A multistage micro-nano biosystem prevents sepsis-associated encephalopathy through coordinated neuroinflammation suppression and gut-brain axis modulation Authors: Xuening Zhang, Jia Dong, Wentao Liu, Shiyuan Hua, Donghu Lin, Jiahua Niu, Min Zhou Published: 2026 ABSTRACT AbstractSepsis-associated encephalopathy (SAE) is a severe neurological complication of sepsis, for which current therapeutic options remain limited. There is an urgent need for innovative strategies that simultaneously target both central and peripheral pathophysiological mechanisms. In this study, we developed a microalgae-nanoparticle multistage biosystem, CV@MT-NPs, by electrostatically assembling melatonin-loaded, quaternary chitosan-modified liposomes (MT-NPs) onto Chlorella vulgaris (CV) microalgae. The resulting CV@MT-NPs enabled sustained melatonin release and significantly prolonged systemic circulation and brain retention of melatonin. In an LPS-induced murine SAE model, CV@MT-NPs markedly improved survival, attenuated cognitive deficits and anxiety-like behaviors, reduced clinical severity, and mitigated body weight loss. Mechanistically, MT-NPs released from CV@MT-NPs entered the brain, where they preserved blood-brain barrier integrity by restoring tight junction proteins and re-establishing astrocyte-endothelial coupling. Furthermore, they suppressed neuroinflammation by inhibiting microglial hyperactivation, promoting microglial homeostasis, attenuating astrogliosis, and protecting mitochondrial structure and function. Beyond the central nervous system, MT-NPs also ameliorated systemic inflammation, restored intestinal barrier integrity, and enhanced M2 macrophage polarization through inhibition of the NF-κB signaling pathway. Notably, the CV component played a complementary role by rebalancing the gut microbiota, specifically enriching beneficial short-chain fatty acid-producing bacteria, and elevating levels of acetic, propionic, and butyric acids in both the gut and the brain. Critically, comprehensive biocompatibility assessments revealed no evidence of acute or chronic toxicity. Collectively, CV@MT-NPs represent a multifunctional, gut-brain axis-modulating therapeutic platform that, when administered prophylactically, demonstrates significant potential to concurrently ameliorate central neuropathology and peripheral immune-metabolic dysfunction, offering a promising preventive strategy for SAE.Keywords: Sepsis-associated encephalopathy, Microalgae-nanoparticle multistage biosystem, Neuroinflammation, Gut-brain axis, Short-chain fatty acids Introduction Sepsis-associated encephalopathy (SAE) is a critical complication of sepsis, and the World Health Organization has identified SAE as a key health focus area [1,2]. This condition arises from systemic inflammatory responses and immune dysregulation that compromise the blood-brain barrier (BBB), leading to a spectrum of neuropsychiatric manifestations ranging from mild confusion to severe coma [1]. The pathophysiology underlying SAE involves complex interactions among pro-inflammatory cytokines, chemokines, and neuronal signaling pathways, which disrupt synaptic transmission, neurovascular coupling, and cognitive processing [3,4]. Emerging evidence highlights the pivotal role of the gut–brain axis in both the onset and progression of SAE [[5], [6], [7]]. This bidirectional communication system links the central nervous system (CNS) with the enteric nervous system and modulates systemic immunity and inflammation. During sepsis, intestinal barrier dysfunction and gut dysbiosis facilitate the translocation of pathogen-associated molecular patterns (PAMPs) [8] and damage-associated molecular patterns (DAMPs) [9,10], amplifying systemic inflammation and exacerbating neuroinflammation and neuronal injury. Consequently, the interplay between systemic inflammation, gut microbiome disruption, and cerebral pathology has opened new avenues for SAE intervention [11,12]. Despite these insights, current therapeutic strategies for SAE remain largely supportive, with no specific pharmacological interventions proven to effectively target both neuroinflammation and gut dysbiosis simultaneously. Melatonin, an endogenous hormone best known for regulating circadian rhythms, has emerged as a promising neuroprotective agent in various brain disorders. Its potent antioxidant and anti-inflammatory properties confer benefits in neurodegenerative conditions such as Alzheimer's [13,14] and Parkinson's disease [15,16], while its ability to mitigate neuronal damage has been demonstrated in models of stroke [[17], [18], [19]] and traumatic brain injury [20,21]. Moreover, melatonin may alleviate mood-related symptoms by modulating serotonergic and other neurotransmitter systems [22]. Critically for SAE, melatonin preserves mitochondrial integrity and attenuates BBB disruption [23], offering multi-faceted protection against sepsis-induced brain injury. However, endogenous melatonin levels are markedly suppressed in SAE patients, and exogenous supplementation faces significant pharmacokinetic challenges, including poor bioavailability, rapid systemic clearance, and limited brain penetration. Encapsulation within phospholipid-based carriers such as liposomes represents a promising strategy to enhance melatonin's stability, prolong its circulation, and improve CNS delivery. Similarly, while modulation of the gut microbiota holds therapeutic promise, conventional probiotic or prebiotic interventions often lack the precision and efficacy required to restore gut-brain homeostasis during acute sepsis. Microalgae, photosynthetic microorganisms abundant in aquatic environments, offer a compelling alternative. Characterized by rapid growth and exceptional nutritional density, microalgae are rich sources of high-quality proteins, complex lipids, polysaccharides, vitamins, and bioactive compounds [[24], [25], [26]]. Therapeutically, they exhibit robust anti-inflammatory, antioxidant, and immunomodulatory activities [27,28]. In gastrointestinal contexts, microalgae modulate microbial composition and suppress intestinal inflammation [29,30]. Their polysaccharides selectively promote the growth of beneficial taxa such as Bifidobacteria and Lactobacillus [31,32], enhance intestinal barrier integrity via upregulation of tight junction proteins, and inhibit pathogen colonization [33]. Notably, microalgae modulate gut microbiota–derived metabolites, including short-chain fatty acids (SCFAs) like acetate, propionate, and butyrate, nourish colonic epithelial cells and support immune homeostasis [34], suggesting broad applicability in SAE-related gut dysfunction. Based on these complementary mechanisms, we hypothesized that a rationally engineered hybrid system, integrating sustained-release melatonin with the gut-reparative and microbiota-modulating functions of microalgae, could synergistically ameliorate SAE through dual targeting of the brain and the gut-brain axis. To test this hypothesis, we developed CV@MT-NPs, a multifunctional therapeutic platform that coordinates neuroprotection and gut modulation (Fig. 1). The system was fabricated in a stepwise manner: melatonin was first encapsulated into liposomal nanoparticles, which were then surface-functionalized with quaternary chitosan to yield positively charged melatonin-loaded nanoparticles (MT-NPs). These MT-NPs were subsequently anchored onto the negatively charged surface of Chlorella vulgaris (CV) microalgae, forming the CV@MT-NPs complex. This design enables sustained melatonin release, enhances systemic circulation, and promotes prolonged retention in the brain. In an LPS-induced murine model of SAE, CV@MT-NPs significantly improved survival, attenuated cognitive deficits and anxiety-like behaviors, reduced clinical severity scores, and prevented body weight loss. Mechanistically, MT-NPs released from CV@MT-NPs crossed the BBB and preserved its integrity by restoring tight junction proteins and reinforcing astrocyte-endothelial interactions. Within the CNS, they suppressed neuroinflammation by inhibiting microglial hyperactivation, promoting microglial homeostasis, reducing astrogliosis, and protecting mitochondrial structure and function. Beyond the brain, CV@MT-NPs mitigated systemic inflammation, restored intestinal barrier integrity, and drove M2 macrophage polarization via NF-κB pathway inhibition. Concurrently, the CV component reshaped the gut microbiota, enriching SCFA-producing bacteria and elevating acetate, propionate, and butyrate levels in both the gut and brain. Comprehensive safety assessments confirmed the absence of acute or chronic toxicity. Collectively, our findings demonstrate that CV@MT-NPs represent a significant advance in SAE prevention, a dual-targeted, gut-brain axis-modulating strategy that concurrently addresses central neurological dysfunction and peripheral immune-metabolic disturbances. Schematic illustration of a multistage micro-nano biosystem (CV@MT-NPs) for preventing sepsis-associated encephalopathy (SAE) through coordinated suppression of neuroinflammation and modulation of the gut-brain axis. Materials and methods Synthesis of MT-NPs and CV@MT-NPs To fabricate MT-NPs, MT-liposomes were initially prepared using the thin-film hydration method. Briefly, a mixture of 200 mg soy lecithin, 20 mg cholesterol, and 20 mg MT was dissolved in 10 mL chloroform and evaporated at 37 °C and 100 rpm to form a thin film. This film was hydrated with 20 mL of ddH2O at 40 °C and 150 rpm. The resulting liposome dispersion underwent sonication in an ice bath, followed by overnight stirring with 8 mg of quaternary chitosan at room temperature to form preliminary MT-NPs. After centrifugation at 8000 rpm for 10 min, the MT-NPs were extruded through polycarbonate membranes with 400 nm and 100 nm pore sizes using a membrane extruder to achieve a uniform particle size distribution. CV (used in its live, non-inactivated form in all experiments) was obtained from Guangyu Biological Technology (Shanghai, China), purified, and subsequently incubated with the prepared MT-NPs for 12 h at room temperature to produce CV@MT-NPs. Characterization of MT-NPs and CV@MT-NPs The morphology of MT-liposomes and MT-NPs was examined via a transmission electron microscope (Cryo-TEM, Talos F200C 200 kV). Images of the CV were captured using an inverted optical microscope (TC-C-TC, Nikon). Size distribution and stability were analyzed using a dynamic light scattering system (DLS, Nano-ZS90, Malvern). Zeta potential measurements to determine the surface charge of CV, MT-liposome, MT-NPs, and CV@MT-NPs were conducted using the same DLS system. Fourier transform infrared spectroscopy (FTIR) analysis was performed with an FTIR Nicolet NEXUS 470 spectrometer (Thermo Fisher Scientific, USA) to investigate the chemical composition of CV, MT, NPs, MT-NPs, and CV@MT-NPs. UV-Vis-NIR spectroscopy was carried out using a UV-2600 spectrophotometer (Shimadzu, Japan) to assess the optical properties of CV, MT, NPs, MT-NPs, and CV@MT-NPs. The morphology characteristics of CV and CV@MT-NPs were examined by scanning electron microscope (SEM, SU-8010, Hitachi, Japan). Fluorescence images of CV@MT-NPs were obtained using a laser scanning confocal microscope (LSM880, Zeiss, Germany). To evaluate melatonin release kinetics, MT, MT-NPs, and CV@MT-NPs (each containing 1 mg of melatonin in 1 mL) were placed into dialysis bags (molecular weight cutoff: 3.5 kDa) and immersed in 50 mL of release medium (PBS, 0.01 M, pH 7.4, 37 °C) containing 0.1% (v/v) Tween 20 to maintain sink conditions under gentle stirring. At predetermined time points, 1 mL of the external medium was withdrawn for UV–Vis analysis and immediately replaced with an equal volume of fresh release medium. The cumulative melatonin release (%) was calculated as: Cumulative release (%) = (50 × Cn+ Σᵢ= 1n−1 Cᵢ × 1)/(Ctotal × 1) × 100%. where Cn represents the melatonin concentration at the sampling time point, and Ctotal denotes the total melatonin concentration measured in each formulation group. To assess the stability of MT-NPs released from CV@MT-NPs, the nanoparticles were suspended in either PBS or PBS supplemented with 10% fetal bovine serum (FBS) and incubated at 37 °C. Changes in particle size and polydispersity index (PDI) were monitored by DLS over a period of 72 h. Drug treatment and SAE model establishment All animal experiments were conducted in strict accordance with protocols reviewed and approved by the Ethics Committee of Zhejiang University (approval number: ZJU20250024). Male C57BL/6 mice (8 weeks old, 20-25 g) were acclimatized for one week under standard housing conditions (12-h light/dark cycle, ad libitum access to food and water) before the experiment. Mice were then randomly assigned to one of six experimental groups: Control, LPS, CV (CV, 12.5 mg/kg), MT (MT, 10 mg/kg), MT-NPs (equivalent to 10 mg/kg MT), and CV@MT-NPs (12.5 mg/kg CV and equivalent to 10 mg/kg for MT). All test formulations were prophylactically administered via intraperitoneal (i.p.) injection once daily for 7 consecutive days. The Control and LPS groups received an equal volume of sterile saline on the same schedule. Twenty-four hours after the final pretreatment dose, all groups except the Control group were challenged with a single i.p. injection of lipopolysaccharide (LPS; 2 mg/kg or 15 mg/kg body weight, depending on the experimental endpoint) to induce SAE. A schematic timeline and detailed design of the pharmacodynamic studies are provided in Fig. S1. Targeted metabolomic analysis of fecal short-chain fatty acids (SCFAs) Fecal samples collected from the colon were also preserved in liquid nitrogen. Targeted analysis of short-chain fatty acids (SCFAs), including acetic, propanoic, and butanoic acids etc, was conducted. Briefly, approximately 20 mg of each fecal sample was homogenized in 800 μL of 0.5% phosphoric acid solution using a cryogenic grinding mill (50 Hz, 3 min). The homogenate was then subjected to ultrasonication for 10 min and centrifuged at 13,000×g for 15 min at 4 °C. A 200 μL aliquot of the resulting supernatant was mixed with 200 μL of n-butanol for solvent extraction. After vortexing for 10 s and a subsequent low-temperature ultrasonication step (10 min), the mixture was centrifuged again at 13,000×g (4 °C, 5 min). The final supernatant was derivatized and analyzed using an Agilent 8890B gas chromatography system coupled with a 5977B mass spectrometer (GC-MS; Agilent Technologies, USA). Data acquisition and initial processing were performed with the Masshunter software suite (Agilent Technologies, USA), following the manufacturer's protocols. A standard curve derived from authentic SCFA standards was used for quantification. Statistical analysis The results from at least three independent experiments were presented as mean ± standard deviation (SD). The difference was assessed using one-way or two-way ANOVA analysis. Statistical significance was indicated as: nsp>0.05; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. All statistical analyses were conducted using GraphPad Prism version 10.2.0 software. Results and discussion Synthesis and characterization of CV@MT-NPs CV@MT-NPs was constructed by electrostatic assembly of positively charged MT-NPs onto the surface of negatively charged CV, as schematically illustrated in Fig. 2A. Transmission electron microscopy (TEM) analysis revealed that the MT-liposomes adopted a unilamellar spherical morphology, characterized by a single lipid bilayer encapsulating the melatonin (Fig. S2). In contrast, the MT-NPs exhibited a distinct multilamellar structure (Fig. 2B), indicating that the introduction of quaternary chitosan led to a significant reorganization of the liposomal architecture. This reassembly process resulted in multiple concentric layers, enhancing the stability and structural complexity of the nanoparticles. The prominent Tyndall effect indicated excellent colloidal stability and dispersibility of MT-NPs (Fig. 2C). Dynamic light scattering (DLS) analysis demonstrated that the MT-NPs solution exhibited no significant change in particle size for up to 30 days and remained stable in ddH2O for one week at 4 °C. However, after one week, melatonin was gradually released from MT-NPs (Fig. 2D and Fig. S3). Synthesis and characterization of CV@MT-NPs. (A) Schematic illustration of the preparation process of CV@MT-NPs. (B) Cryo-TEM images of MT-NPs. (C) Tyndall effect of MT-NPs. (D) Size distribution of MT-NPs from Day 0 to Day 30 (4 °C, in ddH2O). (E) Photograph of CV culture and optical microscope image of CV cells. (F) Photograph of CV and CV@MT-NPs solutions. (G) Drug loading efficiency of MT-NPs. (H) Zeta potential of CV, MT-NPs, and CV@MT-NPs. (I) FTIR spectra and (J) UV-Vis-NIR absorption spectra of CV, NPs, MT, MT-NPs, and CV@MT-NPs. (K) SEM images of CV and CV@MT-NPs. (L) CLSM images of CV@MT-NPs. (M) In vitro release profiles of melatonin from free MT, MT-NPs, and CV@MT-NPs (37 °C, in PBS, n = 3). (N) Serum melatonin concentration over time after intraperitoneal injection of free MT, MT-NPs, and CV@MT-NPs (n = 3). (O) Ex vivo fluorescence images of brains (MT: Cy7 channel) at various time points (0.5, 1, 3, 8, 12, and 24 h) following intraperitoneal injection of CV, MT, MT-NPs, or CV@MT-NPs. (P) Time-dependent relative fluorescence intensity of Cy7 in the brain (n = 3). (Q) Relative melatonin concentration in brain tissue measured by ELISA after intraperitoneal injection of MT, MT-NPs, or CV@MT-NPs (n = 5). Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗∗p < 0.01. CV, a natural green algae, exhibits a diameter ranging from approximately 3 to 8 μm (Fig. 2E) [35]. When MT-NPs were conjugated to the surface of CV, notable alterations in the surface activity of the resulting aqueous suspension were observed (Fig. 2F). Through mixing CV with different concentrations of MT-NPs, it was determined that the drug-loading efficiency of CV reached its peak when the MT concentration within the MT-NPs was set at 200 μg/mL (Fig. 2G). Consequently, this concentration was identified as the maximum drug-loading concentration. Zeta potential measurements validated the scientific rationale behind the synthesis steps of the CV@MT-NPs system. CV inherently possesses a negative surface charge. Although MT-liposomes also exhibited a negative surface charge (Fig. S4), the incorporation of quaternary chitosan during the reassembly process led to the formation of MT-NPs with a positive surface charge. This charge reversal facilitated the electrostatic assembly between the positively charged MT-NPs and the negatively charged CV surface. The resulting CV@MT-NPs complex demonstrates an overall positive zeta potential, confirming the successful conjugation of MT-NPs onto the CV surface (Fig. 2H). This electrostatic interaction not only ensured stable attachment but also enhanced the colloidal stability and dispersibility of the integrated system. Fourier-transform infrared (FTIR) spectroscopy (Fig. 2I) and Ultraviolet-visible-near-infrared (UV-vis-NIR) spectroscopy (Fig. 2J) analyses both demonstrated that the CV@MT-NPs possessed characteristic absorption peaks corresponding to both CV and MT. These results indicated the successful integration of MT-NPs onto the CV surface, preserving the inherent spectral signatures of both components. Scanning electron microscopy (SEM) analysis demonstrated that natural CV possessed a smooth surface characterized by intrinsic wrinkles. However, in the CV@MT-NPs sample, the CV surface exhibited substantial coverage with numerous nanoparticles (Fig. 2K and Fig. S5). This observation aligned with the results obtained from confocal laser scanning microscopy (CLSM) images, which confirmed that Dio-labeled MT-NPs were uniformly distributed across the CV surface (Fig. 2L). Release profile analyses demonstrated that the CV@MT-NPs integrated system promoted a sustained release of MT. Unlike MT-NPs, the CV@MT-NPs complex exhibited a slower release rate because part of the nanoparticles remained associated with the CV surface (Fig. S6A). Such retention creates an additional diffusion barrier and decreases the fraction of freely diffusing nanoparticles, thereby reducing the effective concentration gradient across the dialysis membrane. Consequently, the overall diffusion process is slowed, leading to a more sustained release profile (Fig. 2M). This gradual release mechanism effectively enhanced the pharmacokinetic properties of MT, ensuring prolonged bioavailability and therapeutic efficacy. Besides, the hydrodynamic diameter of MT-NPs released from CV@MT-NPs fluctuated only slightly over 3 days in both PBS and 10% FBS (Fig. S6B), demonstrating that MT-NPs had feasible stability in physiological conditions. CV@MT-NPs enable sustained release and prolonged retention of MT in the serum and brain To evaluate the pharmacokinetic profile and biodistribution of MT, we intraperitoneally administered free MT, MT-NPs, and CV@MT-NPs and monitored serum MT concentrations over time. As shown in Fig. 2N, free MT exhibited rapid clearance from circulation, with a short half-life (T1/2) of 0.67 ± 0.26 h, consistent with its known metabolic instability. In contrast, both MT-NPs and CV@MT-NPs significantly extended systemic exposure to MT, with T1/2 values of 1.45 ± 0.41 h and 2.91 ± 0.40 h, respectively. The area under the curve (AUC0-∞) further confirmed enhanced bioavailability: CV@MT-NPs achieved an AUC of 2979 ± 301.25 h × pg/mL, significantly higher than that of free MT (2354 ± 181.98 h × pg/mL) and MT-NPs (2478 ± 315.90 h × pg/mL) (Fig. 2N and Fig. S7). These results indicate that encapsulation of MT into nanoparticles, particularly when combined with CV as a carrier, effectively delays clearance and prolongs systemic circulation. Although melatonin possesses a relatively long half-life, effective delivery across the BBB remains critical for treating SAE, where neuroprotective agents must reach the central nervous system to exert therapeutic effects [[36], [37], [38]]. To assess brain targeting and retention, we used Cy7, a near-infrared fluorescent dye, as a surrogate tracer to mimic MT and track its in vivo distribution. Ex vivo fluorescence imaging revealed that all formulations accumulated in the brain within 3 h post-injection. Notably, while free MT and MT-NPs exhibited transient brain signals peaking at 3 h, CV@MT-NPs maintained significantly higher fluorescence intensity in the brain up to 24 h (Fig. 2O). Quantitative analysis showed that CV@MT-NPs enhanced brain accumulation by approximately 1.5-fold compared to MT-NPs at both 8 h and 12 h (Fig. 2P), indicating improved retention attributable to sustained release kinetics. To corroborate these imaging findings, we quantified endogenous melatonin levels in brain tissue using ELISA (Fig. 2Q). At 3 h post-administration, no significant differences were observed among the three groups. However, by 8 h, the melatonin concentration in the CV@MT-NPs group was significantly higher than in both the free MT and MT-NPs groups, with a statistically significant difference versus free MT. This temporal shift confirms that CV@MT-NPs not only facilitate initial brain delivery but also sustain drug availability, thereby extending the therapeutic window for SAE intervention. Beyond CNS distribution, we systematically evaluated the systemic biodistribution of both MT and the CV carrier. Fluorescence imaging of major organs, including heart, liver, spleen, lung, kidney, and intestine, revealed that all formulations primarily accumulated in the liver and spleen, with moderate uptake in the intestine and minimal signal in the heart and lungs (Fig. S8). Quantitative analysis indicated that MT-NPs reduced early renal clearance compared to free MT (Fig. S8G), suggesting nanoparticle-mediated protection against rapid excretion. More importantly, CV@MT-NPs exhibited prolonged retention in the liver and spleen (Fig. S8D and S8E), likely due to slowed metabolism and extended systemic circulation, which collectively contribute to sustained drug exposure. We further examined the biodistribution of the CV carrier itself using the Cy5 fluorescence channel. As shown in Fig. S9A–G, free CV and CV@MT-NPs displayed comparable temporal and spatial patterns of organ accumulation, predominantly localizing to the liver, spleen, and intestine. This distribution pattern suggests that, following intraperitoneal administration, CV may influence gut microbiota composition indirectly through modulation of the liver–gut metabolic axis rather than through direct luminal exposure. Critically, neither formulation crossed the BBB efficiently, as evidenced by negligible Cy5 signal in the brain (Fig. S10A–B). This finding indicates that CV does not directly mediate brain delivery; rather, it functions as a stable, biocompatible scaffold that modulates the release kinetics of MT-NPs, enabling sustained systemic and cerebral melatonin delivery without compromising BBB integrity. CV@MT-NPs alleviate cognitive and anxiety-like behaviors and improve overall outcomes in the SAE mouse model SAE is characterized by diffuse neuroinflammation leading to acute and chronic impairments in cognition, mood regulation, and behavioral function [39]. To establish a clinically relevant SAE model, two distinct LPS dosing regimens were employed: a low dose (2 mg/kg) to induce sustained neuroinflammatory responses and behavioral deficits mimicking mild-to-moderate SAE [[40], [41], [42]], and a high dose (15 mg/kg) to trigger severe systemic inflammation, BBB disruption, oxidative stress, and increased mortality [[43], [44], [45]] (Fig. S1). Following the LPS challenge, the successful induction of SAE was systematically validated through multiple biomarkers of neuroinflammation and BBB integrity (Fig. S11). As shown in Fig. S11A–C, brain tissue homogenates from LPS-treated mice exhibited significantly elevated levels of pro-inflammatory cytokines (IL-6, IL-1β, and TNF-α) compared to controls, confirming robust activation of neuroinflammatory pathways. This was further corroborated by increased Evans blue extravasation into the brain parenchyma (Fig. S11D), indicating compromised BBB integrity. Immunofluorescence staining for ionized calcium-binding adapter molecule 1 (IBA1), a marker of microglial activation, revealed pronounced morphological changes in LPS-treated mice, including hypertrophic cell bodies and ramified processes, indicative of microglial polarization toward a pro-inflammatory state (Fig. S11E). Together, these findings confirm that the LPS model effectively recapitulates key pathological features of SAE, including neuroinflammation, BBB dysfunction, and glial activation. To evaluate the therapeutic efficacy of CV@MT-NPs, we conducted a comprehensive battery of behavioral, physiological, and survival assessments over a 7-day prophylactic treatment period before LPS injection (Fig. 3A). The Y-maze test was conducted to assess spatial working memory and spontaneous alternation behavior, which are indicators of hippocampal-dependent cognitive function. The results showed that the percentage of correct alternation was significantly decreased in the LPS group compared to the control group. This decline suggests impaired cognitive function and spatial memory due to LPS-induced inflammation. Notably, treatment with CV@MT-NPs effectively restored spontaneous alternation performance, bringing it close to control levels (Fig. 3B and C). The open field test was used to evaluate anxiety-like behaviors and general locomotor activity. Mice in the LPS group exhibited severe behavioral abnormalities, including a significant increase in immobile time, a marked decrease in total distance traveled, and reduced time spent in the center zone (Fig. 3D–H). These changes are indicative of heightened anxiety, depressive-like states, and diminished exploratory drive. While CV@MT-NPs treatment did not fully normalize all behavioral parameters, it significantly alleviated these deficits. Specifically, mice treated with CV@MT-NPs displayed reduced immobility and increased locomotor activity (Fig. 3F and G), suggesting improved motor function and reduced behavioral despair. Although the time spent in the center did not differ significantly between the CV@MT-NPs (7.55 ± 3.64 s) and LPS (0.25 ± 0.61 s) groups, a clear trend toward improvement was observed (Fig. 3H), implying partial mitigation of anxiety-like behavior. The novel object recognition test was performed to assess recognition memory, a cognitive domain dependent on cortical and hippocampal networks. The results revealed that in the LPS-induced sepsis mouse model, the percentage of recognition index was markedly reduced. However, treatment with CV@MT-NPs effectively ameliorated this cognitive impairment, resulting in a significantly higher recognition index compared to the LPS group (Fig. 3I and J). CV@MT-NPs alleviate cognitive and anxiety-like behaviors and improve overall outcomes. (A) Schematic illustration of the experiments. (B) Representative trajectories from the Y-maze test. (C) Percentage of correct alternations in the Y-maze test (n = 6). (D) Schematic of the open field test. (E) Representative trajectories and heatmaps from the open field test. Quantification of (F) immobile time, (G) distance traveled, and (H) time spent in the center zone during the open field test (n = 6). (I) Schematic of the novel object recognition test. (J) Recognition index in the novel object recognition test (n = 7). (K) Murine sepsis severity scores (n = 6). (L) Percent change in body weight over time (n = 6). (M) Survival rate over time (n = 8). Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Beyond behavioral outcomes, CV@MT-NPs demonstrated broad clinical benefits. The murine sepsis severity score (MSS), a composite measure of posture, activity, respiratory rate, and response to stimuli [46], was significantly reduced from 7.7 ± 0.61 in the LPS group to 5.8 ± 0.82 in the CV@MT-NPs group (Fig. 3K), indicating improvement in overall disease severity. Body weight loss is a hallmark of systemic illness during sepsis. Over 48 h post-LPS challenge, LPS-treated mice lost approximately 20% of their initial body weight, reflecting profound metabolic distress. In contrast, CV@MT-NPs treatment provided the most effective protection against weight loss (Fig. 3L), outperforming free MT, MT-NPs, and CV alone. Most importantly, Kaplan-Meier survival analysis revealed a dramatic improvement in survival rate with CV@MT-NPs treatment. While only 25% of LPS-treated mice survived beyond 48 h, 75% of mice receiving CV@MT-NPs via intraperitoneal injection remained alive at the same time point (Fig. 3M). To evaluate route-dependent efficacy, oral administration of CV@MT-NPs was also tested. However, as shown in Fig. S12, oral gavage failed to confer a significant survival benefit, underscoring the importance of parenteral delivery for better therapeutic outcomes. This negative but informative result suggests that, in the context of acute sepsis, oral delivery may be insufficient to rapidly achieve therapeutic concentrations or overcome systemic inflammatory barriers. Consequently, subsequent experiments employed intraperitoneal administration, which ensured reliable systemic exposure. Collectively, these results demonstrate that CV@MT-NPs effectively alleviate cognitive dysfunction, reduce anxiety-like behaviors, preserve body weight, and significantly enhance survival in a murine SAE model. CV@MT-NPs enhance blood-brain barrier integrity and mitigate neurovascular disruption The BBB, a highly selective interface composed of brain microvascular endothelial cells, pericytes, astrocyte end feet, and the basement membrane, plays a critical role in maintaining central nervous system (CNS) homeostasis by restricting the passage of harmful substances into the brain [47]. During sepsis, systemic inflammation triggers the downregulation of tight junction proteins such as claudin-5, occludin, and ZO-1, leading to increased paracellular permeability and BBB disruption [48]. This breakdown allows inflammatory mediators, endotoxins, and immune cells to infiltrate the CNS, initiating neuroinflammation and neuronal damage. Furthermore, cytokine release within the brain exacerbates endothelial dysfunction, creating a self-perpetuating cycle of vascular injury and neurological deterioration. Therefore, preserving BBB integrity is essential for preventing the progression of SAE. To assess the protective effects of CV@MT-NPs on BBB structure and function, we conducted a comprehensive analysis using TEM, Evans blue extravasation assay, and immunofluorescence staining. First, TEM imaging revealed significant ultrastructural differences across treatment groups (Fig. 4A). In the control group, cerebral microvessels exhibited intact endothelial cells (E), continuous tight junctions (TJ; arrowheads), well-preserved basement membrane (B), and close apposition between astrocytic end feet and the vessel wall, hallmarks of a functional BBB. In contrast, LPS-treated mice displayed severe structural abnormalities: discontinuous tight junctions, swollen endothelial cells, intravascular debris, and detachment of astrocyte end feet from the basement membrane. These changes were partially ameliorated in the CV, MT, and MT-NPs groups, but only the CV@MT-NPs group restored near-normal ultrastructure, with clearly defined tight junctions and preserved cellular architecture, indicating superior protection against neurovascular damage. Next, we quantified BBB permeability using the Evans blue dye extravasation assay. As expected, the LPS group (22.8 ± 6.46 μg/g) showed extensive blue staining throughout the brain parenchyma (Fig. 4B), reflecting massive leakage due to BBB compromise. In contrast, the CV@MT-NPs group (5.86 ± 1.83 μg/g) exhibited significantly reduced Evans blue accumulation, with levels comparable to those in the control group (0.232 ± 0.096 μg/g) (Fig. 4C). While free MT and MT-NPs also reduced dye leakage, their effects were less pronounced than CV@MT-NPs, suggesting enhanced barrier stabilization through synergistic action. CV@MT-NPs enhance BBB integrity and restore microglial homeostasis. (A) TEM images of the BBB ultrastructure in different groups. B, basement membrane; E, endothelial cell; L, lumen; TJ, tight junction. Arrowheads indicate tight junctions. (B) Photographs of Evans blue extravasation in brain tissues. (C) Quantification of Evans blue content in brain homogenates (n = 5). (D) Immunofluorescence staining of AQP4 (green) and CD31 (red) in brain sections. (E) Co-localization analysis of AQP4 and CD31. (F) Immunofluorescence staining of ZO1 (green) and CD31 (red) at BBB sites. (G) Immunofluorescence staining of IBA1 (red) and DAPI (blue) in the cortex, CA1 hippocampus, and dentate gyrus. Quantification of IBA1 fluorescence intensity in the (H) cortex, (I) CA1 hippocampus, and (J) dentate gyrus (n = 6). (K) CLSM images and 3D reconstructions of microglial morphology in brain tissue. Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. To further investigate the interaction between astrocytes and endothelial cells, a key determinant of BBB integrity, we performed immunofluorescence staining for aquaporin-4 (AQP4), a water channel protein localized in astrocytic end feet, and CD31 (PECAM-1), a marker of endothelial cells (Fig. 4D). In the control group, AQP4 (green) and CD31 (red) signals were tightly co-localized along vessel walls, forming a continuous ring-like pattern indicative of normal neurovascular coupling. However, in the LPS group, this co-localization was disrupted, with fragmented and disorganized AQP4 staining, suggesting impaired astrocyte-endothelial interactions. Notably, CV@MT-NPs treatment effectively restored the alignment and co-localization of AQP4 and CD31 (Fig. 4D), as confirmed by line-scan intensity profiles showing sharp peaks of overlapping fluorescence (Fig. 4E). This indicates that CV@MT-NPs promote re-establishment of the neurovascular unit and enhance structural support for the BBB. Finally, we evaluated the expression and localization of zonula occludens-1 (ZO-1), a critical scaffolding protein required for tight junction assembly. Immunofluorescence staining revealed a diffuse and discontinuous ZO-1 signal in the LPS group, consistent with junctional disintegration (Fig. 4F). In contrast, the CV@MT-NPs group exhibited strong, continuous ZO-1 staining along vessel walls, resembling the control group, demonstrating effective preservation of tight junction integrity. CV@MT-NPs prevent neuroinflammation and protect against brain damage by restoring microglial and astrocytic homeostasis Microglia, the primary resident immune cells within the CNS, play pivotal roles in maintaining brain homeostasis, clearing pathogens and dead cells, and repairing damaged tissues [49,50]. Upon disruption of the BBB during systemic inflammation such as sepsis, microglia rapidly become activated and release pro-inflammatory cytokines including TNF-α, IL-6, and IL-1β [51]. Ionized calcium-binding adapter molecule 1 (IBA1) is a widely used marker for identifying both activated microglia and infiltrating peripheral macrophages [52,53], whereas SALL1 is specifically associated with quiescent, homeostatic microglia [54,55]. To assess microglial activation status in the LPS-induced SAE model, we performed IBA1 immunofluorescence staining across key brain regions: the cortex, CA1 hippocampus, and dentate gyrus (DG) (Fig. 4G). Quantitative analysis revealed a significant increase in IBA1 fluorescence intensity and cell number in all three regions following LPS challenge compared to the control group (Fig. 4H–J and Fig. S13A–C). Notably, treatment with CV@MT-NPs markedly reduced both IBA1 expression and the number of IBA1+ cells, indicating suppression of microglial hyperactivation. In contrast, CV, MT, and MT-NPs alone showed less significant effects (Fig. 4H–J and Fig. S13A–C). Conversely, SALL1 expression was significantly downregulated in the LPS group, reflecting a loss of microglial quiescence (Fig. S14A–D). Treatment with CV@MT-NPs effectively restored SALL1 levels in the cortex, CA1, and DG, suggesting a re-establishment of microglial homeostasis. The lack of significant SALL1 recovery in the CV, MT, or MT-NPs groups further underscores the superior efficacy of the combined formulation (Fig. S14). To distinguish between resident microglia and infiltrating peripheral macrophages, we performed flow cytometric analysis of CD45+ immune cells in brain tissue (Fig. S15). No significant difference in CD45high populations was observed between the LPS and control groups, indicating minimal infiltration of peripheral macrophages and confirming that the observed changes in IBA1 expression primarily reflect alterations in microglial activation rather than peripheral immune cell entry [56,57]. Further morphological characterization using CLSM and 3D reconstructions revealed that microglia in the LPS group exhibited enlarged soma and simplified processes, hallmarks of activated and pro-inflammatory states [58] (Fig. 4K). In contrast, CV@MT-NPs-treated animals displayed microglia with smaller cell bodies and more ramified, branched morphology, consistent with a return to a surveillant, resting phenotype. Astrocytes, another critical type of glial cell involved in brain inflammation, were evaluated using glial fibrillary acidic protein (GFAP) as a primary marker [59,60]. Immunofluorescence staining demonstrated robust GFAP upregulation in the CA1 and DG regions after LPS administration (Fig. 5A), indicative of astrocyte reactivity. CV@MT-NPs significantly attenuated this astrogliosis, as evidenced by reduced GFAP fluorescence intensity in both regions (Fig. 5B and C). Importantly, NeuN expression, a marker of neuronal integrity, remained stable across all treatment groups (Fig. 5A and Fig. S16), suggesting that the neurons have not been severely damaged at this time point. CV@MT-NPs attenuate neuroinflammation and mitigate brain damage. (A) Immunofluorescence staining of GFAP (red) and NeuN (green) in the CA1 hippocampus and dentate gyrus. Quantification of GFAP fluorescence intensity in (B) the CA1 hippocampus and (C) the dentate gyrus (n = 6). (D) IL-6 and (E) TNF-α concentration in brain tissue measured by ELISA (n = 5). Relative mRNA expression levels of (F) IL-6, (G) IL-1β, (H) COX2, (I) TNF-α, and (J) iNOS in brain tissue (n = 4). (K) MDA concentration in different treatment groups (n = 5). (L) TEM images of brain tissue sections. Arrows and asterisks indicate mitochondrial swelling and reduced electron density, respectively. Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. To evaluate the broader anti-inflammatory and antioxidant effects of CV@MT-NPs, we measured key inflammatory mediators and oxidative stress markers. ELISA results showed that CV@MT-NPs significantly reduced brain levels of IL-6 and TNF-α (Fig. 5D and E). qRT-PCR confirmed downregulation of pro-inflammatory genes, including IL-6, IL-1β, TNF-α, COX2, and iNOS (Fig. 5F–J). Additionally, malondialdehyde (MDA), a marker of lipid peroxidation, was markedly decreased in the CV@MT-NPs group (Fig. 5K), indicating effective mitigation of oxidative stress. Mitochondrial dysfunction is a critical contributor to SAE pathology, driven by excessive ROS, NO, and impaired energy metabolism [61]. TEM revealed severe mitochondrial damage in the LPS group, including swelling, cristae disorganization, and reduced electron density (Fig. 5L). In contrast, CV@MT-NPs preserved mitochondrial ultrastructure, with significantly fewer damaged mitochondria observed (Fig. 5L and Fig. S17). While MT and MT-NPs also showed some protective effects, CV@MT-NPs provided superior preservation of mitochondrial integrity, likely due to enhanced brain delivery and sustained melatonin release. This protection supports the maintenance of ATP production and calcium buffering capacity, which are vital for neuronal survival. In summary, CV@MT-NPs exert potent neuroprotective effects in SAE by comprehensively targeting multiple pathological pathways, including suppressing microglial and astrocytic activation, restoring microglial homeostasis (as indicated by SALL1 upregulation), reducing pro-inflammatory cytokine production, alleviating oxidative stress, and preserving mitochondrial structure and function. These multi-targeted actions highlight the therapeutic potential of CV@MT-NPs in mitigating neuroinflammation and preventing brain injury during systemic inflammation. CV@MT-NPs protect against intestinal barrier dysfunction and alleviate systemic inflammation in sepsis Sepsis is characterized by a systemic inflammatory response triggered by infection, leading to multi-organ dysfunction and immune dysregulation [1]. Clinical studies have shown that sepsis disrupts circadian melatonin rhythms [62], with markedly reduced melatonin levels strongly correlated with increased mortality [63]. To evaluate the systemic impact of sepsis and the therapeutic potential of CV@MT-NPs, we assessed circulating levels of LPS and melatonin following intraperitoneal administration. As expected, LPS injection induced a significant elevation in serum LPS concentration, confirming successful induction of endotoxemia (Fig. 6A). Notably, serum melatonin levels were drastically reduced in the LPS group, reflecting the disruption of circadian and neuroimmune regulation during sepsis. Treatment with CV@MT-NPs effectively restored serum melatonin to near-normal levels (Fig. 6B), suggesting that this formulation not only delivers melatonin systemically but also enhances its bioavailability and stability. To investigate the broader protective effects of CV@MT-NPs beyond the central nervous system, we examined pathological changes in key peripheral organs involved in immune homeostasis and barrier function, specifically the liver, spleen, and intestine. Histological analysis revealed a severe hepatic inflammatory response in the LPS group (Fig. S18), which was significantly attenuated in mice treated with CV@MT-NPs. In the spleen, TUNEL staining demonstrated markedly increased apoptosis in the LPS group, from 6.122 ± 1.903% to 15.358 ± 2.847%. However, the CV@MT-NPs group showed a significant reduction in apoptotic cells (10.852 ± 1.323%) (Fig. S19A–B). CV@MT-NPs protect against intestinal barrier dysfunction and alleviate systemic inflammation. (A) Serum LPS concentration (n = 8). (B) Serum melatonin concentration (n = 7). (C) H&E staining of small intestine sections. Immunofluorescence staining of tight junction proteins. (D) occludin and (E) claudin-1 in small intestine sections. (F) Flow cytometry analysis of F4/80 and CD11b expression in peritoneal cells. (G) Flow cytometry analysis of CD86 (M1 marker) and CD206 (M2 marker) expression in peritoneal macrophages. Serum concentrations of (H) IL-6 and (I) TNF-α measured by ELISA (n = 6). Immunofluorescence staining of (J) CD86 and (K) CD206 in Raw 264.7 cells treated with different materials. Quantification of mean fluorescence intensity for (L) CD86 and (M) CD206 (n = 3). (N) Relative mRNA expression levels of IL-1β, IL-6, and iNOS in Raw 264.7 cells (n = 3). (O) Western blot analysis of p-p65 and p65 protein expression; β-actin served as loading control. Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Of particular importance is the role of the intestinal barrier in modulating systemic inflammation during sepsis. The gut epithelium serves as a critical physical and immunological barrier preventing bacterial translocation into the bloodstream. However, LPS directly disrupts tight junctions between enterocytes, increasing paracellular permeability and promoting endotoxin leakage [64]. H&E staining revealed severe intestinal pathology in the LPS group, including shortened and disorganized villi, epithelial shedding, and widened intercellular spaces (Fig. 6C). In contrast, CV@MT-NPs treatment preserved intestinal architecture, resembling that of the control group. Immunofluorescence staining for tight junction proteins occludin and claudin-1 confirmed significant downregulation in the LPS group, whereas both CV and CV@MT-NPs significantly enhanced their expression (Fig. 6D–E and Fig. S20A–B), indicating restoration of barrier integrity. This effect aligns with previous evidence that microalgae can support gut health [65]. Sepsis-associated systemic inflammatory response syndrome involves the activation of various immune cells, including neutrophils, macrophages, and natural killer cells. Among these, macrophages are among the most abundant immune cells in many tissues and serve as key first responders to tissue damage. Upon LPS stimulation, proinflammatory (M1) macrophages become hyperactivated, exacerbating inflammation and driving sepsis progression [66,67]. We further explored the immunomodulatory effects of CV@MT-NPs on peritoneal macrophages. Flow cytometry analysis showed a marked decrease in F4/80+CD11b+ peritoneal macrophages in the LPS group (from 42.60 ± 10.17% to 12.776 ± 6.14%), indicative of immune cell depletion (Figs. 6F and Fig. S21). Both CV (33.08 ± 5.64%) and CV@MT-NPs (51.14 ± 6.17%) reversed this loss, with CV@MT-NPs achieving the most robust recovery, highlighting the synergistic benefit of co-delivery. Importantly, macrophage polarization was assessed via surface markers: CD86 (M1, pro-inflammatory) and CD206 (M2, anti-inflammatory/reparative). The LPS group exhibited a dominant M1 phenotype, while MT-NPs and CV@MT-NPs significantly shifted macrophages toward an M2 profile, with CV@MT-NPs inducing the highest M2/M1 ratio (Fig. 6G and Fig. S22), indicating that MT-NPs were more effective in promoting the polarization of macrophages. This polarization shift was corroborated by reduced serum levels of IL-6 and TNF-α in the MT-NPs and CV@MT-NPs groups (Fig. 6H and I), demonstrating effective suppression of systemic inflammation. To validate these findings in vitro, we used LPS-stimulated Raw 264.7 macrophages as a model of M1 polarization. Dose-response experiments revealed that MT-NPs at 80 μg/mL effectively suppressed NO production to baseline levels (Fig. S23), establishing this dose as optimal for subsequent assays. Transwell migration assays showed that both MT-NPs and CV@MT-NPs significantly inhibited macrophage migration, with CV@MT-NPs exhibiting the strongest effect (Fig. S24A–B). Immunofluorescence analysis confirmed that MT-NPs and CV@MT-NPs decreased CD86 expression and increased CD206 expression in activated macrophages (Fig. 6J–M). At the transcriptional level, qRT-PCR revealed that both MT-NPs and CV@MT-NPs downregulated pro-inflammatory genes (IL-1β, IL-6, iNOS, TNF-α) and upregulated Arg1, a marker of M2 polarization (Fig. 6N and Fig. S25A–B). Consistent with gene expression, NO production was most potently suppressed in the CV@MT-NPs group (Fig. S26), reinforcing its superior anti-inflammatory activity. The NF-κB signaling pathway is a key regulator of inflammatory responses, particularly in M1 macrophage activation [68,69]. Western blot analysis showed that both MT-NPs and CV@MT-NPs significantly reduced phosphorylation of the p65 subunit (p-p65), with CV@MT-NPs exerting the greatest inhibition (Fig. 6O and Fig. S27), suggesting suppression of NF-κB-mediated transcription. Notably, all tested formulations (MT, MT-NPs, and CV@MT-NPs) reduced intracellular ROS levels in LPS-stimulated macrophages (Fig. S28A–B), indicating antioxidant activity. While CV alone did not promote macrophage polarization, it significantly lowered ROS, confirming its intrinsic antioxidant properties (Fig. S28A–B). This dual action, anti-inflammatory and antioxidant, may contribute to the overall cytoprotective effects of CV@MT-NPs. Collectively, our data demonstrate that CV@MT-NPs protect against intestinal barrier dysfunction and mitigate systemic inflammation in sepsis through multiple mechanisms, including restoring melatonin levels, preserving intestinal tight junctions, preventing organ-specific injury (liver, spleen), reversing peritoneal macrophage depletion, and promoting M2 polarization via NF-κB inhibition and ROS scavenging. Mechanistically, melatonin activates the NRF2 signaling pathway, upregulates antioxidant enzyme expression, and exerts potent anti-inflammatory and neuroprotective effects [70,71]. It has also been shown to maintain BBB integrity and modulate microglial polarization toward an anti-inflammatory phenotype in LPS-induced sepsis models [72]. However, the clinical translation of free melatonin is hindered by its poor aqueous solubility, short half-life, and low bioavailability. The superior efficacy of CV@MT-NPs, compared to free melatonin or either component alone, likely arises from enhanced cellular uptake and sustained release afforded by liposomal encapsulation, thereby maximizing melatonin's therapeutic potential. CV@MT-NPs restore gut microbial homeostasis and modulate the gut-brain axis via SCFA-mediated signaling Growing evidence highlights the critical role of the gut-brain axis in the pathogenesis of SAE. During sepsis, disruption of intestinal barrier integrity and dysbiosis of the gut microbiota facilitate the translocation of microbial products into systemic circulation, thereby amplifying systemic inflammation and driving neuroinflammation and neuronal injury [5,73]. As previously demonstrated (Fig. 6), CV@MT-NPs effectively preserve intestinal barrier function. Building on this finding, we sought to investigate whether CV@MT-NPs could further modulate gut microbial composition and metabolic output, thereby contributing to the restoration of gut-brain communication. To assess microbial changes, we performed 16S rRNA gene sequencing of fecal samples from septic mice across treatment groups. At the operational taxonomic unit (OTU) level, α-diversity analysis revealed no significant differences in bacterial richness (observed species) or evenness (Chao1 index) among groups (Fig. 7A and B), suggesting that neither LPS-induced sepsis nor therapeutic interventions drastically altered overall microbial diversity. Rarefaction curves approached saturation across all samples, indicating sufficient sequencing depth and high-quality data for robust downstream analyses (Fig. 7C). In contrast, β-diversity analysis revealed distinct clustering patterns, with principal component analysis (PCA) showing clear separation between the LPS group and control group, as well as a notable shift toward the control cluster in the CV@MT-NPs group (Fig. 7D). This indicates that CV@MT-NPs partially restored the structure of the gut microbial community to a healthier state. Community barplot analysis at the genus level showed that the LPS group exhibited a marked expansion of potentially pathogenic bacteria such as Escherichia-Shigella [74] and Akkermansia [75], which are associated with inflammatory conditions and gut barrier dysfunction. Notably, CV, MT, MT-NPs, and CV@MT-NPs significantly reduced the abundance of these genera, particularly CV@MT-NPs, suggesting an anti-pathogenic effect. In contrast, beneficial bacteria such as Faecalibaculum (a butyrate producer) and norank_f_Muribaculaceae were enriched in the CV and CV@MT-NPs groups respectively, highlighting their potential role in restoring microbial homeostasis [76] (Fig. 7E and F, and Fig. S29). To visualize microbial intergroup relationships, Circos plots illustrated the relative abundance and distribution of dominant genera across treatments (Fig. 7G). Furthermore, Linear Discriminant Analysis Effect Size (LEfSe) identified specific bacterial taxa that were differentially enriched across groups, highlighting unique microbial signatures associated with each condition (Fig. 7H). A Sankey diagram illustrated the dynamic shifts in microbial composition from baseline (control) to disease (LPS) and subsequent recovery following CV@MT-NPs intervention, demonstrating a treatment-induced reconfiguration of the gut ecosystem toward a healthier profile (Fig. 7I). To further determine whether the neuroprotective effects of CV@MT-NPs were mediated by the gut microbiota, a pseudo–germ-free mouse model was established. Notably, depletion of the gut microbiota markedly attenuated the protective effects of CV@MT-NPs against neuroinflammation, suggesting that an intact microbial community is required for its full therapeutic efficacy (Fig. S30). CV@MT-NPs restore gut microbial homeostasis and modulate the gut-brain axis via SCFA-mediated signaling. Alpha diversity analysis at the OTU level, assessed by (A) observed species and (B) Chao1 index (n = 6). (C) Rarefaction curves based on the Sobs index. (D) PCA of microbiota composition at the genus level. (E) Relative abundance of differentially abundant bacterial genera across treatment groups. (F) The proportion of representative genera in each treatment group (n = 6). (G)Circos plot illustrating the relative abundance and intergroup distribution of gut microbiota. (H) LDA scores from LEfSe analysis. (I) Sankey diagram showing the flow and relative abundance of differentially abundant microbes at phylum and genus levels. (J) Concentration of fecal SCFAs (n = 6). (K) Spearman correlation analysis of differential microbiota with serum LPS and fecal SCFAs concentration. (L) Concentration of brain SCFAs in different treatment groups, including acetic acid, propionic acid, butanoic acid, and isovaleric acid (n = 6). Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Given that gut microbiota-derived metabolites serve as crucial mediators of host physiology, we focused on SCFAs, key products of bacterial fermentation that maintain intestinal barrier integrity, regulate immune responses, and signal along the gut-brain axis [[77], [78], [79]]. Using gas chromatography-mass spectrometry (GC-MS), we quantified fecal and brain SCFA levels. CV@MT-NPs treatment significantly elevated fecal concentrations of acetic acid, propionic acid, and butanoic acid compared to the LPS group, while markedly reducing isovaleric acid—a branched-chain fatty acid often linked to dysbiosis and metabolic stress (Fig. 7J). Spearman correlation analysis revealed strong associations between serum LPS levels, fecal SCFA profiles, and microbial abundances (Fig. 7K). Notably, Escherichia-Shigella showed a significant positive correlation with LPS, supporting its potential role in promoting endotoxemia and SAE progression [80,81]. Importantly, it also exhibited a negative correlation with key beneficial SCFAs such as acetic acid, propionic acid, and butanoic acid, and showed a positive correlation with isovaleric acid. This dual pattern suggests that Escherichia-Shigella not only contributes to barrier dysfunction through serum LPS translocation but may also disrupt gut metabolic homeostasis by regulating major SCFA production. In contrast, norank_f__[Eubacterium]_coprostanoligenes_group and members of Prevotellaceae exhibited negative correlations with acetic, propionic, and butanoic acids, but showed positive correlations with serum LPS [82,83]. Critically, changes in SCFA levels were mirrored in the brain. CV@MT-NPs treatment increased cerebral concentrations of acetic acid, propionic acid, and butanoic acid, while attenuating the LPS-induced rise in isovaleric acid (Fig. 7L). Importantly, CV alone also elevated brain butanoic acid levels and reduced isovaleric acid accumulation. Butanoic acid, in particular, has been shown to exert potent anti-inflammatory and neuroprotective effects by suppressing microglial activation, enhancing blood-brain barrier integrity, and supporting mitochondrial function in neurons [84,85]. To directly assess whether the elevated SCFA levels contribute to the neurotherapeutic effects observed in SAE, we performed SCFA supplementation experiments. Prophylactic oral administration of SCFAs markedly alleviated neuroinflammation during SAE progression and significantly suppressed excessive activation of glial cells in the brain, further supporting a causal role of SCFAs in mediating neuroprotection (Fig. S31). These findings collectively demonstrate that CV@MT-NPs not only protect the intestinal barrier but also reshape the gut microbiota toward a more favorable composition and enhance the production of beneficial metabolites, particularly SCFAs. By restoring microbial homeostasis and amplifying SCFA-mediated signaling, CV@MT-NPs likely modulate the gut-brain axis to mitigate neuroinflammation and improve outcomes in SAE. The observed synergy between CV and MT may arise from their complementary actions: CV enhances microbial stability and antioxidant capacity, while MT provides direct neuroprotection and immune modulation, together creating a multi-targeted therapeutic strategy. In vitro verification of CV@MT-NPs exerting multi-level neuroprotection by preserving BBB integrity, modulating microglia, and enhancing SCFA signaling To elucidate the neuroprotective mechanisms of CV@MT-NPs at the cellular level, we established an in vitro BBB model using a Transwell co-culture system composed of mouse brain microvascular endothelial cells (bEnd.3) in the upper chamber and BV-2 microglial cells in the lower chamber (Fig. 8A). This configuration mimics the physiological interaction between cerebral endothelial cells and immune-resident microglia, enabling comprehensive assessment of both BBB integrity and neuroinflammatory responses. LPS-induced inflammation significantly disrupted BBB function, as evidenced by a marked decrease in transendothelial electrical resistance (TEER) and increased permeability to FITC-dextran (Fig. 8B and C). Notably, treatment with MT-NPs effectively attenuated these changes, preserving TEER values and reducing paracellular leakage, thereby demonstrating their capacity to reinforce endothelial barrier integrity. Consistent with this, qRT-PCR analysis revealed that MT-NPs suppressed LPS-triggered upregulation of pro-inflammatory genes (IL-1β, IL-6, TNF-α, iNOS, COX-2, CCL3, and MMP9) in bEnd.3 cells (Fig. 8D), indicating suppression of endothelial activation and maintenance of homeostasis. These findings align with our in vivo observations and underscore the critical role of MT-NPs in protecting the BBB against inflammatory insults. In vitro verification of CV@MT-NPs exerting multi-level neuroprotection by preserving BBB integrity, modulating microglia polarization, and enhancing SCFA-mediated signaling. (A) Schematic illustration of the Transwell co-culture system composed of bEnd.3 and BV-2 cells. (B) TEER values (n = 4). (C) BBB permeability (n = 3). (D) Heatmap showing relative mRNA expression levels of inflammatory markers. (E) Melatonin concentration in the lower chamber over time (n = 3). (F) Flow cytometry histograms of Cy5-labeled MT-NPs uptake by BV-2 cells at different time points. (G) CLSM image showing MT-NP distribution relative to lysosomes in BV-2 cells at 24 h. (H) IL-6, (I) TNF-α concentration, and (J) relative NO production in BV-2 cells (n = 3). (K) Flow cytometry analysis of ROS levels using DCFH-DA staining. (L) Immunofluorescence staining and 2.5D reconstruction of CD86 and CD206 in BV-2 cells. Effects of acetate, propionate and butyrate acid on (M) NO production, (N) IL-6, and (O) TNF-α in BV-2 cells (n = 3). (P) Schematic illustration of the Transwell co-culture system composed of BV-2 and HT22 cells. (Q) Flow cytometry analysis of Annexin V-FITC/PI staining for apoptosis detection in HT22 cells. (R) SIM images of MitoTracker-stained mitochondria in HT22 cells. (S) Immunofluorescence staining of Mito-JC-1 aggregates (red) and monomers (green) in HT22 cells. (T) Quantification of Mito-JC-1 aggregate-to-monomer ratio (n = 4). Data are presented as mean ± SD (n = biological replicates), nsp>0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Importantly, MT-NPs exhibited efficient transcytosis across the bEnd.3 monolayer in a time-dependent manner, as demonstrated by increasing melatonin concentrations in the basolateral chamber over 24 h (Fig. 8E). However, while fluorescence signal intensity of Cy5-labeled MT-NPs in BV-2 cells increased initially, it plateaued after 8 h, suggesting intracellular processing or degradation (Fig. 8F). Further validation using PKH26-labeled MT-NPs showed a time-dependent increase in lysosomal colocalization, with Pearson's correlation coefficient rising from ∼0.35 at 3 h to ∼0.65 at 24 h, indicating gradual trafficking to lysosomal compartments for degradation. (Fig. 8G and Fig. S32A–B). This dynamic further underscores the importance of the sustained release of MT-NPs by CV@MT-NPs. Once delivered to microglia, MT-NPs exerted potent immunomodulatory effects. ELISA assays showed significant reductions in IL-6 and TNF-α secretion (Fig. 8H and I), along with markedly suppressed NO production (Fig. 8J) and ROS levels (Fig. 8K), confirming their antioxidant and anti-inflammatory properties. Immunofluorescence staining combined with 2.5D reconstruction further revealed that MT-NPs promoted a phenotypic shift in BV-2 cells: downregulation of the M1 marker CD86 and upregulation of the M2 marker CD206 (Fig. 8L), indicative of polarization toward an anti-inflammatory, reparative state. Together, these results demonstrate that MT-NPs not only protect the BBB but also directly reprogram microglial activation to mitigate neuroinflammation. In parallel, our in vivo metabolomic analyses demonstrated that CV@MT-NPs elevate acetic acid, propionic acid, and butyric acid in both gut and brain (Fig. 7J and L), suggesting enhanced SCFA-mediated gut-brain signaling. To test the functional relevance of these metabolites, we treated LPS-stimulated BV-2 cells with individual SCFAs. Propionate and butyrate reduced NO production (Fig. 8M), reflecting their antioxidant potential. Moreover, all three SCFAs potently inhibited the release of IL-6 and TNF-α (Fig. 8N and O), confirming their role in modulating microglial inflammation. These findings support a dual mechanism: direct anti-inflammatory effects of MT-NPs and indirect modulation via SCFA signaling, both contributing synergistically to neuroinflammation control. Given that chronic neuroinflammation drives neuronal injury and cognitive dysfunction [50,86], we next evaluated the downstream neuroprotective effects of MT-NPs. We employed a Transwell system where conditioned medium from LPS-activated BV-2 cells was applied to HT22 neuronal cells to mimic inflammatory stress (Fig. 8P). Flow cytometry analysis of Annexin V-FITC/PI staining revealed that MT-NPs (1.24 ± 0.28%) dramatically reduced neuronal apoptosis compared to the LPS group (3.39 ± 0.52%) (Fig. 8Q and Fig. S33), nearly restoring survival rates to control levels (0.89 ± 0.21%). This indicates robust protection against inflammation-induced cell death. Further investigation revealed mitochondrial dysfunction in neurons exposed to inflammatory mediators. MitoTracker staining showed fragmented and disorganized mitochondrial networks in the LPS group, which were largely restored upon MT-NP treatment (Fig. 8R). JC-1 staining confirmed improved mitochondrial membrane potential: MT-NPs increased the ratio of red aggregates (indicative of polarized mitochondria) to green monomers (depolarized state), with a significant elevation in aggregate-to-monomer ratio observed in the MT-NPs group (Fig. 8S and T). These data suggest that MT-NPs preserve mitochondrial integrity through both direct and indirect mechanisms. Collectively, these in vitro results reveal a multi-pronged neuroprotective strategy mediated by CV@MT-NPs: preservation of BBB integrity by suppressing endothelial inflammation, targeted delivery of melatonin to microglia, where it modulates polarization and reduces oxidative/nitrosative stress, enhancement of SCFA-mediated signaling, which synergizes with direct anti-inflammatory actions, and protection of neurons from inflammatory damage by reducing apoptosis and maintaining mitochondrial function. Biocompatibility assessment of CV@MT-NPs To comprehensively evaluate the biosafety profile of CV@MT-NPs and their individual components, we conducted a multi-layered assessment encompassing in vitro cytotoxicity, acute systemic immune response, long-term organ histopathology, hematological parameters, and in vivo clearance dynamics. CCK-8 assays were performed to assess the cytotoxicity of CV, MT, MT-NPs, and CV@MT-NPs on two relevant cell types: Raw 264.7 macrophages and bEnd.3 brain endothelial cells. As shown in Fig. S34, all tested materials exhibited excellent biocompatibility across a broad concentration range (0-80 μg/mL). No significant reduction in cell viability was observed in either cell line, indicating that neither the CV nor the MT-NPs exerted cytotoxic effects under experimental conditions. These results confirm the safety of the formulation at therapeutically relevant doses. To evaluate potential inflammatory responses following administration, serum levels of key pro-inflammatory cytokines (TNF-α, IL-1β, and IL-6) were measured at multiple time points (3, 8, 24, and 48 h) after intraperitoneal injection of CV@MT-NPs (Fig. S35). Notably, no significant elevation in cytokine levels was detected compared to baseline controls at any time point, suggesting that CV@MT-NPs do not induce acute systemic inflammation or immune activation. To investigate potential long-term toxicity, a 14-day treatment regimen was administered, and major organs (heart, liver, spleen, lung, and kidney) were harvested two weeks post-treatment for histopathological evaluation via H&E staining (Fig. S36). Histological examination revealed no discernible pathological changes, such as necrosis, infiltration, or structural disruption, in any organ across all treatment groups compared to controls. Furthermore, comprehensive hematological analysis (Fig. S37) showed no statistically significant differences in key parameters between CV@MT-NPs-treated animals and control groups, including white blood cell count (WBC), red blood cell count (RBC), neutrophil percentage (NEU%), lymphocyte percentage (LYM%), monocyte percentage (MON%), hemoglobin (HGB), platelet count (PLT), and mean corpuscular volume (MCV). These findings collectively demonstrate the lack of chronic immune stimulation, hematological disturbance, or organ damage associated with CV@MT-NPs administration. A critical determinant of nanomaterial safety is its biodegradability and clearance pathway. Given that macrophages are the primary phagocytic cells responsible for eliminating foreign particles in vivo, we hypothesized that macrophages would play a central role in the clearance of CV@MT-NPs, particularly the CV. To track the fate of CV in vivo, we exploited the intrinsic autofluorescence of chlorophyll within algal chloroplasts. Flow cytometry analysis of peritoneal macrophages isolated at 6, 12, 24, and 48 h post-injection revealed a dynamic uptake pattern: approximately 13.6% of macrophages were positive for Cy5-labeled CV at 6 h, rising to a peak of 52.4% at 12 h, followed by a gradual decline to 2.46% at 48 h (Fig. S38B–C). This transient accumulation indicates rapid phagocytosis and subsequent intracellular processing. The progressive decrease in fluorescence signal over time likely reflects degradation of algal structures within lysosomes, leading to loss of chlorophyll fluorescence—a hallmark of biological breakdown. This hypothesis was further supported by analysis of peritoneal lavage fluid collected at various time points. UV-Vis-NIR absorption spectra showed a distinct peak at ∼420-450 nm corresponding to chlorophyll (Fig. S39A). Quantitatively, fluorescence intensity peaked at 6 h and declined thereafter (Fig. S39B), corroborating the time-dependent degradation of CV in vivo. Together, these data provide direct evidence that CV is efficiently internalized by macrophages and undergoes enzymatic degradation, confirming its biodegradability and safe metabolic fate. Collectively, the in vitro and in vivo assessments demonstrate that CV@MT-NPs exhibit excellent biocompatibility without inducing acute or chronic toxicity. Importantly, CV is efficiently cleared and degraded in vivo through macrophage-mediated phagocytosis, as evidenced by the time-dependent attenuation of its intrinsic fluorescence and corresponding changes in UV-Vis-NIR absorption spectra in peritoneal fluid. This natural clearance pathway highlights the biodegradable nature of the system and supports its favorable long-term safety profile; however, further investigations are required prior to clinical translation. In particular, repeated dosing studies in large-animal models will be necessary to comprehensively evaluate the therapeutic efficacy, biodegradability, and long-term safety of this system. Admittedly, our study has limitations. While the adopted prophylactic administration strategy may maximize therapeutic efficacy, its clinical translation poses practical challenges, particularly because preemptive treatment is not routinely implemented for all patients in clinical settings. Nevertheless, this strategy may hold meaningful translational potential for well-defined high-risk populations, such as patients undergoing major surgery, experiencing severe trauma, or suffering from underlying immune dysfunction, who are known to have a substantially increased risk of developing sepsis and sepsis-associated encephalopathy (SAE). Importantly, prophylactic or early-intervention strategies are not unprecedented in clinical management of sepsis. Current preventive approaches, including prophylactic antibiotic administration, vaccination, and immunomodulatory therapies, are already employed in selected high-risk populations to reduce the incidence of sepsis and to improve survival outcomes following infection. In this context, our findings provide a proof-of-concept that early modulation of disease-related pathways prior to overt sepsis onset may represent a complementary strategy for mitigating SAE, particularly in patients with predictable or unavoidable risk exposure. Conclusion In conclusion, we have successfully developed a bioinspired and multifunctional drug delivery system (CV@MT-NPs) by electrostatically assembling MT-NPs onto the surface of the natural CV. This integrated platform demonstrates superior colloidal stability, sustained melatonin release kinetics, and enhanced BBB penetration, leading to prolonged systemic circulation and significantly improved brain retention of melatonin. In a prophylactic regimen using an LPS-induced murine model of SAE, CV@MT-NPs demonstrated comprehensive efficacy. The MT-NPs mediated central protective effects by preserving BBB integrity through restoration of tight junction proteins and astrocyte-endothelial coupling, suppressing neuroinflammation via inhibition of microglial hyperactivation and attenuation of astrogliosis, and safeguarding mitochondrial structure and function to mitigate neuronal damage. Concurrently, MT-NPs ameliorated peripheral immune-metabolic dysfunction by inhibiting NF-κB signaling, promoting M2 macrophage polarization, and restoring intestinal barrier integrity. Complementarily, the CV component actively reshaped the gut microbiota by enriching SCFA-producing bacteria and elevating levels of acetic, propionic, and butyric acids in both the gut and brain, thereby reinforcing gut-brain axis communication and amplifying systemic anti-inflammatory and neuroprotective responses. Critically, the synergy between CV and MT-NPs not only enhanced the pharmacokinetics and biodistribution of melatonin but also enabled coordinated, multi-organ protection, spanning the brain, gut, and immune system, without inducing acute or chronic toxicity, as confirmed by rigorous biocompatibility assessments. Collectively, our findings establish CV@MT-NPs as a safe, biodegradable, and highly effective preventive strategy for SAE, wherein MT-NPs and CV act in concert to simultaneously target central neuropathology and peripheral dysregulation. This dual-action platform holds significant promise for treating other inflammatory disorders rooted in gut-brain axis disruption. CRediT authorship contribution statement Xuening Zhang: Writing – original draft, Investigation, Formal analysis, Data curation. Jia Dong: Validation, Investigation, Formal analysis, Data curation. Wentao Liu: Data curation. Shiyuan Hua: Formal analysis. Donghu Lin: Data curation. Jiahua Niu: Data curation. Min Zhou: Writing – review & editing, Supervision, Conceptualization. Ethics approval and consent to participate All animal experiments were conducted in strict accordance with protocols reviewed and approved by the Ethics Committee of Zhejiang University (approval number: ZJU20250024). All animal procedures were performed by the National Institute of Health Guide for the care and use of laboratory animals. Ethics approval and consent to participate All animal experiments were reviewed and approved by the Ethics Committee of Zhejiang University (approval number: ZJU20250848). All animal procedures were performed by the National Institute of Health Guide for the care and use of laboratory animals. Funding This work was supported by the National Key R&D Program of China (2022YFA1104900), the National Natural Science Foundation of China (82402905), the Natural Science and Foundation of Shandong Province (ZR2023ZD30), the Binjiang Institute of Zhejiang University (ZY202205SMKY007), the Medical Science and Technology Project of Zhejiang Province (2022RC024), and the Leading Innovative and Entrepreneur Team Introduction Program of Zhejiang (2022R01002). Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We acknowledge the Core Facility of Zhejiang University School of Medicine for the technical support. Footnotes Peer review under the responsibility of editorial board of Bioactive Materials. Supplementary data to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.02.022. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 References Associated Data Supplementary Materials Multimedia component 1