Human milk oligosaccharides (HMOs) are the third most abundant bioactive component in human breastmilk. More than 150 structurally distinct HMOs have been identified, which are generally divided into three structural groups: neutral fucosylated, neutral non-fucosylated, and acidic (sialylated) HMOs (1). HMOs benefit infant health by modulating the gut microbiome (2–4), strengthening intestinal barrier integrity and function (5, 6), immunomodulation (7, 8), and supporting cognitive development (9, 10). Most HMOs are undigested when they reach the lower gut, where they are fermented by the intestinal microbiome (11). However, a small fraction of HMOs may be absorbed into the systemic circulation (12, 13), indicating that the biological functions of HMOs extend beyond the gut and microbes.

The intestinal epithelium is necessary for gut homeostasis because the cells and overlying mucus layer form a barrier separating the luminal contents from the submucosa and systemic circulation (14, 15). When the intestinal barrier is impaired, luminal components, including bacteria, may translocate into the submucosal tissues causing mucosal and systemic inflammation, and may contribute to gastrointestinal and/or systemic diseases (16). An impaired intestinal barrier is a particular concern in preterm infants because it increases the risk of life-threatening necrotizing enterocolitis (17). Term infants are born with an immature and more permeable intestine, but intestinal integrity normally increases during the first weeks of life (18). Intestinal permeability decreases more rapidly in breastfed compared to formula-fed infants (18, 19), suggesting that breastmilk contains components that contribute to the maturation of the intestine. HMOs have been shown to support the maturation of the intestine in vitro and in vivo by increasing barrier integrity and stimulating mucin expression (5, 20). However, the effect of individual HMO structures on the integrity of the intestinal epithelial barrier is unclear.

Intestinal cells closely interact with the local submucosal immune compartment e.g., by releasing chemokines and cytokines that recruit and/or activate immune cells, including dendritic cells (DCs), macrophages (MØs) and T cells (14). DCs are specialized antigen-presenting cells (APCs) that regulate innate and subsequent adaptive immune responses. Immature DCs in the lamina propria sample the luminal contents, and if they sense pathogen-associated molecular patterns such as lipopolysaccharides (LPS), they become activated, undergo maturation, adjust their cytokine environment, and migrate to lymph nodes where they encounter naïve T cells and induce specific effector T cell responses (21). Similarly, MØs can present antigens to T cells and initiate adaptive immune responses (22, 23). A subtype of MØs (classical or M1 MØs) possess microbicidal activity and promote strong Th1 responses (23, 24). Activated DCs and MØs engage with naïve CD4⁺ T cells and create a cytokine milieu that drives the differentiation of the latter into specific effector T helper (Th) cells, including Th1, Th2 and Th17 subtypes. In the presence of interleukin (IL)-12, naïve CD4⁺ T cells differentiate towards Th1 rather than the Th2 direction. Th1 cells produce large amounts of interferon (IFN)-γ, which promotes cell-mediated immunity. In the presence of IL-4, naïve CD4⁺ T cells favor differentiation into Th2 cells that produce large amounts of IL-4, IL-5, and IL-13, triggering antibody-mediated responses and protection against extracellular parasites (24, 25). Cytokines such as IL-1β, IL-6, IL-23, and tumor necrosis factor (TNF)-α support the differentiation of Th17 cells (26–28), which produce large amounts of IL-17A and drive host-mediated responses against extracellular bacteria, particularly at epithelial surfaces (29).

During pregnancy, the immune system at the maternal–fetal interface is skewed towards a Th2 phenotype, which ensures that the developing fetus is protected from the maternal immune response, hence preventing rejection (30). This Th2 skewing persists into early infanthood, but the excessive production of Th2 cytokines during this period may enhance the risk of allergies later in life (31). Therefore, improved Th1 immunity and the induction of tolerance in early life is important to ensure the proper development of adaptive immunity (31, 32). Pooled HMOs isolated from breastmilk have been shown to inhibit Th1 differentiation in a co-culture model of human blood monocyte-derived DCs and naïve CD4⁺ T cells (33). In another study, the acidic fraction of pooled breastmilk-derived HMOs increased IFN-γ production from cord blood-derived mononuclear cells (34). Using a co-culture model of intestinal epithelial cells and peripheral blood-derived mononuclear cells (PBMCs), 2′-fucosyllactose (2′FL) and 3-fucosyllactose (3FL) were found to inhibit Th2 differentiation (35). Furthermore, the acidic HMO 3′-sialyllactose (3′SL), but not the structurally similar 6′-sialyllactose (6′SL), increased the intestinal expression of IL-17A mRNA in a mouse model of colitis (36).

Although the studies discussed above demonstrated direct effects of HMOs on the intestinal barrier and immune cell responses, the specific role of individual HMOs in these contexts remains poorly understood. Here, we investigated the structure-dependent in vitro effects of fucosylated (2′FL and 3FL), sialylated (3′SL and 6′SL) and neutral non-fucosylated (lacto-N-tetraose (LNT) and lacto-N-triose II (LNT2)) HMOs on intestinal epithelial cells, monocultures of DCs and M1 MØs, and CD4⁺ T cells primed with DCs or MØs cultured with HMOs. Our data suggest that primarily fucosylated and neutral non-fucosylated HMOs trigger a dose-dependent increase in intestinal barrier integrity, whereas sialylated HMOs exert immunomodulatory effects by influencing T cell effector functions and by enhancing the secretion of chemokines and cytokines from the epithelium and immune cells, respectively.

HMOs directly support the maturation of the intestinal barrier and immune system in early life (7, 8). In this study, we focused on HMOs representing different structural classes to gain more insight into their structure-dependent functions on the intestinal epithelium as well as innate and adaptive immunity. We found that individual HMOs affected the intestinal epithelial cell barrier function and chemokine secretion, directly influenced cytokine secretion from innate immune cells (DCs and MØs), and subsequently modulated T cell effector functions, in a structure-dependent manner.

We first studied the effect of the individual HMOs on intestinal barrier integrity and found that fucosylated HMOs (3FL especially, and 2′FL) and neutral non-fucosylated HMOs (LNT and LNT2) displayed the strongest effect by enhancing the barrier both in the presence and absence of a challenge. Notably, we only saw an effect when the intestinal epithelium was exposed to the HMOs on both sides, in contrast to luminal or apical exposure alone. In early life, the intestine is more permeable, which may allow a higher proportion of HMOs to enter the circulation, thereby providing access to the basolateral compartment (18). The effect was most pronounced at the highest dose (20 mg/ml), which exceeds the levels of HMOs found in human milk [5-15g/L (42)], and expectedly also in the circulation of infants. However, the applied doses were selected with the aim of seeing potential effects and compare the individual HMOs. A similar dose has also been applied in previous studies testing HMOs in Caco-2 cells (5, 20). In one of these, it was also found that 2′FL in a blend of six HMOs (2′FL, 3′SL, 6′SL, LNT, difucosyllactose and lacto-N-neotetraose) is the key driver of enhanced intestinal barrier integrity and protection following exposure to pro-inflammatory cytokines (5). LNT also had an effect, but it was less potent than 2′FL (5). In our study, we also tested 3FL, which had a pronounced effect compared to the other HMOs. Thus, in line with earlier studies, our findings confirm that fucosylated and neutral non-fucosylated HMOs also support intestinal barrier integrity during an inflammatory insult.

We investigated the impact of HMOs on DC and M1 MØ monocultures, as well as co-cultures of both cell types with allogeneic CD4⁺ T cells. Given that these immune cells were derived from PBMCs isolated from healthy donors, these models offer insight into the potential systemic effects of HMOs passing into the circulation. We found that the two sialylated HMOs (3′SL and 6′SL) had the most pronounced effect, stimulating the secretion of IL-10, IL-12p70 and IL-23 from LPS-activated DCs and M1 MØs, whereas the secretion of IL-1β was only significantly enhanced in LPS-activated M1 MØs. To the best of our knowledge, these are novel findings in the context of individual HMOs. Notably, we did not observe a significant dose-dependent effect, which could be due to saturation of DC and MØ receptor-mediated signaling responses already at the lowest dose of HMOs we tested (1 mg/mL) (43). We therefore used the same concentration (1 mg/mL) in our co-culture models with naïve CD4⁺ T cells.

Consistent with our monoculture results, we found that 3′SL and 6′SL enhanced the secretion of IFN-γ and IL-17A and reduced the secretion of IL-13 by naïve CD4⁺ T-cells co-cultured with LPS-activated DCs or M1 MØs, demonstrating that both sialylated HMOs induced Th1 and Th17 differentiation while inhibiting Th2 differentiation. Notably, 6′SL appeared to be a slightly more potent inducer of Th1 responses than 3′SL. In comparison, 3FL primed the same pattern of Th differentiation but to a lesser degree, and primarily in co-cultures of CD4⁺ T cells and M1 MØs. These findings are interesting because the neonatal immune system is considered tolerogenic, with enhanced immunosuppressive and tissue-protective mechanisms to maintain balance and prevent collateral damage (44). However, neonatal DCs, MØs and monocytes show impaired antigen-presenting functions, cytokine secretion and T cell stimulation. Consequently, neonatal T cells skew toward Th2 rather than Th1 responses, and coupled with their inexperienced adaptive immune functions, this renders neonates more vulnerable to viral and bacterial infections (45–47). Our findings indicate that certain HMOs, in this case primarily 3′SL and 6′SL, may directly interact with and modulate DC and MØ functions to subsequently support Th1 responses.

The Th17 cell lineage is often described as the third major subset of effector T cells, and Th17 responses during homeostatic and regulated conditions play a critical role in antibacterial and antifungal immunity at epithelial surfaces (48, 49). Importantly, preterm infants are more susceptible to infections at mucosal surfaces, probably in part due to the developmental limitations of neonatal Th17 functions (50). To our knowledge, our study is the first to demonstrate that sialylated HMOs promote Th17-skewed responses in CD4⁺ T cells co-cultured with DCs or M1 MØs. Similar effects of 3′SL and/or 6′SL in clinical settings may reduce the risk of mucosal infections in neonates.

Our findings that sialylated HMOs can modulate the activity of DCs and M1 MØs in monoculture and in co-culture with CD4⁺ T cells by enhancing Th1 and Th17 responses align with previous reports. The acidic HMO fraction (most likely to be rich in sialylated structures) of pooled HMOs isolated from breastmilk increased IFN-γ production from cord blood-derived mononuclear cells (34) and 3′SL was previously shown to induce mesenteric lymph node-derived DCs to produce cytokines driving Th1 and Th17-dependent inflammation in an il10^–/– mouse model of spontaneous colitis (36). Moreover, 3FL, 3′SL and 6′SL enhanced IFN-γ secretion from PBMCs activated with anti-CD3/CD28 antibodies although statistical significance was only achieved for 3FL (35). Notably, 2′FL and 6′SL failed to affect the LPS-induced maturation of DCs and DC cytokine secretion when the HMOs were applied during 6 days of monocyte-to-DC differentiation, suggesting that HMOs have less of an impact on this immunomodulatory process (51). However, pooled human HMOs (isolated from breastmilk) reduced rather than increased the LPS-induced production of IL-12p70, IL-6 and TNF-α from human monocyte-derived DCs, which subsequently reduced Th1 differentiation and IFN-γ secretion (33). This contrasting finding may reflect the small proportion (10–14%) of the HMO pool in human breastmilk made up of sialylated HMOs (52), thus possibly diluting their impact. Moreover, other HMO structures beyond the six included in the present study may have distinct immunomodulatory properties, and some may reduce rather than enhance cytokine secretion from LPS-activated DCs. The aim of the present study was to focus on the host phenotypes induced by individual HMOs, and future/subsequent studies should focus on fully characterizing the underlying mechanisms by identifying the receptors on intestinal epithelial cells and the monocyte-derived DCs and MØ that HMOs engage to promote or suppress signaling. It is already known that different structural classes of HMOs can bind with varying affinity to classes of lectin (glycan-binding protein) receptors expressed on monocytes and other immune cell types. For instance, DC-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN) and sialic acid-binding, immunoglobulin-like lectins (siglecs) have been suggested as receptors for fucosylated and sialylated HMOs, respectively (53). Toll-like receptor (TLR)-4, expressed on the surface of both monocytes and intestinal epithelial cells, has been implicated as a mediator of HMO-induced immune responses (33, 36). Thus, it is possible that 3′SL and 6′SL modulate the cytokine release from DCs and M1 MØs by interacting with specific siglecs. To add more insight into the underlying cellular mechanisms by which the individual HMOs interact with the host, applying transcriptomic and proteomic approaches may be supportive (54).

LPS contaminated HMOs have been addressed as a major concern, suggesting any host effects are caused by LPS rather than the HMOs. In one study the apparent TLR-4 mediated immunomodulatory effects of 3’SL on human monocyte-derived DCs were attributed to LPS contamination (10 EU/mg), thus warranting caution and a need for assessing LPS contamination levels when studying e.g., TLR-mediated effects of HMOs (55). The HMOs studied in the present study had a very low endotoxin level (<0.009 EU/mg, Table 1 ), which is not in the range expected to significantly confound immunological responses, as shown previously (35, 55). Importantly, HMO supplementation to non-activated DCs stimulated minimal cytokine secretion compared to LPS-activated DCs ( Supplementary Figure 3 ).

We also assessed the impact of HMOs on chemokine secretion from intestinal epithelial cells. The intestinal epithelium engages in crosstalk with the innate and adaptive immune systems by secreting chemokines, and thus plays a key role in the regulation of mucosal immune responses (14). We measured the secretion of CXCL10, CCL20 and CXCL8, and found that mainly the sialylated HMOs induced the secretion of all three chemokines, with 6′SL having a slightly greater impact than 3′SL. CXCL10 is a chemoattractant for Th1 cells that acts through CXCR3 expressed on (among others) CD4⁺ Th1 cells and is induced by IFN-γ (56). CCL20 is a chemoattractant for CCR6⁺ cells such as DCs, B cells and CD4⁺ T cells (especially Th17 cells) and CD8⁺ T cells (57, 58). CCL20 also has antimicrobial activity, which reflects its ability to form a cationic antimicrobial peptide that can penetrate bacterial membranes and bind to intracellular components (59). Notably, the upregulation of CCL20 and CXCL8 gene expression has been demonstrated by the transcriptomic analysis of HT29 cells exposed to 3′SL or pooled HMOs, but the structure-dependent effects of individual HMOs were not reported (40).

Given the extensive crosstalk between intestinal epithelial cells and immune cells, it would be informative to investigate the impact of individual HMOs in co-cultures of these cell types. Such models more closely mimic mucosal immune responses, whereas here we have focused on systemic immunomodulatory effects. In a recent study using co-cultures of HT29 human colon epithelial cells conditioned with CpG (a synthetic TLR9 ligand) and PBMCs activated with anti-CD3/CD28 antibodies, 2′FL and 3FL were shown to promote stronger immunomodulatory effects linked to Th1 and regulatory T (T_reg) cell responses than 3′SL and 6′SL (35). Similarly, DCs exposed to 2′FL and CpG-conditioned HT29 cells gained an enhanced ability to promote the secretion of IFN-γ and IL-10 by CD4⁺ T cells (60). Moreover, 2′FL tended to reduce Th2 responses when naïve CD4⁺ T cells were incubated with DCs that had been co-cultured with HT29 cells in the presence of polyinosinic-polycytidylic acid, a synthetic dsRNA analog (61). Future studies evaluating HMOs representing different structural classes should consider model systems closer to the infant state, such as primary intestinal epithelial cells derived from infants or mucosal-derived immune cells. This would provide more insight into the potential structure-dependent effect HMOs on mucosal immunity during early life. Finally, given the complex mixture of HMOs present in breastmilk, it is important to test HMO mixes from different structural groups in these model systems to identify any synergistic and/or antagonistic effects.

In conclusion, our findings highlight the potential structure-dependent effects of six individual HMOs on the intestinal barrier and local and/or systemic immune cells ( Figure 6 ). The fucosylated (2′FL and 3FL) and neutral non-fucosylated (LNT and LNT2) HMOs primarily support intestinal barrier integrity, whereas the sialylated (3′SL and 6′SL) HMOs enhance mucosal and systemic immune responsiveness by promoting DC and MØ-mediated Th1 and Th17 responses and reducing Th2 responses. Furthermore, 3′SL and 6′SL affect chemokine networks in the intestinal epithelium and may facilitate the recruitment of Th1 and Th17 subtypes to the intestinal submucosa. Concurrently, the enhanced secretion of IL-10 by DCs and MØs helps to ensure the immune responses are balanced and not over-reactive. Combining these findings with the fact that HMOs are a significant constituent of breastmilk with more than 150 different structures, our results suggest that individual HMOs play distinct and important roles that support development in early life.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

EB: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing. DL: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing. MT: Investigation, Methodology, Writing – review and editing. SH: Investigation, Methodology, Writing – review and editing. KP: Writing – review and editing. SJ: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing.