induced stress granule formation and related in vivo outcomes were not available in the literature. Instead, sample sizes were based on (i) prior studies of stress granules and nanomaterials reporting robust effect sizes with n = 3 biological replicates for cell-based assays and n = 5–7 animals per group for xenograft models, and (ii) Nature Portfolio’s minimum standards for reproducibility. For cell-based experiments, each condition was repeated in ≥3 independent biological replicates on different days, analyzing ≥200 randomly selected cells per-replicate; the unit of analysis was the per-replicate mean. For animal studies, group sizes (n = 5 for toxicity assays; n = 7 for therapeutic outcome trials) were chosen to allow detection of large effect sizes observed in pilot experiments while minimizing animal use, consistent with ARRIVE guidelines. Baseline tumor volume and body weight were balanced across groups to reduce variability. These sample sizes yielded consistent results across replicates, and raw data are provided in the Source Data file.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Unless otherwise stated, all data supporting the results of this study are available in the article, Supplementary Information, and Source Data files. RNA-seq data are deposited under NCBI BioProject under accession code PRJNA1311971 [https://dataview.ncbi.nlm.nih.gov/object/PRJNA1294231?reviewer=lnradfp6uv0tsh0df2o7g09foa]. Proteomics data have been deposited to PRIDE/ProteomeXchange under accession code PXD067940 [https://www.iprox.cn/page/subproject.html?id=IPX0013238001]. Source Data are provided with this paper.

Code availability

All-atom simulations code is available on the FigShare Project repository [https://doi.org/10.6084/m9.figshare.29995738].

References

Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. Biomolecular condensates: organizers of cellular biochemistry. Nat. Rev. Mol. Cell Biol. 18, 285–298 (2017).
Article CAS PubMed PubMed Central Google Scholar
Anderson, P. & Kedersha, N. Stress granules. Curr. Biol. 19, R397–R398 (2009).
Article CAS PubMed Google Scholar
Protter, D. S. W. & Parker, R. Principles and properties of stress granules. Trends Cell Biol. 26, 668–679 (2016).
Article CAS PubMed PubMed Central Google Scholar
Guillén-Boixet, J. et al. RNA-induced conformational switching and clustering of G3BP drive stress granule assembly by condensation. Cell 181, 346–361 (2020).
Article PubMed PubMed Central Google Scholar
Yang, P. et al. G3BP1 is a tunable switch that triggers phase separation to assemble stress granules. Cell 181, 325–345 (2020).
Article CAS PubMed PubMed Central Google Scholar
Hu, S. et al. Time-resolved proteomic profiling reveals compositional and functional transitions across the stress granule life cycle. Nat. Commun. 14, 7782 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Leśniczak-Staszak, M. et al. Stress granule-mediated sequestration of EGR1 mRNAs correlates with lomustine-induced cell death prevention. J. Cell Sci. 137, jcs261825 (2024).
Article PubMed PubMed Central Google Scholar
Dong, T. et al. G3BP1/2-targeting PROTAC disrupts stress granules dependent ATF4 migracytosis as cancer therapy. J. Am. Chem. Soc. 147, 446–461 (2025).
Article ADS CAS PubMed Google Scholar
Li, X. et al. Stress granules sequester autophagy proteins to facilitate plant recovery from heat stress. Nat. Commun. 15, 10910 (2024).
Article ADS PubMed PubMed Central Google Scholar
Li, H. et al. MG53 suppresses tumor progression and stress granule formation by modulating G3BP2 activity in non-small cell lung cancer. Mol. Cancer 20, 118 (2021).
Article CAS PubMed PubMed Central Google Scholar
Mehta, S. & Zhang, J. Liquid-liquid phase separation drives cellular function and dysfunction in cancer. Nat. Rev. Cancer 22, 239–252 (2022).
Article CAS PubMed PubMed Central Google Scholar
Li, T., Zeng, Z., Fan, C. & Xiong, W. Role of stress granules in tumorigenesis and cancer therapy. Biochim. Biophys. Acta Rev. Cancer 1878, 189006 (2023).
Article CAS PubMed Google Scholar
Ryan, V. H. & Fawzi, N. L. Physiological, pathological, and targetable membraneless organelles in neurons. Trends Neurosci. 42, 693–708 (2019).
Article CAS PubMed PubMed Central Google Scholar
Yang, X., Yang, M., Pang, B., Vara, M. & Xia, Y. Gold nanomaterialas at work in biomedicine. Chem. Rev. 115, 10410–10488 (2015).
Article CAS PubMed Google Scholar
Sperling, R., Gil, P., Zhang, F., Zanella, M. & Parak, W. Biological applications of gold nanoparticles. Chem. Soc. Rev. 37, 1896–1908 (2008).
Article CAS PubMed Google Scholar
Liu, J. et al. Emerging of ultrafine membraneless organelles as the missing piece of nanostress: mechanism of biogenesis and implications at multilevels. ACS Nano 19, 5659–5679 (2025).
Article CAS PubMed Google Scholar
Zheng, Y. et al. Seed-mediated synthesis of single-crystal gold nanospheres with controlled diameters in the range 5-30 nm and their self-assembly upon dilution. Chem. Asian J. 8, 792–799 (2013).
Article ADS CAS PubMed Google Scholar
Li, Q. et al. Decoding the interplay between m⁶A modification and stress granule stability by live-cell imaging. Sci. Adv. 10, eadp5689 (2024).
Article CAS PubMed PubMed Central Google Scholar
Fu, Y. & Zhuang, X. m⁶A-binding YTHDF proteins promote stress granule formation. Nat. Chem. Biol. 16, 955–963 (2020).
Article CAS PubMed PubMed Central Google Scholar
Gasset-Rosa, F. et al. Cytoplasmic TDP-43 de-mixing independent of stress granules drives inhibition of nuclear import, loss of nuclear TDP-43, and cell death. Neuron 102, 339–357 (2019).
Article CAS PubMed PubMed Central Google Scholar
Han, S. et al. Endosomal sorting results in a selective separation of the protein corona from nanoparticles. Nat. Commun. 14, 295 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, S. et al. Mammalian IRE1α dynamically and functionally coalesces with stress granules. Nat. Cell Biol. 26, 917–931 (2024).
Article CAS PubMed Google Scholar
Zheng, W. et al. C9orf72 regulates the unfolded protein response and stress granule formation by interacting with eIF2α. Theranostics 12, 7289–7306 (2022).
Article CAS PubMed PubMed Central Google Scholar
Chen, Z. et al. Screening membraneless organelle participants with machine-learning models that integrate multimodal features. Proc. Natl. Acad. Sci. USA 119, e2115369119 (2022).
Article CAS PubMed PubMed Central Google Scholar
Kozakov, D. et al. The ClusPro web server for protein-protein docking. Nat. Protoc. 12, 255–278 (2017).
Article CAS PubMed PubMed Central Google Scholar
Abramson, J. et al. Addendum: accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 636, E4 (2024).
Article CAS PubMed PubMed Central Google Scholar
Song, D. et al. Yin and yang regulation of stress granules by Caprin-1. Proc. Natl. Acad. Sci. USA 119, e2207975119 (2022).
Article CAS PubMed PubMed Central Google Scholar
Zhao, Z. et al. QKI shuttles internal m⁷G-modified transcripts into stress granules and modulates mRNA metabolism. Cell 186, 3208–3226 (2023).
Article CAS PubMed PubMed Central Google Scholar
Ren, J. et al. Emerging implications of phase separation in cancer. Adv. Sci. 9, e2202855 (2022).
Article Google Scholar
Los, G. V. et al. HaloTag: a novel protein labeling technology for cell imaging and protein analysis. ACS Chem. Biol. 3, 373–382 (2008).
Article CAS PubMed Google Scholar
Soleilhac, A. et al. Temperature response of rhodamine B-doped latex particles: from solution to single particles. Langmuir 32, 4052–4058 (2016).
Article CAS PubMed Google Scholar
Salehian, F. et al. A review: biologically active 3,4-heterocycle-fused coumarins. Eur. J. Med. Chem. 212, 113034 (2021).
Article CAS PubMed Google Scholar
Ye, S. et al. Micropolarity governs the structural organization of biomolecular condensates. Nat. Chem. Biol. 20, 443–451 (2024).
Article CAS PubMed Google Scholar
Shin, Y. & Brangwynne, C. P. Liquid phase condensation in cell physiology and disease. Science 357, eaaf4382 (2017).
Article PubMed Google Scholar
Riback, J. A. et al. Composition-dependent thermodynamics of intracellular phase separation. Nature 581, 209–214 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Gwon, Y. et al. Ubiquitination of G3BP1 mediates stress granule disassembly in a context-specific manner. Science 372, eabf6548 (2021).
Article CAS PubMed PubMed Central Google Scholar
Sahu, I. et al. The 20S as a stand-alone proteasome in cells can degrade the ubiquitin tag. Nat. Commun. 12, 6173 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Böddeker, T. J. et al. Non-specific adhesive forces between filaments and membraneless organelles. Nat. Phys. 18, 571–578 (2022).
Article PubMed PubMed Central Google Scholar
Klein, I. A. et al. Partitioning of cancer therapeutics in nuclear condensates. Science 368, 1386–1392 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Liz-Marzán, L. M. et al. What do we mean when we say nanomedicine? ACS Nano 16, 13257–13259 (2022).
Article Google Scholar
de Lázaro, I. & Mooney, D. J. Obstacles and opportunities in a forward vision for cancer nanomedicine. Nat. Mater. 20, 1469–1479 (2021).
Article ADS PubMed Google Scholar
Monopoli, M. P. et al. Biomolecular coronas provide the biological identity of nanosized materials. Nat. Nanotechnol. 7, 779–786 (2012).
Article ADS CAS PubMed Google Scholar
Lu, X. et al. Tailoring the component of protein corona via simple chemistry. Nat. Commun. 10, 4520 (2019).
Article ADS PubMed PubMed Central Google Scholar
Cong, Y. et al. Protein corona-mediated inhibition of nanozyme activity: Impact of protein shape. J. Am. Chem. Soc. 146, 10478–10488 (2024).
Article ADS CAS PubMed Google Scholar
Portz, B., Lee, B. L. & Shorter, J. FUS and TDP-43 phases in health and disease. Trends Biochem. Sci. 46, 550–563 (2021).
Article CAS PubMed PubMed Central Google Scholar
Boyko, S. & Surewicz, W. K. Tau liquid-liquid phase separation in neurodegenerative diseases. Trends Cell Biol. 32, 611–623 (2022).
Article CAS PubMed PubMed Central Google Scholar
Dahiya, V. et al. The switch from client holding to folding in the Hsp70/Hsp90 chaperone machineries is regulated by a direct interplay between co-chaperones. Mol. Cell 82, 1543–1556 (2022).
Article CAS PubMed Google Scholar
Andrikopoulos, N. et al. Exploring peptido-nanocomposites in the context of amyloid diseases. Angew. Chem. Int. Ed. 63, e202309958 (2023).
Article Google Scholar
Ko, T. W. et al. A fourth-generation high-dimensional neural network potential with accurate electrostatics including non-local charge transfer. Nat. Commun. 12, 398 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Mokashi-Punekar, S., Walsh, T. R. & Rosi, N. L. Tuning the structure and chiroptical properties of gold nanoparticle single helices via peptide sequence variation. J. Am. Chem. Soc. 141, 15710–15716 (2019).
Article ADS CAS PubMed Google Scholar
Biswal, M., Lu, J. & Song, J. SARS-CoV-2 nucleocapsid protein targets a conserved surface groove of the NTF2-like domain of G3BP1. J. Mol. Biol. 434, 167516 (2022).
Article CAS PubMed PubMed Central Google Scholar
Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126, 014101 (2007).
Article ADS PubMed Google Scholar
Abraham, M. J. et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1, 19–25 (2015).
Article ADS Google Scholar
Maier, J. A. et al. ff14SB: Improving the accuracy of protein side chain and backbone parameters from ff99SB. J. Chem. Theory Comput. 11, 3696–3713 (2015).
Article CAS PubMed PubMed Central Google Scholar
Wang, J. et al. Development and testing of a general amber force field. J. Comput. Chem. 25, 1157–1174 (2004).
Article ADS CAS PubMed Google Scholar
Case, D. A. et al. AmberTools. J. Chem. Inf. Model. 63, 6183–6191 (2023).
Article CAS PubMed PubMed Central Google Scholar
Schauperl, M. et al. Non-bonded force field model with advanced restrained electrostatic potential charges (RESP2). Commun. Chem. 3, 44 (2020).
Article CAS PubMed PubMed Central Google Scholar
Lu, T. & Chen, F. Multiwfn: a multifunctional wavefunction analyzer. J. Comput. Chem. 33, 580–592 (2012).
Article ADS PubMed Google Scholar
Heinz, H., Vaia, R., Farmer, B. & Naik, R. Accurate simulation of surfaces and interfaces of face-centered cubic metals using 12-6 and 9-6 Lennard-Jones potentials. J. Phys. Chem. C. 112, 17281–17290 (2008).
Article CAS Google Scholar
Rai, B. et al. Molecular dynamic simulations of self-assembled alkylthiolate monolayers on an Au(III) surface. Langmuir 20, 3138–3144 (2004).
Article CAS PubMed Google Scholar
Hess, B. P-LINCS: A parallel linear constraint solver for molecular simulation. J. Chem. Theory Comput. 4, 116–122 (2008).
Article CAS PubMed Google Scholar
Valdés-Tresanco, M. S. et al. gmx_MMPBSA: a new tool to perform end-state free energy calculations with GROMACS. J. Chem. Theory Comput. 17, 6281–6291 (2021).
Article PubMed Google Scholar
DeLano, W. L. PyMOL: an open-source molecular graphics tool. CCP4 Newsl. Protein Crystallogr. 40, 82–92 (2002).
Google Scholar
Percie du Sert, N. et al. The ARRIVE guidelines 2.0: updated guidelines for reporting animal research. PLoS Biol. 18, e3000410 (2020).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This work was supported by the National Natural Science Foundation of China (12347102, 32371447, 12222506, and 82473182), Jiangsu Provincial Outstanding Youth Fund (BK20230001), and Jiangsu Province “333” project (No. (2022) 3-1-234).

Author information

These authors contributed equally: Liu-Ting Zheng, Zeng-Shuai Yan, Xin-Yue Li.

Authors and Affiliations

School of Pharmacy, Department of Pharmaceutics, Nanjing Medical University, Nanjing, Jiangsu, China
Liu-Ting Zheng, Xin-Yue Li, Jia-Jia Chang, Xiao-Qi Tan, Yu-Xing Wang & Da Huo
Center for Soft Condensed Matter Physics and Interdisciplinary Research, School of Physical Science and Technology, Soochow University, Suzhou, Jiangsu, China
Zeng-Shuai Yan & Hong-Ming Ding
Department of Oncology, Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing, Jiangsu, China
Qin Liu
National Laboratory of Solid State Microstructures and Department of Physics, Collaborative Innovation Center of Advanced Microstructures, Nanjing University, Nanjing, Jiangsu, China
Yu-Qiang Ma

Authors

Liu-Ting Zheng
View author publications
Search author on:PubMed Google Scholar
Zeng-Shuai Yan
View author publications
Search author on:PubMed Google Scholar
Xin-Yue Li
View author publications
Search author on:PubMed Google Scholar
Jia-Jia Chang
View author publications
Search author on:PubMed Google Scholar
Xiao-Qi Tan
View author publications
Search author on:PubMed Google Scholar
Yu-Xing Wang
View author publications
Search author on:PubMed Google Scholar
Hong-Ming Ding
View author publications
Search author on:PubMed Google Scholar
Qin Liu
View author publications
Search author on:PubMed Google Scholar
Yu-Qiang Ma
View author publications
Search author on:PubMed Google Scholar
Da Huo
View author publications
Search author on:PubMed Google Scholar

Contributions

L.Z., Z.Y., and X.L. contributed equally. Conceptualization: L.Z., H.D., Q.L., Y.M., and D.H. raised the idea of investigating nanomaterials' potential to modulate biomolecular condense. Methodology: L.Z. and D.H. developed the chemo-etching-assisted proximity labeling method and synthetic strategy. Investigation: L.Z. performed wet-chemical synthesis, fluorescence imaging, and preparation of a sample for Omics studies. Z.Y. conducted a molecular dynamics simulation. X.L. aided in microenvironment imaging analysis. J.C. and X.T. contributed partially to the bioinformatic analysis. Q.L. conducted the animal studies.; Data visualization: L.Z. contributed to the preparation of all figures appearing in both the main text and supplementary materials with the help of Y.W.; Writing-original draft: L.Z. and Z.Y. drafted the manuscript, which was revised by H.D. and D.H.; Supervision: H.D., Q.L., Y.M., and D.H. supervised the study.

Corresponding authors

Correspondence to Hong-Ming Ding, Qin Liu, Yu-Qiang Ma or Da Huo.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Pu Chun Ke, Ole Jakob Nøstbakken, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information (download PDF )

Reporting Summary (download PDF )

Transparent Peer Review file (download PDF )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Zheng, LT., Yan, ZS., Li, XY. et al. Nanomaterial signatures program biomolecular condensates via triphasic separation for chemoplasticity remodeling. Nat Commun 16, 9554 (2025). https://doi.org/10.1038/s41467-025-64623-4

Download citation

Received: 22 April 2025
Accepted: 23 September 2025
Published: 29 October 2025
Version of record: 29 October 2025
DOI: https://doi.org/10.1038/s41467-025-64623-4