Macrophages in the skin can originate form distinct sources. Tissue-resident macrophages which originate from yolk-sac-derived erythro-myeloid progenitors during early embryonic phase and get renewed frequently (2). The other type are macrophages originating from circulating monocytes, which are recruited to the skin in response to chemotactic signals. In addition to macrophages, monocytes can differentiate into DCs, amplifying the inflammatory response. Monocytes contribute to the inflammatory cycle in AD through production of cytokines and reactive oxygen species, and by their differentiation into active macrophages that drive skin inflammation (3). These macrophages are central players in inflammation, becoming activated at the onset of the inflammatory process. This activation triggers the expression of many genes, enhancing their ability to eliminate bacteria and regulate other cells through cytokine and chemokine secretion (4). However, excessive macrophage activation can be detrimental, contributing to conditions such as septic shock, organ dysfunction syndrome, and chronic inflammatory diseases like psoriasis and AD (3, 5). Research by Kiekens et al. demonstrated that macrophage numbers are significantly increased in the skin of both the acute and chronic inflammation phase of AD compared to non-affected and healthy skin (6).

In acute and chronic AD lesions, the majority of macrophages originate from circulating monocytes that infiltrate the dermis, where they become key players in both immune defense and inflammation ( Figure 1 ). Whereas they are relative scarce in non-lesional skin. These macrophages, function as antigen-presenting cells (APCs), play a role in antimicrobial activity, preventing bacterial invasion, and drive inflammation by releasing pro-inflammatory cytokines (e.g., TNF-α, IL-1β) and chemokines, which recruit additional immune cells (7). In contrast, tissue resident macrophages primarily maintain skin homeostasis and have regulatory functions during inflammation. Moreover, they participate in tissue remodeling and fibrosis by secreting matrix metalloproteinases (MMPs) and other related factors. Toll-like receptor stimuli and IFN-g polarize monocyte-derived macrophages toward the classical M1 activation pathway, resulting in production of high levels of proinflammatory cytokines (including IL-1, TNF-α, IL-12, and IL-23), reactive nitrogen, and oxygen radicals, enhanced microbicidal activity, and promotion of Th1 responses. Signal transducer and activator of transcription (STAT) 1 activation is crucial in this context. Phenotypically, M1 cells express high levels of MHC class II and co-stimulatory molecules, such as CD80 and CD86, and upregulate the expression of intracellular suppressor of cytokine signaling 3 (SOCS3) and inducible nitric oxide synthase (iNOS). Therefore, M1 cells are implicated in initiating and sustaining an inflammatory response (8).

Beyond their inflammatory functions, monocyte-derived macrophages in AD can undergo alternative activation, which is marked by the increased presence of CD163⁺ cells in lesional AD skin (9). These CD163⁺ macrophages, which are indicative of alternatively activated macrophages, are more abundant in lesional AD skin compared to healthy skin, and they share a similar distribution pattern with CD68⁺ cells. The alternative activation indicates a distinct role of these macrophages in the chronic inflammatory environment of AD, where they may exert several functions that support the persistent nature of the disease.

While the M1/M2 framework has long served as a useful model for understanding macrophage function, it is now recognized as an oversimplification, particularly in chronic inflammatory diseases such as atopic dermatitis. Macrophages exhibit considerable phenotypic and functional plasticity, continuously adapting to their microenvironment. In the skin of AD patients, macrophages are exposed to a complex array of stimuli, including type 2 cytokines (e.g., IL-4, IL-13), microbial components (e.g., S. aureus-derived peptidoglycan), neuropeptides (e.g., neuropeptide Y), histamine, lipid mediators, and mechanical stress, which shape their activation states dynamically (10). Among microbial stimuli, S. aureus is a predominant source of PAMPs in AD (11, 12). Its colonization exacerbates barrier dysfunction and drives macrophage activation through TLR2/6 signaling, inducing proinflammatory cytokines such as IL-1β, TNF-α, and IL-6. These cytokines contribute to the inflammatory milieu and promote the differentiation of Th22s and Th17 cells. Consistently, elevated levels of Th2 and Th22 cytokines are observed in both acute and chronic AD lesions, where they further impair barrier integrity and may contribute to microbial dysbiosis (13). Prolonged exposure to S. aureus can also functionally reprogram macrophages, promoting regulatory or tolerogenic phenotypes beyond classical polarization states. Recent single-cell transcriptomic studies have identified macrophage subsets in inflamed human skin that defy traditional M1/M2 classification, instead displaying hybrid or intermediate profiles (14, 15). These include cells co-expressing pro- and anti-inflammatory markers or those with specific metabolic or tissue-remodeling programs. Such macrophage states may be further modulated by epigenetic reprogramming, tolerization due to chronic microbial exposure, or neuroimmune feedback. In AD skin, for instance, CD68⁺CD163⁺ cells co-expressing DC markers like CD1a suggest the presence of a heterogeneous pool of monocyte-derived macrophages with dual immunomodulatory and antigen-presenting functions (6). Understanding this spectrum of macrophage activation is critical for deciphering their roles in disease progression and resolution. It also holds therapeutic potential, as interventions may target distinct macrophage subtypes or reprogramming pathways, such as through TRPV4 activation or cytokine modulation, rather than broadly inhibiting or activating these cells.

Additionally, Chitinase 3-like 1 (CHI3L1), a recognized mediator in Th2-driven inflammation which is also known as breast regression protein 39 (BRP-39), has been shown to mediate the development of AD through the activation of macrophages (16). M2 macrophages, a subtype often associated with anti-inflammatory responses, have been shown to reduce disease severity in a mouse model of AD when selectively activated through the mannose receptor CD206 using the bee venom component phospholipase A2 (17). On the contrary, they have been identified as a major source of IL-31 in AD by the immunohistochemical analysis of skin biopsy samples from AD patients. Their interactions with TSLP, periostin, and basophils further contribute to AD pathogenesis and the perpetuation of the itch-scratch cycle (18). M2 macrophages are also known to produce C–C motif chemokine ligand 18 (CCL18), a chemokine strongly associated with increased morbidity in AD patients (19). IL-4 and IL-13 significantly upregulate CCL18 expression, with IL-10 also contributing, but to a lesser extent. Histamine further enhances the cytokine-induced upregulation of CCL18 mRNA expression by stimulating the histamine receptor 2 (H2R), with the strongest effect observed in IL-10-stimulated macrophages. These combined activations in macrophages drive a substantial increase in CCL18 expression, resulting in its notably high levels in lesional AD skin and in serum of affected individuals. IL-4 upregulates both H2R and H4R, while IL-13 exclusively upregulates H4R without affecting H2R. Conversely, IL-10 upregulates H2R expression but shows a trend toward downregulating H4R (20). Moreover, recent in vitro studies have identified that the transient receptor potential vanilloid 4 (TRPV4) in macrophages, exerts anti-inflammatory properties. Its activation leads to the suppression of IL-1β expression in human macrophages by inhibiting NF-κB signaling and further prevents the differentiation of monocytes into pro-inflammatory macrophages, suggesting a potential therapeutic target for modulating macrophage activity in AD (21). Such findings align with the emerging understanding that macrophage functions in AD are shaped not only by cytokine milieu but also by mechanical and metabolic cues, including ion channels, microbial ligands, and tissue stressors.

In inflamed AD lesions, macrophage markers such as RFD7, which identifies mature tissue phagocytes, and CD68 exhibit similar expression levels and distribution patterns, with CD68⁺ macrophages being more prevalent than RFD7⁺ macrophages. Macrophages with high CD36 expression are also more abundant in inflamed AD lesions. CD36, a membrane glycoprotein involved in the phagocytosis of apoptotic cells such as neutrophils, plays a crucial role in limiting tissue damage and contributing to the resolution of inflammation in AD (22–24). In addition, in inflamed AD skin macrophages and DCs share overlapping phenotypes. Both DCs and macrophages express mannose receptors (MRs) for efficient antigen uptake, with MR expression predominantly found in monocyte-derived macrophages in inflamed AD skin (25). Kiekens et al. demonstrated that some macrophage populations may express macrophage (e.g., CD68) and DC (e.g., CD1a) markers simultaneously, indicating a heterogeneous pool of macrophage/DC-like cells (6). This overlap suggests complex plasticity of macrophages and DCs in AD, contributing to disease pathology.

DCs are key antigen-presenting cells that form a crucial link between innate and adaptive immunity, playing a significant role in the pathogenesis of AD. In AD patients, the numbers of specific myeloid Dendritic Cell (mDC) subsets, including Langerhans cells (LCs) in the epidermis and inflammatory dendritic epidermal cells (IDECs), are elevated in lesional skin ( Figure 1 ). Activation of Toll-like receptors (TLRs) triggers the maturation of DCs, marked by the upregulation of costimulatory molecules and the release of cytokines (26) ( Figure 2 ). Research on DCs in AD skin primarily focuses on TLR2, as it recognizes PAMPs from S. aureus ( Figure 1 ), particularly peptidoglycan, a major component of its cell wall. DCs capture antigens and migrate to lymph nodes, where they present the antigens to naive T cells, thereby driving the differentiation of Th1 and Th2 cells. Th2 cells, which dominate the immune response during acute phases of AD, produce cytokines such as IL-4, IL-5, and IL-13 upon arrival in the inflamed skin, driving the allergic inflammation characteristic of AD. DCs are uniquely specialized to detect antigens from both external and internal sources, responding to various stress signals. Upon stimulation, DCs initiate and regulate adaptive immunity, performing a highly adaptable function that allows them to adjust their behavior to the specific tissue microenvironment, whether in the skin, lung, or gut mucosa. For example, DCs in the skin, such as LCs and IDECs, exhibit distinct marker expression, cytokine production, migration patterns, and metabolic adaptations compared to DC subsets like conventional DCs (cDCs) or plasmacytoid DCs (pDCs) in the lung or gut mucosa. Despite these dynamic and specialized roles in AD, research indicates that the absolute number of DCs remains unchanged in lesional compared to non-lesional AD skin (6, 27–30). In lesional AD skin, the marked dysbiosis of the microbiome, particularly the overgrowth of S. aureus, plays a critical role in shaping DC function. S. aureus-derived components such as peptidoglycan (PGN), lipoteichoic acid (LTA), and enterotoxins activate pattern recognition receptors (especially TLR2 and TLR6) on DCs, leading to the production of IL-6, IL-1β, and IL-23 ( Figure 2 ). This stimulation promotes Th17 and Th2 polarization in a context-dependent manner. However, chronic microbial exposure may lead to DC tolerance or functional exhaustion, characterized by diminished cytokine production and impaired antigen presentation. This paradox, enhanced stimulation coupled with reduced functional responsiveness, highlights a hallmark of the dysregulated immune environment in chronic AD.

MDSCs have long been studied for their crucial role in malignancy and tumor maintenance through immunosuppression. However, they are now increasingly recognized as significant players in inflammation. Originating from immature myeloid cells, MDSCs possess immunoregulatory functions and have the potential to differentiate into mature DCs, macrophages, or granulocytes. These cells are categorized into two subtypes: monocytic MDSCs, characterized by a CD14⁺, CD11b⁺, Ly6C⁺, and Ly6G^- phenotype, and polymorphonuclear MDSCs, identified by a CD14^-, CD11b⁺, Ly6C^low, and Ly6G⁺ phenotype (57, 58). MDSCs exert their immunosuppressive effects through direct cell-to-cell contact and by secreting interleukins and chemokines. Their suppression of T-cell activity is mediated by arginase and iNOS, both of which deplete L-arginine, a molecule essential for T-cell differentiation and proliferation (59) ( Figure 2 ). The depletion of L-arginine results in the downregulation of CD3ζ and MHC class II, along with the inhibition of Janus kinase 3 (JAK3) and STAT5 in T cells, ultimately reducing T-cell proliferation (60). Additionally, iNOS produces nitric oxide (NO), which induces dose-dependent apoptosis in T cells by modulating Bcl-2 expression (61). The role of NO in cutaneous inflammation is closely related to its surrounding conditions and concentration, with both pro-inflammatory and anti-inflammatory effects reported (61). In a mouse model of AD, exposure to S. aureus activated TLR 2–6 in the skin, leading to IL-6 production by skin cells. This IL-6 subsequently stimulated the recruitment of suppressive CD11b⁺Gr1⁺ MDSCs, which inhibited T cell-mediated responses through the iNOS-dependent pathway (62). Ligation of TLR1–2 was observed to not stimulate MDSCs recruitment. MDSCs also induce Tregs response through CTLA4 and membrane-bound TGF-β, which suppresses natural killer cell cytotoxicity, as demonstrated in recent mouse studies (63, 64). Furthermore, MDSCs produce IL-10, which promotes the differentiation of macrophages into predominantly anti-inflammatory M2 cells and fosters a Th2-mediated immune response (65, 66).

The therapeutic potential of MDSCs has been demonstrated in several mouse models, where these cells were attracted to inflamed sites through chemotactic stimuli. MDSCs were observed migrating to the spleen, lungs, lymph nodes, and inflamed skin, suggesting they selectively infiltrate inflamed tissues (67–73). However, infiltration into the skin was observed only in AD mice, reinforcing the idea that MDSCs selectively target sites of inflammation (72, 73). In one study, this was achieved by inducing CCR5 expression, while another study utilized the injection of CXCL17 (74, 75). Furthermore, MDSCs generated from human umbilical cord blood (hUCB) increased IFN-γ expression in the spleen and lymph nodes. IFN-γ is known for its therapeutic effects in AD (76, 77) and its potential to enhance the immunomodulatory capabilities of MDSCs (78, 79). However, it is important to note that mouse-derived IFN-γ cannot bind to the human IFN-γ receptor, rendering it incapable of affecting human MDSCs (80). Therefore, further studies are necessary to elucidate the role of IFN-γ induced by MDSCs in AD mouse models.

It is generally accepted that the numbers of immature granulocytic and monocytic MDSCs are elevated in patients with inflammatory conditions. This observation has been consistently reported in patients with inflammatory skin diseases and inflammatory bowel disease (58, 81–83). hUCB-MDSCs injected into mice with AD-like symptoms induced by Dermatophagoides farinae (Df) alleviated skin lesions in a dose-dependent manner, with higher doses (1 × 10⁵; and 1 × 10⁶ cells) proving more effective than lower doses (1 × 10⁴ cells). The MDSCs also reduced epidermal thickness and decreased inflammatory cell infiltration (11, 64). MDSCs (1 × 10⁵ and/or 1 × 10⁶ cells) restored skin barrier function and improved skin fibrosis, suggesting that the anti-inflammatory effects and wound-healing capacities of MDSC therapy contribute to recovery from skin barrier impairment, dysfunction, and skin fibrosis in Df-induced AD-NC/Nga mice. This likely results from abnormal repair in response to skin damage. Additionally, MDSC treatment reduced IgE production and lowered Th2- and Th17-mediated cytokine levels (64, 84).

Contrasting with findings in AD, MDSCs from psoriasis patients may recruit Tregs, which are less capable of exerting their regulatory functions (82, 85). They also may exhibit reduced expression of surface PD-1 and lower production of the anti-inflammatory cytokine IL-10 (58, 85). To becoming functionally deficient, MDSCs can acquire proinflammatory functions when exposed to an inflammatory environment, especially in psoriasis patients. These proinflammatory functions include the overexpression of IL-1β, IL-6, IL-8, and TNF-α (86, 87); the production of MMPs such as MMP1 and MMP9, which facilitate the transmigration and accumulation of MDSCs in tissues (88); and the secretion of monocyte chemoattractant protein 1 (MCP1), which acts as a chemoattractant for proinflammatory cells (82). Additionally, GRO and IL-8 secreted by MDSCs recruit neutrophils to the skin and inflammatory cells like neutrophils and monocytes to organs beyond the skin (89).

Granulocytes including neutrophils, eosinophils, and basophils, play critical roles in the pathogenesis of AD. Basophils contribute to the initiation of AD by increasing IL-4 expression and interacting with keratinocytes and dermal macrophages, leading to epidermal hyperplasia and skin barrier dysfunction, but play a minor role in chronic lesions (90, 91). It was observed that by stimulation with TSLP basophils interact with neutrophils, sensory neurons and T cells enhancing the inflammation and itch in AD skin (90). The interaction of basophils and neutrophils contributes to the severity of AD whereas the reduction of basophils leads to decreased infiltration of eosinophils and neutrophils, as well as skin thickness (92).