A glycopeptide hydrogel confers protection and treatment in sepsis via recruitment and training of macrophages

TL;DR

A new injectable hydrogel recruits and trains macrophages to treat sepsis and prevent future infections.

Contribution

This is the first β-glucan delivery system using self-assembling peptides to combat sepsis-related immunoparalysis.

Findings

The hydrogel reduces organ damage and sepsis mortality by activating innate and adaptive immune pathways.

It provides long-term protection against secondary infections by training macrophages with immune memory.

The system clears pathogens and balances immune responses in sepsis.

Abstract

Sepsis is a critical global health challenge, characterized by dysregulated host responses and high mortality rates. Current treatment therapies inadequately address persistent immunoparalysis and adverse reactions, leaving survivors susceptible to recurrent infections. Herein, we developed the first β-glucan (BG) delivery system using self-assembling peptide hydrogels to combat immuneparalysis associated with sepsis. This hydrogel combines oxidized BG (BGA) and tuftsin-functionalized self-assembling peptides (t-RADA16) to create an injectable glycopeptide hydrogel (BGA@t-RADA16). Upon subcutaneous injection, the hydrogel forms a dual-functional platform consisting of a “sustained-release depot” of BGA and a local “trained immunity center”, where the three-dimensional network of peptide nanofibers actively recruits macrophages and the BGA educates them in-situ. This endows the macrophages with an immune memory function, generating a rapid and powerful defense against secondary infection, and forming a “targeted-sustained-response” cascade treatment system. The results demonstrate that the glycopeptide hydrogel reduces organ damage, lowers sepsis mortality and confers protection against secondary infections by coordinating the activation of both innate and adaptive immune pathways. Furthermore, BGA-P@t-RADA16 simultaneously clears acute-phase pathogens and provides long-term immune training by loading polymyxin B within a single system, which truly addresses the problem of “immune imbalance” in the pathological process of sepsis. BGA@t-RADA16 provides a new tool for sepsis immunotherapy, pioneering an integrated strategy of “immune training, acute intervention and long-term defense”. This strategy is expected to be an important breakthrough in the treatment of sepsis and related immune disorders.

Image 1

Figures

Preparation and characterization of hydrogel BGA@t-RADA16. (A) Schematic diagram of preparation of BGA@t-RADA16. (B) Representative pictures of BGA@t-RADA16 manufacturing process. (C) Representative SEM images of BGA, t-RADA16 and BGA@t-RADA16. (D) Representative TEM images of BGA, t-RADA16 and BGA@t-RADA16. Rheology of BGA@t-RADA16 as a function of angular frequency (E) CD spectrum of BGA@t-RADA16. (F, G) Rheological analysis of BGA@t-RADA16 as a function of angular frequency (F) and shear strain (G). (H) The self-healing analysis of BGA@t-RADA16 with an alternative large oscillation force (5

BGA@t-RADA16 mediates trained immunity of BMDMs *in vitro*. (A) Schematic of *in vitro* trained immunity experimental setup. The production of (B) IL-1β (C) IL-6 and (D) TNF-α by BMDMs trained with BG and BGA was measured by ELISA after 4 h of LPS treatment. (E, F) Determination of M1 type macrophages (CD11b + F4/80+ CD80^+^ subgroup) in each group by flow cytometry. The production of (G) IL-1β (H) IL-6 and (I) TNF-α by BMDMs trained with BGA@t-RADA16 was measured by ELISA after 4 h of LPS treatment. (J, K) Determination of M1 type macrophages (CD11b^+^ F4/80^+^ CD80^+^ subgroup) in each group

Transcriptomic and metabolomic characteristics of BMDMs after BGA@t-RADA16 training. (A) Differentially expressed genes (DEGs) in control group vs. BGA@t-RADA16 group. (B) DEGs summarized between the BGA@t-RADA16-trained group, the BGA-trained group and the control group. (C) Heatmap comparison of GO enrichment analysis results (*p < *0.05) among BGA@t-RADA16-trained groups. (D) Heatmap comparison of gene set enrichment analysis (GSEA) results (*p < *0.05) among BGA@t-RADA16-trained groups. NES, Normalized enrichment score. (E) Volcano map of the DMs. (F) PCA plot of the qualitative metabolite

Effect of BGA@t-RADA16 on recruitment and activation of macrophages at injection sites. (A) Image and (B) relative FITC fluorescence intensity at different time points after subcutaneous injection of FITC-labeled BGA@t-RADA16 or FITC-labeled BGA in C57BL/6 mice. (C) H&E staining results of injection sites in C57BL/6 mice 3 days after subcutaneous injection of BGA@t-RADA16 or BGA. Proportion of (D, E) macrophages, (F, G) CCR2^+^ macrophages and (H, I) M1 macrophages at the injection site 24 h or 72 h after subcutaneous injection of PBS, BGA or BGA@t-RADA. Data are shown as mean ± SEM (n = 3). n

Effect of BGA@t-RADA16 on preventing sepsis in immunoparalysis mice. (A) Schematic diagram of establishing an *in vivo* immunoparalysis model and inducing trained immunity. Proportion of peritoneal (B, C) monocytes, (D, E) macrophages, (F, G) M1 macrophages and (H, I) M2 macrophages after different treatments. The content of (J) IL-1β, (K) IL-6 and (L) TNF-α in serum was measured by ELISA after different treatment. (M) Representative images of H&E staining of the liver and lungs of mice after different treatment. In the PBS, *E.coli*, t-RADA16 and BGA groups, inflammatory cell infiltration and

Therapeutic efficacy of BGA-P@t-RADA16 against CLP-induced polymicrobial sepsis. (A) Schematic of *in vitro* trained immunity experimental setup. The production of (B) IL-6 and (C) TNF-α by RAW264.7 cells after 24 h of BGA-P@t-RADA16 and LPS treatment. (D, E) Determination of M1 type macrophages in each group by flow cytometry after 4 h of secondary LPS stimuli. (F) Schematic diagram of establishing the CLP model and treatment. (G) Survival curves of mice in each group after treatment with BGA-P@t-RADA16, HC + PMB or PBS (n = 10). (H) Representative petri dishes in which mouse blood was taken

BGA-P@t-RADA16 induces monocyte/macrophage trained immunity and protects against secondary infection after sepsis. (A) Schematic diagram of establishing the CLP model caused by cecum ligation and puncture and its treatment. Proportion of peritoneal (B, F) CCR2^+^ myeloid cells, (C, G) monocytes, (D, H) M1 macrophages and spleen (E, I) CD8^+^ T cells after 24 h of the last treatment. The content of (J) IL-6 and (K) TNF-α in serum was measured by ELISA after 12 h of secondary infection. Mouse blood and peritoneal lavage fluid were taken after 2 h of secondary infection for 16 h of bacterial cult

Full text

Introduction

1

Sepsis is a life-threatening organ dysfunction syndrome driven by a dysregulated host response to infection, characterized by acute onset and rapid progression to multiple organ failure, constituting a critical public health crisis [[1], [2], [3]]. Current standard-of-care intervention, including early respiratory support, fluid resuscitation, broad-spectrum antibiotics and corticosteroids have shown limited efficacy in continuously improving clinical outcomes and cannot address the persistent immunoparalysis after the initial high inflammatory period [[4], [5], [6]]. This immune paralysis state is characterized by impaired innate and adaptive immune functions, weakened phagocytic ability of macrophages and dysregulation of crosstalk in the immune microenvironment [7]. These alterations precipitate significantly elevated post-discharge risk, with 50% of survivors requiring readmission and 16% adult patients dying within one year [8]. Therefore, there is an urgent need to develop immune recovery strategies targeting dysfunctional immune cells and restoring the host’s defense mechanisms.

Trained immunity is the epigenetic and metabolic reprogramming of innate immune cells, including monocytes, macrophages and natural killer (NK) cells, which enabled enhanced, broad-specturm defensive responses upon secondary exposure to homologous or heterologous stimuli [9,10]. This antigen-agnostic memory mechanism provides organisms with accelerated and comprehensive protection, offering novel therapeutic avenues for addressing the persistent immunoparalysis characteristic of sepsis. β-glucan (BG) is a classic training immune inducer that activates innate immune cells by binding to Dectin-1 and bolsters broad-spectrum antimicrobial capacity against bacterial, viral and fungal pathogens, thereby improving therapeutic outcomes [[11], [12], [13], [14], [15]]. However, free BG is prone to rapid clearance, which requires frequent administration and thereby increases the risks of chronic inflammation, off-target cytotoxicity and immune dysfunction [16]. Moreover, in the early stage of sepsis, large amounts of BG entering the bloodstream may exacerbate the inflammatory storm. To overcome the above limitations, current research mainly focuses on the engineering modification of BG, such as polymer nanoparticles (PLGA) [16], superparamagnetic iron oxide nanoparticles (SPIO) [17] and bacterial vectors [18]. Although these vector systems can enhance BG-induced trained immunity and contribute to control infection in the short term, they still have significant limitations. On the one hand, they lack the ability to precisely target specific organs or cell types. Long-term systemic circulation may lead to excessive immune activation, causing immune pathological damage or autoimmune reactions. On the other hand, the biological safety characteristics of some carriers have not been fully elucidated and further systematic reviews are needed.

Given these challenges, there is a growing interest in developing alternative delivery platforms that combine targeted immunomodulation with enhanced biocompatibility. In this context, self-assembling peptides emerge as a highly promising candidate thanks to their programmable architecture and favorable safety profile [19,20]. They can spontaneously organize into nanofibers in aqueous environments and further assemble into supramolecular hydrogels. These hydrogels serve as versatile matrices, enabling facile encapsulation and efficient delivery of diverse therapeutic agents [21,22]. A key feature of these peptide hydrogels is their ability to form a confined drug depot following local injection. This significantly increases the local drug concentration, expands the on-site therapeutic effectiveness and prolongs the treatment window. These advantages of self-assembled peptides make highly customizable and efficient drug carrier systems possible, which is particularly valuable for creating targeted and sustained release platforms in immunotherapy.

In this study, we developed a novel glycopeptide hydrogel based on BG and the self-assembling peptide RADA16 for sepsis management. As a clinically applied self-assembling peptide with FDA approved, RADA16 forms stable hydrogels via electrostatic interactions between cationic arginine and anionic aspartic acid at neutral pH. This sequence not only offers excellent biocompatibility, negligible immunogenicity, and minimal inflammatory risks [23,24], but also provides readily modifiable sites for functionalization and contains primary amine side chains that can dynamically conjugate via Schiff base linkage with oxidized β-glucan (BGA). These properties collectively enable RADA16 to serve as an ideal, programmable scaffold for constructing integrated glycopeptide hydrogels with sustained release and localized immunomodulatory functions. We first functionalized the N-terminus of RADA16 with tuftsin (TKPR), a natural tetrapeptide known for its macrophage-targeting and activating properties, to generate the chimeric peptide t-RADA16. When this peptide is co-assembled with BGA, it can form a glycopeptide hydrogel with a highly ordered β-sheet nanofiber network (BGA@t-RADA16) (Scheme 1A). Upon subcutaneous injection, the hydrogel effectively recruited immune cells (especially macrophages), creating a localized “trained immunity center” and a “sustained release depot” of BGA. Through this process, BGA@t-RADA16 induced immune training in the recruited macrophages, enhancing their phagocytic capacity and pro-inflammatory cytokine secretion, thereby establishing immune memory against sepsis (Scheme 1B). In mouse models of sepsis with immune paralysis, subcutaneous administration of BGA@t-RADA16 prevented secondary infections by reprogramming peritoneal macrophages via trained immunity. Furthermore, the incorporation of polymyxin B (BGA-P@t-RADA16) demonstrated the dual efficacy of rapidly pathogen clearance during the acute phase and inducing sustained trained immunity to prevent secondary infection (Scheme 1C). This platform not only amplified trained immunity responses but also minimized systemic adverse effects and prevented excessive immune activation. Collectively, this multifunctional hydrogel system leverages trained immunity to establish a novel therapeutic paradigm that bridge prophylactic immune training, acute sepsis intervention and sustained long-term defense. This strategy is expected to be an important breakthrough in the treatment of sepsis and related immune disorders.Scheme 1BGA@t-RADA16 recruits and induces trained immunity in macrophages to reverse sepsis-associated immunoparalysis and against recurrent infections. (A) The structure of BGA and t-RADA16 and the diagram of their co-assembly to form glycopeptide hydrogel. (B) Schematic diagram of the induction of trained immunity in macrophages by subcutaneously administered BGA@t-RADA16. (C) Schematic diagram of applying trained immunity to reverse immune paralysis in sepsis and provide protection for secondary infection. This scheme was created with BioRender.com.Scheme 1

Results and discussion

2

Preparation and characterization of BGA@t-RADA16

2.1

BG, a major structural polysaccharide of fungal cell walls composed of repeating glucose units, which primarily consists of a β-(1,3)-linked backbone with β-(1,6)-linked side branches, is a potent inducer of trained immunity. Monocyte/macrophage-mediated BG-induced trained immunity has been reported to partially reverse immunoparalysis in sepsis, highlighting its potential importance for restoring immune homeostasis post-sepsis [17,25]. To enable the hydrogel formation via biocompatible crosslinking, BG was initially oxidized with sodium periodate to generate BGA. Hydroxylamine hydrochloride titration analysis (Fig. S1A, Supplementary Information) confirmed that BGA possessed a 35% oxidation degree, confirming the successful generation of abundant hemiacetal groups. In addition, the chemical structure of BGA was characterized by FT-IR spectra, and it was found that there was an absorption peak representing aldehyde group at 1670 cm^−1^ (Fig. S1B, Supplementary Information). The structural changes in β-glucan after periodate oxidation were further confirmed by ^13^C NMR analysis. The structure of the native BG was confirmed by ^13^C NMR spectroscopy, with the spectrum showing characteristic signals consistent with the reported data for this polysaccharide [26]. Since periodate selectively cleaves vicinal diol groups, oxidation predominantly occurred at the C6 positions of the β-1,6-linked side-chain glucose residues, as the β-1,3-linked backbone lacks such diol configurations. In the ^13^C NMR spectrum of BGA, a new characteristic peak at 90.41 ppm was observed, corresponding to the formation of a hemiacetal, which results from the hydration and cyclization of the initially generated aldehyde groups in aqueous solution. Furthermore, the comparison of NMR spectra of β-1,3/1,6-glucan with reference showed changes in the signal intensities at the C2 (74.13 ppm), C3 (75.63 ppm), and C4 (78.29 ppm) positions of the side-chain glucose units, consistent with oxidative modification at these sites (Fig. S2, Supporting Information) [26]. The above results indicated that β-glucan was successfully oxidized. t-RADA16 peptide serves as a multifunctional self-assembling scaffold that not only provides the structural basis for hydrogel formation and sustained release but also actively contributes to macrophage recruitment and immunomodulation via its TKPR motif, thereby synergizing with BGA to execute the intended therapeutic strategy. The glycopeptide hydrogel (BGA@t-RADA16) was prepared by crosslinking TKPR-modified RADA16 peptide (t-RADA16, Figs. S3 and S4, Supplementary Information) and BGA at a 1:1 mass ratio through Schiff base reaction. Upon addition of NaCl at neutral pH, the glycopeptide mixed solution underwent rapid gelation, resulting in the formation of a turbid hydrogel (Fig. 1A and B). The chemical structure of BGA@t-RADA16 was characterized by FT-IR spectra, and it was found that a new absorption peak appeared at 1630 cm^−1^, which was attributed to the formation of imine bond (Fig. S1B, Supplementary Information). Subsequently, the microstructure of the BGA@t-RADA16 was characterized using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). SEM revealed distinct structures for the precursor components and the hydrogel (Fig. 1C). Specifically, freeze-dried BGA exhibited a fibrous and granular structure, while freeze-dried t-RADA16 presented a flaky, porous morphology. Critically, upon crosslinking to form the BGA@t-RADA16, SEM showed the formation of a loosely porous hydrogel architecture, with residual BGA particles discernible on the gel surface. This interconnected porous network observed by SEM is functionally significant, as it facilitates essential contact and interaction between the BGA component and immune cells. In addition, TEM analysis (Fig. 1D) revealed that both the t-RADA16 peptide and the crosslinked BGA@t-RADA16 formed interwoven nanofibers, indicating robust self-assembly processes occurred during hydrogel formation. The presence of nanofibers observed by TEM suggests the underlying structural mechanism contributing to the porous architecture seen at the mesoscale by SEM. Circular dichroism (CD) spectroscopy further assessed the secondary structure of t-RADA16 and BGA@t-RADA16 gels (Fig. 1E). t-RADA16 displayed characteristic negative peaks at 199 nm and 217 nm, indicating that it may have a composite structure such as β-sheet, α-helix and random coil. Following crosslinking, BGA@t-RADA16 displayed a distinct β-sheet signature that a broad minimum negative peak shifted to ∼219 nm and a positive peak at ∼199 nm. This spectral shift not only confirms the successful glycopeptide crosslinking reaction but also demonstrates the formation of a predominantly β-sheet conformation within the hydrogel network.Fig. 1. Preparation and characterization of hydrogel BGA@t-RADA16. (A) Schematic diagram of preparation of BGA@t-RADA16. (B) Representative pictures of BGA@t-RADA16 manufacturing process. (C) Representative SEM images of BGA, t-RADA16 and BGA@t-RADA16. (D) Representative TEM images of BGA, t-RADA16 and BGA@t-RADA16. Rheology of BGA@t-RADA16 as a function of angular frequency (E) CD spectrum of BGA@t-RADA16. (F, G) Rheological analysis of BGA@t-RADA16 as a function of angular frequency (F) and shear strain (G). (H) The self-healing analysis of BGA@t-RADA16 with an alternative large oscillation force (50%) and a small one (2%).Fig. 1

The viscoelastic properties of BGA@t-RADA16 were further characterized using rheological analysis to evaluate its mechanical stability and injectable potential. Frequency sweep measurements (Fig. 1F) revealed that the storage modulus (G′) consistently exceeded the loss modulus (G″) over an angular frequency range of 1–100 rad/s, demonstrating BGA@t-RADA16’s elastic, solid-like behavior under oscillatory stress, characteristic of a stable hydrogel network. Strain sweep measurements (Fig. 1G) further elucidated the gel’s response to deformation. BGA@t-RADA16 exhibited pronounced shear-thinning behavior, as evidenced by a significant decrease in G′ with increasing strain. Critically, upon reaching a critical strain of 17.6%, G′ fell below G″, marking the yielding point where the hydrogel network collapses reversibly into a solution-like state, confirming its shear-responsive nature. Step-strain measurements (Fig. 1H) directly probed the hydrogel’s self-healing capability. Application of a large destructive strain (50%) caused a sharp decline in both G′ and G″ moduli. Remarkably, immediately after strain reduction back to the linear viscoelastic region (2%), G′ and G″ rapidly recovered to their original values, confirming the gel’s ability to spontaneously reform its network structure. This self-healing capability likely stems from the abundant hydrogen bonds within the t-RADA16 peptide chains and the dynamic nature of the schiff base-derived imine bonds between BGA and t-RADA16. Supporting evidence for injectability and post-shear self-assembly was provided by TEM imaging of BGA@t-RADA16 post-extrusion through a syringe (Fig. S5, Supplementary Information). TEM revealed that the gel dynamically reassembles into its characteristic interwoven nanofiber structure, demonstrating robust self-healing capability at the nanostructural level. Collectively, the rheological results combined with TEM analysis confirm that BGA@t-RADA16 exhibits the key properties of an injectable and self-healing hydrogel, including robust solid-like stability at rest, shear-thinning flow for easy extrusion and rapid network recovery after shear cessation.

Effects of BGA@t-RADA16-induced trained immunity on immune responses and phagocytosis in macrophages

2.2

We next evaluated the biocompatibility of BGA and BGA@t-RADA16 using CCK-8 assays on RAW264.7 cells. The results demonstrated that BGA (50-200 μg/mL) can promote the proliferation of immune cells, which is consistent with previous studies [27]. Moreover, BGA@t-RADA16 (18.7-600 μg/mL) exhibited unaffected cell proliferation. These results confirm that both BGA and BGA@t-RADA16 have excellent cytocompatibility within the tested concentration ranges (Figs. S6 and S7, Supporting Information).

Trained immunity enables immune cells to mount rapid, robust immune responses against secondary challenges such as sterile inflammation or bacterial and viral infection, thereby enhancing protection against immune paralysis diseases [28]. To evaluate whether oxidation affects BG’s capacity to induce trained immunity, RAW264.7 macrophages and bone marrow-derived macrophages (BMDMs) were primed with either BG or BGA for 24 h. After a 5-day rest period, cells were restimulated with LPS for 4 h. ELISA analysis showed that both BG- and BGA-trained macrophages produced significantly elevated levels of IL-1β, IL-6 and TNF-α upon training and LPS challenge, with no statistically significant difference between the two groups (Fig. S8A–D, Supporting Information and Fig. 2B–D). Consistent with the cytokine data, flow cytometry revealed a comparable increase in CD80^+^macrophage populations in both BG- and BGA-primed cells (Fig. S8E and F, Supporting Information and Fig. 2E and F). Together, these data confirm that oxidation does not impair the trained immunity-inducing ability of BG, and BGA retains activity equivalent to that of unmodified BG (*p > *0.05).Fig. 2BGA@t-RADA16 mediates trained immunity of BMDMs in vitro. (A) Schematic of in vitro trained immunity experimental setup. The production of (B) IL-1β (C) IL-6 and (D) TNF-α by BMDMs trained with BG and BGA was measured by ELISA after 4 h of LPS treatment. (E, F) Determination of M1 type macrophages (CD11b + F4/80+ CD80^+^ subgroup) in each group by flow cytometry. The production of (G) IL-1β (H) IL-6 and (I) TNF-α by BMDMs trained with BGA@t-RADA16 was measured by ELISA after 4 h of LPS treatment. (J, K) Determination of M1 type macrophages (CD11b^+^ F4/80^+^ CD80^+^ subgroup) in each group by flow cytometry. (L, M) The phagocytosis of BMDMs in each group was measured by confocal microscopy. Data are shown as mean ± SEM (n = 3). ns: no significant difference, ∗*p < *0.05, ∗∗*p < *0.01, ∗∗∗*p < *0.001, ∗∗∗∗*p < *0.0001.Fig. 2

To determine if BGA@t-RADA16 induces trained immunity in RAW264.7 and BMDMs, we primed RAW264.7 and BMDMs for 24 h with either PBS (Control), BGA (200 μg/mL), t-RADA16 (200 μg/mL) or BGA@t-RADA16 (400 μg/mL, BGA 200 μg/mL composited with t-RADA16 200 μg/mL). Following a 5-day resting period, cells were stimulated with LPS for 4 h. We quantified pro-inflammatory cytokines expression by ELISA and assessed the proportion of CD80^+^ macrophages by flow cytometry. As shown in Fig. 2G–I and Fig. S9A and 9B, Supporting Information, BGA@t-RADA16 significantly enhanced the secretion levels of IL-1β, IL-6 and TNF-α in macrophages. Fig. 2J and K and Fig. S9C and 9D, Supporting Information demonstrated that the BGA@t-RADA16 exhibited the strongest capacity to polarize macrophages toward a pro-inflammatory phenotype with CD80^+^ high expression. These data demonstrate BGA@t-RADA16’s superior efficacy in inducing trained immunity in macrophages, which may be attributed to the fact that tuftsin polypeptide can enhance the endocytosis of macrophages, promote macrophages to more efficiently devour BGA and effectively activate key downstream signaling pathways of trained immunity. In addition, t-RADA16 also has a certain trained immunity effect, which may be due to the tuftsin polypeptide also has a certain activation effect on macrophages [29].

Macrophage phagocytic function critically influences sepsis prognosis, and trained macrophages exhibit enhanced pathogen clearance capacity [30]. To further validate the trained immunity induced by BGA@t-RADA16, we assessed the phagocytosis of BMDMs against E. coli strain expressing mCherry following the established training protocols in Fig. 2A. Confocal microscopy revealed significantly heightened bacterial internalization in BGA@t-RADA16-trained BMDMs compared with the Control group (Fig. 2L and M). Quantitatively, BGA@t-RADA16 induced a 4.8-fold increase in bacteria phagocytosis vs. the PBS control (*p < *0.0001) and a 1.5-fold enhancement over the BGA group (*p < *0.01), confirming its superior efficacy in augmenting pathogen clearance capacity. Collectively, these results demonstrate that BGA@t-RADA16 not only effectively induces trained immunity in macrophages but also significantly enhances pathogen clearance under secondary challenge and markedly surpassing the functional capacity of free BGA.

Transcriptomic and metabolomics characterization of BMDMs after trained immunity induction by BGA@t-RADA16

2.3

Given the robust capacity of BGA@t-RADA16 to induce trained immunity in vitro, we performed transcriptome analysis on trained BMDMs to elucidate the underlying mechanisms. Compared with the PBS control, when macrophages were trained with BGA@t-RADA16, there were 1938 differentially expressed genes (492 upregulated and 1446 downregulated) (Fig. 3A). After BGA@t-RADA16 training, genes related to antigen processing and presentation (H2-Q4/6/7/10) were upregulated, and genes related to inflammatory cytokines (Il-1a, Il-1b) and chemokines (Ccl3/4/5/6/7, Cxcl2/3) were also elevated (Fig. 3B). Trained immunity induces metabolic reprogramming in macrophages, among which the changes in lipid metabolism are very significant. Gene Ontology (GO) analysis of BGA@t-RADA16-trained cells revealed significant enrichment of pathways related to lipid metabolism (e.g. fatty acid oxidation and steroid biosynthesis), suggesting that the hydrogel may enhance immunological memory function through the modulation of lipid metabolic processes. These findings are highly consistent with the established metabolic features of trained immunity [14,25]. Furthermore, GO enrichment analysis also showed that genes related to the positive regulation of leukocyte function, inflammatory response, bacterial defense response and chemokine production were enriched after BGA@t-RADA16 training (Fig. 3C). HIF-1α-mediated aerobic glycolysis serves as the metabolic basis of trained immunity, which is consistent with the results of kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis (Fig. S10, Supplementary Information). Gene set enrichment analysis (GSEA) was used for genome-wide expression profiling of the RNA-seq data and showed that BGA@t-RADA16 training activated immune system-related KEGG pathways, such as the Toll-like receptor (TLR) signaling pathway, NOD-like receptor signaling pathway, NF-κB signaling pathway, IL-17 signaling pathway, TNF signaling pathway and chemokine signaling pathway. Meanwhile, key anti-inflammatory pathways, including Wnt and TGF-β signaling pathway, were significantly inhibited after BGA@t-RADA16 training (Fig. 3D).Fig. 3. Transcriptomic and metabolomic characteristics of BMDMs after BGA@t-RADA16 training. (A) Differentially expressed genes (DEGs) in control group vs. BGA@t-RADA16 group. (B) DEGs summarized between the BGA@t-RADA16-trained group, the BGA-trained group and the control group. (C) Heatmap comparison of GO enrichment analysis results (*p < *0.05) among BGA@t-RADA16-trained groups. (D) Heatmap comparison of gene set enrichment analysis (GSEA) results (*p < *0.05) among BGA@t-RADA16-trained groups. NES, Normalized enrichment score. (E) Volcano map of the DMs. (F) PCA plot of the qualitative metabolite changes. (G) Heatmap of pooled DMs between BGA@t-RADA16-trained groups and control group. (H) KEGG enrichment results of DMs.Fig. 3

Subsequently, untargeted liquid chromatography-tandem mass spectrometry (LC-MS/MS) based metabolomics was employed to characterize the metabolic profile of BMDMs following BGA@t-RADA16 treatment 5d. Principal component analysis (PCA), orthogonal projections to latent structures discriminant analysis (OPLS-DA) of the identified metabolites, and a heatmap of the 116 pooled differential metabolites (DMs) (58 upregulated and 58 downregulated) collectively demonstrated a distinct metabolic profile in BGA@t-RADA16-trained BMDMs compared with the control group (Fig. 3E and Fand Fig. S11, Supplementary Information). It was worth noting that carbohydrate metabolites including L-lactic acid and key lipid substances such as PC (14:0/14:0), PC (18:1/18:1) and PC36: 2 were significantly increased after BGA@t-RADA16 treatment. These changes highlight significant shifts in glycolysis and membrane phosphorus remodeling, which is consistent with the reprogramming of metabolism that is characteristic of trained immunity. KEGG pathway analysis revealed that treatment with BGA@t-RADA16 induced marked upregulation of most DMs involved in glycerophospholipid, purine, and pyrimidine metabolism. Concurrently, key intermediates in pyruvate metabolism, glycolysis, and starch and sucrose metabolism were predominantly upregulated, supporting a metabolic shift toward aerobic glycolysis. Alterations were also observed in amino acid metabolism pathways, including glutamate metabolism and cysteine and methionine metabolism. These changes collectively indicate a state of immunometabolic reprogramming in BMDMs following BGA@t-RADA16 treatment (Fig. 3H). These results suggest that BGA@t-RADA16 induced significant changes in the transcriptome and metabolome associated with BMDMs trained immunity, which could contribute to the prevention and treatment of sepsis.

Local macrophages recruitment and reprogramming by BGA@t-RADA16 at the injection site

2.4

Prior to evaluating the therapeutic or prophylactic efficacy of BGA@t-RADA16 in murine sepsis models, we assessed its biocompatibility, degradation profile, and capacity for local immune cell recruitment and reprogramming following subcutaneous administration in healthy mice. The in vitro hemolysis assay, a standard method for evaluating the hemocompatibility of biomaterials, was conducted to assess the hydrogel’s blood compatibility. The results demonstrated that no hemolytic activity was observed for any test group at the experimental concentrations (Fig. S12, Supplementary Information). This indicates favorable biocompatibility of the hydrogel for intended use, thereby laying the groundwork for its potential in vivo application. 48 h post-subcutaneous injection, analysis of hematological parameters and major organs in mice showed no significant alterations (Figs. S13 and S14, Supplementary Information), further demonstrating its high in vivo biocompatibility. Subsequently, we prepared fluorescently labeled oxidized β -glucan (FITC-BGA) and analyzed the degradation kinetics of FITC-BGA@t-RADA16 after subcutaneous injection using small animal in vivo imaging system to guide the treatment schedule in subsequent sepsis therapy studies. As shown in Fig. 4A and B, in contrast to FITC-BGA, FITC-BGA@t-RADA16 exhibited a significantly slower degradation profile after subcutaneous injection. Specifically, almost all BGA clearance within 12 h post-injection, whereas approximately 36.19% of the BGA@t-RADA16 hydrogel remained at the injection site (*p < *0.0001). Correspondingly, BGA@t-RADA16 maintained 32.43% retention even at the 72 h time point (*p < *0.0001). This enhanced persistence may be attributed to the physically crosslinked hydrogel network.Fig. 4. Effect of BGA@t-RADA16 on recruitment and activation of macrophages at injection sites. (A) Image and (B) relative FITC fluorescence intensity at different time points after subcutaneous injection of FITC-labeled BGA@t-RADA16 or FITC-labeled BGA in C57BL/6 mice. (C) H&E staining results of injection sites in C57BL/6 mice 3 days after subcutaneous injection of BGA@t-RADA16 or BGA. Proportion of (D, E) macrophages, (F, G) CCR2^+^ macrophages and (H, I) M1 macrophages at the injection site 24 h or 72 h after subcutaneous injection of PBS, BGA or BGA@t-RADA. Data are shown as mean ± SEM (n = 3). ns: no significant difference, ∗*p < *0.05, ∗∗*p < *0.01, ∗∗∗*p < *0.001, ∗∗∗∗*p < *0.0001.Fig. 4

Studies have shown that subcutaneous injection of BG can recruit immune cells such as monocytes and macrophages, but its non-specific effect limits the recruitment efficiency [31]. Therefore, we next studied the recruitment effects of macrophage-targeting BGA@t-RADA16 on innate immune cells at different time points histopathological evaluation of injection site tissues firstly demonstrated significantly enhanced immune cells infiltration in BGA@t-RADA16 group versus the BGA control group at 72 h post-injection (Fig. 4C). We then performed immunophenotyping of these recruited cells via flow cytometry analysis. We found that at 24 h after injection, BGA group and BGA@t-RADA16 group recruited a comparable number of myeloid cells (28% vs. 33%). However, in BGA group, the myeloid cells populations at the injection site at 72 h was significantly lower than that at 24 h and almost returned to the control level. While BGA@t-RADA16 group exhibited a significantly higher proportion of myeloid cells at 72 h post-injection than that at 24 h (*p < *0.0001) (Fig. S15, Supplementary Information). Combined with the results of Fig. 4A and B, we speculate that this may be related to the longer retention time of BGA@t-RADA16 in vivo. Furthermore, we analyzed the composition of mouse skin myeloid cells. As shown in Fig. 4D and E, the proportional changes in dermal macrophages exhibited a similar trend to that observed in myeloid cells, probably because BGA@t-RADA16 recruited and retained more macrophages at the injection site. CCR2^+^ macrophages are very important for circulating monocytes to recruit to the injection site. As shown in Fig. 4F and G, CCR2^+^ macrophages exhibited temporal dynamics congruent with the total macrophage population, demonstrating the highest proportion in the BGA@t-RADA16 group at the 72 h (*p < *0.0001 vs. BGA at 72 h, p < 0.05 vs. BGA@t-RADA16 at 24 h). Critically, local administration of BGA@t-RADA16 significantly altered macrophage polarization profiles at the injection site. As shown in Fig. 4H and I, the BGA@t-RADA16 group demonstrated the highest proportion of M1-polarized macrophages (*p < *0.001 vs. BGA at 72 h, *p < *0.05 vs. BGA@t-RADA16 at 24 h). This local shift towards an M1 phenotype indicates successful reprogramming of macrophages at the hydrogel depot. These results demonstrate that subcutaneously injected BGA@t-RADA16 forms a sustained local depot that efficiently recruits circulating monocytes and tissue-resident macrophages and promotes their functional reprogramming towards a pro-inflammatory state. This locally established immune-active microenvironment, characterized by sustained macrophages recruitment and phenotype reprogramming, is posited to serve as the foundational site for the subsequent induction of trained immunity.

Protective effects of BGA@t-RADA16 against bacterial sepsis in immunosuppressed mice

2.5

It is reported that sepsis induced multiple defects in the immune response, leading to immunoparalysis and increasing susceptibility to infection and death. Therefore, sustaining long-term activation of the innate immune system pharmacologically may help the host achieve immune homeostasis and represents a promising strategy for preventing opportunistic infections [32,33]. Moreover, a decline in immune competence occurs in the elderly and in patients undergoing treatments such as immunoparalysis, chemotherapy or immunotherapy, often resulting in therapy-induced immunodeficiency and predisposing them to a higher risk of opportunistic infections [34,35].

Considering the outstanding ability of the BGA@t-RADA16 hydrogel to induce trained immunity and enhance macrophage phagocytosis in vitro, as well as its capacity for good retention in vivo and effective recruitment and training of macrophages, we propose that the BGA@t-RADA16 hydrogel holds promise for protecting patients with immunoparalysis states such as advanced sepsis from opportunistic infections. Therefore, we established a cyclophosphamide (CTX)-induced immunoparalysis mouse model to evaluate the protective efficacy of the BGA@t-RADA16 hydrogel against intraperitoneal E. coli infection under immunoparalysis conditions. After three consecutive days of intraperitoneal CTX injection, the white blood cell and lymphocyte counts in the mice significantly decreased to undetectable levels (*p < *0.0001) (Fig. S16, Supplementary Information), confirming the successful establishment of the immunoparalysis model [36]. Subsequently, the BGA@t-RADA16 hydrogel was administered subcutaneously every two days for a total of three injections. Five days after the last injection, mice were challenged intraperitoneally with 1 × 10^7^ CFU E. coli. Immune status and damage to major organs were analyzed 24 h later (Fig. 5A). Analysis of peritoneal immune cells revealed that compared with control group, BGA@t-RADA16 group significantly increased the proportion of peritoneal monocytes (Fig. 5B and C) (*p < *0.001) and macrophages (Fig. 5D and E) (*p < *0.01). These proportions were 13.8% and 16.8% higher than that of the BGA group, respectively. It indicates that the hydrogel effectively increased the ratio of peritoneal monocytes and macrophages to fight infection. Further analysis of macrophage phenotypes showed that the BGA@t-RADA16 hydrogel significantly increased the proportion of peritoneal M1 macrophages (Fig. 5F and G) (*p < *0.05) and decreased the proportion of peritoneal M2 macrophages (Fig. 5H and I) (*p < *0.001). These findings indicate that following intraperitoneal infection, the innate immune system of mice mounts a rapid response, marked by an expansion of peritoneal monocytes/macrophages and polarization of macrophages toward an M1 phenotype, which collectively facilitate an effective defense against the infection. Furthermore, analysis of serum pro-inflammatory cytokines levels showed that BGA@t-RADA16 group had significantly elevated levels of IL-1β (*p < *0.0001), IL-6 (*p < *0.01) and TNF-α (*p < *0.001) compared with the control group (Fig. 5J–L). Furthermore, treatment with BGA@t-RADA16 alleviated swelling, congestion and inflammatory cell infiltration in the alveolar walls and also provided significant protection against bacterial infection-induced liver damage (Fig. 5M). These data confirm that trained immunity induced by BGA@t-RADA16 can effectively protect immunoparalysis mice against bacterial infection.Fig. 5. Effect of BGA@t-RADA16 on preventing sepsis in immunoparalysis mice. (A) Schematic diagram of establishing an in vivo immunoparalysis model and inducing trained immunity. Proportion of peritoneal (B, C) monocytes, (D, E) macrophages, (F, G) M1 macrophages and (H, I) M2 macrophages after different treatments. The content of (J) IL-1β, (K) IL-6 and (L) TNF-α in serum was measured by ELISA after different treatment. (M) Representative images of H&E staining of the liver and lungs of mice after different treatment. In the PBS, E.coli, t-RADA16 and BGA groups, inflammatory cell infiltration and consequent organ damage were observed at the locations marked by arrowheads. Data are shown as mean ± SEM (n = 3). ∗*p < *0.05, ∗∗*p < *0.01, ∗∗∗*p < *0.001, ∗∗∗∗*p < *0.0001.Fig. 5

Protective effects of BGA-P@t-RADA16 hydrogel against sepsis in mice

2.6

Encouraged by the potent immunostimulatory effects of BGA@t-RADA16 in immunoparalysis mice, we further evaluated its therapeutic potential in a murine model of established septic infection. Given the severity of sepsis, which often leads to multiple organ damage and high mortality, we combined with current clinical treatment options and then added polymyxin B (PMB) to BGA-P@t-RADA16 gel and prepared BGA-P@t-RADA16 gel to more efficiently fight bacterial infection in the early stage of sepsis [[37], [38], [39]]. PMB is a potent antibiotic which is recommended to use in sepsis management guidelines for early therapeutic intervention [[37], [38], [39]]. The efficient loading of PMB to the gel benefited from the abundant charge groups and hydrogen-bond donors provided by the self-assembled polypeptide, which makes it have good drug loading capacity.

We characterized its microstructure using SEM and TEM and found that the BGA-P@t-RADA16 hydrogel also possesses a loose, porous structure similar to the BGA@t-RADA16 hydrogel (Fig. S17, Supplementary Information). TEM results showed that the BGA-P@t-RADA16 hydrogel formed an interwoven nanofiber network (Fig. S18, Supplementary Information), indicating that the introduction of PMB did not significantly alter the hydrogel’s structure. The drug loading efficiency (DLE) of PMB in BGA-P@t-RADA16 was calculated as 84.6 ± 1.2% (n = 3), determined by measuring the unencapsulated PMB in the supernatant after hydrogel formation. The PMB payload in BGA-P@t-RADA16 was 79.3 ± 2.3 μg PMB per mg hydrogel (n = 3). Subsequently, the release profile of PMB from the hydrogel was evaluated in vitro. The PMB release profile demonstrated a rapid release within the first 8 h, followed by a sustained phase where approximately 90% of the loaded drug was released by 12 h, after which the cumulative release reached a plateau (Fig. S19, Supplementary Information). Given the critical need for effective antimicrobial control of infection sources during the initial phase of sepsis, we further evaluated the in vitro bactericidal activity of BGA-P@t-RADA16. After co-incubating 1 × 10^6^ CFU/mL of E. coli with BGA-P@t-RADA16 for 12 h, the supernatant was collected and subjected to bacterial culture for 16 h. The results demonstrated that BGA-P@t-RADA16 exhibited potent bactericidal activity against E. coli, whereas BGA@t-RADA16 without PMB showed no significant antibacterial effect (Fig. S20, Supplementary Information). Since inflammatory factor storms occur early in sepsis, it is necessary to evaluate whether treatment with BGA-P@t-RADA16 can reduce the inflammatory factor storms in the early stage of sepsis. Therefore, we first evaluated the endotoxin adsorption capacity and training immunity induction effect of BGA-P@t-RADA16 hydrogel in vitro. After co-culturing RAW264.7 cells with BGA-P@t-RADA16 and LPS for 24 h, we assessed the levels of pro-inflammatory cytokines secreted by RAW264.7 cells and the trained immunity effect after 24 h of the hydrogel “training” (Fig. 6A). After the first LPS stimulation, the BGA-P@t-RADA16 hydrogel group significantly reduced the levels of IL-6 (*p < *0.01) and TNF-α (*p < *0.05) compared with the control group, (Fig. 6B and C), indicating that the hydrogel can effectively prevented the pro-inflammatory cytokine storm caused by endotoxin. Upon secondary LPS stimulation, BGA-P@t-RADA16 still effectively promoted macrophages polarization towards the M1 phenotype (*p < *0.01) (Fig. 6D and E), demonstrating that the introduction of PMB did not significantly impact the trained immunity effect.Fig. 6. Therapeutic efficacy of BGA-P@t-RADA16 against CLP-induced polymicrobial sepsis. (A) Schematic of in vitro trained immunity experimental setup. The production of (B) IL-6 and (C) TNF-α by RAW264.7 cells after 24 h of BGA-P@t-RADA16 and LPS treatment. (D, E) Determination of M1 type macrophages in each group by flow cytometry after 4 h of secondary LPS stimuli. (F) Schematic diagram of establishing the CLP model and treatment. (G) Survival curves of mice in each group after treatment with BGA-P@t-RADA16, HC + PMB or PBS (n = 10). (H) Representative petri dishes in which mouse blood was taken 24 h after CLP for 16 h of bacterial culture showed bacterial colonies and (I) statistical graphs. (J) Representative images of H&E staining of liver and lungs of mice on day 5 after CLP. In the PBS and HC + PMB groups, inflammatory cell infiltration and consequent organ damage were observed at the locations marked by arrowheads. Data are shown as mean ± SEM. ∗*p < *0.05, ∗∗*p < *0.01, ∗∗∗∗*p < *0.0001.Fig. 6

Before evaluating the protective effect of BGA-P@t-RADA16 against sepsis, we evaluated its biocompatibility in vivo. At 48 h after subcutaneous injection, no significant alterations were observed in the hematological parameters or major organ tissues of mice (Figs. S21 and S22, Supplementary Information), demonstrating its good biocompatibility. We then established a cecal ligation and puncture (CLP) model to investigate the protective effect of the BGA-P@t-RADA16 hydrogel in septic mice (Fig. 6F). As shown in Fig. 6G, compared with the control group, the BGA-P@t-RADA16 hydrogel significantly reduced sepsis mortality from 80% to 30% (*p < *0.05). Its mortality rate was also lower than that of the group treated with the clinically common acute-phase sepsis therapy combining hydrocortisone and PMB (HC + PMB). In addition, the body weight of mice in each group decreased to varying degrees after CLP, with the most obvious decrease in the control group (Fig. S23, Supplementary Information). Blood samples collected on the first day post-modeling were plated for bacterial culture. It was found that the bacterial density in the blood of the BGA-P@t-RADA16 hydrogel group was significantly lower than that of the control group (Fig. 6H and I). Histological analysis of mouse organ sections revealed that, compared with the control group, both BGA-P@t-RADA16 and HC + PMB groups exhibited significant reductions in alveolar wall congestion and inflammatory cell infiltration, as well as markedly attenuated liver and kidney damage (Fig. 6J). Notably, the therapeutic effect of BGA-P@t-RADA16 group was significantly superior to that of HC + PMB group. These results demonstrate that the BGA-P@t-RADA16 provides significant protective effects during the acute phase of sepsis in mice.

Protective effects of BGA-P@t-RADA16 hydrogel against secondary infection in septic mice via inducing of trained immunity

2.7

Subsequently, we evaluated the immune activation status in recovered mice after treatment with BGA-P@t-RADA16. One day after three times of BGA-P@t-RADA16 or HC + PMB treatments, peritoneal immune cells were extracted and analyzed by flow cytometry (Fig. 7A). It was found that compared with the control group, BGA-P@t-RADA16 group showed a significant increase in myeloid cells (Fig. S24, Supplementary Information) (*p < *0.01), indicating that the BGA-P@t-RADA16 hydrogel significantly activated the innate immunity of the mice. Upon further analysis of the myeloid cells, we found that compared with the control group and the HC + PMB group, the proportion of migratory macrophages (CCR2^+^) in the peritoneal cavity of the mice was significantly increased (*p < *0.0001) (Fig. 7B and F). This demonstrates that during infection, BGA-P@t-RADA16 treatment effectively promoted the migration of innate immune cells to the infection sites. Additionally, compared with control group, the proportions of monocytes (*p < *0.05) (Fig. 7C and G) and M1 macrophages (*p < *0.0001) (Fig. 7D and H) were also significantly elevated. Notably, besides activating innate immunity, the BGA-P@t-RADA16 hydrogel also effectively activated the adaptive immune system. By isolating splenocytes and examining the proportion of CD8^+^ T cells, we found that compared with control group and HC + PMB group, BGA-P@t-RADA16 hydrogel significantly increased the proportion of CD8^+^ T cells (*p < *0.0001) (Fig. 7E and I). This may be attributed to enhanced antigen-presenting capability by antigen-presenting cells following innate immune activation, thereby effectively activating adaptive immunity, which is consistent with the results reported by Chen et al. [18].Fig. 7BGA-P@t-RADA16 induces monocyte/macrophage trained immunity and protects against secondary infection after sepsis. (A) Schematic diagram of establishing the CLP model caused by cecum ligation and puncture and its treatment. Proportion of peritoneal (B, F) CCR2^+^ myeloid cells, (C, G) monocytes, (D, H) M1 macrophages and spleen (E, I) CD8^+^ T cells after 24 h of the last treatment. The content of (J) IL-6 and (K) TNF-α in serum was measured by ELISA after 12 h of secondary infection. Mouse blood and peritoneal lavage fluid were taken after 2 h of secondary infection for 16 h of bacterial culture showed statistical graphs of the number of colonies in (L) blood and (M) peritoneal lavage fluid. Data are shown as mean ± SEM. ns: no significant difference, ∗*p < *0.05, ∗∗*p < *0.01, ∗∗∗*p < *0.001, ∗∗∗∗*p < *0.0001.Fig. 7

Considering the effective activation of both the innate and adaptive immune systems by BGA-P@t-RADA16, we then injected E. coli into the cured mice one week later to observe its protective effect against secondary infection. One day before bacterial injection, the cured mice in BGA-P@t-RADA16 group were randomly divided into two groups. One group was administered clodronate liposomes to deplete macrophages, with the aim of impairing trained immune function, and the other group was given no treatment [40]. After 24 h, all mice were then intraperitoneally injected with 5 × 10^6^ CFU E. coli. After 12 h, mice were euthanized, and serum levels of pro-inflammatory cytokines, as well as bacteria density in the blood and peritoneal lavage fluid, were measured. It was found that compared with clodronate liposomes-injected group, the control mice had significantly higher levels of the pro-inflammatory cytokines IL-6 (*p < *0.01) (Fig. 7J) and TNF-α (*p < *0.0001) (Fig. 7K) in their blood. This result indicates that the presence of reprogrammed macrophages is critical for the antibacterial defense observed during secondary challenge. Bacterial plating of blood and peritoneal lavage fluid showed that compared with clodronate liposomes group, mice in control group had significantly lower bacterial density in both their blood (*p < *0.01) and peritoneal cavity (*p < *0.0001) (Fig. 7L and M and Fig. S25, Supplementary Information). The above results indicate that the BGA-P@t-RADA16 hydrogel confers significant protection against secondary bacterial infection in mice, and this protective effect is critically dependent on reprogrammed macrophages, consistent with the establishment of a macrophage-mediated trained immunity response.

These findings demonstrate that the hydrogel effectively establishes macrophage-mediated trained immunity in vivo. Therefore, this injectable platform represents a promising integrated strategy for sepsis management. This injectable glycopeptide hydrogel offers a multifaceted strategy for sepsis management by creating a localized, biodegradable depot that ensures sustained release and targeted delivery of immunomodulators via the tuftsin (TKPR) motif, directly reprogramming myeloid cells on-site. Compared with systemic nanoparticles or cellular vectors, our platform integrates precise spatial control, prolonged retention, and inherent targeting into a single injectable format, which can minimize off-target effects and have good long-term biological safety. Translational potential is underscored by the clinically established components (RADA16 and PMB) and preclinical efficacy, while further development requires addressing scalable manufacturing, long-term safety, and regulatory pathways for combination products.

Conclusion

3

In this study, we developed the first BGA delivery system using self-assembling polypeptide hydrogels to combat immuneparalysis associated with sepsis. This glycopeptide hydrogel (BGA@t-RADA16) was created by co-assembling BGA with tuftsin-functionalized self-assembling peptides. This design created a BGA “sustained-release depot” and a localized “trained immunity center” at the subcutaneous injection site, minimizing the risks of excessive inflammation and systemic adverse effects associated with free BGA administration. Within this depot, the released BGA induced long-term trained immunity in recruited macrophages through glycolipid metabolic reprogramming and activation of inflammatory pathways, enhancing their phagocytic capacity, proinflammatory responses and immune memory. In CTX-induced immune paralysis mice, BGA@t-RADA16 effectively prevented E. coli infection by activating monocytes/macrophages and increasing systemic cytokines. Furthermore, in an acute sepsis mouse model of CLP, incorporation of PMB (forming BGA-P@t-RADA16) enabled simultaneous rapid bacterial clearance and induction of sustained immune memory. Compared with the control group, the BGA-P@t-RADA16 group significantly reduced organ damage and lowered mortality from 80% to 30%. Surviving mice exhibited durable macrophage-mediated immunity, which reversed sepsis-induced immunoparalysis and enhanced defense against secondary infection. In conclusion, this platform acts as an injectable “trained immunity center”, breaking the immune paralysis caused by sepsis through comprehensive preventive training, acute intervention and long-term immune memory training, offering a promising strategy for treating sepsis and other diseases caused by immune imbalances.

Experimental section

4

Materials

4.1

β-glucan was purchased from Yuanye Bio-Technology Co., Ltd (Shanghai, China, S24487, Purity: 80%, derived from yeast, water-soluble), t-RADA16 peptide was purchased from GL Biochem (Shanghai) Ltd (Shanghai, China), Sodium periodate (NaIO_4_), ethylene glycol, hydroxylamine hydrochloride and acetylene glycol hydroxylamine hydrochloride were purchased from Aladdin Chemical Reagent Co., Ltd. (Shanghai, China). LPS was purchased from Solarbio (Beijing, China). Fixable Viability Stain 780, anti-CD45-AF700 (clone: 30F11), anti-CCR2-BV650 (clone:475301), anti-LY6G-PE (clone: 1A8), anti-LY6C-PE-CY7 (clone: AL-21) and anti-F4/80-BV421 (clone: T45-2342) were purchased from BD Pharmingen (New Jersey, USA); anti-CD80-PerCP-cy5.5 (clone: 16-10A1), anti-CD86-FITC (clone: GL-1), anti-CD11b-APC (clone: M1-70), anti-CD206-PE (clone: C068C2) and anti-mouse CD16/32 (clone: S17011E) were purchased from BioLegend (San Diego, USA).

Animals

4.2

5-8 weeks old healthy female C57BL/6 mice were obtained from Vital River Laboratories (Beijing, China) and fed in a germ-free environment. All animal procedures were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals of Peking Union Medical College and experiments were approved by the Animal Experiments and Ethics Review Committee of the Institute of Radiation Medicine, Chinese Academy of Medical Sciences (Ethics Committee No. IRM-DWLL-2023078).

Cell and bacterial culture

4.3

Mouse macrophages RAW264.7 were purchased from ATCC (USA) and cultured in RPMI-1640 (Gibco, Grand Island, NY) medium containing 10% fetal bovine serum (FBS, Gibco) and 1% Penicillin-Streptomycin (Gibco) at 37 °C in 5% CO2. Bone marrow-derived macrophages (BMDMs) were extracted from the tibia of C57BL/6 mice aged 6-8 weeks and cultured 5 days in RPMI-1640 medium containing 20 ng/mL M-CSF (MCE, New Jersey, USA) by adaptation of the previous procedures. Escherichia coli (E. coli) was donated by Nankai University, and E. coli fluorescent strain (ATCC 25922) was purchased from Baosai Biotechnology Co., Ltd. (Hangzhou, China), and was cultured in soybean agar containing tryptone at 37 °C. A single colony was inoculated into tryptone soybean agar medium and cultured overnight at 37 °C. Then, the absorbance at the wavelength of 600 nm was measured to count the number of bacteria.

Synthesis and characterization of BGA

4.4

5 mL solution of 0.5 M sodium periodate was added dropwise to 100 mL of 1% BG solution under continuous stirring. After 2 h, 1 mL of ethylene glycol was added to quench the reaction. The product underwent dialysis for 3 days, followed by lyophilization, and was stored in an oxygen-free dry environment. Hydroxylamine hydrochloride titration analysis was performed using a ZDJ-4B automatic potentiometric titrator (INESA Scientific Instrument Co., Ltd., China). The structure was characterized by Fourier transform infrared spectroscopic (FT-IR) (Elmer Perkin, USA).

Preparation of BGA@t-RADA16

4.5

BGA@t-RADA16 was prepared by simply mixing BGA solution (10 mg/mL) and t-RADA16 polypeptide solution (10 mg/mL) in a volume of 1:1 to form imine bonds. After adjusting the pH to neutrality, an appropriate amount of sodium chloride aqueous solution was added to induce the formation of hydrogel.

Characterization of BGA@t-RADA16

4.6

Transmission Electron Microscopy (TEM) observation. A diluted sample (10 μL) was loaded onto a carbon-coated copper grid and left undisturbed for 3 min. Subsequently, 10 μL of 2% uranyl acetate solution was added for staining (3 min). After drying, TEM images were collected using an electron microscope (FEI, Talos F200C).

Scanning Electron Microscopy (SEM) observation. Samples were flash-frozen in liquid nitrogen and dried in a freeze dryer for 24 h. Specimens were mounted on conductive adhesive tape attached to a sample holder, followed by sputter-coating with gold. SEM images were captured under low-temperature vacuum mode (Beijing Hitachi Scientific Instrument Co., LTD).

Rheology properties of hydrogels. The rheology properties of the hydrogel were evaluated by an AR 2000ex rheometer (TA). The hydrogel samples were placed between parallel plates with a diameter of 10 mm and a gap of 1 mm. The elastic modulus (G′) and viscosity modulus (G″) were measured in a frequency range of 1-100 rad s^−1^, as well as a strain range of 1-100% at 25 °C, respectively. Besides, the self-healing property of BGA@t-RADA16 was investigated through measuring the G′ and G″ under continuous train sweep with an alternative large oscillation force (50%) and a small one (2%). The secondary structure of peptide in BGA@t-RADA16 was used to evaluate by Circular dichroism (CD) (JASCO, Japan).

Cytotoxicity assessment

4.7

The cytotoxicity of BGA and BGA@t-RADA16 on RAW264.7 cells were detected by the cell counting kit-8 (CCK-8, BestBio, China). Briefly, 5 × 10^3^ cells were inoculated into each well of a 96-well plate per well and incubated at 37 °C for 24 h. Next, different concentrations of the BGA or BGA@t-RADA16 solution were added to each well. The solution was then placed in 100 μL of RPMI 1640 culture medium and incubated for an additional 24 h. After washing the cells three times with PBS buffer, 10 μL of CCK-8 reagent was added to each well and then incubated at 37 °C for 4 h. Finally, the absorbance at the wavelength of 450 nm was measured by a Microplate Reader (Thermo Varioskan Flash).

Analysis of the phenotype of RAW264.7 and BMDMs in trained immunity

4.8

RAW264.7 cells or BMDMs were trained with PBS, BG, BGA, t-RADA16 or BGA@t-RADA16 for 24 h. Then, they were stimulated with LPS (100 ng/mL) for 4 h after resting for 5 days. The phenotype of macrophages was then detected. Specifically, RAW264.7 cells or BMDMs were collected and stained with anti-CD11b, anti-F4/80 and anti-CD80 antibody. All antibodies were diluted according to the instructions and incubated with cells at 4 °C for 30 min. The cells were then washed with PBS and analyzed by flow cytometry (BD LSRFortessa X-20).

Evaluation of inflammatory factors

4.9

Inflammatory factors such as IL-1β, IL-6 and TNF-α were detected using the enzyme-linked immunosorbent assay (ELISA) kit (Proteintech, China). The cell culture supernatant and serum were collected. Then, 100 μL of the diluted standards or samples to be tested were added to each well and incubated for 90 min at 37 °C. Discard the liquid and add 100 μL of biotin-labeled detection reagent, and incubate for 60 min at 37 °C. Thoroughly wash each well with washing solution to remove residual solution. Then, add 100 μL of horseradish peroxidase (HRP) labelling reagent and incubate for 30 min at 37 °C. After washing the plate for five times, a mixture of 50 μL of chromogenic solution A and 50 μL of chromogenic solution B was added to each well. After incubating in the dark at 37 °C for 15 min, add 50 μL of stop solution to each well. Finally, the absorbance at a wavelength of 450 nm was immediately measured using a Microplate Reader.

Bacterial phagocytosis experiment of BMDMs

4.10

BMDMs labeled with Cell Tracker Green CMFDA (5-Chloromethylfluorescein Diacetate) (Yeasen Biotechnology, Shanghai Co., Ltd., China) were treated with PBS, BGA, t-RADA16 or BGA@t-RADA16 for 24 h, and then were co-cultured with a fluorescent strain of E. coli for 2 h after resting for 5 days. The undigested E. coli was washed away with PBS, and the phagocytosis of E. coli by BMDMs was analyzed by confocal microscope (Nikon, C2).

RNA extraction, library construction and sequencing

4.11

BMDMs were inoculated into a six-well cell culture plate at a concentration of 2 × 10^6^ cells per well and with PBS, BGA or BGA@t-RADA16 for 24 h. After 5 days, the culture medium was taken out, and the cells were washed twice with cold PBS. Then, total RNA was extracted using Trizol reagent kit (Invitrogen, Carlsbad, CA,USA) according to the manufacturer’s protocol. After total RNA was extracted, eukaryotic mRNA was enriched by Oligo (dT) beads. Then the enriched mRNA was fragmented into short fragments using fragmentation buffer and reversly transcribed into cDNA by using NEBNext Ultra RNA Library Prep Kit for Illumina (NEB #7530, New England Biolabs, Ipswich, MA, USA). The purified double-stranded cDNA fragments were end repaired, a base added and ligated to Illumina sequencing adapters. The ligation reaction was purified with the AMPure XP Beads (1.0X). And polymerase chain reaction (PCR) amplified. The resulting cDNA library was sequenced using Illumina Novaseq6000 (Hangzhou Astrocyte Technology Co,.Ltd, Hangzhou, China).

Analysis of differentially expressed genes (DEGs)

4.12

RNA differential expression analysis was performed by DESeq2 software between two different groups and using the edgeR software between two samples. The genes/transcripts with a false discovery rate (FDR) below 0.05 and an absolute fold change of at least 1.5 were considered DEGs/transcripts.

Metabolomics experiments

4.13

BMDMs were seeded in 10 cm culture dishes (5 × 10^6^ cells per dish) and trained as described above. After 5 days, the cells were washed twice with cold PBS, scraped off, snap-frozen in liquid nitrogen for 15 min, and stored at −80 °C until metabolite analysis. Metabolites were separated using a Waters ACQUITY BEH C18 Column (1.7 μm × 2.1 mm × 100 mm) on a Vanquish Flex UPLC equipped with a refrigerated autosampler (10 °C) and column heater (40 °C) (Hangzhou Astrocyte Technology Co,.Ltd, Hangzhou, China). Two mobile phase conditions were used to improve the metabolite coverage.

Condition 1 (Waters ACQUITY BEH C18 Column): Metabolites were separated using a Waters ACQUITY BEH C18 Column (1.7 μm × 2.1 mm × 100 mm) on a Vanquish Flex UPLC equipped with a refrigerated autosampler (10 °C) and column heater (40 °C). Solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile/methanol = 4/6) were used to elute the metabolites with a 13.5 min gradient, as follows: 10% B at 0 min, 0.25 mL/min; 40% B at 3 min, 0.25 mL/min; 95 % B at 5 min, 0.25 mL/min; 100 % B at 8 min, 0.6 mL/min; 100 % B at 10 min, 0.6 mL/min; and back to 10 % B at 10.5 min, 0.25 mL/min; and equilibrate for 3min. Samples were analyzed using a Q Exactive HF-X (QE-HF-X) mass spectrometry equipped with a heated electro-spray ionization (HESI) source. All the data was acquired in positive and negative mode separately using Full scan/ddMS2 (DDA) to acquire MS1 precursor for quantification and MS2 fragmentation for metabolite annotation. The Full Scan settings were as follows: 60,000 resolution, AGC target, 4e6; Maximum IT, 100 ms; scan range, 60 to 900 m/z. For Full scan/ddMS2(DDA), Top 20 MS/MS spectral (dd-MS2)@15000 were generated with AGC target = 2e5, Maximum IT = 25 ms, and (N)CE/stepped NCE = 10, 40, 80v. Metabolites detection and identification were performed using XCMS and Spectra R package by searching against online database (MoNA, GNPS, HMDB and MS-dial LipidBlast database) and in-house database.

Condition 2 (Waters BEH Amide Column): Metabolites were separated using a Waters BEH Amide Column (1.7 μm × 2.1 mm × 100 mm) on a Vanquish Flex UPLC equipped with a refrigerated autosampler (10 °C) and column heater (40 °C). Solvent A was 25 mM NH_4_CH_3_COOH and 25 mM NH_3_·H_2_O in water, solvent B was acetonitrile. The 12 min gradient was employed: 0 min, 0.5 mL/min, 95% B; 0.5 min, 0.5 mL/min, 95%B; 7 min, 0.5 mL/min, 65%B; 8 min, 0.5 mL/min, 40%B; 9 min, 0.5 mL/min, 40%B; 9.1 min, 0.5 mL/min, 95%B; 12 min, 0.5 mL/min, 95%B. Data acquisition was performed in positive and negative mode separately using Full scan/ddMS2 (DDA) to acquire MS1 precursor for quantification and MS2 fragmentation for metabolite annotation. The Full Scan settings were as follows: 60,000 resolution, AGC target, 4e6; Maximum IT, 100 ms; scan range, 60 to 900 m/z. For Full scan/ddMS2(DDA), Top 20 MS/MS spectral (dd-MS2)@15000 were generated with AGC target = 2e5, Maximum IT = 25 ms, and (N)CE/stepped NCE = 10, 40, 80v. Metabolites detection and identification were performed using XCMS and Spectra R package by searching against online database (MoNA, GNPS, HMDB and MS-dial LipidBlast database) and in-house database.

Data processing and statistical analysis

4.14

MSConvert was used to convert raw MS data into mzXML files, which were subsequently imported into the publicly accessible XCMS program. For isotope and adduct annotation, CAMERA (Collection of Algorithms for Metabolite Profile Annotation) was employed. Metabolite identification was performed by comparing the mass-to-charge ratio (within 10 ppm) and MS/MS spectra with authentic standards. To verify metabolite identification, the integrated mass of the processed data was evaluated against established reference standards.

After processing and normalization, the resulting data were subjected to multivariate analysis using the R software package (ropls), including Pareto-scaled principal component analysis (PCA) and orthogonal partial least-squares discriminant analysis (OPLS-DA). Model robustness was assessed through seven-fold cross-validation and response permutation testing. Variable importance in projection (VIP) values were calculated for each OPLS-DA model to indicate their contribution to group separation. Significance of differences between two independent groups was determined using Student’s t-test. Metabolites with a false discovery rate (FDR) < 0.30 and VIP >0.5 were considered significantly altered. Pearson’s correlation analysis was applied to examine relationships between two variables.

Preparation of FITC-BGA

4.15

BGA (100 mg) was dissolved in 5 mL of NaOH solution (1 M), followed by the addition of 1.5 mL of epichlorohydrin. The reaction mixture was stirred at room temperature for 12 h. Subsequently, 10 mL of ethanol was added, and the reaction was allowed to proceed for an additional 10 min. The solvent was then concentrated under reduced pressure, and the product was precipitated by pouring the concentrate into a large volume of ethanol, followed by filtration. The collected solid was dissolved in 10 mL of deionized water, transferred into a dialysis bag (molecular weight cutoff: 1000 Da), and dialyzed against deionized water for 24 h. The dialyzed solution was lyophilized to obtain epoxy-functionalized BGA (BGA-epoxy). Next, BGA-epoxy (100 mg) was dissolved in 5 mL of ammonia solution and reacted at 40 °C for 1 h. After reaction, the mixture was concentrated under reduced pressure, and the product was precipitated in excess ethanol, collected by filtration, and dried under vacuum to yield aminated BGA (BGA-NH_2_). Finally, BGA-NH_2_ (100 mg) was dissolved in 10 mL of PBS (pH 9.0), followed by dropwise addition of a solution of FITC (10 mg) in DMF. The reaction was carried out at room temperature for 2 h. The resulting mixture was transferred into a dialysis bag (MWCO: 1000 Da) and dialyzed against deionized water for 24 h. The dialyzed product was lyophilized to obtain FITC-labeled BGA (FITC-BGA).

Hemolysis test

4.16

Fresh anticoagulated whole blood is centrifuged and washed with PBS to obtain erythrocytes. The red blood cells (RBCs) are resuspended in PBS to a 4% (v/v) suspension. Test materials at various concentrations are incubated with the RBC suspension at 37 °C for 2 h, using PBS and Triton X-100 (1%) as negative and positive controls, respectively. After incubation, samples are centrifuged. The absorbance of the supernatant is measured at 577 nm. The hemolysis percentage is calculated by comparing the absorbance of test samples to that of the positive control.

Hematological analysis

4.17

To evaluate the biocompatibility of different materials, we conducted hematology analysis. Briefly, healthy C57BL/6 mice were randomly divided into three groups and injected subcutaneously with 200 μL of BGA@t-RADA16 (60 mg/kg), BGA-P@t-RADA16 (70 mg/kg, PMB 10 mg/kg) or PBS per mouse, respectively. After 24h injection, blood was collected for testing using an automatic blood analyzer (Nihon Kohden, MEK-7222K).

Histological analysis

4.18

Mice with different treatments were euthanized and their vital organs (heart, liver, spleen, lungs and kidneys) were collected. Tissue samples were placed in 4% paraformaldehyde fixing solution, subjected to ethanol gradient dehydration, and immersed in paraffin for 12 h. Tissues were then embedded in paraffin and sectioned to prepare 6 μm thick sections. After the slices were dewaxed, they were dehydrated again with ethanol gradient, stained with hematoxylin and eosin (H&E) staining (Solarbio, China), and imaged by light microscopy (Olympus BX51, Japan).

In vivo fluorescence imaging

4.19

To evaluate the degradation kinetics of hydrogel in vivo, the healthy C57BL/6 mice were randomly divided into three groups for fluorescence imaging observation. Each group was injected subcutaneously with FITC-labeled BGA@t-RADA16 (60 mg/kg) or FITC-labeled BGA (30 mg/kg), respectively. Fluorescence imaging was performed at predetermined time points (0, 4, 8, 12, 24, 48 and 72 h) after injection using a small-animal ex/in vivo optical imaging system (CRI, Maestro 2) with an excitation wavelength of 470 nm.

Experiment on immune cell recruitment at the injection site

4.20

Healthy C57BL/6 mice were randomly divided into three groups and then injected subcutaneously with BGA (30 mg/kg), BGA@t-RADA16 (60 mg/kg) or PBS, respectively. The mice were euthanized 24 or 72 h after injection. The tissue at the injection site was extracted and mechanically disrupted, then 1 mL of tissue dissociation fluid (cat: abs9482, Absin) was added. The mixture was incubated on a horizontal shaker at 80 rpm at 37 °C for 40 min, and then filtered through the 70 μm micron cell strainer to obtain a single cell suspension. The cells were stained with anti-CD45-AF700, anti-CD11b-APC, anti-F4/80-BV421, anti-CCR2-BV650, anti-CD80-Percp-Cy5.5 antibodys, respectively, and then detected by flow cytometry.

Drug loading efficacy measurement

4.21

Following gelation, the hydrogel (800 μg) encapsulating PMB (75 μg) was immersed in PBS (1 mL) and subjected to brief low-speed centrifugation (1000 rpm, 1 min). The supernatant was then collected, and the concentration of PMB released into the medium was determined by measuring the absorbance at 220 nm using a UV-Vis spectrophotometer. Drug loading efficacy was calculated based on the amount of PMB detected in the supernatant relative to the total amount initially loaded.

Release kinetics of PMB from hydrogel

4.22

After gelation, the hydrogel (800 μg) encapsulating PMB (75 μg) was immersed in PBS (1 mL) and incubated at 37 °C under gentle agitation. At predetermined time points (1, 2, 4, 8, 12, and 24 h), the release medium was completely collected. The concentration of PMB in the collected medium was determined by measuring the absorbance at 220 nm using a UV-Vis spectrophotometer, with reference to a standard curve of PMB in PBS.

Evaluation the effect of BGA@t-RADA 16 on the prevention of sepsis in immunoparalysis mice

4.23

Based on methods reported in the literature, we created an immunoparalysis mouse model of sepsis [41]. Mice were given CTX at a dose of 100 mg/kg via intraperitoneal injection for three consecutive days. Then, immunoparalysis status was assessed by monitoring changes in the number of white blood cells (WBCs) and lymphocytes (LYMs). After successful modeling, the mice were randomly divided into five groups (n = 8). They were treated with t-RADA16 (30 mg/kg), BGA (30 mg/kg), BGA@t-RADA16 (60 mg/kg) or PBS (Control) every two days for a total of three treatments. After five days of rest after the last treatment, 1 × 10^7^ CFU of E. coli wer e injected intraperitoneally. To assess the immune status of the mice, they were euthanized 24 h after injection of E. coli. Cytokine levels in serum were detected by ELISA, and the proportion of peritoneal monocytes, macrophages, M1 macrophages and M2 macrophages in peritoneal lavage fluid were analyzed by flow cytometry. In addition, organ damage was analyzed by H&E staining.

Evaluation of protective effect of CLP induced polymicrobial sepsis

4.24

A multi-species sepsis model was established by CLP. The CLP model was constructed according to the method described in the literature [42]. Specifically, after anesthesia and abdominal depilation were completed, the abdomen was oincised along the midline, the cecum of the mouse was punctured with a 21-gauge needle, the base was ligated with a 5/0 silk thread at the distal end of the ileocecum, and then the cecum was restored to the abdominal cavity and the wound was sutured. After successful modeling, the mice were randomly divided into three groups (n = 15), and treated with BGA-P@t-RADA16 (70 mg/kg), HC + PMB (HC 15 mg/kg and PMB 10 mg/kg) or PBS every 2 days for a total of three treatments. The weight of mice was assessed daily during the dosing period, and survival conditions were recorded. The protective effect was evaluated at different time points after the end of dosing. To assess the immune status of mice with sepsis, we euthanized the mice 24 h after the last treatment. We analyzed the proportion of peritoneal CCR2^+^ myeloid cells, monocytes and M1 macrophages in peritoneal lavage fluid, as well as the proportion of CD8^+^ T cells in the spleen by flow cytometry. Organ damages were accessed by H&E staining. On the sixth day after the end of dosing, the surviving mice in the BGA-P@t-RADA16 treatment group were randomly divided into two groups, which were intraperitoneally injected with PBS and clodronate liposomes respectively. And 24 h later, 5 × 10^6^ CFU of E. coli were injected intraperitoneally. After12 h, serum cytokine levels were detected by ELISA, and the bacterial load in the blood and abdominal cavity was quantitatively measured.

Statistical analysis

4.25

All statistical analyses were implemented using GraphPad Prism (version 9.5.0). The experimental data were shown as the mean ± Standard Error of the Mean (SEM). Sample size (n) for each statistical analysis was at least 3 samples. Two groups were compared using two-sided Student’s t-tests, and multiple groups were compared using one-way ANOVA with Tukey’s post-hoc tests, and p value < 0.05 was considered statistically significant. ∗ represents *p < *0.05, ∗∗ represents *p < *0.01, ∗∗∗ represents *p < *0.001, and ∗∗∗∗ represents *p < *0.0001.

CRediT authorship contribution statement

Lanbing Zou: Data curation, Formal analysis, Investigation, Software, Writing – original draft. Baixue Fu: Data curation, Methodology, Visualization, Writing – original draft. Dianyu Wang: Formal analysis, Investigation, Methodology. Yanbin Chen: Methodology, Validation. Han Gui: Formal analysis, Methodology. Ganen Mu: Methodology. Xinyi Li: Methodology. Yumin Zhang: Funding acquisition, Supervision, Writing – review & editing. Jianfeng Liu: Resources, Supervision. Cuihong Yang: Conceptualization, Formal analysis, Funding acquisition, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.