Unraveling the regulatory role of intercellular communication in intestinal immune cells mediated by H₂ in sepsis recovery through single-cell RNA sequencing

Abstract

Background

Sepsis is a life-threatening condition with dysregulated immune responses, where the intestinal immune system plays a pivotal role. While hydrogen gas (H₂) has shown therapeutic potential, its precise impact on intestinal immune cell communication remains elusive.

Methods

We employed a murine model of sepsis induced by cecal ligation and puncture (CLP). Mice were divided into control, sepsis, and H₂-treated sepsis groups. A comprehensive approach combining single-cell RNA sequencing (scRNA-seq) of intestinal tissues with histological, biochemical, and functional analyses was used to dissect the immune landscape and H₂’s effects.

Results

H₂ treatment significantly ameliorated intestinal injury, reduced epithelial barrier leakage (evidenced by decreased serum DAO and FITC-dextran), and suppressed systemic inflammation. scRNA-seq revealed that H₂ restored sepsis-induced perturbations in both B and T cell subsets. It normalized the aberrant expansion of non-functional B cells and rectified T cell differentiation trajectories, promoting a functional balance. Mechanistically, H₂ modulated key genes (e.g., Stat3, Ccl5, Foxo1) and critical pathways including oxidative phosphorylation, B cell receptor signaling, and Foxo signaling. Furthermore, cell-cell communication analysis demonstrated that H₂ robustly enhanced the weakened interactions between T cells, B cells, and epithelial cells, particularly via the CCL signaling network.

Conclusion

Our study elucidates that H₂ therapy alleviates sepsis by systemically reprogramming the intestinal immune micro environment at the single-cell level. This is achieved by normalizing immune cell subsets, rewiring intercellular communication, and ultimately rebalancing the inflammatory response, positioning H₂ as a promising immunomodulatory agent for sepsis.

Background

Sepsis presents a critical global health burden, characterized by its high incidence, increasing trend, and devastating mortality. This life-threatening condition, caused by a dysregulated host response to infection, claims the lives of 30% to 50% of patients who advance to the most severe stages of septic shock [1]. The pathogenesis of sepsis is exceedingly complex, involving dysregulation of the immune, inflammatory, and coagulation systems. Despite this understanding, effective targeted therapies remain elusive in clinical practice. Current management primarily relies on supportive care, including antibiotics, fluid resuscitation, and organ function support [2, 3]. While these measures can alleviate symptoms, they often fall short of addressing the underlying pathological mechanisms, thereby failing to significantly reduce the high mortality rate associated with the disease [4–6].

The intestine, being one of the largest immune organs, plays a pivotal role in maintaining both intestinal immune homeostasis and systemic immune balance [7, 8]. During sepsis, intestinal immune cells are a primary site of impact. Alterations in their function and quantity compromise the gut immune barrier, leading to the translocation of gut bacteria and endotoxins. This process, in turn, activates a systemic inflammatory response, perpetuating a vicious cycle that exacerbates the severity of sepsis. Critically, this immune dysregulation manifests as a complex interplay between hyperinflammation and profound immunosuppression. In the hyperinflammatory phase, immune cells within the gut, such as macrophages, T cells, and B cells, become aberrantly activated during sepsis, releasing copious amounts of inflammatory cytokines including tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) [9]. These cytokines not only cause direct damage to the intestinal tissue but also enter the systemic circulation, contributing to multiple organ dysfunction [10–12]. Conversely, the disease is also characterized by severe immunosuppression, driven by sepsis-induced increased apoptosis of intestinal immune cells. This state of immunosuppression diminishes the host’s resistance to pathogens, leading to a heightened susceptibility to secondary infections [13, 14].

In recent years, hydrogen gas (H₂) has emerged as a novel therapeutic agent with promising potential for sepsis management. Owing to its biological properties—including selective antioxidant, anti-inflammatory, and anti-apoptotic effects—H₂ can mitigate oxidative stress and inflammatory responses during sepsis, thereby protecting cells and tissues from damage [15, 16]. Studies have indicated that H₂ improves sepsis outcomes by modulating the gut microbiota and reducing oxidative stress and inflammation [16]. However, the precise molecular mechanisms underlying H₂’s therapeutic effects remain incompletely elucidated. However, the precise molecular mechanisms underlying H₂’s therapeutic effects remain incompletely elucidated. Specifically, while its ROS-scavenging abilities are well-documented [17–19], its impact on the intercellular communication among intestinal immune cells represents a critical knowledge gap. Because excessive oxidative stress and inflammatory cytokines physically disrupt cellular receptors and aberrant intracellular signaling pathways, we hypothesize that H₂ does not merely neutralize toxins, but actively reshapes the biochemical microenvironment. By removing these disruptive elements, H₂ is expected to restore normal receptor-ligand interactions and rewire the intercellular signaling networks necessary for immune homeostasis.

Historically, investigating these complex cellular networks has been hindered by methodological constraints. Traditional bulk tissue sequencing inherently masks the vast cellular heterogeneity and dynamic phenotypic shifts of individual immune subsets, failing to capture the intricate crosstalk between cells [20–22]. As a powerful tool for deconvoluting cellular heterogeneity and transcriptional profiles, single-cell RNA sequencing (scRNA-seq) allows for an in-depth exploration of the changes in intestinal immune cells throughout sepsis and the regulatory mechanisms of H₂ therapy. Our research aims to employ scRNA-seq to elucidate the intricate cell-cell communication network within the intestinal immune compartment of septic mice and to interrogate how H₂ intervention reconstitutes this network. The specific objectives are threefold: First, to define the repertoire and phenotypic characteristics of intestinal immune cell subsets and their sepsis-induced alterations. Second, to reconstruct the intercellular signaling network and identify the key pathways driving disease progression. Third, to determine the modulatory effects of H₂ treatment on this communicative landscape, thereby revealing the molecular basis for its therapeutic efficacy against sepsis.

The study profoundly impacts theory and practice. Theoretically, it pioneers single-cell investigation of H₂-mediated immunomodulation, shifting paradigms in sepsis pathogenesis and invigorating immunology. Practically, it identifies new therapeutic targets and accelerates hydrogen therapy translation, promising improved survival, quality of life, and socio-economic benefits.

Experimental method

Experimental animals and model construction

This study utilized SPF-grade male C57BL/6 mice (6–8 weeks old, 20–25 g) purchased from Shandong Aimaodakang Life Sciences Co., Ltd. Male mice were exclusively selected for this initial investigation to minimize confounding biological variability introduced by the estrous cycle and female sex hormones, which are known to exert distinct immunomodulatory effects in sepsis [23, 24]. After a one-week acclimatization under controlled conditions (22 ± 2 °C, 50–60% humidity, 12 h/12 h light-dark cycle) with ad libitum access to food and water, experiments were commenced.

The cecal ligation and puncture (CLP) procedure was employed to establish the murine sepsis model. The specific protocol was as follows: Mice were fasted for 12 h with free access to water to reduce potential interference from intestinal contents. Anesthesia was induced by intraperitoneal injection of 1% sodium pentobarbital solution (50 mg/kg body weight). The anesthetized mouse was fixed in a supine position on the operating platform, and the surgical area in the lower left abdomen was disinfected with iodophor. Under aseptic conditions, a longitudinal incision of approximately 1–1.5 cm was made in the lower left abdomen. The abdominal muscle layers were bluntly dissected to expose the cecum. A 3 − 0 silk suture was used to ligate the cecum at a position approximately one-half the distance from the cecal tip. The ligation tightness was carefully moderated—sufficient to obstruct the passage of intestinal contents without completely compromising blood flow, thereby preventing premature cecal necrosis. Following ligation, the ligated distal cecum was punctured through-and-through 2–3 times with a 21-gauge needle. A small amount of fecal content was gently extruded to ensure the continuous leakage of bacteria and endotoxins into the peritoneal cavity, thereby inducing a sustained systemic infection and inflammatory response. The cecum was then carefully returned to the abdominal cavity. The peritoneal layer and skin were sequentially sutured closed using a 4 − 0 silk suture. Immediately after surgery, a pre-warmed (37 °C) sterile saline solution was administered subcutaneously (1 ml/10 g body weight) for fluid resuscitation to compensate for perioperative fluid loss, maintain hemodynamic stability, and support recovery [25]. Mice in the sham-operated group underwent identical procedures, including anesthesia, laparotomy, and cecal exposure, but without cecal ligation or puncture. The abdomen was subsequently closed, and the same postoperative fluid resuscitation was provided.

Experimental grouping and treatment

We randomly allocated the mice into four groups of ten: a control group (C), a sepsis group (S), a control group with hydrogen inhalation (CH), and a sepsis group treated with hydrogen (SH). Group C received a sham operation. Group S underwent CLP to induce sepsis. Group SH was subjected to CLP and then, immediately after surgery, received hydrogen therapy (inhaling a 2% H₂ mixture at 1 L/min for 2 h in a sealed chamber); this treatment was repeated once daily for 3 days. Group CH served as a sham-operated control that received the identical hydrogen inhalation protocol. We monitored all mice closely throughout the study for general condition and recorded survival outcomes.

Detection indicators and methods

At the experimental endpoint, mice were euthanized, and intestinal tissue samples were promptly collected for single-cell RNA sequencing (scRNA-seq) analysis. The intestinal tissues were minced and subsequently digested into a single-cell suspension using a combination of enzymatic digestion and mechanical trituration. Single-cell libraries were constructed and sequenced on the 10x Genomics platform, with a target sequencing depth of at least 50,000 reads per cell. Following sequencing, the raw data underwent standard preprocessing steps, including quality control, alignment to the reference genome, and gene annotation. Subsequent data analysis was performed using the Seurat package to identify distinct immune cell subsets within the intestine and to analyze their gene expression profiles. Furthermore, cell-cell communication networks were inferred and constructed using the CellChat tool to decipher intercellular signaling interactions.

Preparation of single-cell suspensions

To obtain single-cell suspensions from the small intestine, tissues were harvested at the experimental endpoint (3 days post-CLP). After rinsing and mincing, tissue fragments were enzymatically digested in Collagenase IV/DNase I solution at 37 °C for 30 min with agitation. The resulting suspension was sequentially filtered (30 μm), centrifuged(500 g, 5 min), and subjected to erythrocyte lysis. The purified cell population was finally resuspended in complete DMEM (10% FBS) at a concentration of 1 × 10⁶ cells/mL, ready for scRNA-seq library preparation.

Single-cell RNA sequencing

Sequencing libraries were prepared using the 10x Genomics Chromium single-cell platform, each group of samples consisted of 3 mice that underwent the same treatment and were all separately subjected to the assay. The prepared cell suspension was co-encapsulated with uniquely barcoded gel beads into GEMs within a microfluidic chip. Inside each GEM, wherein cell lysis and reverse transcription take place, the released mRNA molecules were reverse-transcribed into cDNA, each tagged with a cell-specific barcode. The pooled barcoded cDNA was then purified and used to construct the sequencing library through standard steps: end repair, A-tailing, adapter ligation, and PCR amplification. The resulting libraries, upon passing QC, were subjected to high-throughput sequencing on an Illumina NovaSeq 6000 system using a paired-end 150 bp (PE150) read length.

Data analysis

Data preprocessing: Raw sequencing data were processed using the CellRanger software suite (10x Genomics) for demultiplexing, alignment to the reference genome (mm10), and barcode counting to generate a gene expression matrix per cell. Low-quality cells were filtered out based on the following criteria: cells with fewer than 500 detected genes or with a mitochondrial gene ratio exceeding 20%. Additionally, genes detected in fewer than 3 cells were excluded from subsequent analysis.

Cell clustering and annotation: The filtered gene expression matrix was analyzed using the Seurat package in R. Data were normalized and scaled, and highly variable genes were identified for dimensionality reduction via principal component analysis (PCA). Cells were clustered using a graph-based clustering algorithm (e.g., Louvain) on the significant principal components. Distinct cell clusters were subsequently annotated into specific cell types based on the expression of well-established canonical marker genes.

Differential expression analysis: Differentially expressed genes (DEGs) between experimental groups (e.g., Control vs. CLP vs. CLP + H₂) within specific cell clusters were identified using the FindMarkers function in Seurat. Genes with an absolute log2-fold change (|log2FC|) > 1 and an adjusted p-value < 0.05 were considered statistically significant DEGs.

Functional enrichment analysis: Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed on the identified DEGs using the clusterProfiler R package. This was done to elucidate the involved biological processes, molecular functions, and key signaling pathways.

Analysis of intercellular communication: Cell-cell communication analysis was conducted with the CellChat tool. By leveraging a curated database of ligand-receptor interactions, CellChat was used to infer and compare intercellular communication networks and interaction strengths among intestinal immune cell subsets across different treatment groups, specifically to assess the impact of H₂ intervention.

Experimental verification method

Inflammatory Factor Detection: The concentrations of inflammatory cytokines, including Tumor Necrosis Factor-alpha (TNF-α), Interleukin-6 (IL-6), and Interleukin-1beta (IL-1β), were measured in mouse serum and intestinal tissue homogenates using commercially available Enzyme-Linked Immunosorbent Assay (ELISA) kits, following the manufacturers’ protocols. Absorbance was read with a microplate reader, and cytokine concentrations were calculated based on the standard curve.

Detection of oxidative stress indicators: Oxidative stress levels were evaluated by measuring the content of Malondialdehyde (MDA) and the activities of Superoxide Dismutase (SOD) and Glutathione Peroxidase (GSH-Px) in serum and intestinal tissues. MDA content was determined by the thiobarbituric acid (TBA) method, SOD activity by the xanthine oxidase method, and GSH-Px activity by a colorimetric assay, all according to the specifications of the respective assay kits.

Protein expression detection: Protein expression of tight junction proteins (e.g., ZO-1, Occludin), apoptosis-related proteins (e.g., Bax, Bcl-2), and signaling pathway proteins associated with H₂ action (e.g., Nrf2, HO-1) was assessed by immunohistochemistry (IHC) and Western Blot (WB). For IHC, paraffin-embedded intestinal tissue sections were subjected to deparaffinization, rehydration, antigen retrieval, blocking, and sequential incubation with primary and secondary antibodies, followed by chromogenic development and counterstaining. Staining was visualized and evaluated under a microscope. For WB, total protein was extracted from intestinal tissue homogenates, quantified, separated by SDS-PAGE, and transferred onto a membrane. After blocking, membranes were incubated with specific primary and secondary antibodies. Protein bands were visualized using a chemiluminescence detection system, and band intensities were semi-quantified using ImageJ software.

Result

H₂ alleviates intestinal injury and systemic inflammation in septic mice

In healthy mice, intestinal immune homeostasis is maintained through coordinated interactions among immune cells [26]. T cells, particularly αβ T cells in the epithelium and lamina propria, are central to this process, differentiating into Th1, Th2, Th17, and Treg subsets to eliminate pathogens and preserve immune tolerance [27–33]. B cells in the lamina propria and Peyer’s patches undergo class switching to produce secretory IgA, which neutralizes pathogens and toxins in the gut lumen [34, 35] (21, 22). To investigate the therapeutic potential of hydrogen (H₂) in sepsis, we first evaluated the macroscopic intestinal injury and systemic inflammatory response. We collected small intestinal tissues from the following groups. The experimental groups were: sham operation group (Sham), sham operation plus hydrogen inhalation group (Sham + H₂), sepsis induced by cecal ligation and puncture group (CLP), and sepsis plus hydrogen treatment group (CLP + H₂). The results showed that the survival rate of mice in the CLP group decreased sharply within days after surgery, while the survival rate of mice in the CLP + H₂ group was significantly improved. Log-rank test showed a significant difference in survival curves between the CLP group and the CLP + H₂ group (p = 0.0041) (Fig. 1A). To investigate the therapeutic potential of hydrogen (H₂) in sepsis, we first collected small intestinal tissues from the following groups: normal control (Group C), sepsis (Group S), normal control with H₂ inhalation (Group CH), and sepsis treated with H₂ (Group SH) (Fig. 1B). Histopathological examination via H&E staining revealed significant intestinal injury in septic mice, characterized by local vasodilation, increased permeability, and edema, indicative of plasma extravasation—a physiological defense mechanism aimed at diluting toxins and recruiting immune cells (Fig. 1C). A marked reduction in the vertical distance from the villus tip to the villus-crypt junction was observed, reflecting substantial epithelial cell loss and serving as a direct indicator of damage. Consistent with this, the intestinal wet/dry ratio was significantly elevated in the sepsis group, suggesting severe edema, which was ameliorated by H₂ treatment (Fig. 1D). We further assessed intestinal barrier integrity using several established metrics. Serum levels of diamine oxidase (DAO), a sensitive and specific biomarker for intestinal mucosal injury, were significantly elevated in septic mice and reduced following H₂ therapy (Fig. 1E). Similarly, the intestinal permeability assay using FITC-dextran showed a significant increase in plasma fluorescence in septic mice, indicating a “leaky gut,” which was notably mitigated by H₂ treatment (Fig. 1F). The structural damage was quantitatively evaluated using the Chiu’s score, a semi-quantitative pathological scoring system. Septic mice exhibited significantly higher Chiu’s scores, which were effectively reduced upon H₂ intervention (Fig. 1G). The systemic inflammatory response was evaluated by measuring key pro-inflammatory cytokines. Levels of IL-1β, TNF-α, IL-6, IL-16, and IL-18—core drivers of the systemic inflammatory response syndrome (SIRS)—were substantially elevated in the sepsis group. Specifically within the context of intestinal immunity, these cytokines act as critical mediators of mucosal damage [36]. TNF-α and IL-1β directly induce apoptosis of intestinal epithelial cells and disrupt tight junction proteins, leading to the compromised barrier integrity (as evidenced by our DAO and FITC-dextran results). Concurrently, elevated IL-6 and IL-18 promote the recruitment and pathological hyperactivation of local lamina propria immune cells, perpetuating a vicious cycle of mucosal inflammation [37, 38]. H₂ treatment significantly suppressed the release of these cytokines (Fig. 1H). Collectively, these results demonstrate that H₂ inhalation effectively alleviates sepsis-associated intestinal injury, barrier dysfunction, and systemic inflammation in mice. This prompted us to investigate the underlying mechanisms of H₂-mediated protection. To this end, we performed single-cell RNA sequencing (scRNA-seq) on small intestinal tissues (Fig. 1I). Unsupervised clustering of the scRNA-seq data identified 18 distinct cell populations, which we annotated into five major lineages for focused investigation: Plasma cells, Epithelial cells, NK & T cells, B cells, and T cells (Fig. 1J). These cell types are critically involved in the pathophysiology of sepsis, governing immune responses, maintaining barrier integrity, and mediating inflammation. Their dysfunction is central to the progression from infection to multi-organ failure. Our subsequent analysis aims to decipher how H₂ modulates the interplay among these key cellular players to confer protection against sepsis.

Fig. 1.

Single-cell RNA sequencing of the small intestine in septic mice with or without H₂ therapy. (A) Mouse survival curve. (B) Schematic illustration showing the collection of small intestinal tissues from mice across different experimental groups. (C) Representative H&E-stained sections of the small intestine. Scale bar, 100 μm. D-F. Quantitative assessments of intestinal barrier function and edema: serum diamine oxidase (DAO) concentration, intestinal permeability to FITC-dextran, and the intestinal wet/dry ratio. G. Histopathological evaluation of intestinal injury using Chiu’s score, based on H&E-stained sections. H. Serum levels of key pro-inflammatory cytokines (IL-1β, TNF-α, IL-6, IL-16, IL-18). I. Workflow for single-cell RNA sequencing (scRNA-seq) of the small intestinal tissues. J. UMAP plot displaying the cell populations identified through unsupervised clustering of the scRNA-seq data

Gene expression alterations in major cell lineages following H₂ therapy in sepsis

Gene expression profiles within distinct cell populations critically determine their functional states during sepsis. We analyzed the transcriptomic changes in key lineages—B cells, NK & T cells, epithelial cells, and T cells. In B cells from septic mice, we observed upregulation of H₂-Ab1, Ms4a1 (encoding CD20), and Pax5 compared with the sepsis group treated with hydrogen gas. Notably, a cluster of functionally interrelated genes—Fcer2a, Ly86, Ighd, Bank1, and Ebf1—was highly expressed in the sepsis group. H₂ treatment effectively attenuated these sepsis-induced upregulations, restoring the expression profiles towards baseline levels. Septic NK & T cells exhibited heightened expression of cytotoxic markers, including Gzmb (Granzyme B), Nkg7, and Cd8b1 (Fig. 2A, B). Following H₂ intervention, the expression levels of these markers were significantly reduced, indicating a moderation of their transcriptomic activation state. Intestinal epithelial cells in sepsis showed increased expression of Krt19 (a structural keratin), Cldn3 (Claudin-3, a tight junction protein essential for barrier integrity), and Ace2 (Angiotensin-converting enzyme 2, which exerts anti-inflammatory and antioxidant effects via the Ang-(1–7)/Mas axis). In T cells, sepsis induced the upregulation of Cd4. This may represent an adaptive mechanism to prevent excessive T cell exhaustion. Furthermore, a functionally complementary set of genes—Ccl5, Gzma, Ccl4, Itga1, and Gzmb—was enriched in the sepsis group. Another cluster, including Prkcq, Rgs10, Lef1, A630014C17Rik, and Tcf7, also displayed altered expression patterns(Fig. 2C). Crucially, H₂ treatment broadly normalized these sepsis-driven transcriptomic perturbations across the T cell lineage, suggesting a restoration of basal gene expression networks. (Further functional interpretations of these transcriptomic shifts are detailed in the Discussion section).

Fig. 2.

Single-Cell Transcriptomic Profiling of the Small Intestine in Septic Mice with or without H₂ Therapy. (A) UMAP visualization highlighting the expression of key genes (Gzmb, Nkg7, Cd8b1, Krt19, Cldn3, Ace2, Cd4, Tcf7) in NK & T cells, epithelial cells, and T cells, respectively. (B) Violin plots (or Dot plot) quantifying the expression levels of the selected marker genes shown in (A) across the indicated cell populations. (C) Heatmap displaying the expression of functionally coordinated gene sets in B cells (e.g., Fcer2a, Ly86, Ighd) and T cells (e.g., Ccl5, Gzma, Ccl4, and Prkcq, Rgs10, Lef1), defining their immune status during sepsis. (D) Bar plot illustrating the proportional changes in the abundance of major cell lineages among the four experimental groups (C, CH, S, SH), demonstrating the modulatory effect of H₂ on sepsis-induced B cell expansion. (E) Heatmap showing the average gene expression patterns of key marker genes used to identify and annotate the distinct cell clusters. (F) UMAP plot showing the comprehensive classification of all identified cell populations in the small intestine, including fibroblasts, macrophages, and proliferating T cells

H₂ treatment modulates cell population proportions and ameliorates sepsis-induced B cell dysregulation

We next investigated how H₂ treatment influences the proportions of these cellular lineages (Fig. 2D). In healthy mice (Group C), B cell numbers were relatively low. Interestingly, H₂ inhalation in healthy controls (Group CH) led to a modest increase in B cell abundance, suggesting that H₂ may exert a mild immunomodulatory stimulus on the baseline immune microenvironment. As an exogenous agent with anti-inflammatory properties, H₂ might indirectly promote B cell survival or differentiation by modulating key signaling pathways, rather than directly inducing proliferation. A marked expansion of B cells was observed in the sepsis group (Group S) compared to all other groups. Crucially, H₂ treatment (Group SH) effectively counteracted this sepsis-induced B cell expansion, restoring their abundance towards baseline levels. This preserves functional B cells and precursors, boosts antibody production by shifting differentiation toward plasma cells, and prevents immune paralysis or autoimmunity. Ultimately, it restores humoral function and immune homeostasis (Fig. 2D). Based on detailed gene expression profiling, we identified distinct subpopulations within the septic mouse intestine. Beyond the major lineages discussed, we observed a significant expansion of cells such as Fibroblasts, Macrophages, and Proliferating T cells, suggesting their potential critical roles in sepsis pathogenesis (Fig. 2E-F).

Transcriptional profiling of B cell subsets following H₂ therapy in sepsis

To decipher the specific impact of H₂ on B cell heterogeneity during sepsis, we annotated the intestinal B cells into 11 distinct subsets through scRNA-seq analysis, including ISG_B, GC_prolif, C_meta, B_naïve, PC, GC_LZ, and B_act (Fig. 3A). The dynamic changes of these subsets during sepsis followed a pattern of “compensatory activation in the early phase, followed by exhaustive dysregulation in the late phase.” Early sepsis (0–24 h) was characterized by the activation of B_naïve cells, initiation of the germinal center (GC) reaction (GC_prolif, C_DZ), a rapid response from ISG_B and Brm cells, and a transient increase in IgA_PC and PC. We next characterized the key functional molecules defining major B cell subsets in septic mice (Fig. 3B, C): B_act (Activated B Cells): The high expression of Braf (proliferation), Cd86 (costimulation), and Stat4 (cytokine signaling) underpins their role as core effectors, directly determining their anti-infection capacity and immunomodulatory function. B_naïve (Naïve B Cells): The constitutive expression of Cd79a, Ighm, and Cd79b forms the core functional unit of the B cell receptor (BCR), defining their antigen recognition capability and initial activation potential. GC_LZ (Germinal Center Light Zone B Cells): The specific expression of Bank1 (signal transduction), Ccnd3 (cell cycle control), and Bcl11a (differentiation fate) supports their critical role in antibody affinity maturation and differentiation decisions, directly impacting the quality and efficiency of the humoral response [39]. PC (Plasma Cells): The specific expression of Cd52 (cell surface marker), Rpl34 (protein synthesis), and Iglc3 (antibody structure) underpins their core function as high-volume antibody factories, crucial for pathogen clearance. ISG_mem (IFN-stimulated Memory B Cells): The specific expression of interferon-stimulated genes (ISGs) like Pml, Trim30a, and Rsad2 equips them with dual capabilities for rapid anti-infection responses and immunomodulation, balancing defense against secondary infections and uncontrolled inflammation [40, 41].

Fig. 3.

B cell heterogeneity and transcriptional profiles in the small intestine of septic mice with or without H₂ therapy. (A) UMAP projection of B cells from the small intestine, colored by annotation into 11 distinct subsets, including ISG_B, GC_prolif, C_meta, B_naïve, PC, GC_LZ, and B_act. (B) Violin plots (or Dot plot) displaying the expression levels of key marker genes used to define the major B cell subsets. (C) Heatmap visualizing the scaled expression of signature genes across the identified B cell subsets. (D) Stacked bar plot showing the relative abundance of each B cell subset in the Control (C), Control + H₂ (CH), Sepsis (S), and Sepsis + H₂ (SH) groups. (E) UMAP plots split by experimental group, illustrating the global shifts in B cell subpopulation composition. (F) Bar plots quantifying the proportional changes of selected, critically altered B cell subsets (e.g., B_act, PC) among the four groups

H₂ treatment normalizes sepsis-induced perturbations in B cell subsets

Comparative analysis revealed substantial differences in the proportions of these critical B cell subsets between untreated and H₂-treated septic mice. Subsets such as PC, GC_LZ, and B_act were particularly abundant (Fig. 3D-F). Notably, the abundance of B_act cells was significantly higher in the sepsis group but decreased following H₂ treatment.

Analysis of gene regulatory networks in B cell subsets following H₂ therapy for sepsis

Changes in gene expression induced by H₂ treatment drive a global rewiring of regulatory networks. In septic mice, H₂ therapy resulted in the upregulation of networks related to B cell activation, differentiation, and proliferation (Fig. 4A). This suggests that H₂ mitigates sepsis-associated immunosuppression and repairs B cell functional impairment, thereby contributing to the restoration of anti-infective immunity and immune homeostasis [42]. The core value of this modulation lies in the recovery of humoral immune function, offering a potential immunotherapeutic strategy for sepsis. Concurrently, H₂ treatment led to the downregulation of networks involving primary active transmembrane transport, respiratory chain complexes, oxidative phosphorylation (OXPHOS), and lymphocyte differentiation (Fig. 4A). This should not be interpreted as mere suppression, but rather as an adaptive recalibration in response to sepsis pathophysiology. By reducing excessive metabolic load and correcting aberrant immune differentiation, H₂ helps to protect and restore host homeostasis. Furthermore, we observed an upregulation of pathways including B cell receptor signaling, Viral myocarditis, and the Intestinal immune network (Fig. 4B, C), aligning with H₂’s known roles in immune modulation and organ protection. Conversely, pathways associated with Parkinson’s disease, Alzheimer’s disease, Thermogenesis, Oxidative phosphorylation, and multiple neurodegenerative diseases were significantly downregulated (Fig. 4B, C), consistent with H₂’s potent antioxidant and anti-inflammatory properties. Sepsis can induce brain oxidative stress and inflammation, accelerating neurodegenerative processes [42, 43]. H₂ counters this by reducing ROS and pro-inflammatory cytokines (e.g., TNF-α, IL-6), thereby mitigating neuronal damage. For instance, H₂ treatment in 5xFAD Alzheimer’s model mice lowered oxidative stress and improved cognitive deficits [44]. Mechanistically, H₂ may activate the Nrf2/HO-1 pathway and inhibit Aβ production and tau phosphorylation [45, 46]. Regarding thermogenesis and OXPHOS, H₂ improves mitochondrial function by normalizing dynamics (e.g., downregulating Drp1, upregulating Mfn2) and reducing excessive ROS production from an overactive electron transport chain [47, 48]. It may also modulate energy metabolism and thermogenesis via pathways like AMPK [49, 50]. The chemokine gene CCL5 emerged as a key node in these regulatory networks. We found that CCL5 expression was significantly downregulated by H₂ treatment, returning to levels comparable to those in healthy control mice (Fig. 4E). In sepsis, CCL5 expression is driven by NF-κB activation and ROS. H₂ likely suppresses CCL5 transcription by inhibiting NF-κB nuclear translocation and by directly scavenging cytotoxic ROS like the hydroxyl radical, thereby breaking a pro-inflammatory feedback loop [50, 51].

Fig. 4.

Regulatory signaling pathways in intestinal B cells of septic mice with or without H₂ therapy. A. Gene Ontology (GO) enrichment analysis showing biological processes and pathways altered in B cells following H₂ treatment in sepsis. B-C. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis illustrating the key signaling pathways modulated by H₂ intervention in septic mice. D. Gene Set Enrichment Analysis (GSEA) plots demonstrating the enrichment scores of representative pathways affected by H₂ treatment. E. Expression levels of CCL5, a key target gene co-regulated by multiple signaling pathways, across different experimental groups. Data are presented as mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001

Transcriptional profiling of T cell subsets following H₂ therapy in sepsis

H₂ Therapy Modulates the Composition and Transcriptional Profiles of Intestinal T Cell SubsetsTo evaluate the immunomodulatory effects of H₂ on the adaptive immune response, we characterized the intestinal T cell compartment. Based on gene expression profiles, T cells from the intestinal tissue were categorized into 15 distinct subsets, including Treg, ILC3, iNKT, Cycling_TNK, CD4_Tmem, ISG_TNK, ILC2, Tfh, CD4_Tn, Th1_Eff, CD8_Tn, ILCp, ILC1_trNK, CD8_Teff, and gDT cells (Fig. 5A). Each subset exhibited a unique transcriptional signature (Fig. 5C).Comparative analysis revealed notable differences in the distribution of these T cell subsets across the experimental groups (Fig. 5B, D-F). In septic mice, we observed significant perturbations in T cell subpopulation frequencies. Crucially, H₂ treatment induced a comprehensive recalibration of this landscape. Specifically, H₂ intervention increased the relative abundance of CD8_Tn, ILCp, ILC1_trNK, CD8_Teff, and gDT cells compared to the untreated sepsis group. Conversely, the proportions of Treg, ILC3, iNKT, Cycling_TNK, CD4_Tmem, ISG_TNK, ILC2, and Tfh cells were reduced following H₂ treatment. The detailed dynamic changes, physiological states, and regulatory relevance of these specific T cell subsets are summarized in Table 1. Beyond subset-specific proportional shifts, we identified shared transcriptional features across T cells in sepsis. Differential expression analysis revealed that key genes, including Stat3, Il21r, and Zfp36l2, were consistently upregulated across multiple T cell populations in the sepsis group (Fig. 5G). Following H₂ treatment, the expression levels of these core dysregulated genes were robustly modulated, indicating a transcriptomic shift towards immune homeostasis. (The functional implications of these subset shifts and shared gene alterations are detailed in the Discussion section).

Fig. 5.

Single-cell transcriptomic profiling reveals immune cell composition and dynamics across different groups. (A) UMAP visualization of immune cell subpopulations identified by single-cell RNA sequencing, with cell types annotated (e.g., gDT, Treg, ILC3, NK, CD8_Teff, etc.). (B) UMAP plots showing the distribution of immune cell subpopulations in groups C, CH, S, and SH. (C) Dot plot illustrating the expression levels (color scale) and proportion (dot size) of marker genes across different immune cell subpopulations. (D) Stacked bar charts (top: proportion of cell types; bottom: cell counts) comparing immune cell composition between groups C vs. CH and C vs. S. (E) Bar charts displaying the cell counts of each immune cell subpopulation across all samples, stratified by group (top: C, CH, S, SH; bottom: C, S). (F) UMAP plots highlighting the distribution of immune cell subpopulations in the comparison of groups C vs. S. (G) Volcano plot showing differentially expressed genes (DEGs) between groups, with red dots indicating upregulated genes, blue dots indicating downregulated genes, and gray dots indicating non-significant genes (|log2(Fold Change)| > 1, -log10(p.adjust) > 2 as thresholds, for example)

Table 1.

Definitions and primary biological functions of signaling pathway and intercellular communication abbreviations

Abbreviation	Full Name	Primary Biological Function in Immune Regulation
CCL	CC Chemokine Ligand	Mediates chemotaxis and recruitment of immune cells (e.g., monocytes, memory T cells, and dendritic cells) to sites of inflammation or normal lymphoid tissue homing.
CXCL	CXC Chemokine Ligand	Primarily regulates the migration and activation of neutrophils and lymphocytes during inflammatory responses and tissue repair.
WNT	Wingless-related integration site	Regulates cellular proliferation, differentiation, and cell fate determination. In immunity, it modulates T cell development and inflammatory responses (including both canonical and non-canonical pathways).
MHC-I	Major Histocompatibility Complex class I	Presents endogenous intracellular peptide antigens to CD8 + cytotoxic T cells, playing a critical role in clearing infected or dysregulated cells.
MHC-II	Major Histocompatibility Complex class II	Presents exogenous antigens primarily to CD4 + helper T cells, initiating the adaptive immune response and orchestrating B cell and macrophage activation.
ICAM	Intercellular Adhesion Molecule	Facilitates strong cellular adhesion between immune cells (e.g., T cell-B cell immunological synapse) and endothelial cells, essential for transendothelial migration and robust cellular communication.
CD45	Protein Tyrosine Phosphatase, Receptor Type, C (PTPRC)	A critical receptor-like protein tyrosine phosphatase expressed on all leukocytes; regulates T and B cell antigen receptor signaling thresholds and cellular activation.
CD22	Sialic acid-binding Ig-like lectin 2 (Siglec-2)	An inhibitory co-receptor primarily found on the surface of mature B cells; dampens B cell receptor (BCR) signaling to prevent overactivation and autoimmunity.
BCR / TCR	B cell receptor / T cell receptor	Core membrane receptors responsible for specific antigen recognition and subsequent lymphocyte activation
Nrf2 / HO-1	Nuclear factor erythroid 2-related factor 2 / Heme oxygenase-1	A master signaling pathway protecting against oxidative stress and mitigating inflammatory damage.
NF-κB	Nuclear factor kappa B	A central transcription factor driving the expression of pro-inflammatory cytokines and immune activation.
JAK-STAT	Janus kinase / Signal Transducer and Activator of Transcription	Transmits signals from cytokine receptors to the nucleus, skewing T cell differentiation (e.g., STAT3/STAT4).
PI3K-AKT	Phosphoinositide 3-kinase / Protein Kinase B	Regulates cellular survival, proliferation, and metabolic homeostasis in immune cells.
Foxo	Forkhead box O	Transcription factors essential for T cell survival, memory maintenance, and mitigating metabolic exhaustion.
OXPHOS	Oxidative phosphorylation	The primary mitochondrial metabolic pathway for cellular energy (ATP) production.
ROS	Reactive oxygen species	Highly reactive molecules causing oxidative damage; deeply involved in signaling and sepsis pathogenesis.
AMPK	AMP-activated protein kinase	A central cellular energy sensor regulating energy metabolism, mitochondrial function, and thermogenesis
HMGB1	High Mobility Group Box 1	A central cellular energy sensor regulating energy metabolism, mitochondrial function, and thermogenesis
TNF / TNFα	Tumor Necrosis Factor alpha	A master pro-inflammatory cytokine driving the systemic inflammatory response and inducing tissue damage or apoptosis
IL	Interleukin	A broad family of cytokines orchestrating immune cell activation, differentiation (e.g., Th17), or suppression
TGF-β	Transforming Growth Factor beta	A potent immunosuppressive cytokine secreted by Tregs and Bregs, crucial for maintaining tolerance but capable of causing immunosuppression in late sepsis
TLR / TLR4	Toll-like Receptor (4)	A key innate immune sensor (recognizing bacterial endotoxin) that triggers robust pro-inflammatory signaling cascades
PD-1 / PD-L1	Programmed cell death protein 1 / Ligand 1	A critical immune checkpoint pathway; high expression mediates T cell functional exhaustion during the immunosuppressive phase of sepsis
MAPK	Mitogen-Activated Protein Kinase	Intracellular signaling cascades activated by stress or cytokines, driving inflammation and cellular stress responses

Analysis of gene regulatory networks in T cell subsets following H₂ therapy for sepsis

H₂ treatment induced significant changes in the T cell transcriptional landscape, leading to a rewiring of their regulatory networks. In our septic mouse model, we observed a notable reduction in CD4_Tn (naïve CD4⁺ T cells) and CD4_Tmem (CD4⁺ memory T cells) following H₂ intervention (Fig. 5D). This section focuses on the molecular mechanisms underlying these changes. Downregulation of Key Genes in CD4⁺ Tmem and CD4⁺ Tn Cells. H₂ treatment led to the downregulation of genes including Foxo1, Zfp36l2, Fosl2, Isg15, Tnfrsf1b, and Irak2 in these subsets (Fig. 6A, B). Upregulation of Key Genes in CD4⁺ Tmem Cells. Conversely, H₂ treatment upregulated genes such as Hsph1, Camk2d, and Ltb in CD4⁺ Tmem cells (Fig. 6B), potentially through: Enhanced Cellular Stress Protection: By mitigating oxidative and endoplasmic reticulum stress, H₂ may activate the heat shock response (e.g., via HSF1), increasing Hsph1 expression and promoting cell survival. Improved Calcium Signaling and Activation: H₂ may help restore intracellular calcium homeostasis, enhancing Ca²⁺ signaling and subsequently inducing Camk2d expression. This supports T cell activation and sustains memory function. Strengthened Intercellular Communication: H₂ may help preserve the immune microenvironment, increasing the demand for effective cell-cell communication. This could upregulate Ltb, a key molecule for immune synapse formation and lymphoid tissue homeostasis. Metabolic-Epigenetic Crosstalk: Improved mitochondrial function and metabolic state in CD4⁺ Tmem cells induced by H₂ could influence histone modifications (e.g., H3K4me3) or transcription factor activity (e.g., NFAT), indirectly activating the transcription of these genes. Global Pathway Analysis of Upregulated Genes. GO analysis of upregulated genes in T cells after H₂ treatment revealed a significant enrichment in terms related to positive regulation of leukocyte activation, cell activation, lymphocyte activation, and T cell activation (Fig. 6C). H₂ may coordinate the differentiation of lymphocytes (T and B cells) by modulating key transcription factors (e.g., T-bet, GATA3, RORγt), thereby refining the adaptive immune network and improving post-sepsis immune reconstitution (Fig. 6D). Gene Set Enrichment Analysis (GSEA) of T cell differential genes identified two pathways with high enrichment scores: Ribosome and Foxo Signaling Pathway (Fig. 6E-F). The enrichment of T cell upregulated genes in these pathways post-H₂ treatment leads to the core conclusion that H₂ likely repairs sepsis-induced T cell functional impairment by enhancing protein synthesis and regulating cell survival/metabolism.

Fig. 6.

Analysis of gene regulatory signaling pathways in T cell subsets after H₂ treatment for sepsis. (A) Heatmap (or Violin/Dot plot) showing the expression of key differentially expressed genes (e.g., Foxo1, Zfp36l2, Hsph1, Camk2d) in specific T cell subsets like CD4_Tn and CD4_Tmem. (B) Schematic summary diagram illustrating the potential molecular mechanisms by which H₂ influences gene expression in CD4⁺ Tn and CD4⁺ Tmem cells, including attenuated oxidative stress/inflammation, modulated T cell activation, and epigenetic regulation. (C) Bar plot of Gene Ontology (GO) terms significantly enriched for biological processes related to lymphocyte activation. (D) Bar plot (or Bubble chart) of significantly enriched KEGG pathways, such as those involved in hematopoietic cell lineages and T cell differentiation. E-F. GSEA enrichment plots for the “Ribosome” and “Foxo signaling pathway” gene sets, respectively. The normalized enrichment score (NES) and false discovery rate (FDR) q-value are indicated for each plot.

H₂ treatment alters intercellular communication networks between T cell and B cell subsets in sepsis

Following H₂ treatment for sepsis, the functional and phenotypic changes in T and B cell subsets lead to a remodeling of intercellular signaling. This recalibration of cell-cell communication positively influences immune regulation and contributes to the restoration of normal immune function. We performed a comparative analysis of intercellular communication networks in septic mice with and without H₂ treatment (Fig. 7A-B). The analysis revealed that T cell-B cell communication signals were significantly weakened in septic mice. Notably, H₂ treatment robustly enhanced these communication pathways between T and B cells, thereby reversing the sepsis-induced state of impaired intercellular communication. Furthermore, we identified that other previously weak intercellular signals, such as the CCL-mediated signaling axis between T cells and epithelial cells, were also significantly enhanced by H₂ treatment (Fig. 7C-D). Beyond the CCL signaling network, H₂ treatment significantly influenced other critical immune signaling pathways, including the MHC-I, ICAM, CD45, and CD22 signaling networks (Fig. 7E). To ensure optimal readability and provide broader context, all signaling pathway and intercellular communication abbreviations mentioned in our network analyses, along with their full names and primary biological functions, are explicitly defined in Table 1. This comprehensive impact demonstrates that H₂ treatment induces changes in gene expression, which in turn alters the collective state of cells and rewires the intercellular communication network, ultimately reprogramming the global immune status toward recovery.

Fig. 7.

H₂ treatment alters intercellular communication between T cell and B cell subsets in sepsis. (A) Circle plot or network diagram depicting the overall strength and number of inferred ligand-receptor interactions between major B cell and T cell subsets across experimental groups. The thickness of edges represents communication probability or intensity. (B) Heatmap comparing the differential interaction strength of specific ligand-receptor pairs between B and T cells in the Sepsis group versus the Sepsis + H₂ group. Significantly strengthened pairs in the H₂-treated group are highlighted. (C) Bubble plot showing the contribution of specific sending cell populations (e.g., Epithelial cells) and receiving cell populations (e.g., T cells) to the overall CCL signaling network. The size and color of the bubbles represent the communication probability. (D) Visualization of the specific CCL ligand-receptor interactions (e.g., CCL25-CCR9) that are significantly altered by H₂ treatment, displayed as a differential heatmap or a separate network subgraph. (E) H₂ treatment reprograms key immune signaling pathways between T cells and other cells. A heatmap displays the differential interaction strength of signaling pathways (e.g., MHC-I, ICAM, CD45, CD22) mediated by key ligand-receptor pairs between the indicated cell populations in the Sepsis + H₂ group compared to the Sepsis group

H₂ treatment drives evolutionary divergence in B cell subsets via genetic reprogramming

Following H₂ treatment for sepsis, the altered intercellular communication between T and B cell populations (Fig. 7A-B) is fundamentally rooted in gene expression changes that redefine cellular subsets. We employed pseudotime analysis to reconstruct the differentiation trajectories of B cells and T cells. In septic mice, B cell populations exhibited a blurred differentiation landscape without clear boundaries between subsets (Fig. 8A). In contrast, H₂ treatment induced a more structured topology in septic mice, with clearly demarcated clusters and distinct branching points in the developmental trajectory (Fig. 8B). Healthy control mice displayed the most defined B cell compartmentalization, characterized by two major, well-separated populations (Fig. 8C-D).

Fig. 8.

H₂ treatment drives evolutionary divergence in B cell subsets via genetic reprogramming. (A) Trajectory plot of B cells from the Sepsis group, showing a continuum of states with poorly defined cluster boundaries. (B) Trajectory plot of B cells from the Sepsis + H₂ group, revealing clearer branching and distinct cluster segregation compared to A. (C) Trajectory plot of B cells from the Control group, demonstrating well-separated populations and a defined differentiation path under normal conditions. (D) Integrated trajectory plot comparing all groups (Control, Sepsis, Sepsis + H₂), highlighting the restorative effect of H₂ on B cell subset distribution and differentiation paths. (E) Expression patterns of key driver genes (e.g., Stat4, Klf4, Ccl5, Cd22, Bcl11a) across the B cell pseudotime trajectory or within defined subsets. Expression is visualized as a heatmap or overlaid on the trajectory plot

The trajectory analysis pinpointed several key genes that critically influence B cell differentiation and were simultaneously identified as differentially expressed genes between H₂-treated and untreated septic mice: Stat4, Klf4, Ccl5, Cd22, and Bcl11a (Fig. 8E). The functional implications of these driver genes in determining B cell fate are detailed in the Discussion section.

Conclusion and discussion

In this study, we investigated the therapeutic potential and preliminary mechanisms of hydrogen (H₂) gas for treating sepsis using a murine model. Our findings demonstrate that H₂ treatment significantly alters the immune microenvironment in septic mice. Through single-cell RNA sequencing, we observed substantial changes in the gene expression profiles of intestinal cell populations and a remodeling of intercellular communication networks following H₂ intervention. Mechanistically, the robust modulation of immune signaling pathways such as MHC-II and ICAM-1 by H2 raises an important question regarding its biochemical trigger. It is widely accepted that H2 selectively scavenges highly toxic reactive oxygen species (ROS). This direct antioxidant effect likely prevents the oxidative damage and subsequent internalization or degradation of surface receptors. However, our concurrent protein-level validation revealing the up-regulation of Nrf2 and HO-1 suggests a more targeted molecular cascade. We propose that H2 operates through a synergistic mechanism: direct ROS neutralization limits immediate cellular damage, acting as an upstream trigger that subsequently facilitates the sustained activation of the Nrf2/HO-1 signaling cascade, ultimately promoting the re-balancing of the immune network observed in our scRNA-seq data. These transcriptional and cellular shifts ultimately led to the normalization of key immune factor secretion, contributing to the alleviation of sepsis pathology (Fig. 9). Our transcriptomic data provides deeper insights into the cellular states and potential mechanisms driving sepsis pathology and H₂-mediated protection. In the septic microenvironment, the prominent upregulation of H₂-Ab1, Ms4a1, and Pax5 in B cells highlights an initial robust, yet potentially dysregulated, drive for B cell activation and differentiation [52, 53]. Furthermore, the enrichment of Fcer2a, Ly86, and Bank1 suggests a compensatory activation of innate recognition and inflammatory signaling networks within the humoral compartment [54–63]. Simultaneously, the heightened expression of Gzmb, Nkg7, and Cd8b1 in NK and T cells signifies an acute activation of cytotoxic lymphocytes, which, while crucial for inducing apoptosis in infected cells early on, may contribute to collateral tissue damage if unresolved [64–66]. Within the T cell compartment, the upregulation of Cd4 alongside Tcf7 may represent an adaptive host mechanism attempting to preserve T cell stemness and memory formation to prevent functional exhaustion amidst the intense inflammatory storm [67, 68]. This is further supported by the enrichment of gene clusters associated with chemotaxis (Ccl5, Ccl4), cytotoxicity (Gzma), and Wnt signaling (Prkcq, Lef1), which collectively define a highly activated, yet ultimately detrimental, immune status that accelerates sepsis progression [69–77]. Concurrently, the altered expression of structural and junctional genes (Krt19, Cldn3) in the intestinal epithelium underscores the severe compromise of barrier integrity, whereas the upregulation of Ace2 likely acts as a counter-regulatory attempt to exert anti-inflammatory and antioxidant effects via the Ang-(1–7)/Mas axis (49–51). Crucially, the capacity of H₂ therapy to modulate these aberrant transcriptomic signatures back towards homeostasis suggests that its protective efficacy is deeply rooted in preventing the over-activation and subsequent exhaustion of these key functional pathways.

Fig. 9.

Proposed mechanism of H₂ therapy in modulating the immune microenvironment in sepsis. This schematic diagram summarizes the proposed mechanism by which H₂ treatment alleviates sepsis. In the septic state (left), immune cells exhibit distinct marker gene expression profiles, leading to dysregulated cell subset composition and functional states. This results in aberrant intercellular communication networks. Consequently, the secretion of immune factors is skewed towards a pro-inflammatory and tissue-damaging profile. Following H₂ treatment (right), the expression of key driver genes is altered. This reprograms the identity and state of immune cell subsets (e.g., B cells, T cells), restores physiological cell-cell communication signals, and ultimately normalizes the secretion of immune factors. The collective effect of these changes is a rebalancing of the immune microenvironment and the amelioration of sepsis pathology

Beyond initial transcriptomic shifts, the dynamic alterations in immune cell proportions and subset trajectories heavily dictate sepsis outcomes. Our results showed a dramatic, sepsis-induced explosive expansion of B cells, which initially might appear as a protective response. However, this surge often correlates with immune dysregulation. As evidenced by our subset analysis, this expansion is predominantly driven by ‘non-functional’ immature and marginal zone B cells, which are inefficient at differentiating into protective antibody-producing plasma cells [78]. Furthermore, the potential differentiation of these proliferating cells into regulatory B cells (Bregs) could inadvertently secrete anti-inflammatory cytokines, exacerbating late-stage immunosuppression and suppressing pathogen clearance [79]. This aberrant proliferation also carries the risk of breaking immune tolerance, leading to autoantibody production and immune-mediated tissue injury [80]. The temporal exhaustion of the B_naïve reservoir, stalled germinal center (GC) reactions—likely due to insufficient Tfh help—and the decline in plasma cells signify a profound collapse in humoral immunity, compromising intestinal barrier integrity. Crucially, H₂ treatment normalized this B cell homeostasis, representing a key mechanism through which it alleviates immune dysfunction, mitigates the futile expansion of non-functional subsets, and interrupts the vicious cycle of “hyperinflammation to B cell exhaustion to immunosuppression" [81].

The H₂-driven evolutionary divergence in B cell subsets is fundamentally governed by key regulatory genes identified in our pseudotime analysis. For instance, the modulation of Stat4 [82] and Klf4 [83] likely refines B cell proliferation and shifts differentiation toward functional plasma cell pathways. Concurrently, the regulation of Ccl5 [84] may guide B cell spatial positioning, while the inhibitory co-receptor Cd22 [85] and the essential transcriptional regulator Bcl11a [86] orchestrate BCR signaling thresholds and early developmental fate. Together, the H₂-mediated reprogramming of these crucial genes ensures a structured and highly functional B cell compartmentalization during sepsis recovery.

In the T cell compartment, we observed that H₂ intervention successfully downregulated several critical stress- and activation-related genes (e.g., Foxo1, Stat3, Zfp36l2). We propose several potential mechanisms for this targeted regulation. Through the attenuation of oxidative stress and inflammation: the antioxidant capacity of H₂ may directly reduce ROS levels, thereby diminishing the transcriptional demand on genes like Foxo1 and Stat3, while inhibiting pro-inflammatory pathways (e.g., NF-κB, MAPK) to suppress Isg15 and Irak2 expression. The modulation of excessive T cell activation by H₂ may naturally lead to the downregulation of mRNA stability regulators (Zfp36l2) and transcription complex subunits (Fosl2). We cannot rule out the role of epigenetic regulation, as H₂ might alter the epigenetic landscape—such as modifying histone acetylation or DNA methylation at the promoters of these specific genes—leading to their sustained suppression and the ultimate functional reinvigoration of the T cell pool. Complementing the downregulation of exhaustion markers, H₂ treatment actively upregulates genes driving T cell survival and activation (e.g., Hsph1, Camk2d, Ltb). This is likely achieved through enhanced cellular stress protection, improved calcium signaling, and strengthened immune synapse formation. Crucially, our GSEA and KEGG pathway analyses underscore that H₂ likely repairs sepsis-induced T cell functional impairment by enhancing protein synthesis and regulating cell metabolism. Sepsis can damage the bone marrow hematopoietic microenvironment, leading to insufficient immune cell production [5]. By upregulating ribosomal genes and the Foxo signaling pathway, H₂ restores the energy and stable microenvironment essential for ribosomal protein synthesis, thereby ameliorating the ‘metabolic exhaustion’ state of T cells and facilitating post-sepsis immune reconstruction. Specifically within the T cell compartment, the dynamic shifts in subset proportions serve as core indicators of immune reconstitution. Our scRNA-seq data (and Table 2) comprehensively illustrate how H₂ recalibrates these subsets. During sepsis, the early hyper-activation of Th17 cells and the excessive proliferation of immunosuppressive Tregs severely impair pathogen clearance [87, 88]. H₂ therapy appears to counter this by mitigating oxidative stress and modulating specific cytokine networks (e.g., IL-33), thereby reducing excessive Treg expansion and restoring the functional CD4⁺ and CD8⁺ T cell pools [89]. Furthermore, the normalization of innate-like subsets such as iNKT and gDT cells by H₂ highlights its role in fine-tuning the balance between rapid mucosal defense and excessive tissue damage. At the molecular level, the shared transcriptomic dysregulation we observed across T cells—such as the uniform upregulation of Stat3 and Il21r in sepsis—reflects the intense systemic cytokine storm. The activation of the JAK-STAT signaling pathway by pro-inflammatory cytokines drives Stat3 transcription, which skews T cell differentiation and amplifies immune imbalance [90]. The concurrent upregulation of Il21r creates an autocrine/paracrine loop that exacerbates this over-activation [91]. By robustly modulating the expression of these core genes (Stat3, Il21r, and the RNA-binding protein Zfp36l2), H₂ disrupts these vicious positive feedback loops, preventing eventual T cell exhaustion and effectively restoring adaptive immune homeostasis in the intestine. This restoration of adaptive immune homeostasis is intricately linked to the H₂-induced rewiring of intercellular communication. The strengthened T cell-B cell interactions we observed likely stem from the restoration of antigen presentation signals (e.g., MHC class II upregulation on B cells) [53] and the reinvigoration of T cell help via recovered cytokine production. Concurrently, enhanced adhesion signals (e.g., ICAM-1) provide the structural stability necessary for these immunological synapses. Furthermore, H₂ significantly enhances the CCL-mediated signaling axis (e.g., CCL25-CCR9) between T cells and epithelial cells. By alleviating inflammatory damage via the Nrf2/HO-1/HMGB1 pathway [92], H₂ likely restores normal chemokine expression, thereby promoting the recruitment of T cells to the gut to maintain barrier function.

Table 2.

H₂ modulation of T cell subsets in sepsis

Cell Subset	Pathological Changes in Sepsis	Regulatory Role of H₂	Biological Significance
Treg (Regulatory T cells)	• Early mild increase, followed by excessive proliferation (> 20% of CD4⁺ T cells). • Over-suppression of effector T cells via IL-10/TGF-β, reducing anti-infection capacity. • Functional impairment, failing to repair the intestinal barrier.	1. Inhibits excessive Treg proliferation (restores proportion to 8%-12%). 2. Enhances Treg’s barrier repair function (upregulates IL-22 receptor expression). 3. Reduces immunosuppressive effect (downregulates IL-10 secretion).	Resolves the conflict between excessive immunosuppression and insufficient barrier repair. Maintains anti-pathogen capacity while suppressing hyperinflammation.
CD4+ Tmem (CD4⁺ Memory T cells)	• Significant numerical reduction (> 50% loss in lamina propria). • Functional exhaustion (high PD-1, Tim-3 expression). • Ineffective response to reinfection, leading to systemic spread.	1. Reduces apoptosis (scavenges ROS, inhibits caspase-3). 2. Reverses exhaustion (downregulates PD-1, restores IFN-γ/IL-2 secretion). 3. Promotes homing to lamina propria (upregulates CCR9).	Re-establishes intestinal “immune memory” for rapid clearance of translocated bacteria, preventing sepsis aggravation.
CD4+ Tn (CD4⁺ Naïve T cells)	• Reduced numbers in Peyer’s patches (impaired hematopoiesis & migration). • Skewed differentiation (excessive Th2, insufficient Th1). • Weakened cellular immunity against intracellular bacteria.	1. Promotes bone marrow-to-gut migration (upregulates S1P1 receptor). 2. Corrects differentiation imbalance (increases IL-12, promotes Th1 over Th2). 3. Maintains proliferative capacity (preserves Peyer’s patch structure).	Replenishes the “immune cell reserve,” enhancing Th1-mediated responses against intracellular bacteria, reducing relapse.
Th1_Eff (Th1 Effector cells)	• Early over-activation (cytokine storm: high IFN-γ/TNF-α) causing epithelial apoptosis. • Late exhaustion and severe depletion (< 30% of normal), failing to clear infected cells.	1. Early: Suppresses over-activation (reduces TNF-α, prevents epithelial damage). 2. Late: Promotes survival (activates Nrf2/HO-1, reduces oxidative stress). 3. Modulates cytotoxicity (restores appropriate perforin/granzyme B expression).	Balances anti-inflammation and anti-infection: prevents tissue damage while preserving core function of eliminating infected cells.
CD8+ Tn (CD8⁺ Naïve T cells)	• Marked decrease in intraepithelial lymphocytes (IELs) (< 15% vs. normal 40%). • Blocked differentiation into CD8⁺ Teff. • Loss of “cytotoxic defense” in the epithelium.	1. Promotes epithelial layer residency (upregulates integrin αEβ7). 2. Accelerates effector differentiation (upregulates IL-2R, promotes CyclinD2-mediated proliferation). 3. Protects from inflammatory damage (e.g., from IL-6).	Rebuilds the epithelial “cytotoxic barrier” to clear infected cells, preventing pathogen entry into circulation.
CD8+ Teff (CD8⁺ Effector cells)	• Early excessive mucosal infiltration causing edema. • Late functional impairment due to energy exhaustion (mitochondrial dysfunction) and high PD-L1 expression.	1. Reduces early infiltration (downregulates CCL5). 2. Improves mitochondrial function (clears mtROS, restores ATP). 3. Downregulates PD-L1 expression (restores killing activity).	Prevents excessive mucosal damage while maintaining killing capacity against translocated bacteria, reducing gut-derived infection risk.
iNKT (Invariant NKT cells)	• Numerical reduction in lamina propria (~ 60%). • Altered functional polarization: from anti-inflammatory (IL-4/IL-10) to pro-inflammatory (IFN-γ/IL-17).	1. Suppresses pro-inflammatory polarization (reduces IL-17, restores IL-4/IL-10). 2. Promotes survival (inhibits Fas/FasL-mediated apoptosis). 3. Enhances tolerance to commensal bacteria.	Maintains gut “immune tolerance-inflammation balance”: prevents mucosal inflammation while preserving pathogen recognition.
gDT (γδ T cells)	• Drastic numerical decline in IELs (< 25% of normal). • Loss of antibacterial function (cannot secrete LL-37, defensins) and epithelial repair signals (e.g., IL-22).	1. Rapidly restores numbers (promotes thymic migration via CXCR6 upregulation). 2. Restores antibacterial & repair functions (upregulates LL-37, IL-22). 3. Suppresses pro-inflammatory differentiation (reduces TNF-α, IL-1β).	Rebuilds the epithelial “first line of defense”: directly kills pathogens and promotes epithelial regeneration to repair barrier leaks.
Tfh (Follicular Helper T cells)	• Numerical reduction in Peyer’s patches (< 30% of normal). • Functional defect (impaired CD40L, IL-21 expression). • Leads to deficient plasma cell differentiation and sIgA production (> 40% reduction).	1. Promotes Tfh residency in Peyer’s patches (preserves follicular structure). 2. Restores helper function (upregulates CD40L, IL-21). 3. Enhances sIgA secretion (via modulating Tfh-B cell interaction).	Restores the “humoral immune barrier”: sIgA coats bacteria, preventing epithelial adhesion and translocation, targeting a root cause of sepsis.

The advantages of H₂ therapy for sepsis are multifaceted, centering on its anti-inflammatory and antioxidant properties, organ-protective effects, and high safety profile. Specifically, H₂ effectively downregulates pro-inflammatory cytokines such as IL-6 and IL-18, thereby curbing excessive immune activation and preventing cytokine storm-induced tissue damage. It uniquely modulates both innate and adaptive immune responses, rebalancing the inflammatory microenvironment—an advantage over single-target anti-inflammatory agents. By selectively scavenging cytotoxic reactive oxygen species (ROS), H₂ mitigates oxidative damage to critical organs, including the gut, lungs, liver, and kidneys. Notably, its ability to preserve intestinal mucosal barrier integrity and reduce bacterial translocation targets a pivotal mechanism in sepsis progression. A critical dynamic in restoring intestinal homeostasis is the causal hierarchy between epithelial repair and immune re-balancing. While our data confirms the synchronized recovery of both systems (evidenced by reduced DAO/FITC-dextran leakage and stabilized ZO-1/Occludin alongside immune subset normalization), deciphering the primary driver is complex. Given the extreme diffusibility of the H₂ molecule, it lacks the spatial limitations of conventional drugs. Therefore, we postulate a concurrent, rather than sequential, mechanism. H₂ rapidly penetrates both the epithelial layer to directly mitigate apoptosis and barrier disruption, and the underlying lamina propria to simultaneously reprogram hyper-active immune subsets. This dual-targeting capability underlines the high efficacy of H₂ in disrupting the vicious cycle of epithelial leakage and immune hyper-activation in sepsis. Its inert nature translates to a high safety margin with minimal risk of allergic reactions or organ toxicity. The flexibility in administration routes (inhalation, intraperitoneal injection, hydrogen-rich water) allows for adaptation to diverse clinical scenarios in septic patients. Furthermore, H₂ exhibits synergistic potential, complementing standard therapies like antibiotics, fluid resuscitation, and immunomodulators to enhance overall efficacy and potentially reduce associated side effects.

Despite its promise, several challenges must be addressed before H₂ therapy can be widely adopted in clinical practice. Standardization of administration and dosing remains a primary hurdle, as the bioavailability varies significantly across different delivery methods (e.g., inhalation vs. oral), and optimal dosage and treatment duration are yet to be established. The heterogeneity of sepsis necessitates personalized treatment protocols based on disease severity and stage, for which no consensus currently exists. Variable treatment efficacy among individuals, influenced by factors like age, comorbidities, and pathogen type, poses another challenge, and the underlying reasons for non-responsiveness in severe cases are unclear. The absence of reliable biomarkers to predict treatment success makes it difficult to identify the patient population most likely to benefit. Moreover, while H₂’s anti-inflammatory and antioxidant effects are recognized, the precise molecular mechanisms governing its regulation of specific immune cells (e.g., macrophages, B cells) and its differential impact on the various phases of sepsis (early hyperinflammation vs. late immunosuppression) require deeper mechanistic investigation. While our scRNA-seq analysis robustly identified distinct transcriptomic shifts in intestinal immune sub-clusters, a methodological limitation of the current study is the reliance on macroscopic and downstream functional assays (e.g., ELISA, intestinal permeability, and histological scoring) rather than direct protein-level quantification. Future studies incorporating high-dimensional flow cytometry or multiplex immunohistochemistry (mIF) are warranted to physically map and validate the exact phenotypic proportions of these specific immune cell subsets at the protein level. We acknowledge certain limitations in our experimental design. This study exclusively utilized male mice to minimize the confounding biological variability associated with the female estrous cycle and fluctuating sex hormones, which are known to exert independent immunomodulatory effects in sepsis. Consequently, the potential sex-specific differences in response to H₂ therapy remain unaddressed. Given the critical role of sex as a biological variable in immune responses, future studies incorporating both male and female cohorts are essential to fully elucidate the comprehensive therapeutic profile of H₂ in sepsis.

In conclusion, our single-cell transcriptomic analysis reveals that the therapeutic efficacy of H₂ in sepsis extends far beyond simple ROS scavenging; it acts as a profound systemic modulator of the intestinal immune microenvironment. Crucially, we demonstrate that H₂-mediated recovery is fundamentally driven by the metabolic and functional reprogramming of B and T lymphocytes. By reversing sepsis-induced metabolic exhaustion—specifically through the restoration of ribosomal protein synthesis and the activation of the Foxo survival pathway—H₂ effectively rescues adaptive immune cells from apoptosis and profound dysfunction. Coupled with the rewiring of critical intercellular communication networks, this transcriptomic recalibration breaks the vicious cycle of hyperinflammation and immunosuppression, ultimately restoring mucosal barrier integrity and systemic immune homeostasis. These insights firmly position H₂ inhalation as a highly promising, multi-target immunomodulatory therapy for severe sepsis. Most supporting data are derived from animal studies, and a scarcity of large-scale, multi-center clinical trials leaves a gap in high-quality evidence validating its efficacy in human sepsis. The development of convenient and reliable methods to monitor H₂’s biological effects in patients also represents a critical unmet need for clinical application.

Future perspectives for clinical translation

Despite the promising immunomodulatory effects observed in our study, translating H₂ inhalation therapy from murine models to human clinical practice involves overcoming substantial logistical and physiological barriers. Future research must prioritize the following practical challenges in the setting of the Intensive Care Unit (ICU):

Dosage Standardization and ICU Delivery: Translating the 2% H₂ inhalation protocol from spontaneously breathing mice to human ICUs presents significant pharmacokinetic challenges. In severe sepsis, patients frequently require invasive mechanical ventilation. Standardizing the inhaled H₂ concentration and gas flow rates for human ventilators requires rigorous safety and compatibility profiling. Future studies must determine how to achieve effective systemic tissue saturation of H₂ without interfering with complex respiratory support protocols or altering the delivered fraction of inspired oxygen (FiO2).

Optimal Timing of Administration: The pathophysiology of human sepsis is highly dynamic and heterogeneous, necessitating personalized treatment protocols. It typically transitions from an acute hyper-inflammatory “cytokine storm” to a protracted state of profound immunosuppression. While our controlled murine model administered H₂ immediately post-surgery, clinical patients present at highly variable disease stages. Therefore, a critical prerequisite for future clinical trials is identifying the optimal therapeutic window—specifically, determining whether H₂ is most efficacious when administered during the initial inflammatory peak to prevent organ damage, or during the late phase to reverse immune cell exhaustion.

Author contribution

Yanhua Luo: Roles/Writing - original draft, Yajuan Zhao: Roles/Writing - original draft, Investigation, Methodology, Jinfeng Liu: Investigation, Bin Liu: Visualization, Zhimei Zhang: Investigation, Yuan Jiang: Investigation, Methodology, Yaopeng Zhang: Investigation, Methodology, Xue-tingWang: Resources, Shilong Li: Data curation, Shuxin Jiang: Methodology, Nian Zhao, Data curation, Xiaokang Li: Visualization, Formal analysis, Software, Dong Xiong: Visualization, Shaoqiang Wang: Project administration, Weihang Zhang: Validation, Conceptualization, Bao Lang: Funding acquisition, Project administration, Supervision, Writing - review & editing.

Funding

National Natural Science Foundation of China International (Regional) Cooperation and Exchange Project(32261160571). Young Elite Sponsorship Program of Shandong Provincial Medical Association. 2024 Annual Project of Medical and Health Sciences of Shandong Province/General Project(20241800073). Project of Shandong Provincial Health and Medical Management Research Center (Project No.: sdwjyj2024lm1036).

Data availability

The datasets generated and/or analysed during the current study are not publicly available due [REASON WHY DATA ARE NOT PUBLIC] but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.