Macrophages, which are found in almost all tissues, play a critical role in host defence, wound repair, and immune regulation in mammals. Initially, these cells were classified as part of the monocyte-phagocyte system (MPS) and were believed to be derived solely from monocytes originating from bone marrow precursor cells (15). However, subsequent studies have challenged this view. It has been discovered that the majority of RTMs originate from multipotent erythroid-myeloid progenitor cells (EMPs), which are derived from yolk sac haematopoietic endothelial cells, and haematopoietic stem cells (HSCs) develop from aorta-gonad-mesonephros (AGM) haematopoietic endothelial cells (16, 17). Lineage tracing experiments have further confirmed these findings (18–20). The identification of embryonic-derived macrophages has revolutionized our understanding of the MPS and revealed the existence of diverse macrophage lineages.

Recent research has highlighted the importance of considering the differentiation process and dynamics in different environments when studying macrophage ontogeny. It has been proposed that the self-renewal of RTMs in situ and monocyte migration and differentiation are the two main sources of the maintenance and expansion of the RTM niche (6). The proportion of RTMs in different organs and tissues tends to change with age (21–26), indicating the necessity to differentiate between subsets and reconsider the functional mechanisms of peripheral blood-derived macrophages. Matthew D. Park and colleagues introduced the concept of nonsteady-state differentiation of monocytes, suggesting that under steady-state conditions, homeostasis-related signals such as M-CSF and GM-CSF contribute to the continuous differentiation of monocytes into RTMs. However, in disease states, disease-associated signals such as interleukin (IL)-33, TSLP, damage-associated molecular patterns (DAMPs), and pathogen-associated molecular patterns (PAMPs) override homeostatic signals, leading to the differentiation of monocytes into disease-associated macrophages (6). Another study demonstrated a similar trend in chronic granulomatous disease (CGD), in which disease progression was promoted by NOX2-producing mo-macs lacking the MHCII⁺CD206⁺CD36⁺ mature phenotype. However, when wild-type (WT) mo-macs and CGD mo-macs were transplanted into CGD mice and WT mice, respectively, their immunophenotype and maturation behaviour were reversed (27). This reversal was attributed to the ability of the inflammatory environment to remodel the phenotype and function of mo-macs.

Moreover, due to gene expression plasticity, macrophages can be polarized into different phenotypes and perform distinct functions in response to various environmental signals (28). The two main subpopulations are classically activated (M1) macrophages and selectively activated (M2) macrophages (29) ( Figure 2A ). Importantly, macrophage polarization is a dynamic process influenced by multiple environmental factors, and evidence suggests that this process is reversible, suggesting that even polarized macrophages can transition into another phenotype under new immune conditions (30, 31). A growing body of research has shown that macrophage polarization does not accurately and completely capture all phenotypic and functional differences in macrophages, and the heterogeneous functions of macrophages are depending not only on their subgroup but also on the function of the organ or tissue in which the cells are located (32). Therefore, a more detailed understanding of macrophage subtypes with distinct phenotypes and biological functions in different conditions requires further investigation.

During sepsis, macrophages are activated and assume a proinflammatory phenotype in response to the recognition of PAMPs by pattern recognition receptors (PRRs) on the cell surface. These cells are then recruited to the site of inflammation by the chemokines CCL3 and CCL4. Various TLR isoforms can be activated by different pathogenic microbial components. For instance, lipopolysaccharide (LPS) is recognized by the LPS binding protein (LBP), which transfers LPS to CD14 and MD-2 for delivery to macrophages. LPS then binds to TLR4 on the macrophage surface. Subsequently, TLR4 interacts with the MyD88 adaptor protein family, leading to the activation of mitogen-activated protein kinase (MAPK) and the nuclear factor-κB (NF-kB) signalling pathway, thereby inducing a shift in macrophages towards an M1-like phenotype (33, 34).

Activation by pathogens enhances phagocytosis, cytotoxicity, and antigen presentation. Macrophages clear pathogens through oxygen-dependent and oxygen-independent bactericidal systems and initiate innate immune defence responses. Concurrently, these cells recruit other immune cells to jointly increase the production of proinflammatory cytokines such as tumour necrosis factor-alpha (TNF-α), IL-1, IL-6, IL-8, IL-12, and migration inhibitory factor (MIF). Among these, IL-12 activates natural killer (NK) cells and promotes the differentiation of T cells into Th1 effector cells, which produce IFN-γ. IFN-γ enhances the phagocytic and bactericidal functions of macrophages by mediating STAT (35). Studies have indicated that IFN-γ can ameliorate sepsis by increasing the expression of HLA-DR in monocytes and the secretion of TNF-α (36). TNF-α and IL-1, which are released during the early stage of the inflammatory response, are potent proinflammatory mediators. These factors not only reinforce the differentiation of progenitor cells into macrophages and promote macrophage activation but also amplify the inflammatory cascade by activating other immune cells to secrete proinflammatory cytokines. Activated macrophages release MIF and upregulate microbial pattern recognition receptor expression, thereby strengthening the systemic inflammatory response. Previous studies have shown that MIF and CCL2 can cooperatively recruit inflammatory monocytes, leading to high levels of inflammatory factors in septic mice, which can result in lethal shock (37). During the early innate immune defence response triggered by PAMPs, the body promptly responds to the invasion of foreign pathogens. However, sustained release of DAMPs can cause an inflammatory cytokine cascade, triggering a rapid and overwhelming immune response that can lead to systemic response syndrome (SIRS). Notably, the distinction between PAMP and DAMP activation has been highlighted in studies. For instance, HMGB1 can act as a proinflammatory mediator when activated by PAMPs or as a DAMP when passively released from damaged cells during sterile tissue injury. Specific blockade of MD-2 binding to HMGB1 (P5779), which is a tetrameric peptide, without inhibiting the MD-2-LPS-TLR4 signalling pathway has been shown to attenuate DAMP-mediated inflammatory injury (38, 39). Then, activated M1 macrophages upregulate MHC II and the costimulatory molecules CD40, CD86, and CD80 while also secreting CXCL9, CXCL10, and CXCL11 to recruit Th1 cells and initiate the adaptive immune response (40). Moreover, M1 macrophages express high levels of inducible nitric oxide synthase (iNOS) and produce increased levels of NO and reactive oxygen species (ROS), which can damage microorganisms. These macrophages effectively activate and strengthen the immune response to protect the body against exogenous pathogen invasion. Macrophages contribute to the overwhelming immune response during the early stage of sepsis, which inevitably damages tissues and organs. In fact, these cells play dual roles.

As sepsis progresses, macrophages can mediate immunosuppression, which is characterized by a decrease in proinflammatory cytokine levels and an increase in anti-inflammatory cytokine production. This transition results in the transformation of M1 macrophages into a distinct immune state known as the M2 phenotype (31, 41). The activation of M2 macrophages is mediated by the Th2 cytokines IL-4 and IL-13, which act through the JAK-STAT6 pathway, as well as IL-10, which uses the JAK-STAT3 pathway. This activation causes Krüppel-like factor (KLF4) and STAT6 to dissociate from the NF-κB coactivator, resulting in the inhibition of NF-κB transcription (7, 42). Furthermore, the binding of IL-13 and IL-33 promotes the accumulation of the hypoxia-inducible transcription factor HIF-2α in M2 macrophages, leading to the upregulation of related genes such as arginase-1 (Arg-1), CCL17, and CCL24 (43). The upregulation of Arg1 and Arg2 expression in M2 macrophages enables the catabolism of arginine into ornithine and polyamines, which supports tissue repair in damaged areas (44, 45). Additionally, M2 macrophages exhibit high expression of the C-type lectin CD206 and the endocytic receptor CD163, enhancing their phagocytic activity and promoting the secretion of CCL17, CCL18, IL-10, transforming growth factor-beta (TGF-β), and IL-1R. This leads to the recruitment of Th2 cells, Treg cells, and other immunosuppressive cells that release significant amounts of anti-inflammatory cytokines. Moreover, M2 macrophages release TGF-β1 and platelet-derived growth factor (PDGF), which promote angiogenesis, tissue injury repair, and the differentiation of fibroblasts into myofibroblasts (46).

Importantly, although these macrophages exhibit marked upregulation of M2 genes, emerging evidence suggests that the reprogramming of these monocytes/macrophages in pathological environments, such as sepsis, differs from classic activation of M2 macrophages (47, 48). For instance, research indicates that P21 plays a crucial regulatory role in programming, and P21 deficiency can lead to an increase in IFN-β. Consequently, the ability of monocytes to be reprogrammed into a hyporesponsive phenotype, which helps avoid inflammatory injury, is compromised (48). Furthermore, several negative regulators of TLR signalling, including MyD88, IRAK-M, and ST2, have been shown to be upregulated in a macrophage tolerance model, resulting in the inhibition of M1 gene expression (49). This phenomenon, which is known as endotoxin tolerance, is characterized by reprogrammed macrophages with a blunted response to endotoxin, unchanged or enhanced release of anti-inflammatory cytokines such as IL-1R and IL-10 and decreased release of proinflammatory cytokines. Davis et al. discovered that impaired wound healing after recovery from sepsis was associated with altered inflammatory phenotypes of monocytes/macrophages. They found that the histone methyltransferase MLL1 could bind to the NF-κB locus, and methylation of the activating histone H3K4me3 altered the monocyte/macrophage phenotype, preventing the initiation of the first stage of wound repair even in the presence of a reduction in inflammatory cytokines. More importantly, studies of bone marrow transplantation have revealed that this epigenetic modification begins in bone marrow progenitor/stem cells and is transmitted to peripheral macrophages, allowing the damage to persist long after sepsis recovery (50). Overall, the reprogramming of macrophage phenotype and function is influenced by immunological homeostasis and the immune microenvironment. There may be a common link between the nonsteady-state differentiation of monocytes mentioned earlier and the endotoxin tolerance observed in reprogrammed macrophages, although further research is needed to establish this connection.

A reduction in the expression of HLA-DR on monocytes/macrophages, which has been associated with impaired monocyte/macrophage function and poor prognosis in sepsis patients, has been observed. In a prospective study, persistent expression of HLA-DR on less than 30% of monocytes was independently associated with mortality in sepsis (51). This downregulation may be linked to the decreased expression of MHC class II molecules and CD86 costimulatory molecules involved in IL-10 and TGF-β signalling, as well as impaired antigen presentation by monocytes/macrophages. Consequently, these cells are unable to bind to T lymphocytes, present pathogen proteins, and activate the adaptive immune response (49). The activation of effector T cells is a critical initial step in the adaptive immune response during sepsis that relies on antigen presentation by macrophages. In addition to their impaired function, macrophages also have increased expression of lymphocyte-negative costimulatory molecules such as PD-L1 and other ligands, thereby negatively regulating the immune response. Increasing evidence suggests a significant increase in T-cell apoptosis during sepsis-induced immunosuppression. Furthermore, the remaining T cells may experience functional exhaustion, which is characterized by impaired production of proinflammatory and anti-inflammatory cytokines, increased expression of negative costimulatory molecules such as PD-1 and TIM-3, and decreased expression of CD127 (IL-7Rα chain) (52, 53). In recent years, progress has been made in targeted therapies for T cells in the context of sepsis (54, 55). In vitro experiments using lymphocytes from sepsis patients have demonstrated that treatment with inhibitors of the immune checkpoints PD-1 or PD-L1 can reduce lymphocyte apoptosis (56, 57). Dual blockade of LAG3 and PD-1 has been shown to improve lymphocytic immunosuppression (58, 59).

Therefore, the reprogramming of macrophages from M1-like to M2-like cells, decreasing antigen presentation, and increasing the expression of negative costimulatory molecules are involved in the anti-inflammatory mechanism of sepsis and serve as a compensatory response to the hyperinflammatory state. However, this pathological anti-inflammatory response can lead to immune paralysis, resulting in more severe consequences, such as secondary infection and an increased risk of death (7).

MSCs were initially discovered in bone marrow but have since been identified in various tissues, such as bone marrow, fat, lung, heart, and muscle (60, 61). These cells possess the unique ability to self-renew and differentiate into mesoderm-derived cells such as adipocytes, osteoblasts, and chondrocytes, making them a valuable source for tissue regeneration (62). However, the lack of specific markers for MSCs and limited understanding of their in vivo biological characteristics pose challenges. Currently, the identification of MSCs relies on nonspecific cell markers (positive for CD29, CD51, CD73, CD90, and CD105; negative for CD31, CD34, and CD45), fibroblast morphology, and tri-lineage differentiation potential (63).

The multipotent differentiation potential of MSCs, combined with their low immunogenicity, safety profile, ease of availability, and ethical acceptance, positions them as promising candidates for stem cell-based regenerative therapies (64–67). The well-established capacity of MSCs to differentiate into various cell types has been extensively reviewed (68–71). Importantly, MSCs obtained from different sources and at different ages exhibit heterogeneous differentiation tendencies and degrees of stemness (72). Additionally, although MSCs lack haematopoietic functions and do not express CD34 (73), they play a unique role in haematopoietic support (74, 75). Perivascular MSCs maintain the self-renewal niche of HSCs by producing stem cell factors (SCFs) and CXCL12 (76). Bone marrow-derived MSCs (BMSCs) promote HSC proliferation and differentiation via the Noch and Wnt signalling pathways (77). A retrospective multicentre study involving 119 patients showed improved survival rates and reduced complications when HSCs and MSCs were cotransplanted (78). Furthermore, MSCs exhibit functional adaptability in tissue injury repair, including promoting angiogenesis, differentiating into myofibroblasts, inhibiting inflammation, and maintaining tissue homeostasis, depending on the type of injured tissue (76).

MSC transplantation has shown potential benefits in treating a wide range of diseases (79–83). However, the large size of MSCs (approximately 15 to 30 μm) makes them prone to entrapment in pulmonary capillaries when administered intravenously, limiting their distribution (84). This suggests the presence of alternative supplementary mechanisms through which MSCs exert therapeutic effects, and the immunomodulatory function of MSCs is believed to play a role in this context (85, 86). MSCs possess chemokine receptors that enable their migration to inflamed tissues, where they regulate the immune response, creating a favourable environment for tissue repair and activating endogenous repair mechanisms (87). It is worth noting that the immunomodulatory ability of MSCs is not constitutive but context-dependent and is influenced by the inflammatory environment. These cells exhibit plastic immunoregulatory functions and exert anti-inflammatory and proinflammatory effects depending on the specific immune microenvironment (88).

Recent studies have shown that MSCs exert rapid immunomodulatory effects that persist even after their depletion, making them advantageous for the treatment of immune-related diseases such as sepsis (83, 89, 90). In the early stage of sepsis, exogenous PAMPs, such as LPS derived from gram-negative bacteria, are recognized by TLR4 on the host cell membrane, and intracellular infection or stress signals activate TLR7 and TLR8 in intracellular host vesicles. Subsequently, the activation of regulatory factors such as MAPK and NF-kB activates immune cells to promote the secretion of inflammatory cytokines, chemokines, and antimicrobial peptides, ultimately initiating the innate immune response to eliminate foreign pathogens. HSP70 has been confirmed to activate the NF-κB signalling pathway through TLR2 on MSCs to induce a proinflammatory phenotype in MSCs (91). TLR4 synergistically recognizes and binds to LPS and has a similar effect (92), while TLR3 activation leads to improved survival in sepsis models (93). MSCs express TLR4 and TLR3, and their activation induces a proinflammatory phenotype (MSC1) or an anti-inflammatory phenotype (MSC2) (94) ( Figure 2B ). The phenotype of MSCs appears to be closely associated with Toll-like receptors, making them potential targets for regulating the immune response in sepsis.

Inflammatory cytokines play a crucial role in stimulating the immunomodulatory activity of MSCs. The upregulation of iNOS or indoleamine 2,3-dioxygenase (IDO) in most mammals serves as the immunosuppressive switch for MSCs during immune regulation in rodents (87). MSCs require activation by IFN-γ and multiple proinflammatory cytokines to upregulate iNOS expression and generate nitric oxide (NO). The secretion of chemokines by MSCs attracts T cells, which are subsequently inhibited by high levels of NO, leading to reduced proliferation and immune activity. In anti-inflammatory conditions, the immune microenvironment is reversed. The expression of iNOS is inhibited, and the secretion of chemokines by MSCs is not reduced, which enhances the immune activity of T cells. Overall, this reversal is influenced by inflammatory cytokines, which can modulate the direction of MSC immune regulation.

T lymphocyte apoptosis is a significant factor in sepsis-induced immunosuppression (95–97). The lymphocyte count serves as an important indicator of early sepsis and is associated with mortality (98, 99), particularly in elderly patients (100, 101). MSCs have emerged as a potential target for regulating the immune response by targeting T cells. Through the secretion of chemokines and adhesion proteins, MSCs recruit immune cells such as T cells (88). In the presence of high levels of inflammatory factors, MSCs directly inhibit T-cell proliferation and differentiation through the production of various immunosuppressive mediators, such as NO, IDO, prostaglandin E2 (PGE2), and TGF-β (102). However, this immunosuppressive effect decreases as the levels of inflammatory cytokines decrease ( Figure 3 ). Coculturing MSCs with activated T cells and adding dexamethasone to inhibit T-cell activation has been shown to reduce the production of proinflammatory cytokines and subsequently decrease T-cell immunosuppression (76).

As previously mentioned, macrophage heterogeneity is determined by their origin, immunophenotype, and tissue specificity. In the context of inflammatory injury, peripheral monocytes migrate to the site of injury and differentiate into various phenotypes based on the homeostatic or nonhomeostatic inflammatory environment. RTMs can also be reversibly polarized into different immunophenotypes and reprogrammed in an inflammatory setting. MSCs can modulate the immune phenotype and function of macrophages through paracrine immunomodulatory mediators, direct cell−cell contact, and the secretion of EVs, thereby restoring immune balance during sepsis.

Actually, the heterogeneity of macrophages poses challenges in studying the immunomodulatory effects and mechanisms of MSCs targeting macrophages in the context of sepsis. The incomplete immune phenotyping and functional definition of macrophages contribute to the difficulty, with oversimplified M1/M2 polarization paradigms prevailing. This oversimplification neglects the true and dynamic immunokinetics of macrophages in the immune microenvironment, hindering a comprehensive exploration of the mechanisms by which MSCs regulate immunity targeting macrophages. Macrophage heterogeneity also arises from their diverse origins within the same tissue. Tissue-resident macrophages maintain immune homeostasis and tissue repair under steady-state conditions, supplemented by peripheral blood monocyte-derived macrophages in varying proportions across organs. Distinguishing between these populations becomes crucial during infections or diseases, where a rapid immune response requires recruitment of macrophages from peripheral monocytes. Current disease models often fail to differentiate between these populations, impacting our understanding of their distinct functions. Recent studies attempted to discern the differences between tissue-resident alveolar macrophages (TRAM) and monocyte-derived macrophages (MoAM) by assessing their impact on lung fibrosis post-injury (143). In a severe pneumonia model, Placental mesenchymal stem cell (PMSCs) treatment downregulated TNF-α expression, preventing the recruitment and M1 polarization of bone marrow-derived macrophages (BMMΦ), while preserving the quantity and function of alveolar macrophages (AMΦs). Further investigation revealed that PMSCs treatment promoted M2 polarization in both BMMΦ and AMΦs. But surprisingly, it enhanced antibacterial functions in AMΦs while inhibiting bacterial phagocytosis and reactive oxygen species (ROS) production in BMMΦ (144). This innovative study underscores the immunoregulatory differences of MSCs on tissue-resident AMΦs and recruited BMMΦ populations. As the concept of non-steady-state differentiation of monocyte-derived macrophages emerges (6), future research should precisely define disease-related mo-macs, explore common pathways across diseases, unravel conservative molecular programs in disease participation, and elucidate specific mechanisms of MSCs targeting mo-macs. Furthermore, clarifying the temporal or spatial relationship between steady-state and non-steady-state differentiation of monocyte-derived macrophages and M1/M2 polarization of macrophages will further enhance our understanding and regulation of MSCs’ immunomodulatory pathways and mechanisms targeting distinct macrophage subpopulations, guiding the direction of MSC therapy for sepsis.

The occurrence and progression of sepsis are closely related to immune imbalance, and MSCs possess immunomodulatory abilities. Previous studies have indicated that the ability of MSCs to reduce mortality in mice is dependent on the presence of macrophages, while the depletion of T cells, B cells, or NK cells does not affect the beneficial effects of MSCs (109). Furthermore, the anti-inflammatory effects of MSCs are hindered when macrophages are depleted, as demonstrated by Cutler et al (145). These findings suggest a significant connection between the immunomodulatory effects of MSCs and host macrophages. Consequently, it is crucial to investigate the immune interactions between MSCs and macrophages during sepsis.