One would expect that faced with such a high level of bacterial threat, the udder would have developed an imposing immune defense system. It is not the case at all. Compared with the digestive tract or the upper airways, the immune defenses of the uninfected healthy udder seem weak. The MG epithelium is devoid of cells specialized in the production of mucus or antimicrobial peptides, such as the goblet and Paneth cells present in the upper airways or gut epithelia. Consequently, the MG epithelial lining is not protected by a mucus layer that concentrates antimicrobial molecules or secretory immunoglobulin A (IgA) and isolates the cell surface from bacteria [30].

Accordingly, the MG is not a mucosal organ, even though it can be considered a member of the mucosal-associated lymphoid system due to its links with the gut-associated lymphoid tissue in some species (mouse or human MG) but not in others (MG of ruminants) [10, 31, 32]. In healthy MGs, most of the leucocytes are found in association with the epithelium, mainly ductal macrophages and intraepithelial cluster of differentiation 8 (CD8^pos) T cells [33]. A few CD4^pos T cells can be found, but there are no organized immune formations [34, 35]. In milk, leucocytes are few, mainly macrophages and CD4^pos and CD8^pos lymphocytes that have a memory phenotype [36–38]. In dairy cows, the physiological baseline of cell concentration is usually less than 20,000 cells/mL at peak lactation, and there is reason to believe that more than 50,000 cells/mL in cow milk indicates some degree of inflammation [39], as loss of milk production begins above this threshold [1, 40, 41].

The MG is equipped to sense the intrusion of bacteria and react promptly. This has been shown experimentally with the instillation of various bacterial agonists of the innate immune system (microbe-associated molecular patterns; MAMPs) into the lumen of the MG, which elicits a dose-dependent inflammatory response [42–45]. It has long been known that the MG is very sensitive to E. coli lipopolysaccharide (LPS) [46]. Expression of the Toll-like receptors (TLRs) 2, 4, and 9 in the mammary tissue has been documented [47]. MECs express TLR2 and TLR4 at their apical membrane [48] and bovine MECs in culture express TLR1, TLR2, TLR4, TLR6 (but not TLR5), and the oligonucleotide domain receptors nucleotide oligomerization domain 1 (NOD1) and NOD2 [44]. MECs do not express membrane CD14 when grown in vitro, but they secrete in milk-soluble CD14 which contributes to the innate recognition of bacteria by TLR and reduces the severity of MG infection by E. coli [49–51]. CD14 protein expression in vivo is low in MECs of uninflamed mouse MG, but high quickly after intramammary infusion of E. coli LPS [52]. However, apical membrane expression remains to be documented.

The expression of several pattern recognition receptors (PRRs) by MECs enables them to play the role of a sentinel of the MG. However, the level of expression is not at its peak in healthy MGs and is upregulated under inflammatory conditions, such as the expression of CD14 in mouse MG during LPS-induced mastitis [30, 52]. It is likely that MECs are assisted in their sentinel role by intraepithelial and alveolar macrophages. The numerous ductal macrophages that are in close contact with luminal MECs are good candidates for this function, but their role remains to be established [53, 54]. The fact that MECs do not react to Streptococcus uberis but macrophages does argue in this direction since S. uberis triggers inflammation in the MG [55]. Besides MAMPs, the MG could sense bacterial metabolites, as mucosal-associated invariant T (MAIT) cells have been identified in human and bovine milk [56, 57]. These cells recognize metabolites produced by bacteria and fungi and possess antimicrobial capacities.

In a way, the immune system of the healthy MG can be considered dormant since the mammary tissue does not harbor lymphoid formations and hosts few tissue and milk leukocytes unless infection induces tertiary lymphoid structures [34, 58, 59]. It even appears rather disarmed, with little epithelial protection against bacteria: no mucus layer, few antimicrobial peptides, dilution, and quenching of immune effectors by milk. On the other hand, it is equipped for sensing bacteria and bacterial products, which indicates that it can be awakened. Many observational and experimental studies have established how the MG reacts to and manages bacterial intrusions, mainly in dairy ruminants but also in mouse mastitis models, as discussed in the following.

Bacteria proliferating in the MG lumen release MAMPs and metabolites that the mammary epithelium senses. This triggers a self-defense response on the part of MECs and intraepithelial leucocytes. This response involves the production of antimicrobial peptides (β-defensins and cathelicidins), lactoferrin, complement components [C3, factor B, complement C4b-binding protein (C4BP)], acute phase proteins [serum amyloid protein 3 (SAA3), pentraxin 3 (PTX3)], and calgranulins (S100A8, S100A9, S100A12) [60–65]. The reaction participates in the protection of the epithelium lining from invasion by bacteria and in slowing down their proliferation in milk. An efficient response relies on two additional and complementary components: the generation of chemokines and the lowering of the blood-milk barrier. Among chemokines, those recruiting neutrophils [chemokine C-X-C motif ligand 1 (CXCL1), CXCL2, CXCL3, CXCL8] are prominent, but others [such as chemokine C-C motif ligand 2 (CCL2), CCL5, or CCL20] are also produced, responsible for the influx of mononuclear leukocytes [51, 52, 66–68]. The modulation of barrier leakiness allows blood complement components and Igs (including specific antibodies) to access the gland lumen. These responses concur to make possible an efficient phagocytosis of bacteria by neutrophils in the lumen and the epithelium [12, 69, 70].

There are different ways the MG reacts to different pathogens. However, there are also commonalities in these responses [71]. Neutrophilic inflammation is an indispensable element of an efficient response to MG infection. Obstruction or delay in neutrophil recruitment results in life-threatening mastitis [72–74]. Neutrophilic inflammation is also a harbinger and a fixture of mastitis, and as such is used as a diagnostic tool of MG infection in dairy ruminants [75]. A massive influx of neutrophils is necessary to oppose the proliferation of bacteria, which may cause collateral damage to the mammary epithelium [76, 77]. Milk contains opsonins that help macrophages and neutrophils kill bacteria such as E. coli or S. aureus, although some encapsulated bacteria are not efficiently opsonized [78, 79]. Neutrophils contribute to the partial collapse of the blood-milk barrier, which is beneficial as this allows blood defense components (such as complement, antibodies, transferrin, and lectins) to access the MG lumen [80], but has also deleterious effects on the epithelium [12, 77]. The modulation of mammary neutrophilic inflammation by the immune system is thus of consequence for the secretory function of the MG.

The most numerous cells in a lactating MG are the MECs by far, and these cells can sense bacteria, mount self-defense, and trigger inflammation. They have a moderate capacity to produce inflammatory cytokines, but they express receptors for these cytokines [interleukin-6 (IL-6), IL-1β, tumor necrosis factor alpha (TNF-α)] [81–83]. They also respond to the lymphokines interferon-gamma (IFN-γ) and IL-17 [83, 84]. Thus, their defense activity can be modulated by leucocytes. Resident MG leucocytes are not abundant, but because they are positioned at the frontline of infection, they can respond as soon as bacterial intrusion is detected before inflammation recruits circulating leucocytes.

Besides stromal, ductal, and alveolar macrophages and dendritic-like cells, the MG harbors lymphoid cells. The proportion of B cells is variable across species and the lactation cycle. In the lactating mouse MG, only 2% of leucocytes are B lymphocytes, and 10% T lymphocytes [54]. This issue will not be developed here because B cells and locally produced antibodies have not been shown to modulate appreciably the response of the MG to bacterial intrusion. This role is mainly devoted to T lymphocytes. Unfortunately, knowledge of the localization and nature of T lymphocytes in healthy mammary tissue is limited. It has been reported that in the bovine MG, T lymphocytes are present in close contact with the epithelium and in the connective tissue, with a predominance of CD8^pos over CD4^pos cells, in particular within the epithelium [85]. A proportion of the CD8^pos intraepithelial cells could be γδ T cells [86]. T cells, mainly CD8^pos, and macrophages are also intimately associated with the human breast epithelium [33, 87]. Their presence in healthy glands suggests that they play a role in the epithelium homeostasis and integrity, but this remains to be documented. In the murine MG, immature dendritic cells and retinoic acid receptor (RAR)-related orphan receptor gamma t (RORγt^pos) CD4^pos (T helper 17; Th17) lymphocytes are found in nulliparous and lactating glands, with a transient increase at the onset of involution, and a proportion of these lymphocytes also express forkhead box P3 [FoxP3; Th17/regulatory T (Treg) cells] [88]. Little is known about the activities of MG lymphocytes, and most knowledge is derived from milk cells. In human, mouse, or cow milk, T lymphocytes display the phenotype (CD45RO^pos) and functional characteristics of memory T cells [36, 37, 89–91]. In cattle, a subpopulation of CD8 T cells has been shown to play a major immunoregulatory role [92]. In healthy lactating glands and during involution, most CD8^pos T cells may have an immunoregulatory function, maybe to avoid self-reaction to milk components.

Unconventional T cells, which are not restricted to the classical major histocompatibility complex (MHC) molecules, are also present in the MG [93]. MAIT cells are activated by riboflavin synthesis metabolites presented by the antigen-presenting MHC-related protein 1 molecule (MR1) [93]. These cells contribute to the immune response to bacteria and fungi competent in the synthesis of vitamin B2 riboflavin. MAIT cells have been detected in breast milk where they represent a minor proportion of T cells [56]. In cow milk, only 0.8% of CD3^pos T cells are MAIT cells, but this proportion increases fivefold in mastitis milk [57]. Moreover, bovine MAIT cells respond to E. coli, riboflavin-proficient bacteria responsible for mastitis, by producing IFN-γ and TNF-α, suggesting that they could participate in the defense of the MG [57].

γδ T cells can be considered as unconventional as most are unrestrained by classical MHC restriction [93]. Most γδ T cells produce IL-17 or IFN-γ [94]. γδ T cells from mice, humans, or cows can respond to microbial antigens and can produce IL-17 very rapidly in the absence of clonal expansion, contrary to αβ T cells [95]. This ability would make these cells the major initial IL-17 producers in acute infections [94]. γδ T cells have been shown to interact with epithelial barriers and contribute to the homeostasis and antimicrobial response intestinal intraepithelial γδ lymphocytes do [96]. γδ T cells are present in the tissue of uninfected MGs, but their functions and precise localization have not been determined [74]. γδ T cells are also found in colostrum and milk [97, 98].

At the onset of infection, neutrophils are the first cells to appear in milk in high numbers. Neutrophilic inflammation is characteristic of the MG response to bacterial intrusion, and the massive recruitment of neutrophils may mask the influx of other immune cells. However, mononuclear immune cells are also recruited [35]. All types of lymphocytes can be found in MG secretion. In general, CD4^pos T cells became predominant over CD8^pos cells, both displaying the CD45RO^pos phenotype of effector/memory cells [99]. γδ T cell numbers also increase, suggesting that they could participate in the immune defense of the MG, notably by producing IL-17 [74, 100]. The relative contribution of resident intraepithelial and subepithelial immune cells remains to be clarified.

Although knowledge is limited in this area, it can be hypothesized that resident immune cells play an important role in early responses to infection, as they do in other epithelia exposed to bacteria [101, 102]. The cells recruited by the local inflammatory reaction complete the immune response. Adaptive immune responses will also modify the response of the MG to infections. The immune memory induced by infections, chronic or recurrent, will largely depend on the pathogen in question. This area, which is very broad and complex, will not be treated here. The other mode of induction of immunological memory, deliberate this time, is vaccination. The objective is to induce a population of antigen-specific tissue-resident memory T lymphocytes capable of adaptive immunosurveillance [103]. Antigen-specific neutrophilic inflammation can be induced in the MG by systemic or local (intramammary) immunization [35]. It has been shown that luminal injection of a model antigen (ovalbumin) elicits a neutrophil influx in milk in sensitized but not naive animals [104, 105]. A comparable reaction can be induced with killed bacteria or bacterial extracts [106, 107]. The neutrophilic inflammation can accelerate the cure of infection in relation to IL-17-associated pathways detected in the tissue of immunized MGs [108, 109]. The mammary antigen-specific neutrophilic inflammation depending on the generation of Th17 cells is amplified by innate immunity and correlates with the production of IL-17 and IFN-γ in the milk of immunized cows [105, 110]. It can be put forward that mammary macrophages or dendritic cells present bacterial antigens to tissue-resident Th17 cells which, in response, secrete cytokines (IL-17A, IL-17F, IFN-γ) that stimulate MECs to amplify neutrophilic inflammation [35].

From the above, it is tempting to assume that type 3 immunity is involved during MG infections and adaptive responses. Type 3 immunity is mediated by the signature cytokines IL-17A, IL-17F, and IL-22 that are produced by a diversity of lymphoid cells such as innate lymphoid cell type 3 (ILC3), γδ T cells, CD4 helper (Th17) and CD8 (T cytotoxic 17; Tc17) αβ T cells [111, 112]. Type 3 immunity can elicit neutrophilic inflammation at sites of infection, stimulate epithelial cell self-defense, and contribute to epithelial homeostasis [113]. For these reasons, type 3 immunity appears to be particularly suited to protecting the MG against bacterial species that cause neutrophilic inflammation, such as staphylococci, streptococci, or coliforms [114].