Platinum-doped emodin carbon dots mitigate sepsis-induced lung injury by targeting the gut-lung axis

Abstract

Sepsis-induced acute lung injury is a life-threatening complication with limited therapeutic options. Although the gut-lung axis is crucial in sepsis pathogenesis, effective interventions targeting this pathway remain scarce. Here, we developed multi-enzymatic platinum-doped emodin carbon dots (Pt-ECDs) via a hydrothermal method. Pt-ECDs exhibited superior catalase, superoxide dismutase, glutathione peroxidase and peroxidase-like activities, enabling potent reactive oxygen species (ROS) scavenging. In a murine sepsis model, oral Pt-ECDs significantly improved survival, reduced systemic inflammation, and ameliorated lung injury. Transcriptomic analysis revealed that Pt-ECDs suppressed oxidative stress and macrophage pyroptosis in lung tissues. Mechanistically, integrated metabolomic and microbiome analyses demonstrated that Pt-ECDs modulated the gut microbiota, specifically inhibiting g_Bacteroides-derived palmitic acid (PA) production. We further confirmed that PA exacerbates macrophage pyroptosis and pro-inflammatory polarization by directly binding to NOX2 and NLRP3. Crucially, fecal microbiota transplantation from Pt-ECDs-treated mice attenuated septic lung injury, whereas microbiota depletion abolished the therapeutic benefits. Collectively, our findings identify Pt-ECDs as a promising nanotherapeutic that alleviates septic lung injury by targeting the gut microbiota-palmitic acid-pyroptosis axis.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-025-03972-0.

Introduction

Sepsis, a heterogeneous clinical syndrome precipitated by uncontrolled infections, causes substantial global mortality through life-threatening multi-organ dysfunction [1]. The lungs represent the most commonly affected organs [2], with acute lung injury (ALI) or acute respiratory distress syndrome (ARDS) in severe cases [3, 4], clinical cohorts reporting mortality rates of 34–45% for these pulmonary complications [5]. Despite mounting research, the molecular pathogenesis of sepsis-induced acute lung injury remains unclear, and innovative therapeutic modalities are critically lacking [6–8]. This mechanistic knowledge gap impedes therapeutic breakthroughs, urgently necessitating exploration of non-conventional targets beyond canonical inflammatory pathways.

Recent advances in sepsis management highlight microbiota homeostasis restoration as a promising therapeutic avenue [9, 10]. Bidirectional communication between the intestinal and respiratory systems, termed the gut-lung axis, occurs via circulatory and lymphatic transport of microbial metabolites [11, 12]. Patients with ARDS or ALI demonstrate intestinal reactive oxygen species (ROS) overload, which contributes to intestinal dysbiosis characterized by diminished beneficial taxa (e.g., Faecalibacterium prausnitzii, Bifidobacterium spp.) concurrent with pathogenic enrichment (Escherichia coli, Streptococcus spp.) [13]. Such compositional shifts impair gut barrier function, increasing permeability that enables translocation of microbial components to pulmonary sites. This process initiates sustained inflammatory cascades that aggravate pulmonary damage and accelerate ARDS development [14]. Consequently, rebalancing gut microbial communities represents an essential approach for interrupting pathological gut-lung axis signaling in respiratory failure syndromes.

Pulmonary macrophages are essential for bacterial pathogen defense in sepsis [15]. These immune cells predominantly activate NLRP3 inflammasomes when encountering bacterial stimuli. Uncontrolled stimulation, however, can induce pyroptosis, a highly inflammatory programmed cell death process frequently observed in septic macrophages [16]. Pyroptotic macrophages release intracellular contents including IL-1β and other pro-inflammatory mediators, which amplify inflammatory cascades and drive macrophage polarization toward pro-inflammatory states, thereby aggravating the severity of ALI [17, 18]. Consequently, macrophage pyroptotic pathways present a pivotal pulmonary focus for investigating gut-lung axis modulation in sepsis-related respiratory compromise.

Currently, a gap of studies exists regarding nanomaterials’ ability to modulate the intestinal microbiota-host metabolic axis for mitigating sepsis-induced lung injury. Among them, carbon dot nano-enzymes, characterized by precise redox regulation, structural robustness, and physiological compatibility, exhibit unique capabilities in mimicking endogenous antioxidant enzymes, catalytically scavenging ROS, and exerting anti-inflammatory effects [19, 20], making them promising therapeutic candidates for mitigating septic lung injury [21]. Accumulating evidence indicates that these nanoenzymes can attenuate systemic inflammation through modulation of gut microbiota composition and function [22], while simultaneously promoting the regeneration of intestinal epithelial cells [23, 24]. The catalytic activity of carbon dots is not only attributed to their surface states but also to the surface polymers formed via polymerization, cross-linking, and carbonization processes, which preserve molecular characteristics of the precursor materials [25–28]. Consequently, carbon dots derived from bioactive compounds have been increasingly explored for biomedical applications [29–31]. Among various biomass sources, traditional Chinese medicine (TCM) stands out as particularly favorable precursors for carbon dot synthesis due to their rich content of bioactive constituents and extensive history of clinical use [32, 33]. Various TCM compounds, such as emodin [34], curcumin [35] and formononetin [36], have been utilized in the management of sepsis. Among them, the research related to emodin has attracted extensive attention and demonstrated higher therapeutic potential. Emodin, a naturally occurring anthraquinone compound primarily extracted from rhubarb, a well-known traditional Chinese medicinal herb, has demonstrated potent anti-inflammatory, antioxidant, and organ-protective properties [37]. It has been shown to substantially inhibit inflammation within the context of cecal ligation and puncture (CLP)-induced sepsis, resulting in decreased intestinal mucosal damage [38, 39]. Furthermore, emodin’s distinctive pharmacological profile has led to its widespread investigation in the contexts of pyroptosis, sepsis, and pulmonary disorders [40–42]. Therefore, the development of emodin-derived carbon dots targeting the gut-lung axis represents a highly promising strategy for the treatment of sepsis-induced lung injury.

In this study, we developed platinum-integrated emodin carbon dots (Pt-ECDs) via hydrothermal synthesis that exhibit remarkable multi-enzyme mimetic activities such as catalase (CAT), superoxide dismutase (SOD), peroxidase (POD), and glutathione peroxidase (GPx) to scavenge ROS. Single-cell transcriptomic analysis revealed increased pyroptosis and pro-inflammatory macrophage polarization in the lungs of septic mice. In vivo studies demonstrated that oral administration of Pt-ECDs significantly improved survival rates, mitigated cytokine storms, and alleviated pulmonary injury in septic animal models. Transcriptomic profiling indicated that Pt-ECDs effectively downregulated genes associated with oxidative stress, pyroptosis, and inflammation in lung tissues. Integrated microbiome-metabolome analyses showed that the protective effects of Pt-ECDs were primarily mediated by inhibiting Bacteroides-derived palmitic acid biosynthesis. In vitro studies revealed that palmitic acid intensifies both pyroptosis and pro-inflammatory macrophage polarization. Complementary investigations employing functional genetic alterations (gain-/loss-of-function), receptor inhibitors, and site-directed mutagenesis established a dual binding mechanism: PA directly interacts with NOX2 and NLRP3 to potentiate pyroptotic pathways. Validation through integrated microbiota ablation and fecal transplantation approaches confirmed gut flora and palmitic acid as critical therapeutic targets for mitigating septic lung injury. Our study establishes Pt-ECDs, a TCM-derived nanozyme, as a promising therapeutic agent that ameliorates septic lung injury by targeting the gut microbiota-palmitic acid-NOX2/NLRP3/CASPASE1/GSDMD signaling axis, thereby suppressing macrophage pyroptosis and inflammatory responses, and advancing the development of precision nano-therapeutics for sepsis management.

Results

Increased macrophage pyroptosis and pro-inflammatory polarization during sepsis-induced lung injury

Single-cell RNA (scRNA) sequencing enables precise measurement of gene expression levels within individual cells, revealing transcriptional heterogeneity across distinct cell types and states. This capability provides critical insights into cellular developmental processes and disease mechanisms, particularly under pathological conditions such as sepsis [43, 44]. To construct a comprehensive cellular atlas of both healthy and septic lungs, we performed scRNA sequencing analysis on lung tissues from three groups of mice [45]. Based on the expression of canonical marker genes, cells were clustered and annotated into nine major cell types. These included T cells (Cd3g, Cd3e, Cxcr6, Pdcd1, Lcn4), stromal cells (Actc1, Susd5, Sost, Xirp1, Ctnna3), NK cells (Klra8, Klra4, Cma1, Klra9, Khdc1a), neutrophils (Retnlg, Prok2, Ceacam10, Mrgpra2b, Ly6g), monocytes (Cenpm, Adgre4, Fabp7, Hist1h3c), macrophages (Atp6v0d2, Krt79, Fabp1, Htr2c, Gm41307), fibroblasts (Clec3b, Rbp4, St8sia2, Sftpa1, Sftpc), endothelial cells (Cdh5, Arhgef15, Grrp1, Pllp), and B cells (Lglc2, Fcmr, Lglc3) (Figures S1A and C, Supporting Information). An uneven distribution of these cell proportions was observed between the control and sepsis groups, indicating a major remodeling of the cellular landscape accompanies the development of sepsis (Figure S1B, Supporting Information).

Early mortality in sepsis is primarily attributed to cytokine storm and subsequent inflammatory-induced multiple organ failure [46]. As key mediators of the innate inflammatory response, macrophages contribute to the inflammatory cascade by releasing large quantities of pro-inflammatory cytokines [47]. Therefore, we focused specifically on macrophage subpopulations (Figs. 1A-B). Based on the expression of canonical markers, macrophages were classified into nine distinct clusters (Fig. 1D): MAC1 (Gm13571, C77080, Egfem1, Gm50143), MAC2 (Prrt1, Gm26917, Sftpc, Ly6c1, Sftpa1, Hspa2), MAC3 (C1qa, Cd163, Ccl12, Ctla2b, Folr2), MAC4 (I14i1, Ccr7, Fscn1, II2ra, St8sia1), MAC5 (Ifit1b, Ly6i, Ifi208, Trim30c, Nos2), MAC6 (Stfa2, Mrgpra2b, Pdlim3, Susd5, Slc22a20), MAC7 (Itgae, Clec9a, Gcsam, Gpr141b, Clnk), MAC8 (Ptgfr, Adamtsl1, Amph, Heph, St8sia2), and MAC9 (Rbm44, Ubash3a, Gm33195, Atcayos, Gm26771). Notably, the abundances of MAC1 and MAC5 were significantly elevated in the CLP-48 h group compared with the control group, suggesting their potential critical roles in the pathogenesis and progression of sepsis (Fig. 1C). Gene expression heatmaps in conjunction with the SCISSOR algorithm revealed that both MAC1 and MAC5 exhibited high expression levels of pyroptosis-related genes, including Gsdmd, Casp1, Nlrp3, Nod1, and Il-18, along with markers associated with pro-inflammatory polarization such as Nos2, Tnf, Il-6, and Cd86 (Figs. 1E and S2-3, Supporting Information). Additionally, KEGG and GO pathway analyses consistently indicated significant enrichment of the NOD-like receptor signaling pathway in these two subsets, which is closely linked to the mechanism of pyroptosis (Fig. 1F-G) [48]. Further supporting a pro-inflammatory phenotype, M1/M2 polarization scoring demonstrated significantly higher M1 scores in both MAC1 and MAC5 (Fig. 1H). Altogether, our findings provide initial evidence that the abundances of MAC1 and MAC5 subsets are elevated within pulmonary macrophage populations of septic mice compared to controls, and that these subsets exhibit heightened expression of markers associated with pyroptosis and pro-inflammatory polarization.

Fig. 1.

Single-cell RNA sequencing of murine samples revealed that sepsis progression was associated with macrophage pyroptosis and a shift toward a pro-inflammatory phenotype. (A) UMAP visualization for murine alveolar cells post unsupervised clustering, color-coded by distinct cell populations. (B) UMAP plot of murine alveolar macrophages subdivided into nine unique subclusters, with each distinguished by a specific color based on subcluster identity. (C) Distribution of macrophage subcluster proportions among the control group, sepsis group, and sepsis-48 h group. (D) Dot plots of representative cell markers in each macrophage subcluster. The size of the dots indicates abundance, and the color indicates the level of expression. (E) Heatmap of gene expression across macrophage clusters. (F-G) GO and KEGG enrichment analysis of (F) MAC1 and (G) MAC5. (H) M1/M2 polarization scores for alveolar macrophage subclusters. (I-J) The number and strength of communication connections between macrophage subpopulations in the (I) NOD and (J) Complement pathway. (K) Incoming pattern of target cells. (L-O) Differentiation pathway in macrophages marked by (L) pseudo-time or (M-N) cell subpopulations, or (O) pyroptosis pathway score. (P-T) The expression patterns of different genes in macrophage subpopulations as marked by pseudotime including (P) Nod1, (Q) Pycard, (R) Nlrp3, (S) Il-18 and (T) Il1a

To investigate the potential regulatory mechanisms underlying sepsis induced-lung injury, An analysis of intercellular crosstalk of macrophage subsets was performed. As shown in Fig. 1I and J, extensive interactions were observed among MAC1, MAC5, and other macrophage subpopulations via the NOD and COMPLEMENT signaling pathways. Signaling crosstalk describes how cells exchange and decode signals to coordinate their activities and reactions—a foundational principle for understanding how cells communicate within tissues [49]. Analysis of incoming signaling identified pyroptosis-related NOD and inflammatory-related COMPLEMENT pathways as critically active in both MAC1 and MAC5 subsets (Fig. 1H). In summary, our findings provide initial evidence that the abundances of MAC1 and MAC5 subsets are elevated within pulmonary macrophage populations of septic mice compared to controls, and that these subsets exhibit heightened expression of markers associated with pyroptosis and pro-inflammatory polarization.

Pseudotime analysis works primarily by reducing dimensionality and constructing a pseudotemporal axis, thereby ordering cells according to their potential dynamic changes and revealing the continuous patterns of cell-state evolution [50]. To further delineate the temporal relationship between macrophage subpopulations, including MAC1 and MAC5, and the progression of sepsis, pseudotime analysis was performed. The results demonstrated that macrophage subsets associated with pyroptosis, such as MAC1 and MAC5, exhibited higher expression levels at later pseudotime points (Figs. 1L-N). Concurrently, pathways related to pyroptosis showed a similar temporal activation pattern (Fig. 1O). Key genes involved in pyroptosis, including Nod1, Pycard, Nlrp3, and Il-18, as well as markers of pro-inflammatory polarization such as Il1a, were also upregulated at advanced pseudotime (Figs. 1P-T). Collectively, these pseudotime trajectory results indicate that macrophage pyroptosis represents a late-stage event in the development of sepsis.

Lung tissues from septic mice were assessed by immunofluorescence (IF) staining and hematoxylin-eosin (H&E) staining. The corroboration between IF staining and scRNA-seq data was evidenced by the colocalization of the macrophage marker F4/80 and the pyroptosis-related marker NLRP3 in lung tissues (Figures S4A-D and S4F-G, Supporting Information). Consistent with these findings, H&E staining revealed exacerbated lung injury with the progression of sepsis (Figure S4E, Supporting Information). These results indicate that macrophage pyroptosis and an acquired pro-inflammatory phenotype are key mechanisms through which sepsis progression exacerbates lung injury.

Additionally, we quantified the serum levels of pyroptosis-related markers in a cohort of sepsis patients stratified by disease severity. Western blot analysis revealed a progressive increase in the levels of pyroptosis-related markers (NLRP3, N-GSDMD, CL-CASPASE1, and CL-IL1β) with higher SOFA scores, suggesting a potential correlation between sepsis severity and pyroptosis activation (Figures S5A, B and Supplementary Table 3, Supporting Information). Together, these data suggest heightened pyroptosis of macrophages in patients with sepsis.

Synthesis and characterization of Pt-ECDs

To obtain high-performance Pt-doped emodin-derived carbon dots (Pt-CDs), a mixture of emodin, potassium platinochloride (K₂PtCl₄), and PEI1800 was reacted in N, N-dimethylformamide (DMF) under a series of synthetic conditions (Figure S6, Supporting Information). For comparison, undoped emodin-derived carbon dots (ECDs) were synthesized under an identical protocol with emodin and PEI1800. Transmission electron microscopy (TEM) revealed that both Pt-ECDs and ECDs are monodisperse nanoparticles with a narrow size distribution (Fig. 2A and C). Both Pt-ECDs and ECDs exhibited outstanding dispersion stability, as supported by characterization of their hydrodynamic size and surface charge (Figs. 2E-F). Importantly, the Pt-ECDs maintained their structural integrity under a broad spectrum of pH environments (Figures S10A, C, Supporting Information). To verify the presence of the graphitic carbon (002) plane and successful platinum doping in the carbon dots, high-resolution TEM (HRTEM) and X-ray diffraction (XRD) were performed. HRTEM images of the ECDs revealed distinct lattice fringes with a spacing of 0.32 nm, corresponding to the graphitic carbon (002) plane of graphitic carbon (Fig. 2B). In contrast, the Pt-ECDs not only exhibited lattice fringes at 0.35 nm assigned to the graphitic carbon (002) plane, but also additional fringes with a spacing of 0.21 nm, attributable to the Pt (111) plane (Fig. 2D). XRD patterns further supported these findings: a prominent peak at approximately 27 2θ was observed in ECDs, whereas Pt-ECDs displayed additional characteristic peaks at around 27 2θ and 41 2θ, consistent with both graphitic carbon and metallic Pt, confirming the successful incorporation of Pt into the carbon dot structure (Fig.2G). Raman spectroscopy further confirmed the structural features of the materials. The Raman spectrum showed the presence of both the defect-related D band and the graphitic G band, identified by peaks at 1291.71 cm⁻¹ and 1667.6 cm⁻¹, respectively (Fig. 2H). Additionally, a distinct Pt-C vibrational band was detected in the Pt-ECDs, indicating metal-carbon coordination. Compared to the ECDs, which exhibited an ID/IG ratio of 0.57, the Pt-ECDs showed a decreased ID/IG ratio of 0.508. This reduction indicates an increase in the degree of graphitization following Pt doping (Fig. 2H). Under a nitrogen atmosphere, TGA revealed a two-stage mass loss profile for the CDs. The first stage, occurring below 200 °C, was attributed to the evaporation of adsorbed water, resulting in a minor mass reduction. The second stage, between 200 °C and 800 °C, involved substantial mass loss due to the decomposition of the carbon-based framework. Notably, compared with ECDs, the Pt-ECDs exhibited reduced mass loss in both stages, along with a slower rate of decomposition, suggesting that successful Pt doping may have impeded the segmental mobility and thermal degradation of the carbon skeleton at elevated temperatures (Figs. 2M and S7, Supporting Information).

Fig. 2.

Characterization of Pt-ECDs and ECDs. (A) TEM of ECDs (scale bars: 10 nm). (B) HR-TEM of ECDs (scale bars: 2 nm, lattice spacing: 0.32 nm). (C) TEM of Pt-ECDs (scale bars: 10 nm). (D) HR-TEM of Pt-ECDs (scale bars: 2 nm, lattice spacing: 0.21 nm). (E) Hydrodynamic diameter of ECDs and Pt-ECDs. (F) Zeta potential of ECDs and Pt-ECDs. (G) X-ray diffraction (XRD) patterns of ECDs and Pt-ECDs. (H) Raman spectrum of ECDs and Pt-ECDs displaying distinctive D and G peaks respectively. (I) UV absorption of ECDs and Pt-ECDs. (J) ECDs’ X-ray photoelectron spectroscopy (XPS) spectra of O1s and C1s regions, with deconvoluted peaks indicating chemical states. (K-L) High-resolution XPS spectra of the (K) O1s region and (L) C1s region. (M) Thermogravimetric analysis (TGA) curve showing a weight loss of ECDs and Pt-ECDs. (N) Pt-ECDs’ XPS spectra of O1s, C1s and Pt4f regions, with deconvoluted peaks indicating chemical states. (O-P) High-resolution XPS spectra of the (O) Pt4f region and (P) C1s region. Data were expressed as mean ± SD

To determine the surface elemental composition and chemical states, X-ray photoelectron spectroscopy (XPS) was used for both Pt-ECDs and ECDs. The survey scan of Pt-ECDs revealed characteristic signals corresponding to C1s (284.6 eV), O1s (531.3 eV) and Pt4f (73.05 eV) (Fig. 2N), whereas ECDs exhibited predominant peaks for C1s (284.8 eV) and O1s (532.8 eV) (Fig. 2J-L). High-resolution XPS of the Pt4f region was deconvoluted into four component peaks, with binding energies observed at 76.35 eV and 73.15 eV, suggesting the presence of Pt–O bonding configurations (Fig. 2O). The successful incorporation and stabilization of Pt within the carbon matrix were further supported by the C1s spectrum, which indicated the coexistence of sp²/sp³ hybridized carbon (C–C/C = C) and oxygen-functionalized moieties (Fig. 2P). Additionally, UV-vis spectroscopy showed a prominent absorption band around 210 nm, consistent with n–π* transitions arising from aromatic C = O/C = N groups within a sp²-conjugated carbon framework (Fig. 2I). In summary, our experiments demonstrate that both Pt-ECDs and ECDs were successfully synthesized, with the specific confirmation that platinum has been effectively doped into the Pt-ECDs.

Multienzyme catalytic activity and ROS scavenging capacity in Pt-ECDs

The introduction of Pt is anticipated to further augment Pt-ECDs’ catalytic performance compared to Cu or Fe doped. We further focused on Pt-ECDs and ECDs’ multi-enzyme mimetic activities and ROS scavenging capabilities (Figure S8, Supporting Information). Both Pt-ECDs and ECDs exhibit ultra-small dimensions and are rich in antioxidant functional groups. The introduction of Pt is anticipated to further augment their catalytic performance. Based on these attributes, we hypothesize that Pt-ECDs and ECDs possess multi-enzyme mimetic activities and demonstrate efficient ROS scavenging capabilities. The ability of these nanoenzymes to catalytically eliminate a wide range of ROS was assessed, encompassing superoxide anion (•O₂⁻), hydrogen peroxide (H₂O₂), hydroxyl radical (•OH), as well as stable radical species such as 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS•), 2,2-diphenyl-1-picrylhydrazyl radical (DPPH•), and 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide radical (PTIO•) (Fig. 3A).

Fig. 3.

Pt-ECDs exhibit multi-enzyme catalytic activity and ROS scavenging ability. (A) Schematic illustration of the enzyme-mimetic mechanism of Pt-ECDs. (B) SOD-like activity of Pt-ECDs, quantified by the SOD inhibition rate (n=3 independent samples per group). (C) CAT-like activity of Pt-ECDs, evaluated by the oxygen production amount (Pt-ECDs concentrations: 0, 50, 100, 200, 400 μg/ml). (D) CAT-like activity of Pt-ECDs, determined by the H₂O₂ scavenging ratio (n=3 independent samples per group). (E) Michaelis–Menten kinetic analysis of CAT-like enzyme activity for Pt-ECDs (n=3 independent samples per group). (F) Comparison of POD-like enzyme activity between Pt-ECDs and ECDs (n=3 independent samples per group). (G) Michaelis–Menten kinetic analysis of POD-like enzyme activity for Pt-ECDs (n=3 independent samples per group). (H) Comparison of GPx-like enzyme activity between Pt-ECDs and ECDs (n=3 independent samples per group). (I) Turnover number (TON) values of representative nanoenzymes toward H₂O₂. (J) Turnover number (TON) values of representative nanoenzymes toward TMB. (K) Absorbance detection of 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl 3-oxide (PTIO•) radicals in Pt-ECDs at various concentrations. (L) Total Antioxidant Capacity (TAC) of Pt-ECDs. (M-N) Scavenging efficiency of (M) superoxide anion radicals and (N) hydroxyl radicals detected by ESR spectroscopy (nanoenzyme concentration: 100 μg/ml). Data are presented as mean ± standard deviation (SD)

Superoxide dismutase (SOD) specifically catalyzes the dismutation of the •O₂⁻ into oxygen (O₂) and H₂O₂ [51]. As the first line of defense against ROS in living organisms, its activity directly determines the cellular capacity to counteract oxidative damage [52]. In comparison with ECDs, the Pt-ECDs exhibited superior •O₂⁻ scavenging efficiency and enhanced SOD-like activity [53], while all samples achieved over 90% elimination at 400 µg/mL and displayed a consistent dose-dependent trend (Fig. 3B ,M and S9, Supporting Information). Remarkably, Pt-ECDs maintained a stable SOD inhibition rate of over 80% under varying pH conditions (Figure S10B, Supporting Information). Taken together, these data indicate that Pt-ECDs possess pronounced SOD enzyme-mimicking activity.

Catalase (CAT) is a hemoprotein that utilizes iron porphyrin (heme) as its prosthetic group. Under physiological conditions, CAT continuously catalyzes the decomposition of H₂O₂, thereby maintaining the balance of intracellular ROS and preventing oxidative stress-induced damage to macromolecules such as cellular membranes, DNA, and proteins [54]. Comparative reaction monitoring confirmed a kinetically synchronized relationship between the oxygen evolution profile and the H₂O₂ degradation pattern (Fig. 3C). To gain further insight into the biocatalytic kinetics of H₂O₂ scavenging, steady-state kinetic parameters were derived from the dependence of reaction velocity on substrate concentration. According to Michaelis-Menten kinetics and the corresponding Lineweaver-Burk plot, the michaelis constant (Km) and maximum reaction velocity (Vmax) were determined to be 300 mM and 1.6 mg/L·min, respectively. To situate the CAT-like catalytic performance of our Pt-ECDs within the wider field of nanozyme research, a comparative analysis was carried out against several representative nanocatalysts, such as Co₃O₄ cubes, Mn₃O₄ nanorods, Mn₃O₄ nanoplates, and Pt octahedrons, among others (Fig. 3I) [55, 56], which revealed that Pt-ECDs exhibit markedly enhanced CAT-like catalytic capability compared to other materials tested.

Peroxidases (PODs) represent a unique category of antioxidant enzymes that catalyze the reduction of H₂O₂ to H₂O in the presence of electron donors, thereby serving as essential scavengers of ROS in biological systems [57]. Simultaneously, Pt-ECDs displayed POD-mimicking activity in a time-dependent manner (Fig. 3F). To evaluate their catalytic efficiency, steady-state kinetic assays were performed at various H₂O₂ concentrations in the presence of Pt-ECDs, which followed classical Michaelis-Menten behavior. The Vmax and Km were determined from the corresponding Lineweaver-Burk plots (Figs. 3G and S9D, Supporting Information). Notably, Pt-ECDs achieved a turnover number (TON) of 6.69 × 10^− 3/s and a Vmax of 0.729 × 10^− 6 M/s for their POD-like activity-values significantly higher than those of most conventional nanozymes (Fig. 3J) [58].

Glutathione peroxidase (GPx) is a selenium-containing antioxidant enzyme whose primary function is to catalyze the reduction of H₂O₂ to water using glutathione (GSH) as the reducing substrate, thereby facilitating ROS elimination [59]. The selenocysteine residue within its active site plays a critical role in the catalytic process, enabling efficient ROS scavenging through a redox cycle. The GPx-like activity of the Pt-ECDs was markedly superior to that of the ECDs, achieving substantial catalytic turnover within 20 min (Fig. 3H).

Furthermore, to investigate the reactive nitrogen species (RNS) scavenging ability of Pt-ECDs and ECDs, ABTS+·, a typical RNS model, was selected as the target radical. The scavenging effect was evaluated using ultraviolet-visible spectroscopy [60]. The results demonstrated a concentration-dependent decrease in the absorbance of the ABTS+· solution upon the addition of increasing concentrations of both Pt-ECDs and ECDs, indicating effective radical elimination. Notably, Pt-ECDs exhibited a more pronounced scavenging capability compared to ECDs (Figs. 3L and S9C, Supporting Information).

As a ROS produced from H₂O₂ in Fenton-like reactions, the hydroxyl radical (•OH) exhibits notable persistence. This stability enables it to drive the breakdown of cellular biomolecules [53]. To evaluate the scavenging capacity of Pt-ECDs and ECDs against short-lived free radicals such as •OH, electron spin resonance (ESR) spectroscopy was employed using DMPO. A significant attenuation of the characteristic DMPO- •OH adduct signal was observed following the addition of both materials, with Pt-ECDs demonstrating a markedly stronger scavenging effect compared to ECDs (Fig. 3M).

The generation of reactive oxygen and nitrogen species (RONS) is strongly associated with various pathological conditions [61]. As depicted in Fig. 3K, the absorbance of the PTIO• radical at 557 nm decreased progressively with increasing concentrations of Pt-ECDs. Moreover, in comparison with conventional ECDs, the Pt-ECD exhibited a potent and concentration-dependent scavenging effect against DPPH• radicals (Figure S9E, Supporting Information). These findings confirm the concentration-responsive antioxidant performance of Pt-ECD in effectively neutralizing PTIO• radicals, highlighting their potential as efficient nanozymes for mitigating oxidative stress.

DFT-elucidated multi-enzyme catalytic and ROS scavenging capabilities in Pt-ECDs and ECDs

As a foundational quantum mechanical method employed in fields from materials science to catalysis, Density Functional Theory (DFT) serves as a key computational tool in physics and chemistry. It enables accurate predictions of key properties including geometric structures, electronic band structures, and adsorption energies [62]. To further elucidate the origin of the enhanced catalytic performance of Pt-ECDs, atomic-scale structural models of both pristine ECDs and Pt-ECDs were constructed (Fig. 4). The ECDs model was built with 78 carbon, 20 hydrogen, and 5 oxygen atoms, while the Pt-ECDs model incorporated 33 platinum atoms in addition to 78 carbon, 20 hydrogen, and 5 oxygen atoms (Fig. 4A). Planar charge density difference analysis revealed a significant redistribution of electrons along the Z-axis in the Pt-ECDs, mainly occurring between the platinum atom and neighboring carbon atoms (Fig. 4B). This electronic reorganization was accompanied by notable charge transfer from the carbon framework to the Pt site (Fig. 4C). These computational outcomes suggest that the incorporation of Pt facilitates more efficient electron transfer, thereby establishing an optimized electronic environment that enhances catalytic redox reactivity in Pt-ECDs.

Fig. 4.

The multi-enzyme catalytic and ROS scavenging mechanisms of Pt-ECDs. (A) Illustrated atomic structures of ECDs and Pt-ECDs, where the atoms are color-coded as follows: Pt (grey), C (brown), H (pink), and O (red). (B) Visualized charge redistribution upon Pt adsorption on the carbon dots monolayer. (C) Charge redistribution upon Pt adsorption on the carbon dots monolayer. The loss (purple) and gain (green) of electron density are clearly shown. (D, F, H, J) Energy diagrams illustrating the possible catalytic reaction pathways of (D) SOD-like enzymes, (F) CAT-like enzymes, (H) GPx-like enzymes and (J) OH·removal on ECD and Pt-ECD surfaces, respectively. Corresponding insets display the initial, multiple plausible intermediate, and final configurations. (E, G, I, K) Mechanisms for ECDs of (E) SOD-like enzymes, (G) CAT-like enzymes, (I) GPx-like enzymes and (K) OH· removal reaction respectively

In terms of SOD-mimetic activity, Pt-ECDs exhibit a markedly stronger adsorption affinity for the hydroperoxyl radical (*OOH), reflected in a binding energy of −1.23 eV, which is substantially greater than that of ECDs. This suggests a more efficient initial binding and activation of the reactant facilitated by the Pt active site through enhanced electron transfer. Furthermore, the subsequent protonation step leading to the formation of the *H₂O₂ intermediate is significantly promoted on Pt-ECDs, as indicated by its considerably stabilized energy state of −1.74 eV compared to −0.51 eV on undoped ECDs. These results imply a more stabilized transition state and a reduced energy barrier for hydrogen peroxide generation within the Pt-incorporated catalytic framework (Figs. 4D-E and S11A, Supporting Information).

In the context of catalase-like performance, H₂O₂ adsorption at the platinum-active centers in Pt-ECDs was notably more favorable, exhibiting a binding energy of −0.81 eV, substantially stronger than that observed in ECDs (−0.34 eV), reflecting enhanced substrate capture and catalytic propensity. The dissociation of H₂O₂ into adsorbed H₂O* and O* intermediates occurred spontaneously with a pronounced energy release of −4.17 eV. Subsequent protonation of O* represented the rate-limiting step with a positive energy barrier, underscoring the critical influence of Pt doping on promoting this energetically uphill process. Comparative analysis revealed superior energy modulation within Pt-ECDs relative to ECDs. The catalytic cycle was ultimately completed through water desorption, an endothermic step demanding − 3.55 eV in both systems. Overall, these computational results highlight the crucial role of Pt incorporation in facilitating both adsorption and catalytic turnover, demonstrating the superior potential of Pt-ECDs as efficient nanozymes with enhanced CAT-like antioxidant functionality (Figs. 4F-G and S11B, Supporting Information).

Throughout the GPX-mimetic reaction mechanisms in Pt-ECDs and ECDs, hydrogen peroxide initially binds to the catalytic centers. Pt-ECDs show a markedly stronger adsorption strength, with an energy of −1.16 eV, in contrast to the weakly positive adsorption energy of 0.22 eV observed for pure ECDs. After proton dissociation, the *OOH intermediate is generated, which then interacts with CH₃SH, resulting in the release of the first water molecule and producing a stabilized OOHCH₃SH intermediate. This intermediate is more stable on Pt-ECDs (−3.87 eV) than on ECDs (−3.61 eV). Subsequent dissociation yields *O and *CH₃SH fragments, recorded at −3.41 eV for Pt-ECDs and − 3.37 eV for ECDs. This is followed by the adsorption and sulfur-assisted coupling of a second CH₃SH molecule, which generates an *O-*CH₃SHSCH₃ complex with energies of −2.53 eV on Pt-ECDs and − 2.42 eV on ECDs. The final desorption step produces CH₃SSCH₃ and a second water molecule, with both systems reaching an energy of −2.12 eV at this stage. The net reaction summarizing this GPX-like pathway can be expressed as: 2 GSH + H₂O₂ → GSSG + 2 H₂O (Figs. 4H-I and S11C, Supporting Information).

During the •OH radical elimination process, both Pt-ECDs and ECDs initially adsorb hydroxyl radicals, forming an *OH intermediate. This species is stabilized at −1.74 eV on Pt-ECDs, substantially more stable than that on ECDs (−0.20 eV). The *OH intermediate then dissociates into *H₂O and *O fragments, with Pt-ECDs exhibiting a more exothermic energy of −4.68 eV, surpassing that of ECDs (−4.11 eV). The resulting *O species is further stabilized on Pt-ECDs at −4.07 eV, indicating stronger binding compared to ECDs (−3.07 eV). Subsequent recombination leads to the formation of *H₂O and *O₂ products, with Pt-ECDs reaching a significantly lower energy state (−8.51 eV) relative to ECDs (−7.79 eV) (Figs. 4J-K and S11D, Supporting Information).

In summary, compared to undoped ECDs, Pt-ECDs exhibit enhanced ROS scavenging and multi-enzyme catalysis, as reflected in the consistently more favorable energetics of their key reaction intermediates.

Pt-ECDs mitigated the sepsis-induced systemic clinical manifestations in mice

The hallmark pathological features of sepsis encompass a cytokine storm and oxidative stress. The inflammatory response in sepsis generates ROS, which in turn activate additional inflammatory signaling pathways [63]. This process perpetuates a vicious cycle, rapidly culminating in multiple organ failure. Carbon dots doped with emodin, which possess multi-enzyme catalytic activities and ROS-scavenging capacity, appear to represent a promising therapeutic strategy for this condition. Based on the characterization and properties of both ECDs and Pt-ECDs, preliminary DFT calculations indicate that Pt-ECDs exhibit superior structural stability and enhanced multi-enzyme mimetic activities compared to ECDs.

To further evaluate their therapeutic efficacy in a murine model of sepsis, mice were treated with Pt-ECDs, ECDs, emodin, or K₂PtCl₄ following cecal ligation and puncture (CLP) [64]. Our preliminary time-course screening following CLP surgery identified the 24-hour time point as the optimal therapeutic window, based on the observation that TNF-α levels reached their peak compared to other time points (Figure S12, Supporting Information). The results demonstrated that treatment with Pt-ECDs, ECDs, and emodin all conferred a protective effect against sepsis. Notably, the Pt-ECDs treatment group showed a significant reduction in mortality compared to other groups (Figure S13A, Supporting Information). Furthermore, parameters including the lung wet-to-dry weight ratio and the levels of serum inflammatory cytokines were markedly attenuated in Pt-ECDs-treated mice (Figures S13B-D, Supporting Information). These findings collectively suggest that Pt-ECDs offer a significantly superior therapeutic outcome over the other treatment regimens evaluated.

Integrated biological and material characterization confirms the therapeutic advantage of Pt-ECDs in a murine sepsis model. In a dose-response study, Pt-ECDs administration resulted in a progressive reduction in the concentrations of key pro-inflammatory mediators (TNF-α and IL-1β) in septic mice across increasing dosage levels (0, 1, 2.5, 5, and 7.5 µg/kg) (Figure S14A, Supporting Information). Notably, the anti-inflammatory effects reached a plateau between 5 and 7.5 µg/kg, with no significant additional benefit observed at the higher dose (Figures S14B-C, Supporting Information). Therefore, the lowest concentration demonstrating maximal efficacy (5 µg/kg) was chosen for further evaluation. In addition, histopathological assessment of the lung, heart, liver, kidney and intestine tissues by H&E staining and serum biochemical indices such as ALT, AST and BUN revealed no evidence of toxicity induced by Pt-ECDs (5 µg/kg) (Figure S15A-H, Supporting Information).

Consequently, mice were administered Pt-ECDs (5 µg/kg) via oral gavage prior to CLP induction to evaluate their potential protective effects against subsequent sepsis. Various physiological and biochemical parameters were assessed 24 h post-modeling to determine treatment efficacy (Fig. 5A). Biodistribution results showed that Pt-ECDs, unlike free Cy5.5, were rapidly retained in the intestines within 4 h after oral gavage and cleared gradually over 48 h, indicating their superior ability to accumulate and persist in intestinal tissues (Figs. 5B and S16, Supporting Information).

Fig. 5.

Pt-ECDs Mitigate Systemic Pathological Manifestations in Septic Mice. (A) Schematic illustration of CLP-induced sepsis establishment and Pt-ECDs therapeutic protocol. (B) Representative intestinal fluorescence imaging in mice following oral gavage of Cy5.5-labeled Pt-ECDs at 4, 8, 12, 24, and 48 h post-CLP (n = 3 biological replicates per group). (C-F) Physiological indicators at 24 h post-CLP: (C) Survival rate, (D) Body weight variation, (E) Clinical score, and (F) Body temperature in the control group, sepsis group (24 h post-CLP), and Pt-ECDs treatment group (5 µg/kg) (n = 10 biological replicates per group). (G-R) Expression profiles of serum biochemical markers among three groups including (G) WBC, (H) Neut, (I) ALT, (J) AST, (K) BUN, (L) Cr, (M) TNF-α, and (N) IL-1β, (O) IL-6, (P) IL-10, (Q), IFN-γ, (R) IFN-β (n = 6 biological replicates per group). (S) Lung wet/dry weight ratio among three groups (n = 6 biological replicates per group). (T) Lung injury scores among three groups (n = 6 biological replicates per group). (U) Quantitative analysis of LDH activity among three groups (n = 6 biological replicates per group). (V) Representative H&E-stained lung tissue sections from the three groups (scale bar: 100 μm; n = 6 biological replicates per group). (W-X) Representative images of intestinal reactive oxygen species (ROS) fluorescence staining (W) and quantitative analysis (X) of ROS production capacity (n = 3 biological replicates per group). (Y) Representative western blot analysis of intestinal tight junction proteins (ZO-1 and CLAUDIN-1) (n = 3 biological replicates per group). (Z-AA) Representative immunofluorescence (IF) images of (Z) CLDN1 and (AA) ZO-1 proteins (scale bar: 100 μm; n = 3 biological replicates per group). Data were expressed as mean ± SD (Significance was calculated by one-way ANOVA with Sidakʼs multiple-comparisons test, * P-value < 0.05; ** P-value < 0.01; *** P-value < 0.001;**** P-value < 0.0001)

Subsequent analysis of the therapeutic efficacy of Pt-ECDs in septic mice demonstrated that treatment not only ameliorated sepsis-induced alterations in body temperature, body weight, and clinical scores (Figs. 5D-F), but also significantly prolonged survival (Fig. 5C). Concurrently, complete blood count and hepatic/renal function tests were performed to evaluate systemic inflammatory damage. Hematological and biochemical results revealed that Pt-ECDs administration markedly improved the levels of white blood cells and neutrophils, while also reducing malondialdehyde levels in the CLP model, indicating an alleviation of immune dysregulation caused by hyperactivation of the inflammatory response during sepsis (Figs. 5G-H and S18A, Supporting Information). Furthermore, biochemical indices of liver and kidney function showed that Pt-ECDs significantly reduced the levels of AST, BUN, and Cr in septic mice, suggesting a protective effect on hepatic and renal function (Fig. 5J-L).

To further assess the development of cytokine storm in septic mice, serum levels of pro- and anti-inflammatory cytokines were measured using enzyme-linked immunosorbent assay (ELISA). Notably, Pt-ECDs treatment substantially suppressed the production of key inflammatory cytokines, including TNF-α, IL-1β, IL-6, IFN-γ, and IFN-β (Fig. 5M-O and Q-R). In contrast, a significant increase in IL-10 levels was observed following Pt-ECDs administration, underscoring its robust anti-inflammatory potential (Fig. 5P).

Finally, the impact of Pt-ECDs on sepsis-associated organ injury was systematically evaluated. Lung injury was assessed through pulmonary evaluations including the wet/dry weight ratio, lung injury scoring, lactate dehydrogenase (LDH) activity, and H&E staining (Fig. 5S-V). Compared to the sepsis group, Pt-ECDs treatment resulted in reduced lung W/D ratio (Fig. 5S), lower injury scores (Fig. 5T), and decreased LDH activity (Fig. 5U). H&E-stained sections showed attenuated inflammatory cell infiltration, diminished hemorrhage, and restored alveolar architecture (Fig. 5V). Given that Pt-ECDs enter the system via the intestinal tract, we further evaluated intestinal damage. ROS staining indicated a reduction in fluorescence intensity in the Pt-ECDs-treated group compared to septic controls (Fig. 5W-X). Western blot and immunofluorescence analyses demonstrated that Pt-ECDs administration up-regulated the expression of key intestinal barrier markers (CLDN1 and ZO1), suggesting a protective role against sepsis-induced gut barrier dysfunction (Figs. 5Y-AA, S18B-C and S19, Supporting Information). Analysis of the biodistribution of Pt-ECDs across major organs including the heart, lung, intestine, and kidney revealed significant accumulation primarily in the liver and kidneys, with negligible levels detected in the lungs and heart (Figure S17, Supporting Information). Considering oral administration of Pt-ECDs effectively ameliorated sepsis-induced lung injury, these findings suggest that the therapeutic benefits are likely mediated through indirect mechanisms, potentially involving modulation of the gut microbiota.

Pt-ECDs ameliorate sepsis-induced lung injury by suppressing macrophage pyroptosis and inhibiting pro-inflammatory polarization

To gain a mechanistic understanding of how Pt-ECDs alleviate lung injury in septic mice at the genetic level, RNA transcriptome sequencing was performed on mice treated with Pt-ECDs. Analysis revealed distinct differences in gene expression between the sepsis group (24 h after CLP) and the Pt-ECDs-treated group (Pt-ECDs: 5 µg/kg) (Fig. 6A). Strong correlation coefficients among samples within each group demonstrated the robustness and reproducibility of the data (Figures S21A-C, Supporting Information). GO enrichment analysis indicated that differentially expressed genes following Pt-ECDs treatment were primarily enriched in pathways related to inflammatory response, positive regulation of NLRP3 inflammasome complex assembly, protein binding. Similarly, KEGG enrichment analysis showed significant enrichment in NOD-like receptor signaling pathway, TNF signaling pathway, tight junction compared with sepsis group (Figs. 6B and S21D-E, Supporting Information). Notably, GSEA analysis showed pathways associated with inflammatory response, pyroptosis, regulation of the NLRP3 inflammasome complex, and IL-1β signaling were significantly downregulated after Pt-ECDs administration (Fig. 6C), which were significantly upregulated in septic mice (Figures S20A-F). These findings provide further genetic evidence supporting the mechanistic role of Pt-ECDs in modulating macrophage function.

Fig. 6.

Pt-ECDs attenuate sepsis-induced lung injury via suppressing macrophage pyroptosis and pro-inflammatory polarization. (A) Transcriptomic heatmap comparing sepsis (24 h post-CLP) and treatment groups (Pt-ECDs: 5 µg/kg). Red indicates significantly upregulated genes, while blue indicates significantly downregulated genes. (B) Bubble plots of significantly upregulated GO and KEGG pathways between sepsis and control groups. (C) GSEA analysis between sepsis and treatment groups. (D-E) Representative IF co-staining image of (D) NLRP3/F4/80 and (E) GSDMD/F4/80 among three groups in vivo (scale bars: 100 μm; n = 3 biological replicates per group). (F) Expression of mRNA including NLRP3, GSDMD, Caspase1, IL-6, and TNF-α detected by qPCR among three groups (n = 3 biological replicates per group). (G) Representative western blot of pyroptosis-associated markers (n = 3 biological replicates per group). (H-K) Quantitative analysis of (H) NLRP3, (I) N-GSDMD, (J) CL-CASPASE1 and (K) CL-IL1β protein expression (n = 3 biological replicates per group). (L-M) Representative IF co-staining image of (L) iNOS/F4/80 and (M) CD86/F4/80 among three groups in vivo (scale bars: 100 μm; n = 3 biological replicates per group). (N) Expression of mRNA including inos, IL-1β, CD86, and CD163 detected by qPCR among three groups (n = 3 biological replicates per group). (O) Representative western blot of macrophage polarization and inflammation markers (n = 3 biological replicates per group). (P-Q) Quantitative analysis of (P) iNOS and (Q) p-P65 protein expression (n = 3 biological replicates per group). Data were expressed as mean ± SD (Significance was calculated by one-way ANOVA with Sidakʼs multiple-comparisons test, * P-value < 0.05; ** P-value < 0.01; *** P-value < 0.001; **** P-value < 0.0001)

The NLRP3 inflammasome is a cytosolic multiprotein complex closely associated with the canonical pyroptosis pathway. Upon activation, it promotes the cleavage of gasdermin D (GSDMD), generating the active N-terminal fragment (GSDMD-N), which oligomerizes and forms pores in the cell membrane. This leads to membrane rupture, release of cellular contents, and initiation of an inflammatory response. Based on the transcriptome results, we hypothesized that Pt-ECDs mitigate the pathogenesis and progression of septic lung injury by modulating macrophage pyroptosis and pro-inflammatory polarization. To test this, we conducted IF, western blot, and qPCR experiments. IF staining revealed that Pt-ECDs significantly suppressed NLRP3 inflammasome activation and subsequent processing of GSDMD and IL-1β, alongside a reduction in the specific M1 macrophage polarization markers iNOS and CD86 (Fig. 6D-E, L-M and S22-S23, Supporting Information). qPCR results showed that Pt-ECDs treatment downregulated key genes involved in pyroptosis (e.g., Nlrp3, Gsdmd, Caspase1) and pro-inflammatory responses (e.g., Il-6, Tnf-α, inos, Il-1β, Cd86, Cd80) (Figs. 6F and N). WB analysis further confirmed decreased protein expression of NLRP3, N-GSDMD, CL-CASPASE1, CL-IL1β, iNOS, and phosphorylated P65 following Pt-ECDs treatment (Fig. 6G-K and O-Q).

Overall, both transcriptomic profiling and experimental validation support the conclusion that Pt-ECDs alleviate septic lung injury likely through inhibiting macrophage pyroptosis and pro-inflammatory polarization.

Regulatory effects of Pt-ECDs on gut microbiota in mice with sepsis-induced lung injury

Samples were collected from both murine and human subjects with sepsis, as outlined in the flowchart in Fig. 7A. Fecal samples from mice were subjected to 16S rRNA sequencing. The α-diversity, as measured by the Shannon index, Simpson index, and Chao1 index, was significantly reduced in the Sepsis group compared to the Control group. However, intervention with Pt-ECDs effectively reversed this decline across all three indices, indicating that Pt-ECDs may have the potential to enhance the richness and diversity of the gut microbiota in cases of Sepsis (Figs. 7B-D). Principal Coordinates Analysis (PCoA) highlighted distinct compositions of the gut microbiota among the groups (Fig. 7E), with notable differences observed at the phylum, genus, and family levels (Figs. 7F-H). Compared to the control group, septic mice exhibited an increased abundance of opportunistic pathogens such as g_Escherichia_Shigella, g_Staphylococcus, g_Klebsiella, and g_Enterobacter, while treatment with Pt-ECDs increased the proportion of beneficial probiotics such as g_Ligilactobacillus (Fig. 7I). Further statistical analysis of significantly differentiated taxa was performed to clarify the impact of Pt-ECDs on the gut microbiota. The results demonstrated that, compared to the sepsis group, Pt-ECDs treatment led to an decrease in taxa including g_Ruminococcus, g_Prevotella, g_Enterococcus and g_Bacteroides, while levels of g_Lactobacillus were increased. (Fig. 7J-N).   Analysis of the KEGG enrichment pathways following treatment revealed that the top-ranked pathways were primarily associated with amino acid metabolism, carbohydrate metabolism, lipid metabolism, and related processes (Fig. 7O). These findings suggest that Pt-ECDs may promote the expansion of beneficial probiotics while simultaneously reducing opportunistic pathogens, which are closely tied to the metabolism of carbohydrates, amino acids, and lipids.

Fig. 7.

Gut microbiota dysbiosis observed in septic patients and a murine model is effectively reversed by Pt-ECDs in mice. (A) Schematic diagram of the gut microbiota analysis workflow, including the collection of stool samples from both human subjects and mice. (B-D) Gut microbiota α-diversity indices across control group, sepsis group and treatment group: (B) Shannon, (C) Simpson, (D) Chao1 index (n=5 biological replicates per group). (E) Principal Coordinates Analysis (PCoA) of microbial communities from mouse samples across control group, sepsis group and treatment group (n=5 biological replicates per group). (F-H) Analysis of gut microbiota relative abundance at the family (F), phylum (G) and genus (H) levels (n=5 biological replicates per group). (I) Linear Discriminant Analysis Effect Size analysis across control group, sepsis group and treatment group (LDA=3, n=5 biological replicates per group). (J-N) Several main differential bacterial communities among the three groups including (J) g_Ruminococcus, (K) g_Prevotella, (L) g_Enterococcus, (M) g_Bacteroides and (N) g_Lactobacillus. (O) KEGG microbiota functional prediction analysis. (P-R) Gut microbiota α-diversity indices between sepsis patients and healthy individuals: (P) Chao1, (Q) Shannon and (R) Simpson index (n=7 biological replicates per group). (S) PCoA of microbial communities from human samples between sepsis and control groups (n=7 biological replicates per group). (T-U) Analysis of human gut microbiota relative abundance at the (T) phylum and (U) genus levels. (Significance was calculated by t-test and one-way ANOVA, * P-value < 0.05; ** P-value < 0.01; *** P-value < 0.001; **** P-value < 0.0001)

To further validate our analysis of the gut microbiota in septic mice, we collected fecal samples from human septic patients and conducted 16S rRNA sequencing as well. The results demonstrated significant differences in α-diversity indices, including Chao1, Shannon, and Simpson index, between septic patients and healthy controls (Figs. 7P-R). PCoA further highlighted distinct compositional differences in the gut microbial communities (Fig. 7S). Changes in microbial composition at the phylum, genus, and family levels were consistent with those observed in septic mice (Figs. 7T-U and S24-S25, Supporting Information). To sum up, integrated data from both murine and human studies indicate that sepsis is associated with gut microbiota dysbiosis, characterized by an increase in harmful pathogenic bacteria and a decrease in beneficial bacteria. Treatment with Pt-ECDs in septic mice alleviated this dysbiosis, as evidenced by a reduction in detrimental genera such as g_Escherichia–Shigella, g_Staphylococcus, g_Klebsiella, and g_Enterobacter, along with an increase in the beneficial g_Ligilactobacillus.

The combined gut microbiota and metabolomic analysis of Pt-ECDs in treating mice with sepsis-induced lung injury

The gut microbiota produces a diverse array of metabolites through its metabolic activities, including short-chain fatty acids (SCFAs) and bile acids, which can enter the systemic circulation and influence host metabolism [65]. In conditions of gut dysbiosis, the metabolome also exhibits significant alterations, collectively contributing to the pathogenesis of various diseases [65]. Therefore, we performed serum metabolomic profiling in both murine and human subjects and conducted integrative analysis with gut microbiome data (Fig. 8A). Untargeted metabolomic analysis in mice demonstrated that, compared to the sepsis group, treatment with Pt-ECDs led to the up-regulation of 150 metabolites and the down-regulation of 326 metabolites, out of a total of 1,078 metabolites detected. Principal Component Analysis (PCA) demonstrated distinct metabolite profiles among the three groups, which was further corroborated by orthogonal partial least squares discriminant analysis (OPLS-DA) showing significant separations between each pair of groups (Figs. 8B-D). Volcano plots highlighted a substantial number of differentially abundant metabolites in each comparison (Fig. 8E). KEGG pathway enrichment analysis of these differential metabolites indicated significant involvement in β-alanine metabolism, glycerophospholipid metabolism, arachidonic acid metabolism, among others (Fig. 8F). Building upon the results of the KEGG enrichment analysis and the identification of differential metabolites, we performed a targeted metabolomics analysis focusing on key representative differential metabolites, such as sugars, fatty acids, and amino acids. Both PCA and OPLS-DA again revealed clear separations among the three groups based on targeted metabolic profiles (Figs. 8G-I). Heatmap analysis demonstrated that after treatment, harmful metabolites such as PA, dihydroxybutyric acid, glutaric acid, and 2-hydroxybutyric acid were downregulated, while beneficial metabolites including L-malic acid, betaine, L-carnitine, and L-glycine were upregulated (Fig. 8J). Among these, PA was one of the most significantly altered metabolites (Fig. 8K).

Fig. 8.

Metabolite dysregulation observed in septic patients and a murine model is effectively reversed by Pt-ECDs in mice. (A) Schematic workflow of the metabolomic analysis strategy. (B) PCA of metabolites across control group, sepsis group and treatment group detected by untargeted metabolomics (n = 5 biological independent animals per group). (C) Orthogonal projections to latent structures -discriminant analysis (OPLS-DA) of metabolites between control and sepsis groups detected by untargeted metabolomics (n = 5 biological independent animals per group). (D) OPLS-DA of metabolites between sepsis and treatment groups detected by untargeted metabolomics (n = 5 biological independent animals per group). (E) Volcano plots of differential metabolites across control group, sepsis group and treatment group. (F) Enriched metabolic pathways in treatment vs. sepsis groups. (G) PCA of metabolites across control group, sepsis group and treatment group detected by targeted metabolomics (n = 5 biological independent animals per group). (H) OPLS-DA of metabolites between control and sepsis groups detected by targeted metabolomics (n = 5 biological independent animals per group). (I) OPLS-DA of metabolites between sepsis and treatment groups detected by targeted metabolomics (n = 5 biological independent animals per group). (J) Heatmap of metabolite profiles between treatment and sepsis groups (n = 5 biological independent animals per group). (K) Mouse serum PA levels across control group, sepsis group and treatment group (n = 5 biological independent animals per group). (L) Spearman correlation heatmap linking microbiota, metabolites, and sepsis parameters. (M-P) Correlation of serum PA with (M) leukocytes, (N) creatinine (CREA), (O) SOFA score and (P) APACHE-II score in a septic cohort. (Significance was calculated by t-test and one-way ANOVA, * P-value < 0.05; ** P-value < 0.01; *** P-value < 0.001; **** P-value < 0.0001)

Previous studies have shown that Bacteroides thetaiotaomicron could synthesize PA in vitro and vivo, which further inhibits activated protein C and promotes platelet activation, thereby inducing thrombogenesis [66]. Correspondingly, our gut microbiome data showed a significant reduction in g_Bacteroides after treatment (Fig. 7J). Correlation analysis further revealed that the level of PA was positively correlated with the abundances of g_Bacteroides, g_Prevotella, g_Enterococcus, and g_Ruminococcus, but negatively correlated with g_Lactobacillus (Fig. 8L). To further substantiate the association between g_Bacteroides and PA, we administered g_Bacteroides to mice. Subsequent targeted metabolomics analysis demonstrated a significant elevation in PA levels (Figures S26, Supporting Information). These findings suggest that a microbiota–palmitic acid axis may play an important role in the progression of sepsis. Based on these results, we hypothesize that Pt-ECDs alleviate sepsis-induced lung injury primarily by inhibiting the production of PA by specific microbiota such as g_Bacteroides.

Furthermore, we analyzed the relationship between serum PA levels and clinical parameters including white blood cell count, serum creatinine, SOFA score, and APACHE-II score in septic patients (Figs. 8M-P and Supplementary Table 1). Linear regression was employed to assess the association between PA and these clinical indicators (Figs. 8M-P). The results showed well-fitted curves, and comprehensive curve modeling revealed a positive correlation between PA levels and WBC count, serum creatinine, SOFA score, and APACHE-II score in the sepsis cohort (Figs. 8M-P). These findings suggest that PA may serve as a key biomarker for disease severity in septic patients and could play a critical role in the progression of sepsis.

PA exacerbated macrophage pyroptosis via dual-target binding to NOX2 I411 and NLRP3 D804 with NOX2-mediated ROS amplification

Based on integrated transcriptomic, 16S rRNA sequencing, and metabolomic analyses, we preliminarily propose that Pt-ECDs exert therapeutic effects on septic lung injury in mice, likely through inhibiting the production of PA by gut microbiota such as g_Bacteroides, thereby suppressing macrophage pyroptosis and pro-inflammatory polarization. However, the precise mechanisms by which PA influences downstream signaling pathways remain unclear. To address this, further investigation via PA supplementation experiments (Fig. 9), gain- and loss-of-function genetic studies, and receptor inhibition experiments was conducted in vitro. BMDMs were divided into three experimental groups: control group, LPS + ATP group, and LPS + ATP + PA group (Fig. 9A). Cell viability was assessed under PA concentrations ranging from 0 to 200 µM (Figure S27, Supporting Information). Results indicated that a PA concentration of 25 µM yielded approximately 80% viability in BMDMs, which was selected as a non-cytotoxic concentration for subsequent experiments.

Fig. 9.

Enhanced ROS generation, pyroptosis, and M1-like polarization in macrophages by PA. (A) Schematic of the experiment presented. (B) Visualization of the molecular docking between PA-NOX2 and PA-NLRP3 proteins. (C) Representative fluorescence staining result image of ROS among control group, LPS + ATP group (LPS: 1 µg/ml; ATP: 10mM) and LPS + ATP + PA group (LPS: 1 µg/ml; ATP: 10mM; PA: 25 µM) (scale bars: 20 μm; n = 3 biological replicates per group). (D) Representative flow cytometric detection of ROS generation capacity (n = 3 biological replicates per group). (E) Quantitative analysis of the q-PCR results for Cybb (n = 3 biological replicates per group). (F) Densitometric quantification of DCFH-DA fluorescence intensity (n = 3 biological replicates per group). (G) Transmission electron microscopy (TEM) image among three groups (scale bars: 2 μm for 8000× and 500 nm for 20000×). (H) Scanning electron microscopy (SEM) image among three groups (scale bars: 20 μm, 2.5k × magnification; scale bars: 5 μm, 7.0 k× magnification). (I-K) Representative flow cytometry profiling of (I) NLRP3⁺, (J) cleaved-GSDMD⁺, and (K) CD86⁺ macrophages among three groups (n = 3 biological replicates per group). (L) Expression of mRNA including Nlrp3, Gsdmd, Caspase1, Il-6, and Tnf-α detected by qPCR among three groups (n = 3 biological replicates per group). (M) Representative western blot of pyroptosis-associated markers (n = 3 biological replicates per group). (N-Q) Quantitative analysis of (N) NLRP3, (O) N-GSDMD, (P) CL-CASPASE1 and (Q) CL-IL1β protein expression (n = 3 biological replicates per group). (R) Expression of mRNA including inos, Il-1β, Cd86, and Cd163 detected by qPCR among three groups (n = 3 biological replicates per group). (S) Representative western blot of macrophage polarization and inflammation markers (n = 3 biological replicates per group). (T-U) Quantitative analysis of (T) iNOS and (U) p-P65 protein expression (n = 3 biological replicates per group). (V-X) Representative IF co-staining image of (V) NLRP3/F4/80, (W) GSDMD/F4/80 and (X) iNOS/F4/80 in vitro (scale bars: 20 μm; n = 3 biological replicates per group). Data were expressed as mean ± SD (Significance was calculated by one-way ANOVA with Sidakʼs multiple-comparisons test, * P-value < 0.05; ** P-value < 0.01; ***P-value < 0.001; **** P-value < 0.0001)

Pyroptosis was next evaluated by measuring lactate dehydrogenase (LDH) release, an indicator of membrane rupture. LPS stimulation significantly increased LDH release, an effect further augmented by PA co-treatment (Figure S28, Supporting Information). Scanning electron microscopy (SEM) revealed characteristic pyroptotic morphology including membrane disintegration, cellular swelling, and organelle disruption in LPS-treated cells, with more severe damage observed in the LPS + ATP + PA group (Fig. 9G). Flow cytometry showed elevated counts of NLRP3⁺ and cleaved GSDMD⁺ cells following PA co-treatment (Figs. 9I-J). At the molecular level, PA upregulated mRNA expression of Nlrp3, Gsdmd, Caspase1, Il-6, and Tnf-α (Fig. 9L) and protein levels of NLRP3, N-GSDMD, CL-CASPASE1, and CL-IL1β (Fig. 9M), as confirmed statistically (Figs. 9N-Q). Immunofluorescence further reinforced the enhanced NLRP3 and GSDMD signals in PA-treated cells (Figs. 9V-W) with statistical significance (Figures S33A-B, Supporting Information). Together, these data indicate that PA robustly amplifies pyroptotic signaling in macrophages.

Given that pyroptosis promotes M1 macrophage polarization during sepsis [67, 68], whether PA enhances this phenotype was explored. Flow cytometry revealed a higher percentage of CD86⁺ cells in the LPS + ATP + PA group than in the LPS group (Fig. 9K). Consistent with this, qPCR showed notable upregulation of iNos, Il-1β, Cd80, and Cd86 mRNA (Fig. 9R), and western blot confirmed significantly increased iNOS and p-P65 protein expression (Fig. 9S, quantified in Figs. 9T-U). Immunofluorescence co-staining further amplified iNOS signal in the LPS + PA group (Fig. 9X, quantified in Figure S33C). These results suggest that PA enhances M1 polarization in the context of pyroptosis.

The role of ROS was subsequently investigated, which are known to promote pyroptosis via caspase activation and GSDMD cleavage [69]. Transcriptomic analysis suggested upregulation of Cybb (also known as Nox2), which encodes NOX2, a protein with key role in ROS generation. q-PCR confirmed that PA co-treatment significantly elevated Cybb expression compared to LPS alone (Fig. 9E). Accordingly, both IF and flow cytometry indicated substantially higher ROS levels in the LPS + ATP + PA group (Figs. 9C-D), a finding corroborated by DCFH-DA assay (Fig. 9F). TEM revealed ROS-associated ultrastructural damage—mitochondrial swelling, vacuolization, membrane disintegration, and cellular shrinkage—that was exacerbated by PA (Fig. 9G). To functionally validate the role of NOX2, we performed gain- and loss-of-function experiments. Overexpression of Cybb increased mRNA levels of Cybb, Nlrp3, Gsdmd, and Il-1β, while knockdown suppressed them (Figures S29A-D, Supporting Information), confirming that NOX2-mediated ROS production enhances NLRP3 inflammasome activation and pyroptosis.

Integrating transcriptomic and metabolomic data, we further investigated whether NOX2 and NLRP3 are central to PA-induced pyroptosis by employing selective inhibitors: GSK2795039 (NOX2 inhibitor) and MCC950 (NLRP3 inhibitor) (Figure S30A, Supporting Information). The results showed that LPS + ATP + PA co-treatment significantly enhanced the expression of NOX2 and NLRP3, while GSK2795039 and MCC950 effectively reversed this effect, respectively. Moreover, LPS + ATP + PA significantly upregulated the mRNA and protein levels of markers related to pyroptosis and M1 polarization These enhancements were substantially attenuated by individual inhibition of either NOX2 or NLRP3, with the most pronounced suppression observed upon dual inhibition (Figures S30B-J, Supporting Information). Collectively, these results suggest that NOX2 and NLRP3 are key downstream targets of PA-induced pyroptosis.

To elucidate the direct binding mechanisms, molecular docking analyses were performed, which revealed high-affinity interactions between PA and both NOX2 and NLRP3 (Fig. 9B). Site-directed mutagenesis was then employed to identify the functional binding residues on both proteins. As shown in Fig. 9B, isoleucine (I411), tyrosine (Y324), and glutamate (E410) in NOX2 were predicted as potential PA-binding residues. Transfection with I411A, Y324A, or E410A mutant plasmids did not affect basal NOX2 expression (Figures S31A-B, Supporting Information). However, functional assays demonstrated that the I411A mutation, but not Y324A or E410A, significantly attenuated PA-induced cytotoxicity, ROS accumulation, and NLRP3/GSDMD/caspase-1/IL-1β activation upon LPS + PA stimulation (Figures S31C-H, Supporting Information). These findings suggest that I411 is a critical functional residue mediating PA binding to NOX2. Notably, the I411A mutation not only attenuated PA-induced exacerbation but also partially alleviated LPS-induced damage, indicating that the I411 site in NOX2 may serve as a key mediator of ROS production. Similarly, D804, Y918, and Y861 in NLRP3 were identified as putative binding sites. While none of the point mutations (D804A, Y918A, Y861A) altered NLRP3 expression (Figures S32A-B, Supporting Information), the D804A mutation significantly reduced cytotoxicity and suppressed GSDMD/caspase-1/IL-1β activation under LPS + PA stimulation, unlike Y918A and Y861A (Figures S32C-F, Supporting Information), identifying D804 as a critical residue for PA binding to NLRP3.

In summary, our findings demonstrate that PA directly interacts with NOX2-I411 and NLRP3-D804, thereby promoting ROS overproduction and pyroptosis, respectively. Moreover, ROS overload functions as a positive feedback regulator that further enhances NLRP3-mediated pyroptosis, establishing a self-amplifying cycle that exacerbates inflammatory cell death and pro-inflammatory polarization.

Restoration of the gut microbiota ameliorated septic lung injury by suppressing macrophage pyroptosis and pro-inflammatory polarization.

To further substantiate the essential role of the gut microbiota in the therapeutic mechanism of Pt-ECDs, mice underwent gut microbiota depletion (via antibiotic treatment) (Figure S34A, Supporting Information) and received fecal microbiota transplantation (FMT) (Fig. 10). Specific pathogen-free (SPF) mice were subjected to oral gavage of broad-spectrum antibiotics (ABX) for one week prior to CLP induction to deplete the intestinal microbiota (Figure S34A, Supporting Information). The results revealed that mortality in the Treatment + ABX group began to occur at 24 h post-CLP and reached a level comparable to that of the sepsis group by 168 h, with no statistically significant difference between the two groups (Figure S34B, Supporting Information). Moreover, lung injury scores and serum levels of inflammatory cytokines were also similar to those in the sepsis group (Figures S34C-D, Supporting Information). These findings collectively demonstrate that depletion of the gut microbiota abrogates the therapeutic benefits of Pt-ECDs in sepsis.

Fig. 10.

FMT pretreated with Pt-ECDs exerts a mitigating effect on sepsis-induced lung injury. (A) Schematic diagram illustrating the experimental design of Pt-ECDs-pretreated FMT. (B-C) Physiological parameters 24 h post-CLP including (B) Survival rates (n = 20 biological independent animals per group) and (C) Clinical score (n = 6 biological replicates per group) among control group, sepsis group (24 h post-CLP), sepsis + septic mice FMT group (24 h post-CLP), and sepsis + Pt-ECDs pretreated FMT group (24 h post-CLP). (D) IF of intestinal ROS among four groups (scale bars: 100 μm; n = 3 biological replicates per group). (E) IF of intestinal tight junction protein CLAUDIN-1 among four groups (scale bars: 100 μm; n = 3 biological replicates per group). (F) IF of lung ROS among four groups (scale bars: 100 μm; n = 3 biological replicates per group). (G-I) IF co-staining of (G) NLRP3/F4/80, (H) GSDMD/F4/80 and (I) iNOS/F4/80 in lung macrophages (scale bar: 100 μm; n = 3 biological replicates per group). (J-L) Representative western blot analyses of (J) intestinal barrier proteins (CLAUDIN-1 and ZO-1), (K) pyroptosis markers (NLRP3, GSDMD, CASPASE1, CL-IL1β), and (L) macrophage polarization and inflammation markers (iNOS, p65, p-P65) (n = 3 biological replicates per group). (M-U) Quantification of protein expression of (M) ZO-1, (N) CLAUDIN-1, (O) NLRP3, (P) N-GSDMD, (Q) CL-CASPASE1, (R) CL-IL1β, (S) iNOS, (T) p-P65 and (U) p-P65/P65 among four groups (n = 3 biological replicates per group). (V-W) Expression of mRNAs among four groups (n = 3 biological replicates per group). Data were expressed as mean ± SD (Significance was calculated by one-way ANOVA with Sidakʼs multiple-comparisons test, * P-value < 0.05; ** P-value < 0.01; *** P-value < 0.001)

FMT is a therapeutic approach designed to reconstitute the gut ecosystem through the transfer of intestinal microbiota from a donor to a recipient. To assess whether the gut microbiota contributes to the therapeutic efficacy of Pt-ECDs in septic mice, fecal samples obtained from Pt-ECDs-treated septic mice were transplanted into recipient mice prior to the induction of sepsis via CLP. Recipient mice were pretreated with an antibiotic cocktail to deplete their endogenous gut microbiota (Fig. 10A). The experimental groups included: a control group (untreated), a sepsis group (subjected to CLP and sacrificed 24 h post-CLP), a sepsis/FMT group (receiving fecal transplant from septic mice prior to CLP), and a sepsis/Pt-ECDs FMT group (receiving fecal transplant from Pt-ECDs-treated septic mice prior to CLP) (Fig. 10A). Alterations in the relative abundances of the major bacterial phyla indicate effective FMT (Figure S35, Supporting Information). The results indicated that mice in the sepsis/FMT group exhibited reduced survival, higher clinical scores, and elevated serum levels of the inflammatory cytokines TNF-α and IL-1β compared to the sepsis group. In contrast, the sepsis/Pt-ECDs FMT group showed a significant improvement in survival (Fig. 10B), accompanied by markedly reduced clinical scores and lower levels of serum TNF-α and IL-1β (Figs. 10C and S36A-B, Supporting Information). These findings suggest that fecal microbiota derived from Pt-ECDs-treated septic mice alleviated systemic symptoms in septic recipients.

Further evaluation of intestinal injury revealed that the sepsis/Pt-ECDs FMT group displayed significantly decreased intestinal ROS levels (Figs. 10D and S37A, Supporting Information) and enhanced expression of gut barrier markers CLDN-1 and ZO-1 relative to the sepsis group (Figs. 10E, J, M, N, S36D and S37B-C, Supporting Information), indicating that FMT from Pt-ECDs-treated mice suppressed intestinal radical generation and attenuated gut damage. Assessment of lung injury through histopathological examination demonstrated reduced inflammatory cell infiltration, diminished hemorrhage in alveolar spaces, and partial restoration of tissue architecture in the sepsis/Pt-ECDs FMT group compared to the sepsis group (Figure S36C, Supporting Information). To investigate whether Pt-ECDs mitigate lung injury via gut microbiota-mediated modulation of macrophage pyroptosis and pro-inflammatory polarization, additional analyses including IF, western blot, and qPCR were performed. IF staining revealed reduced intensity of lung ROS, NLRP3 and GSDMD in the sepsis/Pt-ECDs FMT group compared to the sepsis group (Figs. 10F-H and S37D-F, Supporting Information). Western blot analysis confirmed decreased protein expression of NLRP3, FL-GSDMD, N-GSDMD, CL-CASPASE1, and CL-IL1β (Fig. 10K and O-R). qPCR results further indicated downregulation of Nlrp3, Gsdmd, Caspase1, IL-6, and TNF-α in the sepsis/Pt-ECDs FMT group (Fig. 10V). Similarly, IF analysis showed reduced intensity of iNOS and CD86 in the sepsis/Pt-ECDs FMT group relative to the sepsis group (Figs. 10I, S36E and S37G-H, Supporting Information). Western blot indicated decreased expression of iNOS and a reduced p-P65/P65 ratio (Figs. 10L, S-U), while qPCR revealed downregulation of inos, Il-1β, Cd86, and Cd80 (Fig. 10W). Together, these results suggest that FMT from Pt-ECDs-treated septic mice inhibited pyroptosis and pro-inflammatory polarization in pulmonary macrophages.

Conclusion

In this study, emodin was utilized as a carbon source to successfully synthesize Pt-ECDs via a hydrothermal method. The therapeutic potential of Pt-ECDs against sepsis-induced lung injury was comprehensively evaluated. Owing to their robust multi-enzyme catalytic activity and potent ROS scavenging capacity, Pt-ECDs effectively eliminated intracellular free radicals and alleviated oxidative stress, thereby significantly mitigating oxidative cellular damage. Compared with emodin alone, Pt-ECDs exhibited superior therapeutic efficacy in a murine model of sepsis, as evidenced by improvements in survival rates, inflammatory cytokine levels, lung injury markers, and intestinal barrier integrity.

Transcriptomic analysis of lung tissues revealed that oral administration of Pt-ECDs downregulated the expression of genes associated with oxidative stress, pyroptosis, and pro-inflammatory polarization in septic mice. Given the crucial role of gut microbiota in disease modulation, we performed 16S rRNA sequencing and metabolomic profiling. Integrated analysis indicated that Pt-ECDs promoted a beneficial shift in the gut microbial composition, characterized by an increase in beneficial taxa and a reduction in harmful bacteria, particularly g_Bacteroides, which was correlated with the production of PA, a saturated fatty acid. These findings suggest that Pt-ECDs may alleviate sepsis-induced lung injury by inhibiting the production of PA by harmful bacteria such as g_Bacteroides.

To further validate this hypothesis, BMDMs were treated with PA, which resulted in upregulated expression of genes related to oxidative stress, pyroptosis, and pro-inflammatory polarization. Functional studies involving PA supplementation experiments, gain- and loss-of-function genetic studies, and receptor inhibition experiments confirmed that PA exacerbated macrophage pyroptosis via dual-target binding to NOX2 I411 and NLRP3 D804 with NOX2-mediated ROS amplification. Furthermore, FMT from Pt-ECD-treated septic mice into untreated septic mice markedly attenuated lung injury and suppressed the expression of oxidative stress-, pyroptosis-, and pro-inflammatory polarization-related genes. Conversely, ablation of the gut microbiota nearly abolished the therapeutic benefits of Pt-ECDs.

In summary, this study demonstrates that Pt-ECDs, a novel carbon nanomaterial derived from traditional Chinese medicine with multi-enzyme mimicking and ROS scavenging properties, can mitigate septic lung injury by modulating the gut microbiota-palmitic acid-NOX2/NLRP3/CASPASE1/GSDMD pathway, thereby inhibiting macrophage pyroptosis and pro-inflammatory polarization. These findings highlight the potential of Pt-ECDs as a promising therapeutic agent for sepsis-associated lung injury and provide a valuable framework for future clinical strategies in this field.

Discussion

In this work, we demonstrate that Pt-ECDs, derived from a traditional Chinese medicine precursor, alleviate septic lung injury via the gut microbiota-palmitic acid-NOX2/NLRP3/CASPASE1/GSDMD axis, thereby suppressing macrophage pyroptosis and pro-inflammatory polarization. The proposed mechanistic pathway is illustrated in Fig. 11.

Fig. 11.

Mechanistic diagram of Pt-ECD treatment in sepsis-induced lung injury. In a murine model of sepsis, intestinal dysbiosis develops, promoting the release of pathogenic microbial metabolites that enter the systemic circulation and translocate to the lungs. There, these metabolites exacerbate oxidative stress and induce pyroptosis, while simultaneously driving the pro-inflammatory polarization of macrophages. Oral Pt-ECDs preferentially accumulate in the intestinal tract, where they promote restoration of gut barrier integrity and reduce the abundance of detrimental bacteria, particularly g_Bacteroides, leading to a marked decrease in PA levels. The subsequent reduction in PA alleviates oxidative stress in macrophages and suppresses pyroptosis, ultimately attenuating sepsis-induced lung injury

Unlike inert precursors, carbon dots derived from bioactive molecules often retain the functional groups of their parent molecules, thereby inheriting inherent pharmacological activities. This design yields a synergistic effect: Pt-ECDs not only possess the catalytic activity of metal doping but also retain the intrinsic pharmacological benefits of emodin. However, emodin itself exhibits extremely poor aqueous solubility and a slow oral dissolution rate. Moreover, it undergoes extensive glucuronidation in the intestines and liver, leading to rapid systemic elimination and consequently low bioavailability, which significantly limits its clinical application [41]. In contrast, Pt-ECDs effectively circumvent these limitations [70]. Pt-ECDs possess multiple advantageous properties: their ultra-small size (approximately 5 nm) facilitates penetration across various physiological barriers, including the intestinal epithelium, thereby significantly improving therapeutic accessibility. The material demonstrates negligible biotoxicity, and its unique hydrothermal synthesis endows it with excellent biocompatibility in both in vitro and in vivo settings. These attributes, coupled with flexible routes of administration, highlight the strong translational potential of Pt-ECDs for clinical use. The superior performance of Pt-ECDs has been rigorously validated through multiple experimental approaches. First, compared to other metal-doped carbon-based materials, Pt-ECDs exhibited higher TON and Vmax in CAT and POD activity assays, indicating enhanced catalytic efficiency [56, 58]. This improvement is likely attributable to the richer surface functional groups on Pt-ECDs, which facilitate more efficient catalytic reactions. Second, metal doping endowed the emodin-derived carbon dots with novel CAT and POD activities, while also significantly augmenting their SOD, GPx-like activities, and free radical scavenging capacity relative to ECDs [71]. Third, molecular computations further revealed that Pt-ECDs possess lower intermediate-state energy barriers, supporting more thermodynamically favorable and kinetically efficient catalytic processes.

The gut microbiota is now recognized as a vital metabolic organ. Its impact on sepsis-induced lung injury, mediated by specific microbial metabolites, has been well-established [72]. Alterations in the gut microbial community such as changes in the abundance of g_Proteobacteria, g_Enterococcus, g_Bifidobacterium, and g_Lactobacillus have been correlated with the development of septic lung injury [65, 73]. In our study, we observed a significant reduction in g_Bacteroides within the gut microbiota of mice following Pt-ECD treatment. In humans, g_Bacteroides represent one of the predominant bacterial groups in the adult colon, accounting for approximately 25% of gut bacteria. Notably, Lai et al. demonstrated that g_Bacteroides are capable of producing PA in vitro [66]. As an end product of fatty acid synthesis, PA can accumulate intracellularly and induce lipotoxicity. Research indicates that PA promotes the acquisition of enhanced metastatic capabilities in cancer cells via Schwann cells, and this increased metastatic potential does not require sustained exposure to PA [74]. In addition, studies have shown that exogenous PA upregulates the expression of adhesion molecules (e.g., VCAM-1 and E-selectin) and coagulation-related factors (e.g., PAI-1 and TF) in endothelial cells, thereby promoting leukocyte adhesion and thrombus formation, which directly contribute to the pathogenesis of sepsis-induced lung injury [75]. Our metabolomic analysis revealed a decrease in serum PA levels in Pt-ECD-treated mice, supporting the hypothesis that Pt-ECDs exert their protective effects via suppression of the gut microbiota–palmitic acid axis. Furthermore, while previous case-control studies often reported general dysbiosis rather than disease-specific microbial changes [76], our integrated human and animal data successfully identified distinct septic host-specific gut microbial and metabolomic signatures associated with sepsis. In this study, we have demonstrated the therapeutic efficacy of Pt-ECDs in mice with gut microbiota dysbiosis, thereby laying the groundwork for their clinical application.

Pt-ECDs exhibit considerable clinical potential for the management of sepsis-induced lung injury. Compared with commercially available NOX2 inhibitors or pyroptosis inhibitors, Pt-ECDs demonstrate significantly enhanced efficacy in preventing the onset and progression of septic lung injury. Current clinical strategies primarily involve broad-spectrum antimicrobial agents to control infection [77], corticosteroids to suppress inflammation [78], and mechanical ventilation for respiratory support [79]. However, these approaches are associated with limitations such as increased risk of secondary infections with prolonged use and the emergence of drug-resistant pathogens. In contrast, Pt-ECDs possess an ultrasmall particle size, favorable biocompatibility, negligible toxicity, and potent anti-inflammatory properties, thereby offering a promising and versatile therapeutic alternative for sepsis-induced lung injury.

There are some limitations in this study. First, although the current study focused on the efficacy and mechanisms of Pt-ECDs within the 7-day therapeutic window for sepsis-induced lung injury, the assessment of long-term toxicity remains outstanding and will be prioritized in future investigations. Second, only male mice were used to establish animal models in this study, as they are more suitable for modeling sepsis-related lung injury. Consequently, the influence of PA levels in female versus male subjects warrants further validation. Third, although we have successfully demonstrated that PA exacerbates macrophage pyroptosis primarily via direct binding to the NOX2 and NLRP3 proteins, a critical mechanistic facet remains unaddressed: the subsequent post-translational modification termed palmitoylation (S-acylation). As a reversible covalent modification involving the attachment of PA to cysteine residues, palmitoylation plays a pivotal role in regulating protein trafficking and stability [80]. Given its substrate (PA) is central to our findings, the contribution of PA-mediated palmitoylation to the inflammatory response remains an important direction for future functional proteomics research. Finally, given the vast diversity of gut microbiota, which includes numerous other potential producers of PA, more comprehensive studies are needed to fully elucidate the microbial contributions to sepsis pathogenesis.

Experimental section

Single-Cell RNA sequencing analysis

Single-cell RNA sequencing data derived from a murine model of sepsis (GSE207651, GEO repository) were analyzed with the 10× Genomics workflow. Initial quality control involved filtering out low-quality genes and cells by including only genes detected in a minimum of three cells and cells containing at least 200 detected genes. Additional thresholds excluded cells where mitochondrial genes constituted over 30% of total counts or where the total number of detected genes fell outside the range of 200 to 7500. Normalization was carried out with the log-normalization method implemented in Seurat. A set of 2,000 variably expressed genes was selected using the FindVariableFeatures method, and subsequent scaling was applied to the dataset. Using these variable genes, we performed dimensionality reduction via principal component analysis (PCA) and retained the first 30 principal components for downstream analysis.

Hematoxylin and Eosin staining (H&E)

Tissue samples underwent fixation in 10% neutral-buffered formalin followed by paraffin embedding. Sections (5 μm) were mounted on glass slides, stained with hematoxylin and eosin, dehydrated through graded ethanol series, and coverslipped for bright-field microscopic evaluation.

Immunofluorescence Staining (IF): Paraffin-embedded tissue sections were subjected to sequential dewaxing using eco-friendly solutions I-III (10 min per solution), followed by hydration through graded absolute ethanol concentrations (5 min each) and immersion in distilled water. Antigen retrieval was performed prior to passive cooling of slides; subsequent rinsing involved three 5-minute phosphate-buffered saline (PBS: pH 7.4) washes under orbital shaking. Target regions were delineated and blocked with 3% bovine serum albumin (BSA) for 30 min. Primary antibodies were introduced into humidity-controlled chambers for overnight incubation at 4 °C. After PBS rinses, species-matched secondary antibodies underwent 50-min incubation at ambient temperature. Nuclear counterstaining employed 4′,6-diamidino-2-phenylindole (DAPI, Servicebio G1012, China) for 10 min, with subsequent autofluorescence quenching via Reagent B (5 min) and exhaustive rinsing. Slides were coverslipped employing anti-fade mounting medium. Wavelengths are as follows: DAPI (Ex 330-380 nm/Em 420 nm), Alexa Fluor 488 (Ex 465-495 nm/Em 515-555 nm) and Cy5 (Ex 608-648 nm/Em 672-712 nm). Triplicate replicates were conducted per experimental condition. Comprehensive antibody specifications appear in Supplementary Table 1.

Quantitative assessment of inflammatory mediators and hepatorenal injury markers

Circulating concentrations of pro-inflammatory cytokines (WBC, Neut, IL-6, IL-10, TNF-α, IL-1β, IFN-γ, IFN-β) and hepatic/kidney injury biomarkers (ALT, AST, BUN, Cr) in murine blood were measured. Analyses employed specific enzyme-linked immunosorbent assay (ELISA) kits alongside blood gas analyzers (Shenzhen Leidu Life Science Co., Ltd, China). Manufacturer protocols were rigorously followed to ensure methodological reproducibility. All resultant data underwent normalization relative to respective sample dilution factors.

Synthesis of Pt-ECDs and ECDs

A precisely prepared precursor solution was formulated by dissolving emodin (59.2 µmol, T2896), potassium tetrachloroplatinate (K₂PtCl₄, 144.5 µmol, Beyotime Y001037, China), and branched polyethylenimine (PEI, 160.0 mg, Beyotime Y268770, China) in anhydrous N, N-dimethylformamide (DMF, 30.0 mL, Beyotime Y238893, China). The mixture was subjected to pulsed ultrasonication (300 W, 5 s pulse on/5 s pulse off) in an ice-water bath for 10 min until a clear, amber-colored solution was obtained. This solution was subsequently transferred into a 50 mL Teflon-lined stainless-steel autoclave, purged with argon for 5 min to minimize dissolved oxygen, sealed, and heated to 120.0 °C in a forced convection oven (Memmert UNE series) for 6 h. A controlled heating rate of 3 °C per minute was applied to reach the target temperature. After the reaction was complete, the autoclave was allowed to cool to room temperature naturally over no less than 8 h to ensure gradual crystallization. The resulting deep brown suspension was filtered under vacuum through a sterile 0.22 μm polyethersulfone (PES) membrane filter to remove any macroscopic aggregates. The filtrate was immediately transferred into pre-treated dialysis tubing (3,500 Da) and dialyzed against 4.0 L of ultrapure deionized water under gentle stirring (200 rpm). The dialysis water was replaced at precise intervals (every 6 h for the first three changes, then every 12 h) for a total duration of 72 h to ensure complete removal of ionic byproducts, unreacted precursors, and organic solvent. The final deep solution was collected, passed through a second 0.22 μm sterile filter, and stored in an amber glass vial at 4 °C for a maximum of one week. The synthesis method of ECDs is similar to this, except that the raw materials have been replaced with emodin (59.2 µmol, T2896) and branched polyethylenimine (PEI, 160.0 mg, Beyotime Y268770, China).

Characterization of Pt-ECDs and ECDs

For TEM and HRTEM characterization, Pt-ECD suspensions were subjected to ultrasonication before being applied onto copper grids and air-dried. Imaging and lattice parameter measurements were carried out using an electron microscope (Thermofisher, U.S.). Particle size distribution and surface charge were evaluated via DLS and zeta potential measurements performed on a Zetasizer Nano instrument (Malvern, UK). XRD measurements were obtained from solid samples mounted on glass substrates using an X-ray diffractometer (Rigaku, Japan). The UV-Vis absorption spectrum of the sample was recorded using a UV-Vis spectrophotometer (GENESY™ 40/50 Vis/UV-Vis). Thermal stability was assessed by thermogravimetric analysis (TGA) on a thermal analyzer (Netzsch, Germany) under nitrogen atmosphere with a heating rate of 10 °C/min. Raman spectroscopy was performed on liquid specimens placed on glass slides using a spectrometer (Thermofisher, U.S.) with optimized acquisition durations. XPS examination involved survey scans and high-resolution analysis of Pt 4f, C 1 s and O 1 s regions, with subsequent spectral fitting to determine chemical bonding states using an XPS system (Thermofisher, U.S.).

Enzyme Activity and Enzyme Kinetics: (1) SOD-like Activity: The superoxide anion (O₂•⁻) scavenging ability was evaluated using a WST-8 inhibition assay (Beyotime, S0101S). Pt-ECDs or ECDs were introduced into a superoxide system generated by xanthine/xanthine oxidase and incubated (37 °C, 5% CO₂). Absorbance was measured at 450 nm (BioTek Synergy H1). (2) CAT Activity and Enzyme Kinetics: Catalase-mimetic behavior was determined by tracking H₂O₂ decomposition via UV-vis spectroscopy. Reactions containing Pt-ECDs and H₂O₂ were monitored at 240 nm. To quantify oxygen generation, dissolved oxygen levels were recorded every 10 s for 600 s using an oxygen electrode (INESA, JPSJ-605 F) after mixing Pt-ECDs with 100 mM H₂O₂ in PBS (pH 7.4). Real-time O₂ release was also traced with a clark-type microelectrode. For kinetic analysis, various H₂O₂ concentrations were reacted with Pt-ECDs (50 µg/mL) for 1 min. The Michaelis-Menten constant (K_m) and maximum velocity (Vmax) were derived from the equation: v = Vmax·[S]/(K_m + [S]). (3) POD Activity and Enzyme Kinetics: The POD-like activity of Pt-ECDs was evaluated by monitoring the oxidation of TMB in the presence of H₂O₂ using UV-vis spectroscopy. Reaction mixtures containing Pt-ECDs at varying concentrations, H₂O₂, and TMB were prepared in a cuvette, and the increase in absorbance at 652 nm was recorded to quantify catalytic turnover. The K_m and Vmax were derived from the equation: v = Vmax·[S]/(K_m + [S]). (4) GPx Activity: A GR-coupled assay kit was used to evaluate the GPx-like activity of Pt-ECDs. The assay measures the catalytic oxidation of GSH (8.4 mM) to GSSG in the presence of H₂O₂ (5 mM) and Pt-ECDs, followed by NADPH (300 µM)-dependent recycling by GR. The corresponding consumption of NADPH, monitored at 340 nm by UV-vis-NIR spectroscopy, served as a proxy for enzymatic activity. Furthermore, the TON was determined from the ratio of Vmax to the enzyme concentration ([E0]).

Total Antioxidant Capacity (TAC) and Free Radical Scavenging Ability: (1) The TAC of Pt-ECDs and ECDs was evaluated with a commercial assay kit utilizing the decolorization of ABTS•+. The ABTS•+ radical was generated by reacting ABTS stock solution with potassium persulfate, followed by 12-hour incubation in darkness at room temperature. The solution was then diluted to an absorbance of 0.70 at 734 nm to obtain the working solution. Calibration curves were constructed using Trolox (0.15-1.2 mM) and vitamin C (0.15-1.5 mM) as reference standards. The ABTS•+ working solution was mixed with different volumes of standards or carbon dot samples and incubated for 10 min. Absorbance was measured at 734 nm, and the antioxidant capacity was quantified based on the Trolox standard curve and expressed as mM Trolox-equivalent antioxidant capacity. (2) DPPH• and PTIO• scavenging capacity: Electron spin resonance (ESR) spectroscopy was employed to assess the scavenging capacity of Pt-ECDs or ECDs toward hydroxyl (•OH) and superoxide (O₂•⁻) radicals. •OH was generated via the Fenton reaction using a Fe²⁺/H₂O₂ system, and O₂•⁻ was produced by the xanthine/xanthine oxidase reaction. Pt-ECDs or ECDs were introduced into the radical solutions, followed by the addition of the spin trap agent DMPO. ESR measurements were then conducted to evaluate the radical elimination efficiency. (3) DPPH• and PTIO• scavenging capacity: The free radical scavenging activities of Pt-ECDs and ECDs were evaluated using DPPH• and PTIO• assays. For the DPPH• assay, samples were mixed with a 50 µg/mL DPPH• ethanolic solution, and absorbance was measured at 519 nm using UV-vis spectroscopy. For the PTIO• assay, a solution of PTIO• in PBS (5 mM, pH 7.4) was prepared with an initial absorbance of 0.7 ± 0.02 at 557 nm. Pt-ECDs or ECDs (0-100 µg/mL) were mixed with the PTIO• solution at a 1:9 volume ratio. Following a 30-minute incubation at room temperature, the mixture was centrifuged to remove nanoparticles. The supernatant’s absorbance was then measured at 557 nm using a microplate reader. All measurements were performed in triplicate.

Computational Details: All calculations were conducted using density functional theory (DFT) under the GGA-PBE framework; geometries were first optimized, and their energies were subsequently evaluated. Plane-wave basis sets with a cutoff energy of 500 eV were employed. The slab models employed a Γ-centered 3 × 3 × 1 k-point mesh, with convergence thresholds set at 1.0 × 10^− 6 eV/atom for energy and 0.05 eV/Å for forces. A 25 Å vacuum layer was applied normal to the slab to avoid periodic interactions. Free energies included zero-point energy (ZPE) and entropy corrections, calculated via: ΔG = E + ΔZPE + ΔH - TΔS, where E is electronic energy, ΔZPE the ZPE correction, ΔH the thermal enthalpy, T temperature, and ΔS the entropy change.

Sepsis model induction and treatment intervention

Male C57BL/6J mice (8 weeks old; 22 ± 2 g) were obtained from the Naval Medical University Animal Experiment Center (Shanghai, China) and Cyagen Biotechnology Co. under specific pathogen-free (SPF) conditions. Polymicrobial sepsis was established using a standardized cecal ligation and puncture (CLP) protocol. Briefly, anesthetized subjects underwent midline laparotomy, followed by ligation of the distal cecum just below the ileocecal valve with a 4 − 0 silk suture. The ligated segment was then punctured once through-and-through with a 21-gauge needle to induce fecal leakage, and a small amount of fecal content was gently extruded. Subsequently, the abdominal wall and skin were closed in layers with interrupted 4− 0 silk sutures. For therapeutic assessment, Pt-ECDs suspension (5 µg/kg) was administered orally (100 µl/mouse) before CLP. All procedures followed ARRIVE guidelines with approval from Shanghai Changzheng Hospital’s Institutional Animal Care and Use Committee.

In vivo biodistribution analysis

Following 24-hour fasting to minimize gastrointestinal content interference, mice received oral gavage of Cy5.5-conjugated Pt-ECDs or PBS formulations. Probe preparation involved vigorous mixing of Pt-ECDs or PBS with near-infrared dye Cy5.5 (MedChemExpress HY-D0924, USA) followed by overnight incubation at 4 °C. Subjects were euthanized at scheduled intervals (4, 8, 12, 24, 48 h post-administration). Excised organs underwent fluorescence quantification using a 676/705 nm excitation/emission filter system (LABATECH, China).

Body temperature and body weight

Core temperature and body mass were recorded daily for all mice (n = 10). To ensure accuracy, every measurement was conducted in triplicate.

Western Blot: Protein concentrations were determined via bicinchoninic acid (BCA) assay. Lysates and plasma specimens underwent separation on 4–12% Bis-Tris gradient gels under denaturing conditions, followed by electrophoretic transfer to nitrocellulose membranes. Membranes were blocked with TBST (Tris-buffered saline + 0.1% Tween-20) containing 5% non-fat milk, then incubated overnight at 4 °C with primary antibodies including anti-β-actin (1:10000), anti-claudin-1 (1:1000), anti-ZO-1 (1:1000), anti-GSDMD (1:1000), anti-NOX2 (1:1000), anti-iNOS (1:1000), anti-CASPASE1 (1:1000), anti-CL-IL1β (1:1000) and anti-NLRP3 (1:1000), all obtained from commercial sources. After TBST washing, immunoblotting was performed using HRP-conjugated secondary antibodies with detection by enhanced chemiluminescence. ImageJ software quantified band intensities, normalized to β-ACTIN, followed by statistical analysis. Comprehensive antibody details appear in Supplementary Table 1.

Quantitative Real-Time PCR analysis

Total RNA was isolated using RNAfast200 reagent, followed by cDNA synthesis with PrimeScript RT Reagent Kit (TaKaRa RR037A; integrates gDNA Eraser). Gene-specific primers were designed in accordance with MIQE guidelines using PrimerBank resources, with sequences detailed in Supplementary Table 1. qRT-PCR amplifications were performed using these primers and SYBR Premix Ex Taq (TaKaRa RR420A) on the ABI 7900HT Fast Real-Time PCR System from Applied Biosystems. Relative expression of target genes normalized to the endogenous reference GAPDH was quantified using the 2^(-ΔCt) method, where ΔCt = Ct(target) - Ct (GAPDH), yielding expression values relative to the reference gene within each biological replicate.

Bulk transcriptome profiling and analysis

Murine pulmonary tissue underwent RNA sequencing for genome-wide transcriptional assessment. Library construction and processing followed rigorous quality standards, with technical replicates exhibiting > 95% concordance (CV < 5%). Differential expression analysis employed DESeq2 (R package) using thresholds of |fold change| > 1.5 and FDR < 0.05. Sample relationships and transcriptomic patterns were visualized through hierarchical clustering heatmaps and ggplot2-derived volcano plots.

ROS Assay: For ROS detection in intestinal and lung tissues, hydrophobic barriers were drawn around frozen tissue sections. The sections were incubated with a ROS probe (Service G1746, China) at 37 °C for 30 min under dark conditions. Following PBS washes, nuclear counterstaining was performed using DAPI (Service G1012, China) at room temperature for 10 min. Slides were subjected to an additional wash and then mounted with anti-fade mounting medium. Image acquisition was carried out with the following filter configurations: DAPI (excitation: 330–380 nm, emission: 420 nm); FITC/488 (excitation: 465–495 nm, emission: 515–555 nm). For intracellular ROS detection, the Reactive Oxygen Species Detection Kit (Beyotime S0033, China) was utilized in accordance with the manufacturer’s instructions. All experiments were performed in triplicate.

Human Samples: Human blood and fecal samples were collected from control subjects without sepsis (n = 19) and patients with sepsis (n = 19) under the approved study protocol (license number: 2025SL032), which was authorized by the Ethics Committee of Changzheng Hospital, Naval Medical University. Patients’ information was presented in Supplementary Table 2. Prior to biospecimen acquisition, written informed consent was secured from every individual or their legally authorized representatives.

Gut microbiota analysis

For the purpose of exploring the associations between the gut microbiota of septic mice and clinical sepsis, along with the functional variations of intestinal bacterial communities in septic mice after Pt-ECDs intervention, cecal contents were harvested from each experimental group and subjected to 16S rRNA gene sequencing. Genomic DNA was isolated using a commercially available kit and utilized as the template for PCR amplification of the V3-V4 hypervariable regions of the 16S rRNA gene, with the specific primers 341 F and 806R. Sequencing was performed on the Illumina MiSeq platform (Novogene Biotech Co., Ltd., Beijing, China).

Raw sequencing data were analyzed within the QIIME2 workflow. With the DADA2 plugin, the reads were subjected to quality control, denoising, read merging, and chimeric sequence elimination. High-quality reads were grouped into Amplicon Sequence Variants (ASVs) with a 97% similarity cutoff, generating an ASV abundance matrix. Alpha diversity was evaluated via the Shannon index at the ASV level, and statistical significance was determined by one-way ANOVA. Beta diversity was analyzed through Principal Coordinates Analysis (PCoA) based on bray_curtis dissimilarity and visualized using R software. Differential microbial abundance among groups was detected by LEfSe analysis, with a linear discriminant analysis (LDA) score threshold set at ≥ 3.Functional potential was predicted from Operational Taxonomic Units (OTUs) using PICRUSt, which inferred the abundance of KEGG orthologs.

Untargeted metabolomics

For untargeted metabolomic profiling of mouse or human serum samples, 100 µL serum sample was combined with 20 µL internal standard (IS) (0.3 µg/mL, 2-Chloro-L-phenylalanine) and a mixture consisting of 200 µL methanol and 100 µL acetonitrile. The resultant mixture was homogenized for 30 s and subsequently subjected to centrifugation at 13,000 rpm for 15 min at 4 °C to facilitate the precipitation of denatured proteins. The supernatant was gathered and stored at −80 °C for subsequent analysis. Metabolite analysis employed a UHPLC system (SCIEX, USA) coupled to an ZenoTOF 7600 system. Chromatographic separation was performed on a Waters ACQUITY UPLC HSS T3 column, with mobile phase A (0.1% aqueous formic acid) and mobile phase B (0.1% formic acid in acetonitrile) employed. Mass spectrometric analysis adopted Information Dependent Acquisition (IDA) and Dynamic Background Subtraction (DBS) technologies, operating in both positive and negative ionization modes. Data acquisition and processing were conducted using SCIEX MExplorer Ultimate for subsequent analysis.

Targeted metabolomics

Targeted metabolomic analysis employed a Agilent 1290 Infinity II (Agilent, USA) coupled to an SCIEX 6500 QTRAP + triple quadrupole-mass spectrometry system. Separation occurred on a ACQUITY UPLC BEH C18 Column using mobile phase A (0.1% formic acid in water) and phase B (10 mmol/L formic acid ammonium salt in methanol). The MS analysis was conducted in positive ion mode utilizing ESI and multiple reaction monitoring (MRM) scans. Quantitative analysis, including MS data acquisition and processing, was conducted using SCIEX Analyst Work Station (v1.7.32) and BIOTREE Bio Bud (v2.0.3). All calibration curves demonstrated linearity with R² > 0.95.

Metabolic profiling analysis

Data processing and statistical analysis were conducted with MetaboAnalyst 6.0. The raw data matrix underwent sum normalization, logarithmic transformation, and Pareto scaling. To explore inherent clustering, PCA was performed, while orthogonal partial least squares discriminant analysis (OPLS-DA) was applied to enhance group separation. Features with a variable importance in projection (VIP) score exceeding 1.0 were selected as relevant for discrimination. Statistical significance was assessed using Student’s t-test, with a threshold of p < 0.05 that was further adjusted for multiple comparisons via false discovery rate (FDR) correction. Metabolites exhibiting a fold change ≥ 2 and meeting the combined criteria of VIP > 1.0 and adjusted p-value were included in the final analysis.

Cell culture

Bone marrow-derived macrophages (BMDMs) were isolated from female C57BL/6 mice following established protocols [81]. Following euthanasia, femurs and tibias were excised, and bone marrow cells were harvested by flushing the bones with ice-cold Dulbecco’s modified Eagle’s medium (DMEM; Gibco, NY, USA). After centrifugation, the resulting cell pellets were resuspended and maintained for 7 days in DMEM containing 10% fetal bovine serum (FBS; Gibco), 1% penicillin-streptomycin (PS; Gibco), and 20 ng/mL recombinant murine macrophage colony-stimulating factor (M-CSF; Miltenyi Biotec, Germany) to promote macrophage differentiation.

Preparation of BMDMs: For stimulation experiments, cells were treated with 1ug/ml LPS (HY-B2176) for 2 h, after which the supernatant was discarded and treated with 10mM ATP (CAT No. 56-65-5) for 3 h. When assessing PA, varying concentrations of the compound were administered to the culture medium prior to LPS + ATP stimulation. Inhibitor studies utilized BMDMs seeded in 6-well plates at a density of 2 × 10⁶ cells per well. Cells were allocated into the following treatment groups: control, LPS + ATP, LPS + ATP + PA (25 µM; Selleck-S3794, USA), LPS + ATP + PA + GSK2795039 (5 µM; Selleck-S8974, USA), LPS + ATP + PA + MCC950 (5 µM; Selleck-S8930, USA).

Molecular Docking: To validate the molecular docking protocol, the three-dimensional structures of key active compounds were retrieved in SDF format from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). The crystallographic structure of the target protein was acquired from the RCSB PDB database (http://www.rcsb.org/) in PDB format. Prior to docking, the protein structure was processed using PyMOL to remove water molecules and native ligands, followed by saving the modified structure in PDB format. The binding site parameters were defined using the Getbox plugin. Both the prepared protein and ligand structures were then loaded into AutoDock Tools 1.5.6 for conversion into the PDBQT format. Docking simulations were carried out with AutoDock Vina 1.1.2. The resulting docking poses were analyzed and visualized using PyMOL version 2.6.0.

Cell counting Kit-8 (CCK8)

Optimal PA concentrations for BMDMs were determined employing the Cell Counting Kit-8 (Beyotime C0037, China) per manufacturer specifications. Cells underwent 24-hour exposure to logarithmically diluted PA solutions at concentrations of 3.125, 6.25, 12.5, 25, 50, 100, and 200 µM prior to assessment.

Flow cytometry

A 100 µL aliquot of cell suspension (approximately 1 × 10⁵ cells) was incubated with a mixture of fluorescent antibodies comprising CD86 (Invitrogen 11-0862-82, China), NLRP3 (Invitrogen MA5-23919, China), and GSDMD (Invitrogen PA5-115330) for 30 min under dark conditions with mild agitation. Post-staining, the cells were subjected to two rounds of PBS washing (1 mL per wash, centrifugation at 300×g for 5 min) and then resuspended in 300 µL ice-cold PBS. A BD CytoFLEX flow cytometer (United States) was used for the quantification of distinct cell subsets.

LDH release

Upon establishing the pyroptosis model in mouse BMDMs in vitro, the cell culture supernatant was collected. Membrane integrity was evaluated via the detection of lactate dehydrogenase (LDH) release with a commercially available LDH assay kit (Beyotime C0019S, China).

SEM analysis

Macrophages were harvested by centrifugation, after which the supernatant was carefully removed. The cell pellet was subsequently washed with phosphate-buffered saline (PBS), and the PBS was aspirated completely prior to re-suspending the cells in the appropriate fixative. Following fixation, samples were rinsed three times with 0.1 M phosphate buffer (PB, pH 7.4), post-fixed with 1% osmium tetroxide for a duration of 1–2 h at room temperature, and then washed an additional three times with PB. Dehydration was carried out using a graded ethanol series (30%, 50%, 70%, 80%, 90%, 95%, and two changes of 100% ethanol for 15 min each), followed by incubation in isoamyl acetate. The dehydrated sample was then placed onto a coverslip and dried using a critical point dryer. Finally, the sample was mounted on a metal stub using conductive carbon adhesive tape, sputter-coated with gold for 30 s, and examined using scanning electron microscopy.

TEM analysis

For TEM analysis, macrophage pellets were collected via centrifugation, resuspended in TEM fixative, and stored at 4 °C. For agarose pre-embedding, the fixed pellets were centrifuged again, washed three times with 0.1 M phosphate buffer (PB, pH 7.4), and subsequently embedded in pre-cooled 1% agarose prior to solidification.

Post-fixation was carried out by immersing the agarose-embedded samples in a solution of 1% osmium tetroxide in PB for 2 h at room temperature under light-protected conditions. Following this, the samples were rinsed three times with PB. Dehydration was performed using a graded ethanol series (30% to 100%, with 20 min per step), followed by two changes of acetone (15 min each), all conducted at room temperature. Resin infiltration was conducted sequentially using acetone and EMBed 812 resin mixtures (1:1 ratio for 2-4 h, followed by 1:2 ratio overnight), and finally with pure EMBed 812 resin for 5-8 h at 37 °C. The samples were then transferred into embedding molds filled with fresh resin and polymerized overnight at 37 °C, followed by an additional 48 h at 60 °C. The resulting polymerized resin blocks were sectioned into ultrathin slices (60-80 nm), which were collected onto formvar-coated copper grids. These sections were stained with 2% uranyl acetate for 8 min under light-protected conditions and subsequently with 2.6% lead citrate for 8 min under CO₂-free conditions, with intermediate rinsing steps in between. After drying, the grids were examined using a transmission electron microscope.

Cybb Silencing and NOX2/NLRP3 Mutant Plasmid Transfection In Vitro: BMDMs received 50 nM Cybb-targeting siRNA (sequence: 5′-GCTGAATGTCTTCCTCTTT-3′; GenePharma, China) or non-targeting control siRNA (GenePharma) via GP-transfer-Mate Transfection Reagent, incubated for 24 h. Parallel experiments introduced wild-type (WT), mutant NOX2 plasmids or mutant NLRP3 plasmids into BMDMs using the same transfection system, followed by 48-hour incubation.

Fecal Microbiota Transplantation (FMT): To determine whether Pt-ECDs mitigates septic lung injury via gut microbiota modulation, FMT was implemented according to established methodologies as previously described. Donor fecal samples from Pt-ECDs-treated septic mice were collected under specific pathogen-free (SPF) conditions. Immediately following collection, aliquots (100 mg/mL) were homogenized in 1 mL sterile saline and incubated for 15 min. Subsequent processing involved vortexing followed by centrifugation at 1000 × g (4 °C, 5 min). Supernatants were aseptically harvested and cryopreserved at -80 °C until transplantation. Recipient animals received a 3-day oral antibiotic cocktail (ABX: 1 g/L amoxicillin, 1 g/L neomycin sulfate, 1 g/L metronidazole, 0.5 g/L vancomycin) in drinking water to establish pseudo-germ-free conditions through endogenous microbiome ablation. Antibiotic water was subsequently replaced with standard hydration. Transplantation commenced 24 h later, with recipients receiving daily 100 µL oral administrations of bacterial suspensions or saline controls for 14 consecutive days.

Statistical analysis

All experimental data originated from independent replicates, with sample sizes ranging between 3 and 50, and were expressed as mean ± standard deviation (SD). For parametric data, statistical analyses were conducted via one-way or two-way analysis of variance (ANOVA), complemented by Sidak’s or Tukey’s multiple comparison tests respectively; the unpaired t-test was utilized for direct pairwise comparisons. Ordinal nonparametric data were assessed using the Kruskal-Wallis test, followed by Dunn’s correction to account for multiple comparisons. Fluorescence colocalization was quantified through the calculation of Pearson’s correlation coefficient (R-value ranging from − 1 to + 1). All statistical analyses were conducted using GraphPad Prism 10 software, and significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Abbreviations

ALI

Acute lung injury

ARDS

Acute respiratory distress syndrome

CLP

Cecal ligation and puncture

scRNA-seq

Single-cell RNA sequencing

Pt-ECDs

Pt-doped emodin-derived carbon dots

ECDs

Emodin-derived carbon dots

ROS

Reactive oxygen species

SOD

Superoxide dismutase

CAT

Catalase

POD

Peroxidase

GPx

Glutathione peroxidase

TEM

Transmission electron microscopy

HRTEM

High-resolution transmission electron microscopy

XRD

X-ray diffraction

XPS

X-ray photoelectron spectroscopy

TGA

Thermogravimetric analysis

DLS

Dynamic light scattering

TON

Turnover number

Km

Michaelis constant

Vmax

Maximum reaction velocity

FMT

Fecal microbiota transplantation

GSDMD

Gasdermin D

ABx

Antibiotic cocktail

SPF

Specific pathogen-free

GO

Gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

GSEA

Gene set enrichment analysis

PCA

Principal component analysis

UMAP

Uniform Manifold Approximation and Projection

IF

Immunofluorescence

WB

Western blot

qPCR

Quantitative polymerase chain reaction

ELISA

Enzyme-linked immunosorbent assay

UPLC-MS/MS

Ultra-performance liquid chromatography-tandem mass spectrometry

DMEM

Dulbecco’s modified eagle medium

SCFA

Short-chain fatty acids

DMF

N, N-dimethylformamide

K₂PtCl₄

Potassium platinochloride

FBS

Fetal bovine serum

CCK-8

Cell Counting Kit-8

CFSE

Carboxyfluorescein succinimidyl ester

RBCs

Red blood cells

DAPI

4′,6-diamidino-2-phenylindole

BCA

Bicinchoninic acid

TBST

Tris-buffered saline with Tween-20

HRP

Horseradish peroxidase

ESR

Electron spin resonance

ABTS

2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid)

DPPH

2,2-diphenyl-1-picrylhydrazyl

PTIO

2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide

TAC

Total antioxidant capacity

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

TNF-α

Tumor necrosis factor-alpha

OPLS-DA

Orthogonal projections to latent structures -discriminant analysis

PA

Palmitic acid

IL

Interleukin

ZO-1

Zonula occludens-1

CLDN-1

Claudin-1

Arg1

Arginase 1

iNOS

Inducible nitric oxide synthase

Author contributions

Honghao Song: Conceptualization, Methodology, Investigation, Writing-Original Draft, Visualization; Yuqing Ma: Methodology, Validation, Formal Analysis, Investigation, Data Curation; Lei Peng: Software, Validation, Formal Analysis, Resources, Writing-Original Draft, Data Curation; Fangyuan Gao: Investigation, Resources, Writing-Review & Editing, Project Administration; Xiaoyi Fan: Investigation, Data Curation, Visualization; Mei Yang: Investigation, Resources; Tong Hua: Resources, Supervision; Yutong Yang: Resources; Rongrong Fan: Resources; Zhenjie Li: Conceptualization, Writing-Review & Editing, Supervision, Project Administration, Funding Acquisition; Hongbin Yuan: Conceptualization, Writing-Review & Editing, Supervision, Project Administration, Funding Acquisition.

Funding

This study was supported by the National Natural Science Foundation of China (Grant No. 82471239) and Military clinical key specialty project fund (JDLCZDZK).

Data availability

The original contributions presented in the study are included in the article/Supplementary Material; further inquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

All animal experiments were subject to review and approval by the Animal Care and Use Committee of Shanghai Changzheng Hospital affiliated with the Naval Medical University (license numbers: SYXK Hu, 2022-0011). The experiments followed the guidelines for responsible animal research as set forth by the International Association for the Study of Pain. All human feces and serum samples were subject to review and approval by the Animal Care and Use Committee of Shanghai Changzheng Hospital affiliated with the Naval Medical University (license numbers: 2025SL032), and Chinese Clinical Trial Registry (license numbers: ChiCTR2500108440).

Competing interests

The authors declare no competing interests.