Abstract
The mammary gland is not classically considered a mucosal organ, although it exhibits some features common to mucosal tissues. Notably, the mammary epithelium is contiguous with the external environment, is exposed to bacteria during lactation, and displays antimicrobial features. Nonetheless, immunological hallmarks predictive of mucosal function have not been demonstrated in the mammary gland, including immune tolerance to foreign Ags under homeostasis. This inquiry is important, as mucosal immunity in the mammary gland may assure infant and women’s health during lactation. Further, such mucosal immune programs may protect mammary function at the expense of breast cancer promotion via decreased immune surveillance. In this study, using murine models, we evaluated mammary specific mucosal attributes focusing on two reproductive states at increased risk for foreign and self-antigen exposure: lactation and weaning-induced involution. We find a baseline mucosal program of RORγT+ CD4+ T cells that is elevated within lactating and involuting mammary glands and is extended during involution to include tolerogenic dendritic cell phenotypes, barrier-supportive antimicrobials, and immunosuppressive Foxp3+ CD4+ T cells. Further, we demonstrate suppression of Ag-dependent CD4+ T cell activation, data consistent with immune tolerance. We also find Ag-independent accumulation of memory RORγT+ Foxp3+ CD4+ T cells specifically within the involution mammary gland consistent with an active immune process. Overall, these data elucidate strong mucosal immune programs within lactating and involuting mammary glands. Our findings support the classification of the mammary gland as a temporal mucosal organ and open new avenues for exploration into breast pathologic conditions, including compromised lactation and breast cancer.
Introduction
Under conditions of homeostasis, classical mucosal organs, such as the lung and gut, harbor unique immunological properties in which epithelial and immune cells function as a unit to protect the organ from external insult (1). Specifically, subsets of Th17 CD4+ T cells and various antimicrobial products support epithelial barrier function and limit infection (2). Another key attribute of mucosal immunity is the presence of tolerogenic dendritic cells and regulatory CD4+ T cells, which promote immune tolerance and dampen response to frequently encountered Ags (3, 4). Although not classically considered mucosal, the mammary gland has a mucin-containing barrier to the external environment and is at increased risk of infection during nursing. Further anecdotal evidence for mucosal classification is the dependence of the mammary epithelium on immune cells during development. Specifically, dendritic cells and CD4+ T cells coordinate pubertal branching (5), and macrophages are essential for pregnancy-dependent alveolar expansion (6) and weaning-induced epithelial cell death (7). Because of these potential mucosal attributes, we elected to systematically study the murine mammary gland using a mucosal immunology framework. We focused on two developmental states that impact infant and mother health: lactation and weaning-induced mammary gland involution. This work may lead to new avenues of investigation into lactation failure and postpartum breast cancer, two critical and understudied public health concerns (8–10).
To date, studies supportive of mucosal biology in the mammary gland have focused on lactation because an increased risk of mastitis in dairy cows is a significant health as well as economic problem. One proposed mechanism of increased infection in lactating cows is active immune suppression, a biology that would be consistent with mucosal function. However, active immune suppression has not been explicitly demonstrated (11–13). Further, immune suppression is not the only possible explanation for the increased infection rate observed in lactating cows. Notably, heightened pathogen exposure because of teat damage from mechanical milking (14) could also contribute to increased infection rates, independent of immune suppression.
Supporting evidence for mucosal function in the lactating mammary gland has also been reported in the context of human neonatal health and been corroborated in murine studies. Specifically, expression of antimicrobial molecules found at mucosal epithelial borders, including IgA and mucins, is present in milk (15–17). In mice, milk IgA is the product of developmentally regulated B cell influx into the mammary gland via the chemokine CCL28 (17). Importantly, milk IgA has been demonstrated to play a critical role in the maintenance of infant gut health by providing maternal-derived antimicrobial function (18, 19). However, it is unknown whether IgA also plays a protective, antimicrobial role in the lactating mammary epithelium, which is a role consistent with mammary mucosal function. Indeed, mammary epithelium may require additional barrier function and immune tolerance because of the bioactive components of milk, including lactoferrin, bacteria, and leukocytes (20). Although there is strong rationale for proposing the presence of mucosal immunologic programs in the lactating gland, definitive demonstration is lacking, especially for active induction of immune tolerance.
In contrast to lactation, the reproductive state of weaning-induced involution has not been studied in the context of mucosal immunology. Weaning is a developmentally regulated process characterized by the death of ∼80–90% of secretory mammary epithelial cells followed by wound-healing–like tissue repair and immune cell influx (21–24). We predict that weaning-induced mammary gland involution will be characterized by mucosal immune features similar to, but likely distinct from, lactation. One rationale for this is that the risk of self-Ag exposure as a consequence of weaning-induced epithelial cell death likely necessitates the mucosal hallmark of immune tolerance. Further, tissue restoration and maintenance of barrier function are expected to be particularly prominent during involution. Study of the involuting mammary gland with a mucosal framework will improve understanding of baseline immune programs during involution and may inform the development or progression of postpartum breast cancer, which is a particularly life-threatening form of young women’s breast cancer (10).
In this study, we find evidence for baseline mucosal features within the mammary gland at all reproductive stages evaluated. This is evidenced by immature dendritic cell populations and phenotypes as well as RORγT+ CD4+ T cells, consistent with Th17 polarization, within the nulliparous mammary gland. Further, during lactation, we find evidence for enhanced mucosal features, chiefly the presence of dendritic cells with tolerogenic phenotypes and decreased Ag-dependent T cell expansion. This immunologic program is anticipated to assure successful lactation by providing enhanced protection against bacterial penetration and suppressing immunity to milk. With weaning, we find an extension of immune tolerance with the addition of immune regulatory programs, including the inflammation based recruitment of memory RORγT+ Foxp3+ CD4+ T cells. During involution, these mucosal hallmarks are anticipated to suppress immunity to self-antigens that may be released during epithelial cell death as well as support tissue repair.
In sum, the data presented in this study support the classification of the mammary gland as an immune-tolerant mucosal organ. Further investigation of the mammary gland within a mucosal framework may lead to new understanding of lactation failure and provide insight into known links between lactation, involution, and breast cancer risk.
Materials and Methods
Human breast tissue collection
Formalin-fixed tissues and paraffin-embedded breast tissue specimens from premenopausal women who underwent clinically indicated biopsies were obtained with Oregon Health and Science University (OHSU) Institutional Review Board approval. All cases were deidentified to the research team.
Preclinical mouse models and tissue collection
The OHSU Institutional Animal Care and Use Committee approved all mouse procedures in compliance with current National Institutes of Health guidelines. To obtain mice of various reproductive stages, female BALB/c mice were outbred with male C57BL/6 mice (Jackson Laboratory) in ventilated cages with 12 h light/dark cycles. For ex vivo Ag cross-presentation studies (described below), C57BL/6 females were mated to C57BL/6 males. Pup number was normalized 2–3 d after parturition to six to seven pups per dam to ensure equal lactational load across groups. Synchronized weaning was initiated at lactation day 9–14 (denoted involution day [InvD] 0) to generate groups of animals for various postpartum (involution) time points. Age-matched nulliparous animals were either never bred or bred without detectable pregnancy. Mice were sacrificed across reproductive groups, with CO2 and cervical dislocation, and tissues were collected. Left and right numbers 4+5 mammary glands, inguinal lymph nodes (LNs), and spleens were harvested and fixed in formalin for immunohistochemistry (IHC) applications or placed in HBSS on ice to be processed for ex vivo assays and flow cytometry.
H&E and IHC
Formalin-fixed tissues were processed and paraffin embedded, sectioned to 4 μm, and subjected to H&E staining or IHC for CD45 (clones 2B11+PD7/26 Dako for human, clone 30-F11 BD for mouse), CK18 (polyclonal; Abcam), or E-cadherin (24E10; Cell Signaling) and were visualized with a DAB chromogen (Dako). H&E- and IHC-stained slides were scanned on an Aperio ScanScope AT, and images were captured using Aperio ImageScope Software (Leica Biosystems). CD45 IHC images were pseudocolored in Aperio using a color deconvolution algorithm, in which blue represents epithelium negative for DAB signal, and yellow, orange, and red represent increasing intensity of DAB staining (in CD45+ immune cells).
Analysis of publicly available microarray data set
A publicly available data set of microarray gene expression in wild type BALB/c mice at various reproductive stages (16) was analyzed for various genes. Each data point represents the average of three replicates, each composed of mammary glands from three BALB/c female mice.
Tissue digestion and staining for flow cytometry
Fresh tissues were collected as above and finely minced and then digested with collagenases II and IV (2.5 mg ml−1 Worthington) and DNase I (0.5–2.5 mg ml−1 Worthington) in HBSS with calcium and magnesium for 30 min with agitation at 37°C. Digests were filtered (100 μm) and RBCs lysed (eBioscience) per manufacturer’s instructions. Samples were counted and blocked with CD16/32 (1:100; eBioscience) and stained with Live Dead (aqua, 1:500; Invitrogen) for 30 min and then washed 1× with PBS. Cells were stained with Abs for extracellular proteins (CD45, 30-F11; MHC class II [MHCII], M5/114.15.2; B220, RA3-6B2; CD11b, M1/70; CD11c, N418; F480, BM8; Ly6C, HK1.4; Ly6G, 1A8; CD103, 2E7; CD80, 16-10A1; CD86, GL-1; CD4, RM4-5; PD-1, 29F.1A12; CD3e, 145-2C11) and diluted in buffer (1% BSA in 1× PBS) for 30 min at room temperature. Cells were fixed (BD Cytofix) and permeabilized (eBioscience) per manufacturer’s instructions, stained with Abs for intracellular proteins (Gata3, 16E10A23; RORγT, AFKJS-9; Foxp3, FJK-16s), and diluted in permeabilization buffer (eBioscience) for 2 h at 37°C or overnight at 4°C. Cells were fixed again and run on an 18-color flow cytometer (Fortessa BD) in the OHSU flow cytometry shared resource. Data were analyzed using FlowJo software.
Ex vivo Ag uptake and processing assays
Mammary digests from reproductively distinct mice were incubated for 4 h at 37°C with Alexa-Fluor 488–labeled OVA (100 μg ml−1; Thermo Fisher Scientific) or DQ-OVA (100 μg ml−1; Thermo Fisher Scientific) to test Ag-binding/uptake and Ag processing, respectively. Cells were stained with extracellular Abs (CD11c, N418; F480, BM8; Ly6C, HK1.4; MHCII, M5/114.15.2; CD45, 30-F11), and flow cytometry was performed.
Ex vivo Ag cross-presentation assay
C57BL/6 females were used for this assay to use a strain-specific Ab (clone eBio25-D1.16; eBioscience) that can specifically detect OVA Ag peptide (SIINFEKL) loaded into MHC class I (MHCI; Kb). Mammary digests were incubated with unlabeled whole OVA protein (100 μg ml−1) for 4 h at 37°C, then stained with extracellular Abs (same as above with the addition of eBio25-D1.16), and flow cytometry was performed.
Adoptive transfer in vivo T cell experiments
CD4+ splenocytes were isolated from DO11.10 transgenic female mice (stock no. 003303, stock name C.CG-Tg[DO11.10]10Dlo/J; Jackson Laboratory) and were enriched (to >95%) using a CD4+ negative selection kit (MACS; Miltenyi Biotec). CD4+ T cells were either unstimulated (putative naive T cells) or stimulated for 10 d in vitro with plate bound CD3 (2 μg ml−1; eBioscience) and CD28 (5 μg ml−1; eBioscience) in IMDM base media containing TGF-β (1 ng ml−1), IL-1B (10 ng ml−1), and IL-6 (50 ng ml−1) (25). This method of activation generated a cell population enriched for long-lived CD44+ cells (Supplemental Fig. 4E, 4F), consistent with resting effector memory T cells (26), referred to as memory T cells in this article. Further, the resulting T cell pool was >95% RORγT+ Foxp3+ (Th17/regulatory T cells [Tregs]) (Supplemental Fig. 4G, 4H). Naive and memory T cells were adoptively transferred (150,000 T cells) via tail vein into syngeneic InvD0 or InvD4, lactation day 10, regressed 6 wk, or nulliparous age-matched hosts. Two days later, OVA protein Ag (10 μg in 10 μl; Worthington Biochemical Corporation) was injected into either the left or right no. 4 mammary fat pad, and the opposite mammary fat pad received equal volume PBS. Five days later, tissues were harvested, digested, and stained with extracellular Abs (CD45, 30-F11; CD4, RM4-5; D011.10 TCR, KJ1-26) and intracellular Abs (Gata3, 16E10A23; RORγT, AFKJS-9; Foxp3, FJK-16s) as described above. A known quantity of absolute counting beads was added to samples (C36950; Invitrogen), and flow cytometry was performed. Naive or memory T cell counts in inguinal LNs, mammary glands, and spleens were calculated by normalizing to the known abundance of counting beads. Data are either represented as transgenic T cell number or by ratio of T cell number (with Ag: without). Statistical outliers were identified using a GraphPad Prism outlier test with an α value of 0.05 and were removed from the data set. This resulted in the removal of 3 out of 70 total data points in Fig. 5D as explained in the figure legend.
Statistics
If not indicated otherwise, data are represented as mean ± SEM. Significance was determined by performing one-way ANOVA, two-way ANOVA, or two-tailed unpaired or paired t tests in GraphPad Prism. The statistical test employed and key for significance values are described in each figure legend.
Results
CD45+ immune cells and antimicrobial barrier attributes in the normal mammary gland
Close physical interactions between epithelium and immune cells are found at mucosal surfaces; hence, we predict this physical proximity in the normal mammary gland. In this article, we find, even in quiescent nulliparous glands, that CD45+ leukocytes are closely associated with mammary epithelial (CK18 or E-cadherin positive) cells in human (Fig. 1A) and murine (Fig. 1B) mammary glands. These observations support mucosal biology and are consistent with previous reports (23, 27, 28). We next assessed if CD45+ abundance within the mammary gland changes with reproductive state, as murine mammary gland structure and function are dynamic across the reproductive cycle. The nulliparous murine mammary gland is composed of sparse epithelium, an adipose-rich stroma, and a collagen-rich extracellular matrix (23, 29). During pregnancy, the epithelium proliferates, differentiates, and replaces the fat pad. With lactation, secretory epithelium dominates, and milk is secreted into the ductal lumen. Upon weaning, up to 90% of the secretory epithelium undergoes programmed cell death, and the fat pad re-emerges in a tissue-remodeling process referred to as involution. By histologic criteria, involution is essentially complete by 8–10 d postweaning, with 6 wk postweaning considered a full return to a prepregnant-like state (16, 30–32). Of note, similar hormone-dependent structural and functional changes occur in the human breast (27).
Assessing mammary CD45+ cells across a reproductive cycle, we observe a close association between CD45+ cells and the epithelium at all reproductive stages, as determined by IHC staining (Fig. 1C, data not shown). Using flow cytometry, we found an increased abundance of total immune cells (CD45+ per milligram of mammary tissue) during lactation and an even further increase during involution (Fig. 1D, 1E). Quantitation of total CD45+ cell abundance as a percentage of all live cells by flow cytometry shows a decreased abundance during lactation, possibly due to the increased cellularity of the lactating gland (Supplemental Fig. 1A). These data confirm regulation of mammary immune cell abundance by reproductive state.
To further investigate the mucosal state of the mammary gland, we assessed for expression of various mRNA transcripts associated with mucosal function using publicly available data (16). We find that IgA H chain mRNA is increased in the murine mammary gland during lactation (Fig. 1F), as expected based upon previous characterizations of milk (17, 20). Somewhat unexpectedly, in postweaning, we find even further increases in IgA H chain mRNA expression, suggesting extension and expansion of antimicrobial function within the involuting gland (Fig. 1F). In addition, mRNA for mucins, which aid in barrier function, are dynamically but distinctly upregulated during lactation and involution (Fig. 1G). Further, orosomucoid 1 (Orm1) and serum amyloid A1 (Saa1) mRNAs, acute early-phase inflammatory mediators induced at liver and gut barrier surfaces (33, 34), are more abundant in lactating and involuting glands (Fig. 1H). In total, these gene expression features are consistent with enhanced but distinct barrier function in the mammary gland during lactation and weaning-induced involution.
Dynamic regulation of dendritic cell abundance, differentiation, and maturation with reproductive state
Mucosal-mediated immune suppression is dependent on active suppression of T cells by APCs, of which tolerogenic dendritic cells are a major contributor. Thus, increased dendritic cell influx is anticipated under conditions of increased need for immune tolerance, such as lactation and weaning. To assess for mammary dendritic cell abundance as a function of reproductive state, we used flow cytometry to examine dendritic cells present in the mammary tissue. Mammary dendritic cells were identified as CD45+, CD11c+, MHCII+, Ly6C−, Ly6G−, B220−, and F480− (Fig. 2A, Supplemental Fig. 1B). Dendritic cell abundance, normalized to the milligram of mammary tissue, increased during early involution and peaked 6 d postweaning (Fig. 2B). These data are consistent with the increases in total CD45 abundance we observe during lactation and involution (Fig. 1E). To determine whether differences in immune cell composition also occur, dendritic cell abundance was calculated as a percentage of the entire immune cell pool (all CD45+ cells). In nulliparous and lactating mice, the relative abundance of mammary dendritic cells to total CD45 cells was comparable (Fig. 2B, 2C). After weaning, dendritic cell abundance increased slightly 2–4 d postweaning (Fig. 2B) and then increased greatly by ∼5 fold at 6–8 d postweaning, accounting for ∼10% of the total immune cell population (Fig. 2C). These observed increases in dendritic cell abundance with involution were transient; at 6 wk postweaning (i.e., regressed state), dendritic cell numbers returned to baseline nulliparous levels (Fig. 2B, 2C). The abundance of other myeloid immune cells, including monocytes and macrophages, was also dependent on reproductive state (Supplemental Fig. 2A, 2B), as previously reported (23, 24). In sum, our data show for the first time, to our knowledge, that the numbers of dendritic cells increase in the mammary gland during involution. We next assessed for the tolerogenic character of mammary dendritic cells, as anticipated for a mucosal organ.
A hallmark of tolerogenic dendritic cells is decreased maturation state as measured by reduced cell surface expression of MHCII and costimulatory molecules CD80 and CD86 (35–37). Thus, we analyzed for the expression of these molecules by flow cytometry. During lactation, dendritic cell expression of CD80 and CD86 was greatly reduced. In the early postweaning (InvD2–InvD4) mammary glands, CD86 remained reduced on dendritic cells, returning to baseline nulliparous levels by InvD6 (Fig. 2D, Supplemental Fig. 1C). Dendritic cell expression of CD80 returned to nulliparous levels earlier, on InvD2 (Fig. 2F, Supplemental Fig. 1D). Unlike CD86 and CD80, MHCII expression was not diminished during lactation but was reduced only during early involution (InvD2-InvD4) before returning to nulliparous levels by InvD6 (Fig. 2E). These data indicate dendritic cells with tolerogenic phenotypes in the mammary gland during lactation and early stages of involution. These data also suggest distinct mechanisms of dendritic cell tolerance in the lactating, compared with the involuting, mammary gland.
To further delineate the regulation of mammary dendritic cell populations by reproductive state, the total mammary dendritic cell pool was evaluated for CD11b and CD103, which are differentiation markers that delineate dendritic cell subsets in nonlymphoid tissues (38). Determining subset abundance is important because CD11b and CD103 dendritic cells have differing propensities for T cell activation and tolerance (35, 39–41). For example, CD103+ dendritic cells have a superior ability to initiate immune tolerance in the intestinal mucosa (39, 42). To our surprise, we found that ∼50% of the dendritic cells in the nulliparous mammary gland did not express CD11b or CD103 (Fig. 3A, lower left quadrant). Our analyses do not support the classification of these dual negative dendritic cells (CD11b− and CD103−) as Ly6C+ monocytes or F480+ macrophages (Supplemental Fig. 2A, 2B). Further, these dual negative dendritic cells do not appear to contain plasmacytoid dendritic cells because they express similar levels of CD11c to the CD11b+ and CD103+ dendritic cell subsets and do not express the plasmacytoid dendritic cell specific marker, PDCA-1 (Supplemental Fig. 2C, 2D). During lactation and early involution, these dual negative dendritic cells constituted almost the entire mammary dendritic cell pool before returning to baseline nulliparous levels by day 4 postweaning (Fig. 3B, 3E, dark blue).
For the singly positive CD11b and CD103 dendritic cell populations, we also find dynamic regulation of abundance across the reproductive cycle. Notably, CD11b+ dendritic cells dominated over CD103+ dendritic cells at all reproductive stages (Fig. 3C–E), similar to subset ratios reported in the homeostatic gut (43). Also, the abundance of mammary CD11b+ dendritic cells during lactation was very low and only modestly increased during early involution before returning to baseline levels by InvD6 (Fig. 3A, 3C, 3E, light blue). In contrast, the CD103+ dendritic cells, although a smaller subset of the overall mammary dendritic cell population, were specifically increased during involution (Fig. 3A, 3D, 3E, orange). We also analyzed each dendritic cell subset for maturation state. Overall, during lactation and involution, we observed lower maturation state within these dendritic cell subsets (Supplemental Fig. 3A–C). Combined, these data suggest dampened dendritic cell differentiation and maturation during lactation and involution compared with nulliparous or regressed hosts. Further, these data are predictive of tolerogenic dendritic cell functions during lactation and weaning-induced involution, which we evaluated next.
Dendritic cell Ag binding, uptake, and presentation are not blocked during lactation and involution
One mechanism by which dendritic cells can exert tolerogenic programs is through active T cell suppression, which is dependent on dendritic cell Ag uptake, processing, and presentation. Thus, we assessed these functional capabilities in mammary dendritic cells from lactating and involuting glands in comparison with nulliparous glands. LN-free mammary gland digests prepared from nulliparous, lactating, and involuting hosts were incubated with fluorescently labeled OVA proteins. Dendritic cells were then assessed for Ag binding and uptake by fluorescent OVA (AF488) signal and for Ag processing by DQ-OVA signal, which is produced upon Ag cleavage in the lysosome because of reduced pH. Ag uptake and binding (AF488-ova) were evident at all reproductive stages; however, lactation and involution (day 2) dendritic cells had ∼2-fold reduction in binding and uptake compared with nulliparous mammary dendritic cells (Supplemental Fig. 3D). With respect to Ag processing, we found that dendritic cells had an equivalent ability to process Ag at all reproductive stages tested, consistent with intact dendritic cell functionality (Supplemental Fig. 3E).
To assess the ability of mammary dendritic cells to cross-present Ag, we used an ex vivo Ag presentation assay that detects the final step in the process: the presence of OVA peptide–MHCI complex (MHCI-SIINFEKL) at the cell surface (Fig. 4A cartoon). Representative flow plots of MHCI-SIINFEKL expression from lactation and InvD6 stages are shown (Fig. 4B, positive staining shaded red). We found that mammary dendritic cells from all reproductive stages tested had the ability to cross-present Ag (Fig. 4C, all values above zero), with dendritic cells from InvD6 having the highest level of Ag presentation (Fig. 4C). In sum, mammary dendritic cells retain the ability to present Ag at all reproductive stages and may be superior at this ability during midinvolution. We next tested for the ability of APCs to support Ag-dependent naive T cell activation. Based on our evidence that mammary dendritic cells have reduced expression of T cell activation markers (MHCII, CD80, and CD86, Fig. 2D–F, Supplemental Fig. 3A–C) and mucosal organs are characterized by tolerogenic dendritic cells, we predicted that lactating and involuting hosts are skewed toward inducing T cell tolerance.
Mammary APCs have reduced ability to activate naive T cells during involution
To address dendritic cell functionality, we employed an in vivo T cell activation assay. This assay evaluates the ability of mammary APCs, with dendritic cells presumably being the most potent, to present Ag to adoptively transferred CD4+ naive T cells within the mammary draining LNs (Fig. 5A, Supplemental Fig. 4A). This assay used naive DO11.10 TCR transgenic CD4+ T cells, which specifically recognize OVA Ag. Transgenic CD4+ T cells were purified (Supplemental Fig. 4B) and transferred by tail vein injection into syngeneic, wild type nulliparous, lactation, involution, or regressed hosts. Two days later, OVA Ag was injected unilaterally into the no. 4 mammary fat pad with PBS injected into the contralateral gland to serve as the internal negative control. Finally, CD4+ transgenic T cells in the mammary draining LNs were then detected via specific staining for DO11.10 TCR by flow cytometry (Fig. 5B, Supplemental Fig. 4C). To determine absolute transgenic T cell count, we used counting beads (Fig. 5B, top left panel). In nulliparous and regressed hosts, we observed an Ag-dependent increase in DO11.10 T cell abundance in the draining LNs, whereas increased T cell abundance was observed to a lower degree in lactation hosts and was absent in involution hosts (Fig. 5C). To assess the magnitude of T cell accumulation, we calculated the ratio of T cell abundance in OVA to PBS conditions. We observed an Ag-dependent ∼12-fold increase in T cells in the draining LNs of nulliparous and regressed hosts and only a ∼3–5-fold increase in Ag-dependent T cell accumulation in lactation and involution hosts (Fig. 5D). In contrast to these data observed in mammary draining LNs, the abundance of DO11.10 CD4+ T cells in the spleens of animals did not differ based on reproductive state or Ag treatment (Supplemental Fig. 4D), data consistent with localized, mammary-specific immune suppression. In sum, these data support mammary-specific suppression of Ag-dependent T cell activation during lactation and involution, consistent with mucosal immune programs.
Phenotyping of mammary tissue CD4+ T cells
Because we observe a decrease in CD4+ T cell expansion in the mammary draining LNs of lactation and involution hosts, we next sought to determine if the endogenous CD4+ T cell compartment in the mammary gland was similarly reduced. Compared to nulliparous mammary glands, lactating glands had decreased abundance of total CD4+ T cells (CD45+ CD4+ CD11b−) as a percentage of all immune cells (Fig. 6A, 6B). However, we observed an increase in CD4+ T cells during involution (Fig. 6B).
We next assessed programming of CD4+ T cells in the mammary gland at various reproductive states by staining for transcription factors indicative of T cell subsets commonly found at mucosal barriers, including RORγT+ (Th17), Gata3+ (Th2), and Foxp3+ (Treg). We found the existence of a baseline RORγT+ CD4+ T cell infiltrate indicative of a mucosal state in the nulliparous mammary gland (Fig. 6D, 6J). Further, we found evidence for activation and expansion of a mucosal state during lactation, as measured by increased RORγT+ and PD-1+ CD4+ T cell populations (Fig. 6D, 6I, 6J, PD-1+ T cell populations are outlined in red). With the onset of weaning (InvD2), we found further expansion of the RORγT+ population (Fig. 6C, 6D) consistent with the need for increased barrier protection as well as evidence for tissue inflammation. By InvD6, the gland is characterized by the presence of RORγT+, Gata3+, and Foxp3+ populations with high PD-1 positivity (Fig. 6D, 6E, 6G, 6I, 6J). Of note, many of the Gata3+ CD4+ T cells were also positive for RORγT, indicating a potentially plastic Th17–Th2 cell subset specific to late involution. Because our previous result described reduced naive T cell activation in the mammary LN during involution (Fig. 5) and given that we observe increased T cell abundance in the mammary gland 2 d following cessation of weaning (not long enough for a naive response to be mounted), we hypothesized that a putative source of CD4+ T cells in the involution mammary gland may be recruitment from a systemic memory pool.
Th17 and Treg memory CD4+ T cells are recruited to involution mammary tissue
To test the hypothesis that systemic memory CD4+ T cells accumulate specifically in the involuting mammary gland, we devised an in vivo, adoptive T cell transfer experiment in which memory T cells could be enumerated. Also, we tested for Ag dependence of memory T cell accumulation, as Ag-independent accumulation is consistent with an inflammatory environment. OVA-specific (DO11.10 TCR transgenic) CD4+ T cells were activated in vitro with anti-CD3 and anti-CD28 in the presence of TGF-β1, IL-1β, IL-6, and IL-2 to recapitulate the Th17-biased programming we observed in vivo (Fig. 6). Following 10 d of culture, the long-lived T cells were ∼85% positive for CD44, a marker of T cell activation (Supplemental Fig. 4E, 4F). Further, these cells were largely RORγT+ Foxp3+ (Supplemental Fig. 4G), indicative of a Th17–Treg polarization state. Next, we transferred these in vitro–activated long-lived (memory-enriched) D011.10 CD4+ T cells into involution (day 0) or nulliparous age-matched hosts. Two days later, OVA protein (i.e., Ag) was injected into the mammary fat pad with PBS injected into the contralateral gland (Fig. 7A, Supplemental Fig. 4A). Abundance of the adoptively transferred memory CD4+ T cells was assessed 5 d later using flow cytometry. We found significantly more memory CD4+ T cells in the involution mammary glands compared with the nulliparous glands (Fig. 7B). Further, this memory CD4+ T cell recruitment was independent of the presence of OVA Ag (Fig. 7C).
We next determined if memory CD4+ T cells in the involuting host were capable of expanding in the LN in response to mammary Ag or were suppressed, similar to what we observed with naive T cells (Fig. 5D). For these experiments, LNs and spleens were assessed for memory CD4+ T cell numbers following memory CD4+ T cell adoptive transfer in the presence and absence of mammary Ag (Fig. 7D). We observed Ag-dependent accumulation of memory T cells in the LNs of nulliparous hosts but not in involution hosts (Fig. 7E). Finally, we did not observe T cell expansion in the spleens of nulliparous or involution hosts (Fig. 7F), even in the presence of mammary Ag, data consistent with immune modulations being restricted to the mammary gland and its draining LN. Altogether, these data show that the involuting mammary gland is uniquely capable of accumulating Th17- and Treg-skewed memory T cells via an Ag-independent mechanism and support the existence of tightly regulated, localized inflammatory immune milieu.
Discussion
Our study is the first, to our knowledge, to describe a baseline mucosal immune profile in the murine mammary gland and provide evidence that immune tolerance to foreign Ags is reproductive state–dependent. Specifically, we find the nulliparous mammary gland is dominated by mucosal-associated immature dendritic cells and RORγT+ CD4+ T cells. During lactation, we find increases in antimicrobial IgA and mucin mRNAs, dendritic cells with tolerogenic features, RORγT+ CD4+ T cells, and functional evidence for Ag-dependent immune tolerance. During involution, these mucosal features are enhanced to include increased antimicrobial gene expression and extended to include Ag-independent accumulation of memory Th17–Treg CD4+ T cells and elevated levels of Gata3+, Foxp3+, and PD-1+ CD4+ T cells. The enhanced mucosal attributes of involution are transient; at 6 wk postweaning, the immune milieu resembles the nulliparous state (Fig. 8). These studies support the mammary gland as a mucosal organ and further identify reproductive control over these mucosal attributes.
Our data are consistent with mammary barrier function and immune tolerance being essential for maintaining women’s health during lactation, which is an argument pioneered by others (12, 44). One reason that enhanced barrier function and immune tolerance may be needed is because milk represents a foreign Ag risk. Specifically, although milk is a natural endogenous product of the mammary gland, milk proteins may represent rarely-encountered Ags and challenge both central and peripheral self-tolerance mechanisms (15, 45–47). Further, there is an increased risk of foreign Ag exposure during lactation because of the presence of a wide variety of viable bacteria in milk (20, 48), likening the lactating mammary gland to the gut. These bacteria are supported by milk sugars and are thought to play important roles in infant gut health (49). Thus, throughout lactation, it can be argued that it is important to prevent an immune reaction toward milk as well as minimize reactivity to milk-associated bacteria. This argument is congruent with the increases in IgA and mucin mRNAs, tolerogenic dendritic cells, and RORγT+ CD4+ T cells that we describe during lactation because these are key factors in maintaining immune tolerance and healthy epithelial barriers in mucosal organs (1–4).
To our knowledge, this study is also the first to identify mucosal barrier and immune tolerance programs that extend past lactation and into weaning-induced involution. Weaning-induced mucosal programs likely support the remodeling of the lactation-competent mammary gland to its nonsecretory, prepregnant-like state. During involution, we propose that the stimulus for tolerance mechanisms is the massive alveolar cell death that occurs postweaning (50). This is because the ingestion of dying cells induces tolerogenic dendritic cell functions and is thought to be a primary mechanism to avoid the generation of self-reactive immunity (51–53). Specifically, dendritic cell phagocytosis of apoptotic cells is known to inhibit dendritic cell maturation, as measured by reduced CD86, which altogether leads to a reduced capacity to activate naive T cells (51–53).
The primary mechanism by which APCs, with dendritic cells being most potent, suppress naive T cell activation is through Ag presentation in the absence of costimulatory signals. The results of this dendritic cell/T cell interaction are deletion of T cells via apoptosis, T cell anergy (or paralyzation), or T cell polarization to a regulatory phenotype (Treg) (35). Although we show Ag-dependent naive T cell accumulation is suppressed during lactation and involution and demonstrate T cell polarization to regulatory phenotypes, the particular role of the dendritic cell in executing these T cell responses is unknown, as other immune cells are also capable of these functions. However, decreased dendritic cell expression of MHCII and CD86 during lactation and involution is consistent with reduced Ag presentation to T cells and reduced costimulation. Furthermore, although the cytokine milieu of the lactating murine gland is poorly understood, the cytokine milieu within the involution mammary gland has high levels of TGF-β, COX2, IL-10, IL-4, and IL-13 (23, 24, 54–56). This cytokine milieu is consistent with the promotion of tolerogenic dendritic cell functions and/or direct suppression of T cells.
Surprisingly, despite observing T cell suppression after cognate Ag exposure during lactation and involution, we also observed Ag-independent accumulation of memory T cells to the mammary gland specifically during involution. The function of these memory T cells might be to skew the environment toward immune suppression (Treg) and maintain epithelial barrier integrity (Th17) during gland collapse. However, Treg–Th17-skewed CD4+ T cells have also been reported to have proinflammatory roles in intestinal mucosa (57). Thus, the involution-associated CD4+ T cell populations may provide tissue remodeling and regulatory programs balanced with proinflammatory immune programs as the gland returns to a nonlactating state. Although these programs are also expected to limit Th1 damaging inflammation that could be elicited during the massive epithelial cell loss and gland remodeling that occurs postweaning, future studies are needed to assess the specific roles of CD4+ T cell populations during gland involution.
The mechanism of how memory CD4+ T cells are recruited during involution is unknown, although activated vasculature may be predicted (58). Indeed, mammary vasculature is highly dynamic across a reproductive cycle (59, 60) and could play a role in differential recruitment and drainage of immune cells. Further, the source of these memory CD4+ T cells is unknown. Recruitment from other mucosal sites is possible, as integration across mucosal tissues is reported for lung and gut and is evidenced by alteration of systemic immune responses by the gut microbiome (61, 62). One interesting developmental window that we did not assess for mucosal attributes is the mammary gland during pregnancy. However, it is known that systemic immune modulation/suppression occurs to permit tolerance to fetal nonself Ags (63). How systemic immune suppression during pregnancy impacts the mammary immune milieu remains to be determined.
Our research on the mammary gland immune milieu may have far-reaching impacts on understanding pathologic conditions that arise during lactation and involution. For example, these two reproductive windows are characterized by increased self and foreign Ag exposure, and autoimmunity may occur if immune tolerance is not executed properly. Although a role for autoimmunity in lactation failure has not been explored, tissue destruction is a common result of autoimmune disease, and insufficient glandular tissue correlates with lactation failure in women (64, 65). The potential implications warrant further investigation, as lactation failure occurs in 5–15% of all parous women, leading to serious infant malnutrition if formula supplementation is not possible (64). Further, Th1 immunity specific to milk proteins has been shown to elicit autoimmune-like lactation failure, supporting a potential need for immune tolerance for successful lactation (66). Our study also may provide new avenues to investigate breast cancer risk and promotion because immune tolerance is the antithesis to immune surveillance. Of potential relevance, the risk of developing breast cancer is increased in the postpartum window, and cancers that develop postpartum have poorer prognoses than breast cancers in age-matched nulliparous women, even after matching for tumor stage and biologic subtype (9). Further, in rodent models, the immune milieu of the involuting gland has been implicated in progression of postpartum tumors (24, 54, 67). Our study finds that immune tolerance, which may be executed by tolerogenic dendritic cells, is enhanced during weaning-induced involution, implicating immune tolerance as one potential mechanism of mammary tumor promotion.
Altogether, this study describes for the first time, to our knowledge, the immune milieu of the normal murine mammary gland to be mucosal in nature at all reproductive stages assessed, as indicated by baseline immature dendritic cell phenotypes and RORγT+ CD4+ T cells. These mucosal immune programs are enhanced and extended to include immunosuppressive and immune-tolerant programs during lactation and involution. This study provides rationale for including the mammary gland as a mucosal organ, a perspective that may lead to new investigative insight within the mammary gland and possibly within the greater mucosal organ system.
Acknowledgements
We thank the OHSU flow cytometry shared resource and members of Dr. Lisa Coussens’ laboratory for helpful discussions. We thank Drs. Peter Henson and Jeff Nolz for critical review of the manuscript. We also thank Weston Anderson for support in manuscript editing.
Footnotes
This work was supported by National Institutes of Health (NIH)/National Cancer Institute (NCI) National Research Service Award Grant F31CA196052 to C.B.B. and NIH/NCI Grant R01CA169175 to P.S. and V.F.B.
The online version of this article contains supplemental material.
References
Disclosures
The authors have no financial conflicts of interest.