Metabolically engineered probiotic OMVs as nanovaccine mediating sequential immunomodulation for chronic bone infection immunotherapy Authors: Jie Lei, Bide Tong, Shihao Zhang, Zixuan Ou, Dingchao Zhu, Xingyu Zhou, Huaizhen Liang, Zhengdong Zhang, Lei Tan, Cao Yang, Xiaobo Feng Published: 2026 ABSTRACT SummaryChronic, drug-resistant bone infection caused by methicillin-resistant Staphylococcus aureus (MRSA) features an immunosuppressive niche enabling persistent infection and impaired bone healing. A specific treatment requires initial antibacterial immune activation against infection, followed by reshaping an anti-inflammatory microenvironment for later bone repair. Here, we screen Lactococcus lactis outer membrane vesicles (Lac-OMVs) with first-stage dendritic cell (DC) activation and later stage inflammatory macrophage repolarization potential. Mechanistically, enrichment of the nicotinamide metabolism pathway within Lac-OMVs is discovered, with nicotinamide adenine dinucleotide (NAD+) as a key anti-inflammatory mediator. NAD+-enriched Lac-OMVs (NAD+-Lac-OMVs) are thus metabolically engineered via targeted culture condition optimization, exerting biphasic immunomodulatory effects: (1) early-stage DC activation establishes robust humoral immunity, protecting against primary and recurrent MRSA challenge (99.35% and 98.07% bacterial clearance rates, respectively); and (2) later stage targeted NAD+ delivery reprograms inflammatory macrophages at the defect site, resolving inflammation and establishing a pro-osteogenic microenvironment (∼10 times higher bone-repair rate).Keywords: chronic bone infection, probiotic outer membrane vesicles, OMVs, nicotinamide adenine dinucleotide, NAD+, antibacterial nanovaccine, immune microenvironment regulation Introduction Unlike infections in other systems (e.g., respiratory), the gold-standard management of bone infection involves not only intensive antibiotic therapy but also surgical interventions: initial debridement to remove necrotic bone followed by secondary reconstruction.1 Ineffective control of infection and inflammation severely compromises bone repair during later stages. The rise of antibiotic-resistant pathogens, notably methicillin-resistant Staphylococcus aureus (MRSA), poses a significant challenge. High recurrence rates prolong infection into a chronic state, persistently hindering bone regeneration and leading to non-unions in nearly 40% of patients.2,3,4 Chronic bone infection is a dynamic, complex disease. While antimicrobials are crucial, resistance of pathogens like MRSA leads to inconsistent efficacy.5 Furthermore, antibiotics cause adverse effects such as gut dysbiosis and organ toxicity and in turn contribute to the emergence of resistance.6 Resistant pathogens such as MRSA employ evasion strategies: generating persisting bacteria resistant to antibiotics and immune clearance and actively secreting metabolites that induce an immunosuppressive network, particularly affecting antigen-presenting cells (APCs).7,8,9,10 Upon waning host defenses, these persisting bacteria reactivate, causing relapse and further delaying bone repair, potentially leading to amputation or death.11,12 Therefore, for effective long-term management of chronic, drug-resistant bone infection, sustained activation of host antibacterial immunity to generate specific, durable protection against recurrence is paramount. However, excessive immune activation impedes osteogenesis13,14; thus, maintaining a long-term anti-inflammatory microenvironment is equally essential for bone repair that often requires a long period. Thus, a central therapeutic challenge lies in reconciling the need for rapid antimicrobial immune activation with the requirement for sustained anti-inflammatory conditions conducive to bone regeneration. Colonization of various gut probiotics inhibits proliferation of pathogenic bacteria15,16 while relieving intestinal inflammation against diseases like inflammatory bowel disease (IBD).17 Inspired by this foundation, we screened outer membrane vesicles (OMVs) from several common natural gut probiotics with potential anti-inflammatory efficacy for their therapeutic utility in chronic bone infection. Co-culture with MRSA, mouse bone marrow-derived mesenchymal stem cells (mBMSCs), dendritic cells (DCs), and M1-polarized macrophages assessed bacterial inhibition, cytotoxicity, and immunomodulatory functions (DC activation/maturation and macrophage repolarization). Results revealed slight variations in intrinsic antibacterial activity among the probiotic OMVs, broadly comparable DC activation capacity, and marked differences in anti-inflammatory efficacy (macrophage M2 reprogramming). OMVs from Lactococcus lactis (Lac-OMVs) emerged as the lead candidate, exhibiting antibacterial activity, potent DC activation, and strong induction of macrophage anti-inflammatory repolarization. To discover the anti-inflammatory mechanism and explore possible engineering improvements, vesicular metabolomics was employed, and significant enrichment of the nicotinamide (NAM) metabolism pathway within Lac-OMVs was identified, suggesting that their anti-inflammatory effects stem from encapsulated metabolites. This implies a biphasic and sequential therapeutic action: at the first stage, Lac-OMVs drive rapid DC activation via vesicular membrane pathogen-associated molecular patterns (PAMPs), followed by the second-stage sustained macrophage anti-inflammatory repolarization mediated by targeted intracellular delivery of interior NAM-related metabolites. Metabolic engineering yielded NAD+-enriched Lac-OMVs (NAD+-Lac-OMVs), and in vitro studies demonstrated that at the first stage, NAD+-Lac-OMVs maintain antibacterial activity and DC activation properties against MRSA immunosuppression, while at the later inflammatory stage, NAD+-Lac-OMVs strongly induce macrophage anti-inflammatory repolarization, thereby reshaping a pro-osteogenic microenvironment. In vivo evaluation confirmed that a single local dose of NAD+-Lac-OMVs activates durable humoral antibacterial immunity at the early stage, protecting against primary and recurrent infection, while later inducing potent macrophage anti-inflammatory repolarization at the defect site, ensuring a microenvironment conducive to sustained bone regeneration (graphical abstract). Results Validation and screening of the antimicrobial, immunomodulatory, and anti-inflammatory regulatory effects of probiotic OMVs Probiotics with established safety profiles and therapeutic efficacy include Lactobacillus, Bifidobacterium, Saccharomyces, etc.18 Their metabolites, including OMVs, demonstrate therapeutic potential against diseases such as dyspepsia,17 infections,19 gastrointestinal tissue damage,20 and immunodeficiency.21 We selected candidate probiotics from the aforementioned list to isolate OMVs and investigated whether their OMVs concurrently retain the immune-activating properties characteristic of OMVs and the anti-inflammatory/tissue-repair benefits of their parental probiotics (Figure 1A).22,23 Seven common probiotics were included: Lactococcus lactis (Lac, ATCC 19435) and Lactobacillus rhamnosus (LGG, ATCC 7469) from Lactobacillus; Bifidobacterium longum (B. longum, ATCC 15707); Saccharomyces cerevisiae (S. cerevisiae, ATCC 9763); the probiotic Escherichia coli strain Nissle 1917 (EcN); Akkermansia muciniphila (AKK, ATCC BAA-835); and Bacillus coagulans (B. coagulans, ATCC 7050). Identification and screening of antibacterial, immunomodulatory, and anti-inflammatory reprogramming effects of common intestinal probiotic OMVs (A) Schematic of the work done in this section (partly created in BioRender). Seven common intestinal probiotics were cultured and ultracentrifuged to collect different OMVs separately. Subsequently, biocompatibility test was employed to determine the optimal concentration used in following investigation. Lastly, antibacterial test with MRSA, DC activation test, and macrophage polarization test were employed to explore the optimal OMVs for following experiments. (B) Seven candidate probiotic OMVs’ 7-day biocompatibility test at 5, 10, 20, and 40 μg/mL, n = 3. (C and D) The 10-h growth curve of MRSA (C) and the 10-h MRSA survival rate (D), n = 3. (E and F) One-hour enrichment rate of DiO-labeled Lac-OMVs and BMSC-EVs in macrophages, presented by confocal fluorescence representation picture (E) and flow cytometry results (F), n = 3. Scale bar: 40 μm. (G and J) At the first stage, seven candidate probiotic OMVs induced DC cell differentiation, n = 3. (H, K, and L) At the first stage, seven candidate probiotic OMVs induced DC cell maturation, n = 3. (I and M) At the later stage, seven candidate probiotic OMVs induced anti-inflammatory macrophage repolarization, n = 3. At least three independent experiments were performed. Data represent means ± SD, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, no significance, by Student’s t test. Biocompatibility and safety profiles were first evaluated.24 Viability of mBMSCs co-cultured with OMVs (5, 10, 20, and 40 μg/mL, vesicle protein concentration) for 7 days was measured. At 5 μg/mL, all seven candidate OMVs showed biocompatibility comparable to the control group. At ≥20 μg/mL, all OMVs except B. coagulans-OMVs exhibited cytotoxicity (Figures S1A–S1G). However, at 10 μg/mL, all OMVs maintained high mBMSC viability (>85%) after 7-day co-culture. Notably, Lac-OMVs and B. coagulans-OMVs enhanced mBMSC bioactivity beyond the control group (Figure 1B). Based on optimal biocompatibility, 10 μg/mL was selected for subsequent experiments. The antimicrobial activity of candidate probiotic OMVs (10 μg/mL) revealed general MRSA growth inhibition during 10 h co-culture across all OMVs (Figure 1C), while EcN-OMVs, Lac-OMVs, B. longum-OMVs, and AKK-OMVs exhibited the highest and most consistent inhibition rates, with no significant differences. LGG-OMVs and S. cerevisiae-OMVs showed moderate efficacy (statistically comparable to Lac-OMVs), and B. coagulans-OMVs demonstrated the weakest effect (Figure 1D). To demonstrate superior APC uptake efficiency of probiotic OMVs compared to other vesicles, 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO) labeled vesicles were co-cultured with macrophages. After 1-hour co-culture, confocal microscopy revealed significantly stronger DiO signal in CD11b+ macrophages incubated with Lac-OMVs versus bone marrow mesenchymal stem cell-derived extracellular vesicles (BMSC-EVs) (Figure 1E). Flow cytometry reconfirmed that only ∼1% of macrophages were DiO positive for BMSC-EVs, compared to ∼5% for Lac-OMVs (Figure 1F). Probiotic OMVs inherit a diverse array of PAMPs from their parental bacteria, facilitating the targeted enrichment and efficient internalization of OMVs into immune cells, underscoring the potential of OMVs as targeted immunostimulants.25 DC differentiation and maturation are crucial for their function and downstream adaptive immunity.26,27 Specifically, conventional DCs (cDCs) differentiate into cDC1s (priming cytotoxic CD8+ T cell responses) and cDC2s (driving humoral immunity and plasma/memory B cell generation), distinguishable by surface markers CD103 and CD11b.26,28 Maturation, marked by upregulated co-stimulatory molecules CD80/CD86, signifies functional competence.27 Flow cytometry analysis of differentiation markers in primary mouse bone marrow-derived DCs cultured with or without 10 μg/mL candidate probiotic OMVs revealed that DCs predominantly differentiated into cDC2s (gated as CD11b+CD103−) under probiotic OMV stimulation, and probiotic OMVs (excluding AKK-OMVs) significantly enhanced cDC2 differentiation (Figures 1G and 1J). EcN-OMVs, Lac-OMVs, and B. coagulans-OMVs were the top inducers, with no significant difference between EcN-OMVs and Lac-OMVs. Maturation marker analysis of these cDC2s showed that EcN-OMVs induced the strongest expression, although the increase was modest (Figures 1H, 1K, and 1L). The other probiotic OMVs induced cDC2s maturation with relatively similar efficacy. We further investigated the immunomodulatory effects of probiotic OMVs on macrophage. Macrophage actively regulates the function and state of tissue cells, plays a critical role in a variety of inflammatory diseases, and is of great significance for tissue homeostasis and repair.29 Macrophage phenotypes are broadly defined as anti-inflammatory (M2, CD206+CD86−) or pro-inflammatory (M1, CD206−CD86+). Therefore, whether OMVs can sequentially reprogram inflammatory macrophages induced by first-stage antibacterial therapy to an anti-inflammatory phenotype is key to reshaping an anti-inflammatory microenvironment for later stage bone tissue repair. Lipopolysaccharide (LPS) and hydrogen peroxide (H2O2) were applied to treat primary mouse bone marrow-derived macrophages, simulating the inflammatory state after first-stage antibacterial therapy initiated by OMVs. Under normal conditions, macrophages were predominantly CD206+CD86− and CD206−CD86− phenotypes (Figures 1I and 1K). LPS/H2O2 stimulation significantly activated macrophages, increasing CD86+ cells and CD206+CD86+ double-positive cells, while decreasing the CD206+CD86−/CD206−CD86+ subset ratio (Figure 1K). All seven candidate probiotic OMVs partially reversed this LPS/H2O2-induced pro-inflammatory phenotype, increasing the CD206+CD86− population and decreasing the CD206−CD86+ population (Figures 1K and 1M), among which Lac-OMVs, LGG-OMVs, and B. longum-OMVs induced the most substantial increases in CD206+CD86− macrophages. Quantifying the overall phenotypic shift (ratio of CD206+CD86− to CD206−CD86+ cells) confirmed that Lac-OMVs, LGG-OMVs, and B. longum-OMVs most potently reprogramed an anti-inflammatory macrophage phenotype (Figure 1M). Lac-OMVs and B. longum-OMVs were the strongest, followed by LGG-OMVs and then S. cerevisiae-OMVs. EcN-OMVs, AKK-OMVs, and B. coagulans-OMVs showed weaker effects. Among the OMVs exhibiting strong macrophage anti-inflammatory reprogramming properties (Lac-OMVs, LGG-OMVs, and B. longum-OMVs), Lac-OMVs demonstrated the highest capacity for inducing cDC2 differentiation (Figure 1J) and possessed antimicrobial activity equivalent to LGG-OMVs and B. longum-OMVs (Figure 1D). While its induction of DC maturation (CD80/CD86) was slightly lower than LGG-OMVs and B. longum-OMVs (Figures 1K and 1L), Lac-OMVs specifically combined potent first-stage immunostimulation with significant later stage anti-inflammatory macrophage repolarization. Although LGG-OMVs and B. longum-OMVs possessed nearly equivalent capabilities to Lac-OMVs in repolarizing inflammatory macrophages toward an anti-inflammatory phenotype and exhibited comparable potency in inducing DC maturation, a key distinction emerged in their ability to promote DC differentiation into specialized antigen-presenting subsets (namely cDC1s and cDC2s)—a critical process for initiating robust antibacterial immunity. In this specific aspect, Lac-OMVs demonstrated markedly superior performance compared to LGG-OMVs and B. longum-OMVs. Therefore, based on a comprehensive evaluation, Lac-OMVs were selected as the lead candidate in this study, demonstrating the desired integration of sequential immune-activating properties and anti-inflammatory/tissue-repair potential characteristic of a probiotic-derived nanovaccine. Potential mechanistic elucidation of Lac-OMVs’ immunomodulation via proteomics and metabolomics To decipher Lac-OMVs’ immunomodulatory mechanisms, quantitative proteomic sequencing and enrichment analysis were first performed on Lac-OMV-treated DCs and inflammatory macrophages. Quantitative proteomics comparing Lac-OMV-treated DCs to untreated controls revealed differentially expressed proteins significantly enriched in Gene Ontology (GO) terms related to metabolic processes, DNA translation, cell replication, DNA replication/repair, and cell cycle regulation (Figure 2A), indicating broad cellular effects. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment highlighted alterations in infection-related pathways (e.g., COVID-19 and Salmonella infection) and, crucially, pathways intimately linked to DC function, activation, and maturation: fatty acid metabolism, FcγR-mediated phagocytosis, and the PPAR signaling pathway30 (Figure 2B). Volcano plot analysis identified specific upregulated proteins within these pathways (SCP2, ACADM, ACADL, CPT2, CPT1A, and ACAA1A) (Figure 2C), primarily associated with enhanced fatty acid degradation. As reduced lipid accumulation via fatty acid oxidation promotes DC activation, maturation, and cross-presentation,30 these results suggest that the first-stage antibacterial therapy is mainly established by membrane components of Lac-OMVs that enhance DC functionality and maturation through activation of fatty acid oxidative degradation. Elucidating potential DC activation and macrophage phenotypic reprograming mechanism of Lac-OMVs by proteomics and metabolomics (A–C) GO, KEGG, and volcano plot analyses of proteomic sequencing result in DCs under ± Lac-OMV treatment, n = 3. (D–F) GO, KEGG, and volcano plot analyses of proteomic sequencing result in macrophage under LPS/H2O2 pathological conditions ± Lac-OMV treatment, n = 3. (G and H) Pathway enrichment and volcano plot analysis of untargeted metabolomics sequencing, n = 5. (I) Biosynthesis process of NAD+ in Lac. nadR, NMN adenylyltransferase; PncC, NMN aminohydrolase; FtnadE∗, NMN synthase from Francisella tularensis; nadE, NAD synthetase; nadD, NaMN adenylyltransferase; pncB, nicotinic acid phosphoribosyltransferase. (J) Other upregulated metabolites with inflammatory/antioxidant properties in Lac-OMVs sorted by p value and fold change, n = 5. At least three independent experiments were performed. Data represent means ± SD, ∗∗∗p < 0.001, by Student’s t test. Given the central role of macrophages in regulating inflammatory microenvironments, we focused on Lac-OMVs’ mechanism of macrophage phenotype modulation for potential engineering. Quantitative proteomics of primary mouse bone marrow-derived macrophages under LPS/H2O2 inflammatory conditions ± Lac-OMV treatment revealed differential protein enrichment in GO terms including cytokine activity, oxidative stress response, inflammatory response, and tissue remodeling (Figure 2D), suggesting that Lac-OMVs impact cytokine secretion in inflammatory milieus. KEGG enrichment identified cytokine-cytokine receptor interaction and the relaxin signaling pathway as the most significantly altered (Figure 2E). Volcano plot analysis confirmed significant upregulation of NOS2 and downregulation of COL1A1/COL1A2, indicative of relaxin pathway activation, which is correlated with macrophage M2 polarization.31 Critically, MRC1 (CD206) was significantly upregulated (Figure 2F), corroborating flow cytometry results showing increased CD206+ anti-inflammatory macrophages. Additionally, Lac-OMVs upregulated several anti-inflammatory phenotype-associated proteins such as PTGS2, HMOX1, IL1RN, PEX14, and FASTKD1,32,33 while also moderately downregulated the pro-inflammatory factor FGL2.34 Collectively, the proteomics indicates that Lac-OMVs (1) significantly enhance expression of anti-inflammatory proteins in macrophages within inflammatory microenvironments, mediating downstream anti-inflammatory phenotype repolarization and functions; and (2) strongly activate the relaxin signaling pathway, suggesting that this pathway might contribute to the observed phenotypic and functional shifts. As prokaryotic-derived vesicles, Lac-OMVs carry numerous parental Lac metabolites, nucleic acids, and proteins. While nucleic acids/proteins may not directly function in eukaryotic cells and require specific conditions to affect downstream biological processes, metabolites retain their biological activity across species.35 Thus, at the later stage anti-inflammatory therapy, interior metabolites likely represent the primary mode of Lac-OMVs’ interaction with host cells, driving phenotypic and functional changes. Leveraging Lac’s well-established genetic tractability and scalability for microbial engineering, untargeted metabolomics was next employed to reveal the metabolite profile of Lac-OMVs and therefore elucidate the potential immunomodulatory mechanism from the perspective of metabolites, providing a basis for easy targeted engineering to magnify the anti-inflammatory potential of Lac-OMVs and therefore further enhance the bone repair efficacy. Given the functional similarities and differences between Lac-OMVs and EcN-OMVs (comparable antimicrobial activity, Figure 1D; cDC2 differentiation, Figure 1J; EcN-OMVs induced slightly higher DC maturation, Figures 1K and 1L; but significantly weaker macrophage anti-inflammatory reprogramming, Figure 1M), these two OMVs confer similar antimicrobial activity and antibacterial immunity induction ability, while the anti-inflammatory potential of EcN-OMVs is significantly weaker than that of Lac-OMVs. Therefore, comparative metabolomics of these two OMVs was performed to identify metabolites specifically associated with Lac-OMVs’ potent anti-inflammatory macrophage repolarization. Pathway enrichment analysis revealed that Lac-OMVs’ differential metabolites were significantly enriched in three pathways: starch and sucrose metabolism, nicotinate and NAM metabolism, and pantothenate and CoA biosynthesis (Figure 2G). While starch/sucrose and pantothenate/CoA metabolism primarily affect cellular sugar balance and energy metabolism, which lack direct immune regulation or tissue repair links, nicotinate and NAM metabolism is crucial for cellular energy balance, antioxidant defense, and genomic stability, establishing a critical impact on immunomodulation and inflammatory regulation. The core metabolite of this pathway, NAD+, plays vital roles in antioxidant responses, DNA repair, inflammation/immune regulation, and energy metabolism and is a key driver of macrophage anti-inflammatory polarization.36 Volcano plot analysis confirmed significant upregulation of key nicotinate and NAM metabolism pathway metabolites in Lac-OMVs: NAD+, nicotinamide riboside (NR), nicotinamide mononucleotide (NMN), dihydroxyacetone phosphate (DHAP), and NAM. Conversely, nicotinic acid (NA) and nicotinate mononucleotide (NaMN) were significantly downregulated (Figure 2H). Recent research underscores NAD+ as a central therapeutic target in inflammatory, aging-related, acute injury, and infectious diseases. Declining intracellular NAD+ levels strongly promote pro-inflammatory macrophage phenotypes.37,38 Targeted supplementation with NAD+ precursors (e.g., NMN) not only reverses this phenotype but also boosts ATP production, enhances cellular metabolism, and promotes clearance of apoptotic cells and inflammatory debris, facilitating tissue repair.36,39 Concurrently, elevated intracellular NAD+ levels are closely linked to DC activation and maturation, promoting cytokine production by CD11b+CD11c+ DCs and influencing downstream CD4+ T cell differentiation.40 This suggests that NAD+-targeted anti-inflammatory therapy does not compromise DC-mediated immune activation. Combining Lac-OMVs’ inherent anti-inflammatory metabolites and membrane components with strategies to boost NAD+ may thus be an effective approach to further enhance their sequential-phase immunomodulatory capacity, promoting DC activation/maturation at the first stage and then tuning macrophage anti-inflammatory reprogramming at the later stage. Microbial NAD+ biosynthesis slightly differs from mammalian cells. Microbes convert NR directly to NMN and then to NAD+. They can also convert NA to NaMN, which is then converted either to nicotinic acid adenine dinucleotide (NaAD) and then NAD+, or to NMN and then NAD+41 (Figure 2I). The upregulation of NR, NMN, and NAD+ in Lac-OMVs indicates predominant utilization of the NR→NMN→NAD+ pathway in Lac, rather than the NA-initiated route. Beyond the pathway-enriched metabolites, volcano plot analysis identified other significantly upregulated metabolites in Lac-OMVs with known anti-inflammatory/antioxidant properties when sorted by p value (e.g., 6-formylpterin, D-allose, and raffinose)42,43 and fold change (e.g., robinin, daidzin, and astaxanthin)44 (Figure 2J). Notably, trigonelline was also significantly upregulated. Besides its antioxidant/anti-inflammatory effects, trigonelline serves as an important NAD+ precursor and enhancer in mammalian cells.45 Enhanced anti-inflammatory repolarization and osteoimmunological repair by NAD+-Lac-OMVs Direct cellular uptake of free NAD+ is hindered by the lack of specific transmembrane transporters in mammalian cells and rapid degradation by ectoenzymes like CD38.37 Consequently, utilizing Lac-OMVs as targeted carriers for NAD+ enrichment in APCs offers significant advantages. NMN, a large polar molecule, requires specific transporters typically absent in microbes, while NR, a smaller molecule, can passively diffuse across membranes via concentration gradients.41,46 Therefore, the NAD+ content within Lac-OMVs could be enhanced by supplementing the Lac culture medium with NR. Microbial NAD+ biosynthesis primarily relies on the bifunctional enzyme NadR (or PnuC in some systems) to convert NR to NMN and subsequently NMN to NAD+,41 with maximal NadR activity reported at 2 mM NR and 10 mM ATP.47 Leveraging the commercial availability and low cost of both precursors, we supplemented the culture medium to amplify the NR→NMN→NAD+ pathway in Lac, thereby increasing NAD+ levels within the derived OMVs (NAD+-Lac-OMVs) (Figure 3A). The enhanced macrophage-related anti-inflammatory repolarization effect of the metabolically engineered NAD+-Lac-OMVs (A) Schematic of the metabolic engineering process of NAD+-Lac-OMVs (partly created in BioRender). NR passively diffuses across membranes to elevate intracellular NAD+ levels, which can be enhanced in Lac-OMVs by NR supplementation. Microbial NAD+ biosynthesis, primarily mediated by the bifunctional enzyme NadR (or PnuC), converts NR to NMN and subsequently to NAD+, though its activity is subject to substrate inhibition and depends on NR and ATP concentrations. By supplementing culture medium with commercially available NR (2 mM) and ATP (10 mM), we amplified the discovered NR→NMN→NAD+ pathway in Lac to boost NAD+ levels within the resulting OMVs. (B) Representative transmission electron microscopy (TEM) images of Lac-OMVs and NAD+-Lac-OMVs. Scale bar: 100 nm. (C) Nanoparticle tracking analysis (NTA) results of Lac-OMVs and NAD+-Lac-OMVs. (D) NAD+ content measurement in Lac-OMVs and NAD+-Lac-OMVs, n = 4. (E) Lactate content measurement in Lac-OMVs and NAD+-Lac-OMVs, n = 5. (F) Antibacterial effect of Lac-OMVs and NAD+-Lac-OMVs, n = 3. (G) ELISA analysis of nisin from Lac-OMVs and NAD+-Lac-OMVs, n = 3. (H and I) Flow cytometry results of macrophage repolarization induced by Lac-OMVs and NAD+-Lac-OMVs, n = 3. (J) RT-qPCR results of macrophage repolarization induced by Lac-OMVs and NAD+-Lac-OMVs, n = 3. (K) ELISA analysis of IL-10, TGF-β, TNF-α, and IL-6 secreted by macrophage, n = 3. (L) Western blotting analysis of key protein expression during anti-inflammatory repolarization. At least three independent experiments were performed. Data represent means ± SD, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, no significance, by Student’s t test. Transmission electron microscopy confirmed that both Lac-OMVs and NAD+-Lac-OMVs exhibited the characteristic cup-shaped morphology of bacterial OMVs (Figure 3B). Nanoparticle tracking analysis (NTA) revealed no significant difference in particle size distribution between the two vesicle types (Figure 3C). Crucially, NAD+ quantification demonstrated that the optimized culture conditions significantly boosted (∼6-fold, per μg OMV protein) the NAD+ content within NAD+-Lac-OMVs compared to Lac-OMVs (Figure 3D), confirming successful metabolic engineering for targeted NAD+ enrichment. To assess potential metabolic perturbations, lactate, pH, and H2O2 levels were measured. NAD+-Lac-OMVs showed a non-significant increase in lactate content compared to Lac-OMVs, though levels in both were low overall (Figure 3E). The pH of OMV suspensions was comparable to phosphate-buffered saline (PBS) (Figure S2A), and H2O2 content was minimal in both preparations (Figure S2B). These results indicate that neither OMV types exert direct antibacterial effects or impact general cell viability through lactate or H2O2. Consistent with this, co-culture with MRSA revealed that NAD+ enrichment did not enhance the inherent antibacterial activity of Lac-OMVs (Figure 3F). Agarose gel electrophoresis confirmed the presence of conserved Lac-synthesized proteins in both OMV types, including a protein polymer band at approximately 14.4 kDa (Figure S2C), likely representing an unprocessed bacteriocin with retained antibacterial activity.48 Furthermore, quantitative enzyme-linked immunosorbent assay (ELISA) of nisin, a natural bacteriocin produced by Lac that exhibits significant antibacterial activity against Gram-positive bacteria,49,50 indicated that both Lac-OMVs and NAD+-Lac-OMVs contained approximately 3.47 and 3.65 ng nisin per 109 OMVs, respectively, with no statistically significant difference (Figure 3G), suggesting that the metabolic engineering process did not substantially alter the bacteriocin secretion profile of Lac, thereby preserving the direct antimicrobial contribution while primarily affecting NAM metabolism. Given the potent anti-inflammatory role of NAD+, we evaluated the impact of engineering on DC activation and maturation. NAD+-Lac-OMVs induced a slightly stronger, though not statistically significant, promotion of cDC2 differentiation compared to Lac-OMVs (Figures S2D and S2E). Free NMN/NAD+ supplementation had negligible effects. Importantly, NAD+-Lac-OMVs exhibited significantly enhanced capacity to induce DC maturation, evidenced by increased CD80 and CD86 co-expression (Figure S2F). Free NMN/NAD+ only moderately increased CD80. To determine whether this enhancement is originated from elevated NAD+ content within NAD+-Lac-OMVs or not, we treated DCs with NAD+-Lac-OMVs in the presence of FK866 (10 μM), a specific NAMPT inhibitor that blocks intracellular NAD+ downstream signaling. Inhibiting NAMPT with FK866 leads to the accumulation of NAM, which in turn exerts feedback inhibition on the activity of NAD+-consuming enzymes, obstructing cellular utilization of the delivered NAD+.36,51 Following FK866 addition, marked downregulation of CD80 and CD86 was observed, with CD80 expression reduced to a level comparable to that induced by Lac-OMVs (Figure S2G). Thus, enriching internal NAD+ content does not impair the intrinsic PAMP-dependent early-stage DC activation capacity of OMVs, and the inherent NAD+ potentiates DC activation and maturation. Previous proteomics implicated the relaxin signaling pathway in Lac-OMVs’ anti-inflammatory effects. We therefore investigated its role in NAD+-Lac-OMVs’ function using small interfering RNA (siRNA) knockdown of the relaxin family peptide receptor 1 (RXFP1), the primary receptor for relaxin.52 As shown in Figures 3H and 3I, Lac-OMVs restored the CD206+CD86−/CD206−CD86+ ratio close to physiological levels. NAD+-Lac-OMVs further enhanced this anti-inflammatory shift, normalizing the ratio. siRNA-mediated Rxfp1 knockdown significantly attenuated the anti-inflammatory effects of both OMV types, although the phenotype remained significantly less inflammatory than the LPS/H2O2-only control, indicating that RXFP1-independent mechanisms also contribute. Free NMN/NAD+ also induced anti-inflammatory shifts but was much less effective than Lac-OMVs. To further assess functional later stage anti-inflammatory repolarization, we measured mRNA expression of key markers by quantitative reverse-transcription polymerase chain reaction (RT-qPCR) (Figure 3J). Lac-OMVs significantly upregulated anti-inflammatory markers (Mrc1 [CD206], Il10, Tgfb, Arg1, Fizz1, Ym1, and Il1rn), with NAD+-Lac-OMVs further amplifying this upregulation. Rxfp1 knockdown suppressed this upregulation, though levels remained higher than the LPS/H2O2 control, and free NMN/NAD+ induced weaker upregulation. Conversely, Lac-OMVs and NAD+-Lac-OMVs downregulated pro-inflammatory markers (Cd86, Tnf, and Il1b), an effect potentiated in NAD+-Lac-OMVs and partially reversed by Rxfp1 knockdown. Notably, Nos2, often associated with pro-inflammatory states, was significantly upregulated by both OMV types in an RXFP1-dependent manner, whereas free NMN/NAD+ downregulated Nos2, dissociating its effect from relaxin pathway activation and highlighting a specific, context-dependent role for Nos2 within the vesicle-induced signaling cascade. Next, we assessed anti-inflammatory repolarization at the cytokine level via ELISA (Figure 3K). Compared to the CTRL, LPS/H2O2 stimulation upregulated both anti-inflammatory (interleukin-10 [IL-10] and transforming growth factor β [TGF-β]) and pro-inflammatory (tumor necrosis factor alpha [TNF-α] and IL-6) cytokines, with pro-inflammatory increases substantially greater. Both OMV treatments reversed this trend, and NAD+-Lac-OMVs induced the highest secretion of anti-inflammatory cytokines. RXFP1 knockdown significantly attenuated this repolarization process, although the cytokine profile remained more anti-inflammatory than the pathological control. Direct NMN/NAD+ supplementation also exhibited modest repolarization capacity. To further elucidate underlying mechanisms, we performed western blotting to examine activation of key signaling pathways downstream of both the relaxin receptor and NAD+ metabolism (Figure 3L).53 Both Lac-OMVs and NAD+-Lac-OMVs induced comparable activation of the PI3K-Akt and STAT3 pathways, which was slightly attenuated upon Rxfp1 knockdown, suggesting that the immunomodulatory effects may involve activation of these pathways. Furthermore, NAD+-Lac-OMVs mediated the highest SIRT1 expression, which was unaffected by Rxfp1 knockdown. Subsequently, to assess whether NAD+-Lac-OMVs alter the phagocytic and bactericidal capacity of macrophages, we performed a co-culture assay using macrophages and attenuated MRSA particles labeled with CypHer5E, a pH-sensitive dye that exhibits maximal fluorescence in acidic environments, allowing easy discrimination between actual phagocytosis and mere particle adhesion,54 for 12 h. Figures S2H and S2I revealed that macrophages activated under pathological conditions showed enhanced bactericidal capacity, indicated by increases in both the CypHer5E-positive rate (+) and strongly positive rate (++). Lac-OMVs did not markedly increase the overall CypHer5E-positive rate but significantly elevated the strongly positive rate (++), suggesting increased phagocytic load per cell. NAD+-Lac-OMVs significantly increased both parameters, indicating enhancements in both overall phagocytic population and phagocytic load per cell. Importantly, Rxfp1 knockdown abolished these increases, confirming the crucial role of relaxin signaling in modulating macrophage phagocytic and bactericidal function.31 Direct NMN/NAD+ supplementation also boosted macrophage phagocytic capacity, yielding effects comparable to Lac-OMVs.36,55 These flow cytometry findings were corroborated by immunofluorescence imaging, which visually confirmed stronger CypHer5E fluorescence signals in the Lac-OMV group compared to the pathological control, with the NAD+-Lac-OMV group displaying the most intense signal (Figure S2J). Direct co-culture of mBMSCs with Lac-OMVs or NAD+-Lac-OMVs for 14 and 21 days revealed limited intrinsic osteoinductive capacity, as assessed by alkaline phosphatase (ALP) and alizarin red S (ARS) staining (Figure S3A). The slightly deeper staining observed in the NAD+-Lac-OMV group likely reflects enhanced cell proliferation. Consistent with this, RT-qPCR analysis showed no significant upregulation of key osteogenic genes (Alp, runt-related transcription factor 2 [Runx2], and collagen type I [Col1]) in mBMSCs treated with either OMV type compared to controls (Figure S3B). To evaluate their osteoimmunomodulatory potential, we collected conditioned media (CM) from macrophages or inflammatory macrophages treated under different conditions and then used it to culture mBMSCs under pathological conditions (Figure 4A). Inflammatory stimulation significantly impaired mBMSCs’ ALP activity and expression of osteogenic genes. CM from Lac-OMV-treated macrophages (Lac-OMVs-CM) significantly restored ALP activity and osteogenic gene expression. CM from NAD+-Lac-OMV-treated macrophages (NAD+-Lac-OMVs-CM) demonstrated the most potent restorative effect, which was attenuated after Rxfp1 knockdown but still remained superior to the disease model control. CM from macrophages treated with free NMN/NAD+ also modestly enhanced ALP activity and gene expression (Figures 4B–4E). Osteoimmunomodulatory effect of NAD+-Lac-OMVs (A) Schematic of osteoimmunomodulatory effect of Lac-OMVs and NAD+-Lac-OMVs (partly created in BioRender). (B) ALP activity measurement of mBMSCs treated with different CM, n = 3. (C–E) RT-qPCR results of Alp, Runx2, and Col1 in mBMSCs treated with different CM, n = 3. (F) ALP and ARS staining (mineralized nodules) of mBMSCs treated with different CM. Scale bar: 5 mm. (G–I) Protein expression of ALP, RUNX2, and COL1 in mBMSCs treated with different CM. Scale bars: 30 μm. At least three independent experiments were performed. Data represent means ± SD, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, no significance, by Student’s t test. Inflammatory conditions severely reduced mBMSC ALP functionality and calcium deposition (mineralization). Lac-OMVs-CM and NAD+-Lac-OMVs-CM robustly rescued ALP function and mineralization, with NAD+-Lac-OMVs-CM showing the strongest effect. CM from the Rxfp1 knockdown group significantly diminished this rescue, though mineralization remained higher than the pathological control. CM from free NMN/NAD+-treated macrophages improved ALP function but failed to enhance mineralization significantly (Figure 4F). Inflammatory stimulation markedly decreased protein levels of key osteogenic markers in mBMSCs. Lac-OMVs-CM and NAD+-Lac-OMVs-CM effectively restored the expression of these proteins under inflammatory stress, with NAD+-Lac-OMVs-CM being the most potent. CM from the Rxfp1 knockdown group significantly reduced this restorative capacity, although expression levels were still higher than in the pathological model. CM from free NMN/NAD+-treated macrophages also promoted osteogenic protein expression compared to the pathological control but was less effective than Lac-OMVs-CM (Figures 4G–4I). NAD+-Lac-OMVs loaded alginate hydrogel for treating primary and recurrent MRSA bone infection To evaluate the translational potential of NAD+-Lac-OMVs, we first assessed their therapeutic efficacy in a murine model of primary bone infection, focusing on short-term antibacterial activity and immunomodulation following local administration (Figure 5A). The mouse calvarial infected bone defect model was primarily employed.56,57,58,59 NAD+-Lac-OMV-loaded alginate hydrogel for primary bone infection treatment (A) Schematic of primary bone infection treatment (partly created in BioRender). Firstly, a 3 mm calvarial infected bone defect was established by drills and MRSA suspension inoculation, followed by wound sewing. Then, 3 days after establishment of bone infection, the sewed wound was reopened, and debris at the defect site was replaced by hydrogel of different groups. (B) Lac-OMVs’ distribution within alginate hydrogel. Scale bar: 50 μm. (C and D) μCT reconstruction of bone tissue and analysis of new-born bone 5 weeks post-surgery, n = 3. (E) Masson staining of calvarial bone tissue 5 weeks post-surgery. Scale bar: 500 μm. (F and G) WBC and neutrophil counts 3 days post-surgery, n = 3. (H) Serum IgG concentration measurement 10 days post-surgery, n = 3. (I–K) Serum IgG1 and IgG2c concentration measurement and IgG2c/IgG1 ratio analysis 10 days post-surgery, n = 3. (L and M) Serum IgG3 and mucosal IgA concentration measurement 10 days post-surgery, n = 3. (N) Colony-forming unit of MRSA from calvarial defect site tissue homogenate, n = 3. (O) Schematic of flow cytometry experiment (partly created in BioRender). (P and U) Flow cytometry results of DC differentiation in DLNs, n = 3. (Q and V) Flow cytometry results of DC maturation in DLNs, n = 3. (R and W) Flow cytometry results of macrophage polarization in calvarial defect site, n = 3. (S and X) Flow cytometry results of CD4+ and CD8+ T cells in DLNs, n = 3. (T and Y) Flow cytometry results of plasma cell in DLNs, n = 3. At least three independent experiments were performed. Data represent means ± SD, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, no significance, by Student’s t test. NAD+-Lac-OMVs were encapsulated within an alginate hydrogel (ALG@NAD+-Lac-OMVs) to create an implantable sustained-release system. Confocal microscopy confirmed the uniform distribution of PKH26-labeled Lac-OMVs within the alginate matrix (Figure 5B). Hematological and serological analysis of liver and kidney function post-implantation revealed transient, non-significant elevations in most parameters at day 7, corresponding to a metabolic clearance phase (Figures S4A–S4E). These minor elevations showed no statistically significant difference compared with the CTRL group. By day 14, the values had largely returned to baseline levels (Figures S4J–S4N). Serum ELISA quantification of pro-inflammatory cytokines (IFN-γ, IL-1β, IL-6, and TNF-α) indicated no significant systemic inflammatory overactivation at either day 7 or day 14 (Figures S4F–S4I and S4O–S4R). Histopathological examination via H&E staining of major organs harvested at days 7 and 14 further supported the biosafety, with tissue architecture remaining typical and comparable to controls, showing no evidence of inflammatory cell infiltration or overt toxicity (Figures S5A and S5B). To specifically address safety within the context of the disease model, brain tissues from primary and recurrent cranial infection models were also examined (Figures S7A and S7B). Throughout the experimental period, no animals exhibited overt neurological signs of infection (e.g., neurological deficits or seizures), and H&E analysis revealed no gross or cytological alterations or inflammatory cell accumulation within the brain parenchyma, confirming the safety of the model.60,61 To track in vivo distribution and clearance, fluorescently labeled Lac-OMVs and NAD+-Lac-OMVs were loaded into the hydrogel and implanted (Figures S6A and S6B). At 12 h, intense fluorescence was observed at the defect site, liver, and kidneys, indicating hepatic sequestration and renal clearance. By day 7, signals had markedly diminished, although residual signals remained at the defect site, demonstrating systemic clearance. Micro-computed tomography (μCT) analysis of calvarial defects 5 weeks post-implantation demonstrated that MRSA infection severely impaired bone healing. The alginate-only (ALG) group showed minor improvement. Vancomycin facilitated better bone repair. Both ALG@Lac-OMVs and ALG@NAD+-Lac-OMVs significantly enhanced bone regeneration within infected defects, achieving the highest bone volume fraction (BV/TV) (Figures 5C and 5D). Masson’s trichrome staining revealed robust bridging with new bone formation in both OMV-treated groups, while vancomycin showed only moderate regenerative capacity (Figure 5E). Immunohistochemical (IHC) staining for osteocalcin (OCN) and RUNX2, key bone repair markers, showed relatively higher expression in the OMV-treated groups, with ALG@NAD+-Lac-OMVs exhibiting the strongest immunostaining (Figure S8A). H&E staining around the defect site indicated that vancomycin, ALG@Lac-OMVs, and ALG@NAD+-Lac-OMVs effectively suppressed local inflammation, with significantly reduced infiltration of polymorphonuclear leukocytes (PMNs) (Figure S8B). Complete blood count analysis 3 days post-surgery revealed elevated white blood cell (WBC) and neutrophil counts in the MRSA group. Vancomycin normalized these counts to near-physiological levels. ALG@Lac-OMVs reduced neutrophil counts to levels comparable to vancomycin, while WBC counts remained slightly elevated. ALG@NAD+-Lac-OMVs normalized both WBC and neutrophil counts to levels similar to vancomycin, demonstrating rapid antibacterial and anti-inflammatory efficacy (Figures 5F and 5G). Serum total IgG ELISA analysis at day 10 revealed that vancomycin induced a modest humoral response. Both ALG@Lac-OMVs and ALG@NAD+-Lac-OMVs significantly enhanced serum IgG production compared to vancomycin, with ALG@NAD+-Lac-OMVs inducing the highest level (Figure 5H). Furthermore, both IgG1 and IgG2c subtypes showed a modest elevation in the MRSA-infected group and the vancomycin-treated group, likely stemming from the host’s natural anti-infective response and, in the case of vancomycin treatment, may also result from enhanced bacterial antigen exposure and immune activation due to antibiotic-mediated bacterial killing. Marked upregulation of both IgG1 and IgG2c was observed in the ALG@Lac-OMVs group, further amplified in the ALG@NAD+-Lac-OMVs group (Figures 5I and 5J). The ratio of IgG2c to IgG1 remained similar between both groups, with IgG1 dominance indicating that the metabolic engineering process augmented the intensity but not the type of the induced response, a Th2-polarized response (Figure 5K).62 IgG3 subtype, an antibody subtype recognized for its potent anti-infective properties,63 followed a similar trend, confirming potent anti-infective properties (Figure 5L). Mucosal IgA levels showed only minimal, non-significant increases across all treatments, suggesting that IgA is likely not a primary therapeutic antibody induced by the OMV-based antibacterial vaccine. (Figure 5M). To directly evaluate antibacterial efficacy, tissue from the calvarial defect site was homogenized and subjected to plate counting (Figures 5N and S8C). Both OMV groups demonstrated potent bacterial clearance, with ALG@NAD+-Lac-OMVs exhibiting the strongest effect, consistent with the observed humoral immune response. Vancomycin also showed considerable efficacy, achieving a bacterial reduction close to that of ALG@Lac-OMVs. Flow cytometry analysis of cells from the defect site and draining lymph nodes (DLNs) elucidated the immune responses (Figure 5O). In DLNs, both OMV groups significantly increased the cDC2/cDC1 ratio, with ALG@NAD+-Lac-OMVs achieving the highest value (Figures 5P, 5U, S8D, and S8E), suggesting induction of cDC2 differentiation for humoral immunity. Both OMVs significantly increased the proportions of mature DCs (mDCs) in DLNs (Figures 5Q and 5V). At the defect site, MRSA infection decreased anti-inflammatory CD206+CD86− macrophages and increased pro-inflammatory CD206−CD86+ macrophages. Vancomycin had limited capacity to reverse this polarization. Both OMVs significantly increased the CD206+CD86− population and decreased the CD206−CD86+ population, restoring the CD206+CD86−/CD206−CD86+ ratio to physiological levels, with ALG@NAD+-Lac-OMVs significantly more effective (Figures 5R, 5W, S8F, and S8G). In DLNs, MRSA infection primarily elicited a CD4+ T cell response associated with humoral immunity. Vancomycin slightly amplified this trend, while both OMVs significantly enhanced the CD4+ T cell proportion (Figures 5S and 5X). Assessment of plasma cells (CD19−CD138+) in DLNs revealed that ALG@Lac-OMVs induced a slightly higher proportion, while ALG@NAD+-Lac-OMV treatment triggered a significant increase64,65 (Figures 5T and 5Y). To validate durable antibacterial immunity, we established a recurrent infectious bone defect model.64 Ten days after initial surgery and treatment, wounds were reopened and reinoculated with MRSA suspension to simulate recurrent infection (Figure 6A). Quantitative μCT analysis revealed that ALG@Lac-OMVs still promoted bone regeneration under recurrent infection, but ALG@NAD+-Lac-OMVs facilitated the most significant bone repair, achieving the highest bone volume fraction (BV/TV) at the defect site. The MRSA control group exhibited further expansion of the bone defect area due to recurrent infections. Vancomycin failed to adequately manage the recurrent MRSA challenge and associated inflammation, resulting in inferior bone repair compared to both OMV groups (Figures 6B and 6C). Masson’s trichrome staining revealed that ALG@NAD+-Lac-OMVs induced robust bone bridging with thick new-born bone formation, and ALG@Lac-OMVs also promoted bone bridging, though less substantially, while vancomycin-treated defects were primarily filled with soft connective tissue (Figure 6D). IHC staining for OCN and RUNX2 revealed congruent trends (Figure S9A), with the highest expression in the ALG@NAD+-Lac-OMVs group. Notably, in this recurrent infection model, the vancomycin treatment group also exhibited modest expression of OCN and RUNX2, likely attributable to its early bacterial clearance activity. H&E staining around the defect site showed significantly reduced PMN infiltration in both OMV groups (Figure S9B). NAD+-Lac-OMV-loaded alginate hydrogel for recurrent bone infection treatment (A) Schematic of the establishment of recurrent bone infection model and humoral immune response (partly created in BioRender). Ten days after hydrogel implantation, the wound was reopened and MRSA suspension was inoculated again in the defect to imitate MRSA rechallenging infection. (B and C) μCT reconstruction of bone tissue and analysis of new-born bone, n = 3. (D) Masson staining of calvarial bone tissue. Scale bar: 500 μm. (E) Serum IgG concentration measurement, n = 3. (F–H) Serum IgG1 and IgG2c concentration measurement and IgG2c/IgG1 ratio analysis, n = 3. (I and J) Serum IgG3 and mucosal IgA concentration measurement, n = 3. (K–M) ELISA analysis of serum antibodies against whole bacteria antigen and surface targets (LTA and SasA), n = 3. (N) Colony-forming unit of MRSA from calvarial defect site tissue homogenate, n = 3. (O and P) Flow cytometry results of memory B cells in DLNs, n = 3. At least three independent experiments were performed. Data represent means ± SD, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, no significance, by Student’s t test. Serum total IgG ELISA confirmed sustained elevation in vancomycin and OMV groups during long-term observation. Notably, ALG@NAD+-Lac-OMVs induced the highest serum IgG concentration (Figure 6E). Serum titers of IgG subtypes revealed a pattern consistent with the primary infection model (Figures 6F–6J). Given that antibacterial defense in the recurrent infection model relies almost entirely on the immune response primed by the OMV vaccine implanted during the primary infection, we further assessed the antigen specificity of serum antibodies from this model. ELISA against whole MRSA antigens revealed that ALG@NAD+-Lac-OMVs induced high levels of IgG targeting broad MRSA antigens (Figure 6K). Subsequent ELISA targeting specific MRSA surface antigens—lipoteichoic acid (LTA)66 and Staphylococcus aureus surface protein A (SasA)67—confirmed substantial IgG levels against these defined surface epitopes (Figures 6L and 6M). These findings collectively indicate that the IgG antibodies induced by the OMV-based antibacterial vaccine exhibit specificity against both broad MRSA antigens and key surface antigens.68 Plate counting of tissue from the recurrent infection model revealed that vancomycin demonstrated limited efficacy against recurrent infection. In contrast, both OMV treatments provided superior long-term protection, with ALG@NAD+-Lac-OMVs achieving the lowest bacterial burden (Figures 6N and S9C). As it has been proven in the aforementioned experiments that the ALG@NAD+-Lac-OMV nanovaccine predominantly induced humoral immunity in both in vitro and in vivo model, we quantified memory B cells (gated as IgG+IgD−) in DLNs to directly demonstrate the establishment of long-term antibacterial immunity.69 Vancomycin induced a modest increase, while both OMV treatments significantly enhanced memory B cell proportion, with ALG@NAD+-Lac-OMVs eliciting the most substantial increase (Figures 6O and 6P). Discussion Chronic bone infection, particularly driven by resistant pathogens like MRSA,70,71,72,73 poses a formidable clinical challenge due to its self-perpetuating immunosuppressive niche and high recurrence rate, ultimately leading to catastrophic bone failure.2,10,69 Two temporally distinct therapeutic goals, initial robust immune activation to eradicate infection64 and subsequent resolution of inflammation to enable bone repair, are urgently needed to achieve effective integrated treatment. This study presents a paradigm-shifting therapeutic strategy centered on the innate immunomodulatory capabilities of probiotic-derived OMVs, specifically engineered for sequential, biphasic action. The development of immunomodulatory nanovaccines represents a promising strategy for managing dynamic and protracted diseases. Current nanovaccine paradigms can be broadly categorized into several approaches. The first involves therapeutic modalities that promote antigen release and passive immune activation by directly attacking pathogens, for instance, by inducing apoptosis or pyroptosis. Examples include delivering pyroptosis-inducing nanopores to tumor cells74 or employing dynamic therapies that generate reactive oxygen species to directly kill pathogens.14 The second approach focuses on pre-processing pathogen antigens to directly prime specific immunity, exemplified by lymph node-targeting lipid nanoparticles pre-loaded with tumor antigens,75 or DC-targeting systems integrating cancer stem cell-like antigens shared across tumor types with epigenetic modulators.76 The third strategy utilizes cross-species immunogenic substances, leveraging their inherent immunogenicity to activate immunity. Examples include employing pathogenic bacterial OMVs to ignite trained immunity,77 or using induced pluripotent stem cell membranes to stimulate innate immune responses.78 Our work falls within this third category, employing probiotic-derived natural OMVs, and we enhanced their therapeutic efficacy through metabolic engineering to specifically enrich immunomodulatory metabolites within these vesicles. Compared to the general third strategy, which commonly uses cross-species immunostimulants that primarily focus on boosting immune activation while often overlooking the detrimental effects of prolonged inflammatory side effects on subsequent tissue repair, our work systematically screened OMVs from gut probiotics and revealed Lac-OMVs as a specifically potent natural nanovaccine, demonstrating integrated therapeutic potential of sequential immune regulation. Crucially, vesicular metabolomics identified the profound enrichment of the NR→NMN→NAD+ NAM metabolism pathway within Lac-OMVs, pinpointing NAD+ as the core immunometabolite responsible for mediating macrophage repolarization and establishing the anti-inflammatory milieu essential for bone healing.36,45 Capitalizing on this metabolic insight, we devised a scalable metabolic engineering strategy. By amplifying precursor feeding in the NR→NMN→NAD+ pathway during Lac culture, we successfully generated NAD+-Lac-OMVs, retaining the intrinsic DC-activating properties but exhibiting enhanced anti-inflammatory reprogramming efficacy. In a mouse calvarial model of primary and recurrent MRSA-infected bone defects, a single local dose of NAD+-Lac-OMV demonstrated sequential therapeutic effects. Phase 1 (immune activation): NAD+-Lac-OMVs rapidly elicited potent pathogen-specific humoral immunity, providing robust protection against both primary infection and recurrent MRSA rechallenge, significantly outperforming vancomycin in preventing recurrence. Phase 2 (inflammation resolution and repair): NAD+-Lac-OMVs effectively delivered their NAD+-core payload intracellularly to inflammatory macrophages within the defect site at the later stage, driving potent anti-inflammatory repolarization within the defect site,36 resolving chronic inflammation, re-establishing immune homeostasis, and fostering a pro-osteogenic microenvironment, as quantifiably demonstrated by μCT (BV/TV) and histomorphometry, surpassing the gold-standard vancomycin therapy. Limitations of the study This study has certain limitations, particularly regarding its potential for clinical translation. While one of its strengths lies in the relatively convenient scaled-up production of the metabolically engineered nanovaccine and the established clinical use of Lac and its products, several steps remain before clinical translation can be realized. First, studies in large animal models, such as goats or rhesus monkeys, are necessary. These should be accompanied by standardized biosafety assessments and dose optimization to identify a safe and effective dosage window suitable for large animals and, ultimately, for clinical application. Furthermore, to advance clinical translation, the in vitro stability and optimal storage conditions of the OMVs require further investigation. Although OMVs are generally stable in vitro, they typically require low-temperature storage. While refrigeration at 4°C may preserve stability for only a few days, storage at −20°C can extend the shelf life. Verifying the efficacy of OMVs under various storage conditions is an essential prerequisite for future clinical development. Finally, although this work explores a specific therapeutic paradigm in the context of bone infection, the underlying strategy—early-stage antibacterial action coupled with immune activation, followed by later stage reprogramming of the inflammatory milieu into an anti-inflammatory, pro-repair microenvironment—is not limited to bone infection. This sequential immunomodulatory approach holds significant potential for extension to other conditions, such as infected wound healing (particularly in diabetic patients) and soft tissue infections like cellulitis. Future work may productively adapt and apply this therapeutic framework to these and similar pathological contexts. Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Prof. Xiaobo Feng (fengxiaobo@hust.edu.cn). Materials availability This study did not generate new unique reagents. Data and code availability •The proteomics and metabolomics data have been deposited in iProX and MetaboLights, respectively. The proteomics data are available under iProX: PXD076052 (https://www.iprox.cn/), and the metabolomics data are available under MetaboLights: MTBLS14091 (https://www.ebi.ac.uk/). The proteomics and metabolomics data have been deposited in iProX and MetaboLights, respectively. The proteomics data are available under iProX: PXD076052 (https://www.iprox.cn/), and the metabolomics data are available under MetaboLights: MTBLS14091 (https://www.ebi.ac.uk/). •This study did not involve the reporting of novel or original code. This study did not involve the reporting of novel or original code. •All data essential for evaluating the conclusions drawn in this paper are presented within the paper itself and/or in the supplemental information. All data essential for evaluating the conclusions drawn in this paper are presented within the paper itself and/or in the supplemental information. •Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request. Acknowledgments This work was supported by the National Science Foundation of China (no. 82572766, 82272459, 82130072, 82072505, 82472511, and 22175058), Hubei Provincial Natural Science Foundation (no. 2026AFC1425), Creative Research Groups of Hubei Province (no. 2021CFA007), and Shenzhen Science and Technology Program (SGDX20230116093544006). We thank the Huazhong University of Science & Technology Analytical & Testing Center Medical Subcenter for their technical support. Additionally, we thank BioRender for providing illustrations in graphical abstract, Scheme 1, and Figures 1A, 3A, 4A, 5A, 5O, and 6A in this manuscript (LO28S458TO, UP28S45AWY, and BE28S459J2). Schematic of the screening and engineering strategy Partly created in BioRender. (1) Initial screening of seven gut probiotic-derived OMVs for key functionalities: antibacterial activity, DC activation, and macrophage reprogramming toward an anti-inflammatory phenotype. (2) Identification of Lac-OMVs exhibiting superior performance across all three functionalities. (3) Vesicular metabolomic sequencing revealed NAD+-core-enriched nicotinamide metabolites as potential mediators of Lac-OMVs’ anti-inflammatory efficacy. (4) Metabolic engineering cultivation to produce NAD+-enriched OMVs (NAD+-Lac-OMVs), developed into a potent OMV-based antibacterial nanovaccine with enhanced anti-inflammatory properties. Author contributions J.L. and B.T. conceived the project and designed the experiments. J.L., B.T., S.Z., Z.O., D.Z., X.Z., and H.L. performed the experiments and collected the data. J.L., B.T., Z.O., and L.T. discussed the results. J.L. and L.T. wrote and revised the manuscript. Z.Z., C.Y., L.T., and X.F. supervised the project and edited the manuscript. Declaration of interests The authors declare no competing interests. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work, the authors used DeepSeek in order to polish statements to reduce grammatical errors and spelling mistakes. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesAnti-Mouse CD11b (M1/70)ProteintechCat. No. 65055-1-Ig; RRID: AB_2918363PE-Cy7 anti-mouse CD11bBD PharmingenCat. No. 552850; RRID: AB_394491APC anti-mouse CD11cBD PharmingenCat. No. 550261; RRID: AB_398460PE anti-mouse MHC-IIBD PharmingenCat. No. 557000; RRID: AB_396546BV421 anti-mouse CD103BD PharmingenCat. No. 748259; RRID: AB_2872688FITC anti-mouse CD86BD PharmingenCat. No. 561962; RRID: AB_10896136PerCP-Cy5.5 anti-mouse CD80BD PharmingenCat. No: 560526; RRID: AB_1727514BV421 anti-mouse F4/80BD PharmingenCat. No: 565411; RRID: AB_2734779APC anti-mouse CD206InvitrogenCat. No: 17-2061-82; RRID: AB_2637420FITC anti-mouse CD8BD PharmingenCat. No: 561966; RRID: AB_10896291PE-Cf594 anti-mouse IgGAbsinCat. No: abs20017; RRID: AB_3750995PerCP-Cy5.5 anti-mouse CD3BD PharmingenCat. No: 561108; RRID: AB_394082PE-Cy7 anti-mouse B220BD PharmingenCat. No: 561881; RRID: AB_394458APC-Cy7 anti-mouse CD45BD PharmingenCat. No: 561037; RRID: AB_396774BV421 anti-mouse CD138BD PharmingenCat. No: 566289; RRID: AB_11153126BV605 anti-mouse CD4BD PharmingenCat. No: 740352; RRID: AB_2740085BV711 anti-mouse IgDBD PharmingenCat. No: 405731; RRID: AB_2563342BV786 anti-mouse CD19BD PharmingenCat. No: 563333;RRID: AB_2738141GAPDH Monoclonal antibodyProteintechCat. No: 60004-1-Ig; RRID: AB_2107436Anti-PI3K antibodyProteintechCat. No: 21708-1-AP; RRID: AB_3085666Anti-p-STAT3 antibodyProteintechCat. No: 60479-1-Ig; RRID: AB_3670262Anti-p-AKT antibodyProteintechCat. No: 66444-1-Ig; RRID: AB_2782958Anti-SIRT1 antibodyProteintechCat. No: RMX00009; RRID: AB_3750072Anti-ALP antibodyProteintechCat. No: 33076-1-AP; RRID: AB_3742845Anti-RUNX2 antibodyProteintechCat. No: 82636-2-RR; RRID: AB_3086495Anti-COL1 antibodyProteintechCat. No: 66761-1-Ig; RRID: AB_2882107Anti-OCN antibodyProteintechCat. No: 20277-1-AP; RRID: AB_11232596Bacterial and virus strainsLactococcus lactisChina Center for Type Culture CollectionATCC 19435Lactobacillus rhamnosusChina Center for Type Culture CollectionATCC 7469Bifidobacterium longumChina Center for Type Culture CollectionATCC 15707Saccharomyces cerevisiaeChina Center for Type Culture CollectionATCC 9763Escherichia coli strain Nissle 1917BeNa Culture CollectionCat. No: BNCC361741Akkermansia muciniphilaChina Center for Type Culture CollectionATCC BAA-835Bacillus coagulansChina Center for Type Culture CollectionATCC 7050Chemicals, peptides, and recombinant proteinsM-CSF protein, mouseGenScriptCat. No: Z03275GM-CSF protein, mouseGenScriptCat. No: Z03300LPS, <3% Protein (Lowry)Sigma-AldrichCat. No: L6529-1MGNAD+, 99.96%MedChemExpressCat. No: HY-B0445NMN, 99.94%MedChemExpressCat. No: HY-F0004NR, 99.90%MedChemExpressCat. No: HY-123033AATP, 99.88%MedChemExpressCat. No:HY-B2176PKH26Sigma-AldrichCat. No: PKH-26GLDiOBeyotimeCat. No: C1993SCypher5ECytivaCat. No: PA15401Critical commercial assaysMouse TNF-α ELISA kitElabscienceCat. No: E-EL-M3063Mouse IL-6 ELISA kitElabscienceCat. No: E-EL-M0044Mouse IL-1β ELISA kitElabscienceCat. No: E-EL-M0037Mouse IFN-γ ELISA kitElabscienceCat. No: E-EL-M0048Mouse IgG1 ELISA kitElabscienceCat. No: E-EL-M3035Mouse IgG2c ELISA kitElabscienceCat. No: E-OSEL-M0010Mouse IgG3 ELISA kitElabscienceCat. No: E-EL-M2401Mouse IgA ELISA kitElabscienceCat. No: E-EL-M0690Mouse IL-10 ELISA kitElabscienceCat. No: E-EL-M0046Mouse TGF-β ELISA kitElabscienceCat. No: E-EL-0162Mouse MRSA IgG ELISA kitMeiMianCat. No: MM-49493M1Mouse LTA IgG ELISA kitMeiMianCat. No: MM-49494M1Mouse SasA IgG ELISA kitMeiMianCat. No: MM-49496M1Nisin ELISA KitMeiMianCat. No: MM-1051Z2Deposited dataProteomics dataThis paperiProX: PXD076052Metabolomics dataThis paperMetaboLights: MTBLS14091Experimental models: Cell linesMouse: DCsThis paperN/AMouse: macrophagesThis paperN/AMouse: mBMSCsThis paperN/AExperimental models: Organisms/strainsC57BL/6 miceLaboratory Animal Center of Huazhong University of Science and TechnologyEight-week-old, male, SPF-gradeOligonucleotidesPrimers for real-time PCR analysis, see Table S1This paperN/ASoftware and algorithmsmageJNIHV2.0.0FlowJoBD BiosciencesV10.4.0GraphPad PrismGraphPadV10 Experimental model and study participant details Animals Eight-week-old male C57BL/6 mice were obtained from the Laboratory Animal Center of Huazhong University of Science and Technology and maintained in the specific pathogen–free barrier facility under 12:12 h light:dark cycle at 21°C. All rat experiments were approved by the Institutional Animal Care and Use Committee at Huazhong University of Science and Technology (protocol number IACUC-2024-4622). Method details Culture and expansion of probiotic strains Strains used: Escherichia coli Nissle 1917 (EcN), Lactococcus lactis (Lac, ATCC 19435), Lactobacillus rhamnosus (LGG, ATCC 7469), Bifidobacterium longum (B. longum, ATCC 15707), Akkermansia muciniphila (AKK, ATCC BAA-835), Saccharomyces cerevisiae (S. cerevisiae, ATCC 9763), and Bacillus coagulans (B. coagulans, ATCC 7050) were procured from China Center for Type Culture Collection and BeNa Culture Collection. All strain manipulations were performed within a class II biological safety cabinet (BSL-2 equivalent). For resuscitation, frozen stocks were thawed at room temperature (RT) and inoculated into their respective culture media. Cultures were incubated at 37°C with shaking at 200 rpm for 24 h. Activated cultures were stored short-term at 4°C; long-term stocks were cryopreserved at −20°C. Anaerobic strains (Lac, LGG, B. longum, AKK) were cultured in flasks sealed with Parafilm during shaking. Anaerobic conditions for these strains were maintained using anaerobic jars with gas-generating packs (Hopebio, Qingdao) during static incubation at 37°C for 24 h. Probiotic outer membrane vesicle isolation and quantification Cultures of the seven probiotics were harvested when the OD600 reached 1.5. Supernatants were collected via sequential centrifugation: (A) 300 × g for 10 min (4°C); (B) 2,000 × g for 20 min (4°C); (C) 12,000 × g for 60 min (4°C). The resulting supernatant was filtered through a 0.45 μm membrane to remove debris. OMVs were pelleted by ultracentrifugation: (A) 150,000 × g for 90 min (4°C); the supernatant was discarded. (B) The pellet was washed by resuspension in sterile PBS and centrifuged again at 150,000 × g for 90 min (4°C). The final pellet was resuspended in 500 μL PBS and was filtered through a 0.22 μm membrane to remove potential endotoxin or contamination. Protein concentration was determined using the BCA assay (Beyotime) to indirectly quantify OMVs, and particle size/concentration were analyzed by Nanoparticle Tracking Analysis (NTA). Isolation and culture of mouse primary bone marrow-derived cells All animal experiments were carried out in compliance with the guidelines and approved by the Institutional Animal Care and Use Committee at Huazhong University of Science and Technology (protocol number IACUC-2024-4622). Under sterile conditions, femurs and tibiae were dissected from 5-6-week-old C57BL/6 mice euthanized by cervical dislocation. Bone marrow was flushed into tubes via centrifugation (10,000 × g, 15 s). After washing with cold PBS supplemented with 2% FBS Red blood cells were lysed using ACK lysis buffer (Solarbio) for 5 min at RT. (i) mBMSCs: Washed cells were resuspended in complete growth medium (DMEM/F12, 10% MSC-qualified FBS, 1% Pen/Strep) and plated. Non-adherent cells were removed after 48 h; adherent mBMSCs were expanded in a humidified incubator (37°C, 5% CO2); (ii) DCs and macrophages: Washed cells were resuspended in RPMI-1640 complete medium (10% FBS, 1% Pen/Strep). Cells were plated at 2×105 cells/mL in 100 mm dishes (approximately cells from one mouse hindlimb per dish) supplemented with 20 ng/mL recombinant murine GM-CSF (for DCs) or M-CSF (for macrophages). Medium was refreshed on day 3. Cells were harvested on day 6–7 for experiments. Western--blotting analysis Treated cells were lysed with RIPA (Boster, China) supplemented with 1% PMSF (Boster, China) for 30 min on the ice, centrifuged at 12000 rpm for 15 min, and the protein content of supernatant was measured using a BCA protein assay kit (Boster, China). Proteins (30–40 mg) were separated on a 7.5–12.5% SDS-PAGE and moved onto a polyvinylidene difluoride (PVDF) membrane (Millipore, USA). The membrane was blocked with 5% milk and exposed to primary antibodies at 4°C overnight. Following two washes with Tris-buffered saline (TBS) containing 0.1% Tween 20 and incubation with secondary antibodies, proteins were exposed to chemiluminescence reagents (Affinity, China) and imaged by ChemiDoc MP Imaging System (Bio-Rad, USA). Immunofluorescence staining Cells grown on coverslips were washed with PBS, fixed with 4% paraformaldehyde (RT, 15 min), permeabilized with 0.5% Triton X-100 (RT, 20 min), and blocked with immunofluorescence blocking buffer (Beyotime) (RT, 1–2 h). Primary antibodies, diluted in antibody dilution buffer (Beyotime) according to manufacturer’s instructions, were applied and incubated overnight at 4°C. After washing with PBST, fluorophore-conjugated secondary antibodies were applied and incubated for 2 h at RT (light-protected). Coverslips were washed, counterstained with DAPI (Solarbio), mounted using anti-fade mounting medium (Beyotime), and imaged using a confocal laser scanning microscope. OMVs morphology and size distribution analysis OMVs suspension (10 μL) was adsorbed onto a carbon-coated copper grid (200 mesh) for 1 min. Excess liquid was blotted. The grid was negatively stained with 2% uranyl acetate (10 μL) for 1 min, blotted dry, and air-dried. Grids were imaged using TEM. OMVs samples were diluted appropriately in sterile PBS to achieve optimal particle concentration (20–100 particles per frame). Size distribution and concentration were analyzed using NTA analyzer. siRNA-mediated gene knockdown Target siRNA and negative control siRNA were diluted to a working concentration of 100 nM in Opti-MEM Reduced Serum Medium. Lipofectamine 2000 transfection reagent (1 μL) was diluted in 50 μL Opti-MEM and incubated for 5 min at RT. Diluted siRNA was combined with the diluted Lipofectamine 2000 (1:1 ratio), mixed gently, and incubated for 20 min at RT to form transfection complexes. The complex mixture was added dropwise to cells. Plates were gently rocked and incubated at 37°C, 5% CO2. Medium was replaced with complete medium 6–8 h post-transfection. Knockdown efficiency was assessed 48 h later by qRT-PCR. CypHer5E-labeled MRSA particle phagocytosis Assay The cultured MRSA suspension was inactivated by heating at 60°C for 60–90 min. The bacterial particles were pelleted by centrifugation and resuspended in 1 mL PBS. CypHer5E dye was added to a final concentration of 10 μM, followed by incubation at 37°C in the dark for 30 min. The labeled MRSA particles were then co-cultured with macrophages for 1 h. Fluorescence Microscopy: Cell climbing slices were collected, fixed, and mounted with an anti-fade mounting medium containing DAPI. CypHer5E fluorescence signals were visualized using a fluorescence microscope (excitation wavelength: 488 nm; emission wavelength: 610 nm). Flow Cytometric Quantification: Cells were detached using trypsin, washed with PBS, and resuspended. The CypHer5E fluorescence signal (detected in the APC channel) was analyzed by flow cytometry to determine the percentage of macrophages that had phagocytosed MRSA particles and to quantify the associated fluorescence intensity. Osteogenic induction and assessment Macrophage conditioned medium (CM) collection: Primary macrophages were treated under specified conditions for 72 h. Supernatants were collected, centrifuged (300 × g, 10 min, 4°C), filtered (0.22 μm). Osteogenic differentiation of mBMSCs: mBMSCs were seeded in complete medium. At ∼80% confluence, complete medium was replaced with osteogenic induction medium (Cyagen) mixed 1:1 with the appropriate macrophage CM. Medium was changed every three days. Alkaline Phosphatase (ALP) Activity/Staining: Measured at day 7/14 using an ALP assay kit (Beyotime) and BCIP/NBT staining (Beyotime). Alizarin Red S (ARS) Staining: Mineralized nodules were stained at day 21 and quantified by dye extraction (Solarbio). Osteogenic Gene Expression: Assessed by qRT-PCR at day seven. Synthesis of alginate-calcium hydrogel and OMVs loading A 3% (w/v) sterile sodium alginate solution was prepared by dissolving 3 g sodium alginate in 100 mL deionized water under constant stirring for 4 h at RT, followed by sterile filtration (0.22 μm). OMVs were resuspended in PBS to a protein concentration of 1 mg/mL. The OMVs suspension was mixed with the 3% alginate solution at a 1:9 volume ratio and vortexed thoroughly. The alginate-OMVs mixture was. Mouse calvarial primary and recurrent bone infection model Eight-week-old male C57BL/6 mice were randomly assigned to 6 groups (n = 3/group): (A) CTRL: Sham surgery (defect, no infection/no treatment); (B) MRSA: Defect + MRSA infection (no treatment); (C) ALG: Defect + MRSA + Alginate hydrogel; (D) ALG+VAN: Defect + MRSA + Alginate hydrogel + Systemic Vancomycin (40 mg/kg, i.v.); (E) ALG@Lac-OMVs: Defect + MRSA + Lac-OMVs-loaded Alginate hydrogel; (F) ALG@NAD+-Lac-OMVs: Defect + MRSA + NAD+-Lac-OMVs-loaded Alginate hydrogel. Under anesthesia, a midline scalp incision was made, periosteum reflected, and a 3 mm full-thickness defect was created in the left parietal bone using a trephine drill (avoiding the sagittal suture). The defect was inoculated with 10 μL MRSA suspension (1 × 107 CFU/mL). The wound was closed with 6-0 sutures. On postoperative day three, the wound was reopened, the infected site debrided, and 50 μL of the respective hydrogel was implanted. The wound was resutured. For the recurrent infection model, mice received a second MRSA inoculation (10 μL, 1 × 107 CFU/mL) directly into the defect site on day 10 post-initial surgery.64 Given the invasive nature and potential risks associated with our established animal model of bone infection, we strictly adhered to the 3R principles (Replacement, Reduction, and Refinement) in animal experimentation. To minimize animal usage and suffering while ensuring experimental validity, a sample size of three biological replicates was employed for each group. Furthermore, the data obtained from these triplicate experiments consistently demonstrated statistically significant differences between the final experimental groups and both control groups (the negative control being the infection-only model and the positive control being the vancomycin-treated model). Furthermore, there were no cases of animal deaths during this experiment. Based on this clear and reproducible separation of outcomes, further expansion of animal numbers was considered unnecessary. Micro-computed tomography (μCT) analysis Calvariae were harvested, fixed in 4% paraformaldehyde, and scanned using a high-resolution μCT system (SkyScan 1276). Reconstruction was performed using NRecon software (Bruker) with standardized parameters. Bone volume fraction (BV/TV) and other morphometric parameters were analyzed within a standardized cylindrical volume of interest (VOI) centered on the defect using CTAn software (Bruker). In vivo imaging of OMVs According to the experimental design, randomly selected C57BL/6 mice were sacrificed, and major organs were collected. OMVs in vivo imaging at each time point was performed by using an FX PRO imaging system (Bruker, Germany). Histological analysis Calvarial samples were decalcified for 21 days at 4°C before embedded in paraffin and sectioned. Major organ samples were also embedded in paraffin and sectioned. Sections were stained with standard IHC staining kit, Hematoxylin & Eosin (H&E) and Masson’s Trichrome staining kit. Stained sections were imaged using a brightfield microscope. Quantification and statistical analysis In vitro biocompatibility assessment of probiotic OMVs mBMSCs were seeded in 24-well plates. After 24 h, cells were treated with OMVs diluted in fresh medium to final concentrations of 5, 10, 20, and 40 μg/mL (based on BCA protein quantification). After the designated incubation period, medium was replaced with fresh medium containing 10% CCK-8 reagent (Boster). Plates were incubated for 1 h (37°C, 5% CO2), and 100 μL of supernatant from each well was transferred to a 96-well plate. Absorbance was measured at 450 nm using a microplate reader. In vitro antibacterial activity assessment of probiotic OMVs Assays were performed in sterile 96-well plates. Reaction mixtures contained: (A) Test Group: 100 μL MRSA suspension (diluted 1:5000 in broth, ≈1 × 106 CFU/mL) + 100 μL OMVs (20 μg/mL in PBS); (B) Control Group: 100 μL MRSA suspension +100 μL sterile PBS. Plates were incubated statically at 37°C. Bacterial growth was monitored by measuring OD600 at 2, 4, 6, 8, and 10 h using a microplate reader. Flow cytometry Cells were harvested, washed with PBS, and stained with fluorochrome-conjugated antibodies against surface markers diluted in staining buffer for 60 min at 4°C (light-protected). For intracellular staining, cells were fixed, permeabilized (Fixation/permeabilization kit, BD), and stained with appropriate antibodies for 60 min at 4°C (light-protected). Cells were washed, resuspended in staining buffer, and analyzed immediately by flow cytometer. Data analysis was performed using FlowJo software (v10.8). Quantitative real-time PCR (qRT-PCR) After the cell samples were washed by PBS, an appropriate amount of Trizol lysate was added, 200 μL chloroform was added to each sample, and after vigorous shaking for 20 s, centrifuge at 4°C, 12000 × g for 30 min, aspirate the upper aqueous phase, and add isopropyl alcohol of the same volume. Centrifuge at 12,000 × g for 30 min to collect the RNA pellet, wash with 75% ethanol (prepared with DEPC water), and dissolve the RNA with 20 μL of DEPC water. Primer sequences are listed in Table S1. Gene expression was normalized to Gapdh and calculated using the 2−ΔΔCt method. Spread plate assay One gram granulation tissues were collected from the defects for bacterial examination 10 days after implantation in both primary and recurrent bone infection model. Subsequently, samples were put into 1 mL of sterile PBS and homogenized for 5 min. Then 20 μL of the solution was used for spread plate experiments, and the plates were incubated for 24 h at 37°C. Then viable bacterial counts were taken. Enzyme-linked immunosorbent assays (ELISA) Inflammatory cytokine levels, IgG subtypes and antigen specificity were measured in supernatants or serum using the standard ELISA Kit (Elabscience, China, and MeiMian, China). Statistical analysis Data are presented as mean ± standard deviation (SD) from at least three independent experiments. Statistical analysis was performed using GraphPad Prism software (v9.0). For comparisons between two groups, an unpaired two-tailed Student’s t test was used. Statistical significance was defined as p < 0.05 and indicated as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Published: June 2, 2026 Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.xcrm.2026.102842. Contributor Information Lei Tan, Email: tanlei_hubu@163.com. Cao Yang, Email: caoyangunion@hust.edu.cn. Xiaobo Feng, Email: fengxiaobo@hust.edu.cn. Supplemental information Document S1. Figures S1–S9 and Table S1 Document S2. Article plus supplemental information References Associated Data Supplementary Materials Document S1. Figures S1–S9 and Table S1 Document S2. Article plus supplemental information Data Availability Statement •The proteomics and metabolomics data have been deposited in iProX and MetaboLights, respectively. The proteomics data are available under iProX: PXD076052 (https://www.iprox.cn/), and the metabolomics data are available under MetaboLights: MTBLS14091 (https://www.ebi.ac.uk/). The proteomics and metabolomics data have been deposited in iProX and MetaboLights, respectively. The proteomics data are available under iProX: PXD076052 (https://www.iprox.cn/), and the metabolomics data are available under MetaboLights: MTBLS14091 (https://www.ebi.ac.uk/). •This study did not involve the reporting of novel or original code. This study did not involve the reporting of novel or original code. •All data essential for evaluating the conclusions drawn in this paper are presented within the paper itself and/or in the supplemental information. All data essential for evaluating the conclusions drawn in this paper are presented within the paper itself and/or in the supplemental information. •Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.