Levilactobacillus brevis 23017 (CCTCC AB 2018164) was isolated and identified by our laboratory (38, 39) and cultivated in MRS medium at 37°C. S. aureus CMCC26003 was purchased from the China Institute of Veterinary Drug Control (IVDC, Beijing, China). MRSA strain USA300 (ATCC BAA-1717) was provided by Dr. Yanhua Li at Northeast Agricultural University. TSB medium was used to culture S. aureus at 37°C. Escherichia coli pET28a-GFP/Rosetta, which fluoresces green, is maintained in our laboratory. β-glucan (CAS: 904122-9) was purchased from Sigma-Aldrich (St. Louis, MO, USA).

Female Kunming mice (4–6 weeks old, 18–22 g) were purchased from Changsheng Biological Company (Liaoning, China). All animal experiments in this study were carried out in compliance with the guidelines set by the Laboratory Animal Ethics Committee of Northeast Agricultural University (No. NEAUEC20210326). The animals were free to drink and eat under constant conditions, and the experiments were conducted after the test animals had adapted to the environment.

RAW264.7 cells were purchased from the IVDC. RAW264.7 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin at 37°C.

As previously mentioned (9), BLPs were prepared through hot acid pretreatment. 23017 strains were suspended in trichloroacetic acid (10%) and boiled for 45 min, then washed three times, and finally stored at −80°C. The particles were counted using a cell counter. The number of BLP particles per milliliter was adjusted according to the colony-forming unit/mL determined in the starting culture (1 U = 2.5 × 10⁹ inactive BLPs) (40).

Bioinformatics analysis is becoming more commonly utilized to design multi-epitope vaccines. In our earlier experiments, we predicted and constructed SAMEA (41). Comprehensive physicochemical and structural analyses confirmed SAMEA’s stability, with additional parameters such as its antigenicity, toxicity, and allergenicity aligning with the desired profiles for vaccine efficacy. It includes dominant antigenic epitopes of 10 antigens (clfa, fnbpa, idsb, pnsg, spa, sasx, hla, mntc, seb, and plc), one B cell, one CTL, and one HTL epitope identified through bioinformatic analysis, which were selected from each of the 10 immunodominant antigens (to be published). The SAMEA protein used in this study was an antigen purified using a His-tag protein purification kit (Beyotime, China). After purification, endotoxins were removed using a Protein Endotoxin Removal Kit (Beyotime, China). Subsequently, an endotoxin detection kit was used to measure the LPS content (Beyotime, China).

For immunization, mice were assigned to six groups (n = 5). Both the control and infection groups received PBS through intraperitoneal injections (i.p.) on day 0 and day 21; the SAMEA group received SAMEA (50 μg, i.p.) on day 0 and day 21; the SAMEA+ALU group received SAMEA (50 μg) and an equal volume of alum adjuvant on day 0 and day 21 (i.p.); the SAMEA+β-glucan group received SAMEA (50 μg) and β-glucan (500 μg) on day 0 and day 21 (i.p.); the SAMEA+BLP group received SAMEA (50 μg, i.p.) and oral administration with BLP (1 U) on day 0 and day 21. On days 28, 35, and 42, the serum was then extracted from the tail vein for subsequent analysis. On day 42, mice in all groups except the control group were injected with MRSA (5×10⁸ CFU/100 μL) in the tail vein. Mouse body weight and survival were monitored until day 63. Then, the mice were euthanized by cervical dislocation under anesthesia with 4% isoflurane, with relevant samples collected for subsequent analysis.

Experiments were designed according to the methodology of Ferreira et al. (42). For one dose immunization, mice were assigned to six groups (n = 5), and both the control and infection groups received PBS on day −3 and day 0 (i.p.); the SAMEA group received PBS on day −3 and SAMEA (50 μg) on day 0 (i.p.); the SAMEA+ALU group received PBS on day −3 and received SAMEA (50 μg) and an equal volume of alum adjuvant on day 0 (i.p.); the SAMEA+β-glucan group received β-glucan (500 μg) on day −3 and day 0 and SAMEA (50 μg) on day 0 (i.p.); the SAMEA+BLP group was oral administrated with BLP (1 U) on day −3 and day 0, and intraperitoneally injected with SAMEA (50 μg) on day 0. On days 7, 14, and 21, the serum was then extracted for subsequent analysis. On day 20, the inflammatory response in mice had completely subsided. On day 21, mice in all groups except the control group were injected with MRSA (5×10⁸ CFU/100 μL) in the tail vein. Mouse body weight and survival were monitored until day 42. Then, the mice were euthanized by cervical dislocation under anesthesia with 4% isoflurane, with relevant samples collected for subsequent analysis.

The methodology was based on Paris et al. (43); RAW264.7 cells were stimulated with stimulants. The cells were cultured for 2 h to allow them to adhere to the wall. Initial stimulation of RAW264.7 cells was performed by adding PBS, β-glucan (30 µg/mL, a dose that effectively induced trained immunity in previous studies) (44), and BLP (0.01 U/well). Remove stimulus after 24 h. On day 7, cells in each group were stimulated for 24 h with LPS (100 ng/mL) for secondary stimulation. Cells were harvested in Trizol, and culture supernatants were collected in sterile EP tubes for subsequent experiments.

Inhibition experiments were performed according to the method in Paris et al. (43). We used 1 mM of 5’-deoxy-5’-(methylthio) adenosine (MTA) (S880063, Macklin, Shanghai, China) to inhibit the histone methylation. To inhibit glycolysis, macrophages were pre-incubated with 15 mM of 2-deoxy-D-glucose (2-DG, D807272, Macklin, Shanghai, China) for 1 h. Also, 10 μmol/L rapamycin (RAPA, S115842, Aladdin, Shanghai, China) was used to inhibit the mTOR signaling pathway to explore whether this pathway plays a role in trained immunity induced by BLP.

The methodology was based on Ferreira et al. (42); mice were assigned to four groups (n = 5). The control group and infection group were intraperitoneally injected with PBS on day 0 and day 3; the β-glucan group was intraperitoneally injected with β-glucan (500 μg) on day 0 and day 3; the BLP group was orally administered with BLP (1 U) on day 0 and day 3. On day 7, mice in all groups except the control group were intraperitoneally injected with S. aureus CMCC26003 (1×10⁹ CFU/200 µL). Normal feeding was observed until day 5, monitored, and recorded for body weight, survival rate, and activity score. Then, the mice were euthanized by cervical dislocation under anesthesia with 4% isoflurane.

Mouse activity was scored for damage based on a simple modification of the methodology of Pasquali et al. (45), with healthy mice scoring 0; mice showing slightly abnormal symptoms of wilting scoring 1–2; mice arching their backs, constricting their abdomens, and being inactive for some time or blowing up their fur scoring 3–5; mice being almost inactive, with ragged coats, bunching up, and becoming moribund scoring 6–9; and mice dying scoring 10.

The methodology was based on Pan et al. (44); mice were assigned to three groups (n = 5). The control group was intraperitoneally injected with PBS on day 0 and day 3; the β-glucan group was intraperitoneally injected with β-glucan (500 μg) on day 0 and day 3; the BLP group was orally administered with BLP (1 U) on day 0 and day 3. On day 6, mice were euthanized by cervical dislocation under anesthesia with 4% isoflurane. Peritoneal macrophages were isolated and cultured for 2 h before being stimulated with LPS (100 ng/mL). Peritoneal macrophages were collected at 1, 6, 12, and 24 h post-LPS stimulation for subsequent analysis.

Mouse peritoneal macrophages stimulated with LPS for 6 h were subjected to RNA-seq analysis by NovelBio Bio Pharm Technology Co., Ltd. (Shanghai, China). Total RNA was extracted with Trizol Reagent (Invitrogen, Waltham, MA, USA). Briefly, check the total RNA quality with an Agilent 2200 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA with RIN (RNA Integrity Number) greater than 7.0 can be used to construct cDNA libraries. The Agencourt AMPure XP-PCR Purification Beads (Beckman Coulter, Brea, CA, USA) were used to purify and fragment the RNA. The libraries were quality-controlled with an Agilent 2200 and sequenced by DNBSEQ-T7 on a 150-bp paired-end run. The clean reads were then aligned to the Mouse genome (mm10) using Hisat2 (46). HTseq fields (47) were used to obtain gene counts, and the RPKM method was employed to determine gene expression. We applied the DESeq2 algorithm (48) to filter the differentially expressed genes (DEGs). After the significant analysis, p-value and FDR analysis were subjected to the following criteria: fold change > 1.5 or <0.667, p-value < 0.05, and FDR < 0.05.

We obtained Gene Ontology (GO) annotations from the GO website (http://www.geneontology.org/). Pathway analysis was conducted based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Gene set enrichment analyses (GSEAs) were performed using GSEA software v4.1.0 from the Broad Institute, utilizing the 50 Hallmark gene sets of MSigDB as input (49). Fisher’s exact test was employed to select significant GO categories and pathways, with a significance threshold set at p < 0.05 (50).

To detect cytokines, the method described in Shi et al. (39) was followed. Total RNA was extracted (ER101-01, TransGen Biotech, Beijing, China) and reverse transcribed to cDNA (TRT-101, TOYOBO, Shanghai, China), followed by quantification and gene expression analysis (A2250B, HaiGene Biotech Co., Ltd., Harbin, China), with Actin β as the reference gene. The primers were listed in Supplementary Table S1 as per references (39, 44, 51–56) (Comate Bioscience Co., Ltd., Changchun, Jilin, China). The relative mRNA expression levels were determined using the 2^−ΔΔCT method (57).

The bacterial load was determined using the method described by Nour et al. (58). The tissues of the mouse in each group were aseptically collected, homogenized in PBS, diluted up to 10⁻⁸ times the original dilution, and plated out on TSA plates, where the colonies were counted after 18 h at 37°C.

Under aseptic conditions, the spleens and kidneys of mice were harvested, fixed in 10% formaldehyde, labeled, and sent to Wuhan Servicebio Technology Co., Ltd. (Wuhan, Hubei, China) for pathological section preparation.

Detection of antibody levels was according to the method of Niu et al. (9); indirect ELISA was employed to assess the levels of SAMEA-specific antibodies IgG, IgG1, and IgG2a. In brief, purified SAMEA antigen (2 µg/mL) was added to a 96-well plate overnight for coating. It was subsequently blocked with 5% skim milk, serum samples were diluted, and diluted antibodies IgG (HRP-conjugated Goat anti-Mouse IgG, AS003, Abclonal, Wuhan, China), IgG1 (HRP-conjugated Goat anti-Mouse IgG1, AS066, Abclonal, Wuhan, China), and IgG2a (HRP-conjugated Goat anti-Mouse IgG2a, AS065, Abclonal, Wuhan, China) were added with 5% skim milk. The reaction used TMB substrate (PR1210, Solarbio, Beijing, China). H₂SO₄ (2 M) was used to stop the reaction, which was then measured at 450 nm using a Thermo Fisher spectrophotometer (New York, NY, USA).

The BLP was tested to determine whether it induced changes in phagocytosis of Raw264.7 macrophages according to the method in Pan et al. (44), such as RAW264.7 macrophage in vitro model treatment of the cells, followed by static incubation of the macrophages overnight, and then co-cultivated with GFP-labeled E. coli (E. coli strains that fluoresce green) for 1 h. After washing away the unphagocytosed bacteria, a mixture of double antibiotics, gentamicin and ampicillin, was added to eliminate them from the macrophages. While the cells were later stained using DAPI dye (the nucleus fluoresces blue after staining), the phagocytosis ability of each group of Raw264.7 cells was subsequently observed using fluorescence microscopy.

Lactic acid (LA) and glucose (GLU) levels in the cell supernatants and serum were determined by using the LA and GLU assay kits (Nanjing Jiancheng, Nanjing, China), respectively.

Nitric oxide (NO) was detected according to the kit (Nanjing Jiancheng, Nanjing, China) instructions; reactive oxygen species (ROS) was detected according to the kit (Nanjing Jiancheng, Nanjing, China) instructions, and its luminescence intensity was detected by a fluorescent enzyme marker.

All data were statistically evaluated using GraphPad Prism 9 software. One-way ANOVA was utilized to determine the significance of differences between groups. Survival analysis was conducted with the Kaplan–Meier method, and survival rate differences were examined using the log-rank test. Results are presented as mean ± SD. Statistical significance was defined as p < 0.05 for significant differences and p < 0.01 for highly significant differences, with **** indicating p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.

To enhance the subunit vaccine SAMEA and improve its protective role against MRSA, we co-immunized with BLP in combination with SAMEA. Mice were immunized with a two-dose SAMEA combined with BLP on days 0 and 21; MRSA challenge via intravenous injection was performed on day 42 after the first immunization ( Figure 1A ), and we monitored antibody levels at regular intervals. The SAMEA-specific IgG antibody titers after two immunizations with or without the combination of ALU, β-glucan, and BLP are displayed as endpoint dilution titer in Figure 1B . The results showed that BLP significantly elevated SAMEA-specific IgG antibody levels at all three monitored time points (p < 0.01) ( Figure 1B ). When IgG2a/IgG1 > 1, the response is biased toward a Th1 cell immune response; the opposite is true for Th2 (59). Further subtype analysis showed that compared to the SAMEA group, the levels of IgG1 in the SAMEA+ALU group were increased, and the ratio of IgG2a/IgG1 was decreased. The ratio of IgG2a/IgG1 in the SAMEA+BLP group is increased ( Figure 1C ). The above data show that BLP induced higher levels of IgG antibodies and a more balanced Th response.

Subsequently, we evaluated whether BLP enhanced the vaccine’s protective effect against MRSA. We monitored the status of the mice 21 days after infection. The survival rates of mice in the SAMEA+BLP group and the SAMEA+ALU group were increased from 60% to 80% compared to the SAMEA group ( Figure 1D ). Changes in mouse body weight showed that mice in the BLP group had the slowest weight loss and recovered more quickly from day 0 after the MRSA infection ( Figure 1E ). Bacterial loads in mouse kidneys revealed that SAMEA+ALU, SAMEA+β-glucan, and SAMEA+BLP significantly reduced bacterial loads (p < 0.0001) compared to the SAMEA group ( Figure 1F ). Kidney histopathological sections also showed that, relative to the SAMEA alone group, BLP significantly reduced pathological damage and renal tubular vacuolization, which was attributed to the increased renal microsomes in the kidney caused by MRSA infection ( Figure 1G ).

Furthermore, we analyzed major cytokine and TLR transcript levels in mouse kidneys, finding that in combination with BLP, mouse kidney tissues exhibited superior inflammation resolution compared to the SAMEA group. Specifically, compared to the SAMEA group, we observed significantly lower transcript levels of IL-6, IL-1β, TNF-α, TGF-β, CCL2, and CCL7 and the highest transcript levels of IL-10 and IFN-β in the SAMEA+BLP group. TLR2 expression was significantly upregulated in all three groups compared to the SAMEA group, with the highest levels in the SAMEA+BLP group. Similarly, the SAMEA+BLP group showed the highest HIF-1α transcript levels in mouse kidneys ( Figure 1H , Supplementary Figures S1 , S2 ). To initially assess whether BLP induced changes in glycolysis levels, we examined serum levels of GLU and LA. The results show that the SAMEA+BLP group serum GLU and LA levels changed, but not to a significant degree ( Figure 1I ). All in all, BLP significantly improves vaccine immunization efficacy, elicits a more rapid organismal response to MRSA infection.

We next established a one-dose immunization protocol with SAMEA to evaluate the adjuvant potential of BLP ( Figure 2A ). We also assessed whether a single dose of BLP could enhance antibody production. Dynamic monitoring revealed that the SAMEA+BLP group showed elevated IgG and IgG2a levels compared to the SAMEA group, with a significant difference observed only on day 7 (p < 0.001) ( Figure 2B ). Furthermore, the IgG2a/IgG1 ratio indicated a Th1-biased response in the SAMEA+BLP group, while the SAMEA+ALU group showed a Th2-biased response ( Figure 2C ).

Following intravenous MRSA challenge on day 21, mice were monitored for 21 days. The SAMEA+BLP group showed an improved survival rate of 60%, compared to 40% in the SAMEA group, and a similar rate to the SAMEA+ALU group ( Figure 2D ). Mice in the SAMEA+ALU (p < 0.01), SAMEA+β-glucan (p < 0.05), and SAMEA+BLP (p < 0.01) groups exhibited reduced body weight loss and better final weight retention compared to the SAMEA group. Notably, the SAMEA+BLP group showed less weight loss than the SAMEA+ALU group and recovered more rapidly ( Figure 2E ). Bacterial load analysis revealed that the SAMEA+BLP group had significantly lower kidney bacterial burdens than the SAMEA, SAMEA+ALU, and SAMEA+β-glucan groups ( Figure 2F ).

We further analyzed immune-related cytokines and TLR expression in mouse kidneys by qPCR at day 21 post-MRSA challenge. Compared to the SAMEA group, the SAMEA+BLP group showed significantly lower transcript levels of IL-6, IL-1β, CCL2, and CCL7, and significantly higher levels of IL-10 and IFN-β. TLR2 expression was significantly upregulated in the SAMEA+BLP group, while TLR4 expression remained unchanged ( Figure 2G , Supplementary Figures S3 , S4 ). In addition, serum GLU and LA levels showed no notable changes in this group ( Figure 2H ).

To assess whether BLP induces trained immunity in RAW264.7 cells, we established a trained immunity model following established protocols. Cells were first exposed to β-glucan or BLP, then rested and subsequently stimulated with LPS to evaluate macrophage responses ( Figure 3A ). Trained immunity is characterized by a memory-like phenotype and enhanced responses upon secondary stimulation. Thus, key cytokines associated with trained immunity (IL-6, IL-1β, and TNF-α) were measured. Following BLP treatment (day 1), IL-6 and IL-1β levels were significantly elevated, returned to baseline by day 7, and were again significantly upregulated upon LPS stimulation on day 8 (IL-6: p < 0.05; IL-1β: p < 0.0001) ( Figure 3B ). In contrast, TNF-α showed no significant change after primary stimulation and was reduced following secondary stimulation. To better characterize the inflammatory profile, we expanded the cytokine panel ( Figure 3C , Supplementary Figure S5 ) and observed significantly increased transcript levels of IL-10, IL-12, IFN-β, TGF-β, TLR2, CCL2, and CCL7 in the BLP group compared to controls on day 8. These results suggest that BLP-induced trained immunity is associated with enhanced and diversified cytokine responses upon secondary challenge.

Previous studies have shown that epigenetic and metabolic reprogramming of macrophages through the Akt-mTOR-HIF-1α pathway is crucial for trained immunity (60). To explore the phenotypes of BLP-induced trained immunity, we assessed the transcript levels of HDAC7, mTORC2, and HIF-1α, along with GLU consumption and LA production. After secondary stimulation, the BLP group exhibited significantly higher expression of HDAC7, mTORC2, and HIF-1α (p < 0.01) compared to controls ( Figure 3D ). Additionally, BLP significantly increased GLU consumption (p < 0.05) and LA production (p < 0.01) ( Figure 3E ). These findings suggest that BLP-induced trained immunity may involve metabolic reprogramming.

Macrophage function is highly correlated with its ability to fight infection, and the proinflammatory profile of activated macrophages is usually associated with antimicrobial purposes (61). We investigated the activation status of macrophages after trained immunity by measuring their phagocytic activity, ROS, and NO levels. Immunofluorescence images showed more E. coli in BLP-induced macrophages ( Figure 3F ) than in the control group, showing that the phagocytic capacity was significantly increased. Meanwhile, the BLP-induced trained immunity exhibited decreased ROS levels and increased NO levels (p < 0.001) ( Figures 3G, H ). The above data suggest that BLP-induced trained immunity may be mediated through modulation of cell function.

To confirm the roles of metabolic and epigenetic reprogramming in BLP-induced trained immunity, macrophages were pretreated with the glycolysis inhibitor 2-DG, the histone methylation inhibitor MTA, and the mTOR inhibitor rapamycin (RAPA) ( Figure 3I ). BLP stimulation elevated IL-6 and IL-1β levels in RAW264.7 cells, but these responses were attenuated to varying degrees by all three inhibitors ( Figure 3J ). These results indicate that BLP-induced trained immunity is mediated through glycolysis, epigenetic regulation, and mTOR-dependent pathways.

Because BLP can play a role in inducing RAW264.7 cells to produce trained immunity, we hypothesized that the trained immunity induced by BLP may also play a role in mice. Accordingly, mice trained with or without BLP and β-glucan were intraperitoneally injected with S. aureus CMCC26003 to determine their survival ( Figure 4A ). The results showed that trained immunity induced by BLP increased the survival rate of wild-type mice from 20% to 100% ( Figure 4B ). BLP is highly protective against weight loss and increased injury in mice. Mice in the BLP group lost less body weight and had a faster rate of recovery relative to the infected group (p > 0.05) ( Figure 4C ). The injury score of BLP-treated mice was less than the infected mice and β-glucan treatment mice (p < 0.0001) ( Figure 4D ). The above data suggest that trained immunity induced by BLP in mice increases survival and decreases body weight changes and body damage.

The mice trained with BLP also showed lower bacterial burdens in the kidneys, spleens, and livers (p < 0.0001) compared to the infection group ( Figure 4E ). Spleen and kidney injuries were reduced in the BLP group of mice compared with the infected group. The red and white marrow boundaries were clear, and no pathological changes such as megakaryocytic hyperplasia were seen ( Figure 4F ). Cytokine transcript analyses in the spleens and kidneys revealed a diminished inflammatory response in mice pretreated with BLP. Specifically, spleen cytokine levels, including IL-6, IL-1β, TNF-α, IFN-β, TGF-β, CCL7, and TLR2, were reduced in BLP groups relative to the infected group ( Figure 4G ). Kidney cytokine levels in the BLP group were lower than those in the infected group, whereas levels of CCL2, CCL7, and TNF-α in the β-glucan group tended to be higher than those in the infected group ( Figure 4H , Supplementary Figures S6 , S7 ). The above data suggest that trained immunity induced by BLP in mice can reduce tissue bacterial load, pathological injury, and inflammatory response in mice.

Additionally, in spleen tissue, compared to the control group, the levels of HDAC7 (p < 0.05) and HIF-1α (p < 0.001) were notably lower in BLP-treated mice, but the levels of mTORC2 showed no significant changes (p > 0.05) ( Figure 4I ). In the kidneys, the levels of HIF-1α were significantly reduced in the BLP group compared to the infected group (p < 0.01). Serum GLU shows a downward trend while LA shows an upward trend, but the difference is not significant (p > 0.05) ( Figure 4J ). Overall, pretreatment of mice with BLP provided effective protection against S. aureus infection.

Both in vitro and in vivo experiments have shown that BLP immunization could induce increased proinflammatory cytokines and chemokines after the secondary challenge against heterologous pathogens. We hypothesized that the trained immunity induced by BLP may also mount a faster and greater response in mouse peritoneal macrophages. For this purpose, the peritoneal macrophages of mice trained with or without BLP and β-glucan were extracted and re-stimulated for 24 h in vitro with LPS ( Figure 5A ). We then monitored the transcripts of typical trained immunity markers under continuous LPS stimulation for 1, 6, 12, and 24 h. Here, BLP immunization induced elevated proinflammatory cytokines of IL-6, IL-1β, and TNF-α release rapidly compared to that of the control in peritoneal macrophages of mice in 6 h after ( Figure 5B ). The significantly elevated expression of HDAC7, mTORC2, and HIF-1α persisted in the BLP group in 1, 6, and 12 h (p < 0.05) ( Figure 5C ). Analysis of GLU and LA concentrations in the cell supernatant showed that the decreased levels of GLU and elevated levels of LA were consistently elevated in the β-glucan and BLP groups. Moreover, the decreased levels of GLU and the elevated levels of LA were significantly higher inside BLP-treated cells than the control group in 24 h ( Figure 5D ). In summary, after induction with BLP, LPS stimulation elicits a rapid response in macrophages, characterized by increased secretion of proinflammatory cytokines, peaking at 6 h. Meanwhile, BLP-induced trained immunity exhibits enhanced levels of epigenetic and metabolic-related factors and glycolysis.

Based on the findings that BLP induced trained immunity and provided heterogeneous protection, we collected macrophages from each group that had been continuously stimulated with LPS for 6 h and conducted RNA-seq analyses to elucidate the underlying mechanisms. Heatmap analysis revealed distinct gene expression profiles for each group ( Figure 6A ). Using DESeq2, we identified 1,960 DEGs in the BLP group relative to the control group, comprising 927 upregulated and 1,023 downregulated genes ( Figure 6B ). At the same time, we found changes in epigenetically related upregulated genes such as DNA (cytosine-5)-methyltransferase genes (Dnmt1, Dnmt3c, and Dnmt3l), histone deacetylase (Hdac7 and Hdac5), lysine (K)-specific demethylase 7A (kdm7a), and K (lysine) acetyltransferase 2B (kat2b) ( Figure 6B ). The above data suggest that BLP-induced trained immunity may have influenced epigenetic changes, but the exact role remains unclear. After BLP training, the inflammatory cytokines (IL22, Il4, etc.), chemokines (Ccl2, Cxcl3, Cxcl5, etc.), PRRs (TLR1-4, Nod1, Nod2, etc.), and metabolism-related factors (Hif-1α, Arg1, Acod1, etc.) were elevated after secondary stimulation ( Figure 6C ).

GO analyses indicated that BLP upregulated biological processes associated with immunity, such as immune system process, positive regulation of cytokine production involved in immune response, and innate immune response. Also, processes related to phagocytosis, such as phagocytosis, positive regulation of phagocytosis, positive regulation of xenophagy, and endocytosis, are upregulated ( Figure 6D ). Related genetic changes such as MAPK3, CAV, and CD163 are shown in Figure 6E . BLP-induced trained immunity significantly elevated macrophage phagocytosis and altered macrophage function, similar to the results we obtained in RAW264.7 cells. There was also a significant enrichment of upregulated genes associated with antigen presentation, such as Icam1, Malt1, and Mdk ( Figure 6F ). The results showed that BLP-induced trained immunity significantly activated the innate immune response and modulated cytokines to fight against exogenous infections. This also suggests that our BLP-induced trained immunity may enhance the immunization effect of the vaccine.

KEGG analyses indicated that a large number of genes are enriched in metabolic pathways, including the process of glycolysis, as well as amino acid metabolism and fatty acid metabolism. The enrichment of genes in the glycolytic pathway also corroborates our previous findings ( Figure 6G ). Next, we carried out a comparative transcription factor (TF) enrichment analysis of upregulated DEGs in the BLP and control groups using the HOMER package. Motif analysis was conducted to explore which TFs are involved in regulating the epigenetic modifications associated with BLP training. As shown in Figure 6H , the Sfpi1 motif was found as a regulator of 40.62% of upregulated genes for the first rank genes. It is a function including histone deacetylase binding activity and regulation of myeloid leukocyte-mediated immunity. On the other hand, motifs associated with KLF1 contributed to 35.99% of upregulated genes; they act upstream of or within chromatin remodeling. GSEA using the Hallmark gene set collection from the Molecular Signatures Database (MSigDB) indicated that BLP training leads to the upregulation of various genes related to Inflammatory Response, IL-6-JAK-STAT3 Signaling, Cholesterol Homeostasis, Xenobiotic Metabolism, and Interferon-gamma Response ( Figure 6I , Supplementary Figure S8 ). Based on the above data, we hypothesized that BLP may induce trained immunity by activating the JAK-STAT3 pathway through the activation of the TLR2 receptor secretion of IL-6, which regulates HIF-1α and subsequently modulates macrophage metabolic reprogramming and cellular function.