Preconditioning with Lipopolysaccharide or Lipoteichoic Acid Protects against Staphylococcus aureus Mammary Infection in Mice

Staphylococcus aureus is one of the most causative agents of mastitis and is associated with chronic udder infections. The persistency of the pathogen is believed to be the result of an insufficient triggering of local inflammatory signaling. In this study, the preclinical mastitis model was used, aiming to evaluate if lipopolysaccharide (LPS) or lipoteichoic acid (LTA) preconditioning could aid the host in more effectively clearing or at least limiting a subsequent S. aureus infection. A prototypic Gram-negative virulence factor, i.e., LPS and Gram-positive virulence factor, i.e., LTA were screened whether they were able to boost the local immune compartment. Compared to S. aureus-induced inflammation, both toxins had a remarkable high potency to efficiently induce two novel selected innate immunity biomarkers i.e., lipocalin 2 (LCN2) and chitinase 3-like 1 (CHI3L1). When combining mammary inoculation of LPS or LTA prior to a local S. aureus infection, we were able to modulate the innate immune response, reduce local bacterial loads, and induce either LCN2 or CHI3L1 at 24 h post-infection. Clodronate depletion of mammary macrophages also identified that macrophages contribute only to a limited extend to the LPS/LTA-induced immunomodulation upon S. aureus infection. Based on histological neutrophil influx evaluation, concomitant local cytokine profiles and LCN2/CHI3L1 patterns, the macrophage-independent signaling plays a major role in the LPS- or LTA-pretreated S. aureus-infected mouse mammary gland. Our results highlight the importance of a vigilant microenvironment during the innate immune response of the mammary gland and offer novel insights for new approaches concerning effective immunomodulation against a local bacterial infection.

by a primary insult that enhances the activity of a beneficial antibacterial host response which, in turn, clears the bacterial hazard when it presents itself (3). In contrast to previous literature, it is now more accepted that next to the adaptive immunity, the innate immunity can also harbor some immunological memory (4).
Mastitis is a disease featured by either chronically or acutely inflamed mammary glands and is still a major challenge in the dairy sector as it affects both milk quality and animal welfare (5,6). Staphylococcus aureus (S. aureus) in the udder of dairy cows causes harmful clinical symptoms and is associated with poor cure rates (7). The insufficient recognition of the infecting S. aureus in the mammary compartment is believed to hinder the host in raising an adequate innate immune response and the pathogen may, therefore, persist in the tissue (8,9). In addition, these Gram-positive bacteria may infiltrate the mammary epithelial cell (MEC) layers and induce complicating factors such as biofilms formation (10,11). Altogether, these observations are recognized to be a reason why classical antibiotic therapies have difficulties in eradicating contagious infections from a herd. Moreover, this currently leads to excessive use of antibiotics and even drug-resistant variants of mastitis pathogens (12)(13)(14). Commercial efforts have been made to create S. aureus mastitis vaccines e.g., Lysigin (Boehringer Ingelheim Vetmedica, Inc.) and Startvac (Hipra, Inc.) as an alternative to antibiotics. Unfortunately, many of these strategies have shown limited efficacy under field conditions (15).
The mammary gland is an immune receptive organ that harbors different kinds of pathogen recognition receptors (PRRs) on local cells (16). It is accepted that the outcome of the inflammatory reaction in this niche strongly depends on the efficiency of activating local PRR-expressing cells through the recognition of pathogen-associated molecular patterns (PAMPs) (17). Gram-negative mastitis pathogens contain several highly potent PAMPs that trigger multiple PRRs during an intramammary infection and result in a cytokine storm associated with acute symptoms at the cow's udder. These events eventually induce either a fast recovery or a negative outcome (18,19). By contrast, the comprehensive identification of Gram-positive pathogens by the innate immune system is impeded due to a smaller number of highly potent immunodominant proteins that typically mediate much milder symptoms and chronic infections (20)(21)(22)(23). Unraveling the inflammatory mechanisms that are triggered by different bovine mastitis pathogens and boost the host's immunity for low immunopotent bacterial strains such as S. aureus remain, therefore, mandatory. Such experiments in cows are severely limited with respect to availability of immunological research tools and requested infrastructure. Complementing in vitro bovine models, in vivo mouse models were developed to tackle these major drawbacks (24). The latter preclinical models allow the dynamic interaction between mammary gland cell populations and have been successfully used with different mastitic bacteria, including bovine S. aureus isolates, focusing on the early innate immune response that is highly conserved between both mammalian species (25).
Until recently, the main focus with regard to PRR-containing mammary cells was on the MECs as these were expected to be the key cell type responsible for sensing of mastitis pathogens (17,26). Nevertheless, novel data indicate that the mammary compartment is an immunological microenvironment where multiple innate immune cells are present that are also able to sense virulence factors of pathogens (27). The differential role of the sentinel cell types responsible for immunosurveillance during an infection is gaining increasing importance in the mammary gland immunity field (28)(29)(30)(31)(32). It is also currently, albeit partly, known that the mammary epithelium communicates with other PRR-containing cells in the mammary gland for bacterial recognition, e.g., by shedding PAMPs such as CD14 and likely also through microRNA (33,34). However, whether appropriate activation of this microenvironment may trigger subsequent antibacterial or anti-inflammatory responses to a secondary insult remains largely unknown.
The aim of this paper was to analyze the antibacterial capacity of the mammary microenvironment against S. aureus mastitis using an intraductal inoculation mouse model. First, the innate immune response of the mammary compartment resulting from exposure to virulence factors lipopolysaccharide (LPS) and lipoteichoic acid (LTA) or the well characterized bovine mastitis isolate S. aureus Newbould 305 was investigated. Second, we assessed immunomodulation and the antibacterial response of the complete mammary microenvironment by LPS or LTA on a subsequent insult with S. aureus. Third, we verified whether the preconditioning by either LPS or LTA to a S. aureus infection in our mouse model was unique for the local macrophages or whether other MECs are of essence for this beneficial response.
For the analysis of the host immune response in the different experiments, we preferred to measure lipocalin 2 (LCN2) and chitinase 3-like 1 (CHI3L1) to complement local cytokine profiles, rather than classical acute phase protein, such as serum amyloid A and haptoglobin. Both LCN2 and CHI3L1 are increasingly reported as key players in a wide range of inflammatory diseases. LCN2 binds and sequesters the iron-scavenging siderophore enterobactin, preventing Gram-negative bacterial iron acquisition (35,36). Recently, it has been shown that secreted human LCN2 cloned into bovine mammary epithelial stem cells actively inhibits growth of mastitis-causing S. aureus as well (37). CHI3L1, on the other hand, belongs to the family of chitinase-like proteins and was initially reported in host antifungal responses as an important scavenger of chitin that is present in the fungal cell wall (38). However, further research demonstrated that the enzymatically inactive CHI3L1 has prominent functions in bacterial adhesion and invasion in host tissue, bacterial clearance, and innate immunity processes (39)(40)(41)(42)(43).

MaTerials anD MeThODs intramammary inoculation Model
FigUre 1 | Experimental design. Hsd:ICR (CD1) 8-week-old female outbred mice were mated with 10-week-old male mice. The female mice were isolated for 1 week to give birth to their offspring. To ease the intraductal accessibility of the lactating mothers, the 10-day-old pups were weaned 2 h before the first inoculation. Depending on the preconditioning experiment: lipopolysaccharide (LPS)/lipoteichoic acid (LTA)/PBS or ex vivo LPS-/LTA-/PBS-treated macrophages were injected 24 h prior to the induction of Staphylococcus aureus mastitis in mice. To deplete local alveolar macrophages, preconditioning experiments could be preceded by clodronate liposome treatment. All intraductal injections (100 µl) were performed in the fourth gland pair of each of the mice. At relevant time points, post-inoculation mice were sedated and subsequently euthanized through cervical dislocation. Mammary glands were harvested to determine local bacterial loads, cytokine levels, or histopathology.  Figure 1. Eight-week-old female mice were mated with 10-week-old male mice. To ease the intraductal accessibility of the lactating mothers, the 10-day-old pups were weaned 2 h before the first inoculation. If possible the pups were transferred to foster mothers, the remaining pups were culled by CO2 in a closed compartment as approved by the ethical committee of the faculty. All intraductal injections (100 µl) using a 32-gauge blunt needle were performed in the fourth gland pair of each of the mice under isoflurane anesthesia combined with the long-acting analgesic buprenorphine (10 µg/kg Vetergesic, Patheon UK Ltd., UK) and core body temperature was post-operative monitored with a rectal thermistor. Next to the ones described below, different agents were injected in the mammary gland, such as clodronate and PBS liposomes (Van Rooijen, The Netherlands). Inoculation of mammary glands with clodronate-containing liposomes depletes local alveolar macrophages and indirectly also decreases the recruitment of PMNs following intramammary LPS inoculation (27). At relevant time points post-inoculation (p.i.) mice were sedated by an intraperitoneal administered mixture of ketamine (100 mg/kg Anesketin, Eurovet Animal Health BV, Bladel, The Netherlands) and xylazine (10 mg/kg; Xylazini Hydrochloridum, Val d'Hony-Verdifarm, Belgium) and, subsequently, euthanized through cervical dislocation. Mammary glands were harvested to determine local bacterial loads, cytokine levels, or histopathology.

Bacterial Products
The mastitis isolate S. aureus Newbould 305 (ATCC 29740) was grown in brain heart infusion (Thermo Scientific, Belgium) broth for 5 h from a frozen stock. The CFU/mL was determined after 24 h of incubation at 37°C from the 1/100 diluted bacterial preculture by spectrophotometric measurements at 600 nm (OD600) and were confirmed a posteriori by plating on tryptic soy agar (Thermo Scientific, Belgium) plates. The pathogen was suspended in PBS to achieve the required concentration (10 3 -10 4 CFU/100μl). This 100 µl solution was then used to inoculate the mammary gland in the different mouse experiments (Figure 1). Ultrapure LPS 0111:B4 (InvivoGen, USA) and ultrapure LTA from S. aureus (InvivoGen, USA) were suspended in endotoxin-free water and diluted to the required concentration in PBS. These 100 µl LPS and LTA solutions were injected 24 h prior to the intramammary infection with S. aureus (Figure 1).

cell culture
Peritoneal macrophages were isolated as described in Walachowski et al. (44) and plated out in DMEM/F12 (Gibco) with 10% fetal calf serum for 2 h according to Mosser and Edwards (45).   supernatant was centrifuged for another 30 min to obtain a clear pellet. Protein concentration of the lysate samples was determined as previously described (46). Cytokine quantification (ProcartaPlex from eBioscience), LCN2 (Biotechne, USA), and CHI3L1 ELISA (Biotechne, USA) on the mammary gland lysates were all performed according to company guidelines or previously reported (47).
histopathology Two mammary glands per condition were fixed in buffered 3.5% formaldehyde for 24 h at room temperature, dehydrated, and embedded in paraffin wax. Sections were deparaffinized, hydrated, and stained with hematoxylin and eosin. Subsequently, sections were dehydrated and mounted with a cover glass.

statistical analysis
Statistical differences were determined on normalized data with an independent sample T-test (SPSS 20.0, IBM Corporation, USA). The statistical differences for multiple comparisons were determined using ANOVA tests. Statistical significance between different groups was determined through Tukey post hoc testing.

resUlTs
Both lPs, and to a lesser extent also lTa, induce Mammary immune cell recruitment and local il-8 induction The effect of intramammary challenge with different doses (0.01-10,000 EU) of either LPS or LTA was compared in mice at 24 h p.i. (Figure 2A). Mammary glands subjected to LPS at 10,000 EU had significantly higher IL-8 levels than glands exposed to a same dose of LTA 24 h p.i. (Figure 2B). While significant differences were detected in local IL-8 levels after a low-dose challenge of LPS (10 EU) compared to PBS (sham)-inoculated glands, IL-8 levels only increased in LTA challenged glands starting at a 100-fold higher dose (1,000 EU). This observation was in line with the histological data, showing the recruitment of immune cells in the alveolar lumen at a later time point for the mammary glands exposed to LTA compared to the mammary glands injected with LPS ( Figure 2C).  lPs Treatment More effectively induces local lcn2 and chi3l1 than lTa Treatment, Partially concomitant with Mammary immune cell recruitment Lipopolysaccharide was more effective than LTA to significantly induce both mammary LCN2 and CHI3L1 compared to PBS (sham)-inoculated mammary glands (Figure 3). Nevertheless, it was also observed that local LCN2 and CHI3L1 levels not fully mimicked the previously mentioned IL-8 levels. Indeed, dissimilar to the IL-8 patterns, 10-fold higher doses of both LPS (≥100 EU) and LTA (10,000 EU) were needed to significantly induce CHI3L1 in the mammary gland compared to shaminoculated glands. Furthermore, no significant difference in LCN2 and CHI3L1 levels could be observed between the LPS and LTA inductions at the highest dose (10,000 EU). Interestingly, recruitment of neutrophils was observed in the mammary alveoli at the time point that both the LCN2 and CHI3L1 levels in the LTA and LPS groups were significantly increased compared to the PBS (sham)-inoculated glands.

S. aureus effectively recruits immune cells during Mammary-induced inflammation independently of lcn2
In contrast to sterile mammary gland-induced inflammation, inoculation of 1,000 CFU S. aureus Newbould 305 (S. aureus) in the murine mammary gland resulted in high bacterial counts (9.2 ± 0.3 log10 CFU/g mammary gland), accompanied by a prominent local host response at 24 h p.i. (Figure 4A). This inflammatory response was characterized by the production of multiple cytokines and chemokines as comparing tissue samples of a S. aureus-and a PBS (sham)-inoculated murine mammary gland (data not shown). In addition, the significant increase of IL-8 levels at 24 p.i. in the S. aureus-infected glands compared to sham-inoculated control glands could be demonstrated ( Figure 4B, left). This latter observation could be confirmed upon histological examination through a pronounced presence of immune cells in the lumen of the mammary alveoli with S. aureus, compared to sham-inoculated mammary glands ( Figure 4B, right). Interestingly, the IL-8 levels showed a median fold increase of 14.0 between S. aureus-and sham-inoculated mammary glands, whereas a 34.1 and 14.6 median fold increase was observed for 10,000 EU LPS-or LTA-exposed mammary glands, respectively, compared to the control glands. Although LCN2 levels in these infected glands were not significantly different from the LCN2 levels in sham-inoculated control glands, local induction of CHI3L1 levels by S. aureus could still be observed at 24 h p.i. (Figure 4C). More specifically, the median fold increase of local CHI3L1 levels in S. aureus-inoculated glands compared to sham-inoculated control glands was 2.88, which is an order of magnitude lower compared to the CHI3L1 levels following exposure to 10,000 EU LPS or LTA (8.16 and 5.13, respectively).
Preconditioning of the complete Mammary gland compartment with lPs or lTa strongly immunomodulates S. aureus Mastitis, reducing Bacterial loads, recruiting PMn, and increasing local chi3l1 or lcn2 levels, respectively As shown in Figure 1, mice were intramammary injected with PBS, LPS (10,000 EU), or LTA (10,000 EU) 24 h prior to S. aureus infection. With this setup, we wanted to verify whether LPS or LTA preconditioning is able to boost the murine mammary gland microenvironment against a S. aureus infection ( Figure 5A). All mice pretreated with either LPS or LTA showed a significant decrease in local bacterial loads 24 h p.i. with S. aureus compared to PBS (sham)-pretreated mammary glands ( Figure 5B). Histology also showed that all conditions were characterized by the presence of PMN in the alveoli of the mammary glands at 24 p.i. with S. aureus ( Figure 5C). However, on the molecular level, this observation during LPS and LTA pretreatment was accompanied with a significant reduction in mammary IL-8 levels compared to sham preconditioning ( Figure 5D). As previously shown, preconditioning with PBS prior to S. aureus infection induced low levels of CHI3L1, but did not induce LCN2 levels in the mammary gland. Preconditioning with 10,000 EU LPS or LTA prior to S. aureus infection, on the other hand, boosted either one of these innate immune biomarkers ( Figure 5E). More specifically, a significant increase in mammary CHI3L1 levels was detected with LPS preconditioning prior to S. aureus infection, whereas a significant increase in mammary LCN2 levels was observed with LTA preconditioning prior to S. aureus infection.
Local cytokine profiling showed that the inflammatory response following infection with S. aureus was influenced by the LPS and LTA preconditioning at an inoculum dose of 10,000 EU. IL-1 alpha, IL-1 beta, IL-17A, IL-6, MCP-1, and TNF-alpha followed the same pattern as the neutrophil chemoattractant IL-8 profile and were significantly reduced by LPS and LTA preconditioning compared to PBS preconditioning in the S. aureus-infected mammary glands (Figure 5F). Mammary RANTES/CCL5 levels were significantly increased by LPS preconditioning compared to the other conditions, while LTA preconditioning significantly reduced local BAFF levels compared to PBS preconditioning ( Figure 5G). alveolar Macrophages have an immunomodulatory Function during lPs and lTa Preconditioning against S. aureus-induced Mastitis Alveolar macrophages are a major innate immune cell type in the mammary gland that contain PRRs on the cell surface to sense bacterial antigens (such as LPS or LTA). Macrophages are known to be a crucial cell type during immune tolerance, e.g., endotoxin tolerance. It is a protective mechanism against excessive inflammation where macrophages change their responsiveness to LPS following a first exposure to LPS (50). To test their immunomodulatory potential of macrophages in the mammary gland, isolated murine peritoneal macrophages were exposed ex vivo either to 10   were found to secrete significantly more LCN2 and CHI3L1 than non-activated macrophages, i.e., M(PBS) (Figure 6A). Of notice, the secreted levels of both these innate immune biomarker proteins showed a significantly higher increase with M(LPS) than with M(LTA). In a next step, the immune modulating effect of these M(LPS)-versus M(LTA)-activated macrophages was studied in a mammary gland environment. As shown in  (Figure 6F).
Local cytokine profiling further identified a partly and differential immunosuppressive effect due to the activated macrophages ( Figure 6G) Figure 7A). Histology showed that intramammary inoculation of clodronate-containing liposomes could decrease the recruitment of PMN in the alveoli of the mammary gland upon challenge with LPS, whereas inoculation of PBS-containing liposomes followed by LPS challenge resulted in a typical strong influx of PMN ( Figure 7B). This distinction was not clearly observed when comparing both liposome groups upon challenge with LTA. Luminol-dependent bioluminescence is able to visualize detection of MPO activity and its use in vivo is specified for active PMN rather than macrophages. In vivo imaging with luminol confirmed the effect of clodronate treatment in the murine mammary gland upon LPS challenge by a decreased MPO activity compared to the mammary glands that were treated with PBS-containing liposomes ( Figure 7C). Of note, the low recruitment or activity of PMN by LTA compared to LPS as seen previously, impaired the visualization of the bioluminescent signal from luminol in both conditions. These observations indicate that clodronate treatment has mainly an effect on LPS-rather than LTA-induced inflammation. This was also indirectly shown through the stronger decrease in mammary IL-8 levels following clodronate treatment in the LPS-challenged group compared to the LTAchallenged group (Figure 7D).
To verify whether other PRR-expressing mammary cells-next to macrophages-may be activated and have immunomodulatory effects during a S. aureus infection, mammary glands of mice were challenged with a clodronate-or PBS-containing liposome treatment followed by either LPS or LTA challenge (both 10,000 EU) and a S. aureus infection (Figures 1 and 8A). Local depletion of macrophages by clodronate-containing liposomes did not significantly increase the local bacterial loads when LPS or LTA was intramammarily inoculated prior to S. aureus ( Figure 8B). However, the innate immune response following The white mammary glands represent the PBS liposome inoculated mammary glands, whereas the gray color gradient glands represent the glands that received clodronate. Following this liposome treatment, mammary glands were inoculated once more with either LPS (checkered pattern) or LTA (striped pattern). (B) Representative images of alveoli of mammary glands that received clodronate followed by LPS or LTA. PMN are indicated by an arrow. Scale bar, 100 µm. (c) Luminol-based bioluminescence imaging of mouse mammary glands that received clodronate followed by LPS or LTA (n = 3 mice for each group). (D). Normalized quantitative IL-8 values between PBS and clodronate inoculation in the mammary glands to compare the differences between LPS or LTA treatment (n = 6 glands for each group). Represented data were pooled from three independent experiments. p.i., post-inoculation. S. aureus infection was influenced by this depletion as IL-8 levels significantly increased in infected mammary glands challenged with either LPS or LTA (Figure 8C). In this experimental setup, LCN2 and CHI3L1 also showed different patterns upon S. aureus infection in the LPS-versus LTA-challenged groups. Intramammary inoculation with clodronate-containing liposomes and subsequent LTA, but not LPS preconditioning significantly increased local LCN2 levels compared to the PBS liposome treatment in infected glands. By contrast, local CHI3L1 levels were decreased in mammary glands that received LPS preconditioning and clodronate treatment followed by S. aureus infection. Liposomes in the mammary gland did not affect the local CHI3L1 levels p.i. with LTA preconditioning followed by S. aureus infection (Figure 8D).
Local cytokine profiling also showed that, similar to the neutrophil attractant IL-8 profile, the macrophage chemoattractant MCP-1 as well as IL-6 were increased by the macrophage depletion process both with LPS and LTA preconditioning followed by The bacterial loads in mammary glands that received clodronate before LPS or LTA preconditioning prior to a S. aureus infection (n = 7) glands for the PBS liposome groups (i.e., macrophage +); n = 10 glands for the clodronate liposome groups (i.e., macrophage −). (c) The IL-8 levels in mammary glands that received clodronate injection before LPS or LTA preconditioning prior to a S. aureus infection (PBS liposomes: n = 8 for the LPS preconditioning group, n = 6 for the LTA preconditioning group; clodronate liposomes: n = 10 for both the LPS and LTA preconditioning group). (D) Lipocalin 2 (LCN2) and chitinase 3-like 1 (CHI3L1) levels in mammary glands that received clodronate injection before LPS or LTA preconditioning prior to a S. aureus infection [PBS liposomes: n = 7 (LCN2/CHI3L1) for the LPS preconditioning group, n = 7 (LCN2) and n = 8 (CHI3L1) for the LTA preconditioning group; clodronate liposomes: n = 8 (LCN2) and n = 10 (CHI3L1) for both the LPS preconditioning group, n = 10 (LCN2/CHI3L1) for the LTA preconditioning group]. (e) MCP-1, IL-17A, RANTES/CCL5, BAFF, and IL-6 levels in mammary glands that received clodronate injection before LPS or LTA preconditioning prior to a S. aureus infection (n = 7 glands for the PBS liposome groups; n = 10 glands for the clodronate liposome groups). Represented data were collected from one experiment. Values are expressed as mean ± SEM. ns, not statistically significant; p.i., post-inoculation (*P < 0.05, **P < 0.01, and ***P < 0.001). S. aureus infection ( Figure 8E). BAFF was only affected by clodronate treatment in the LPS-pretreated group with S. aureus infection. IL-17A and RANTES levels, respectively, were significantly increased and decreased, respectively, in the LTA-pretreated group with clodronate-depleted macrophages and S. aureus infection.

DiscUssiOn
Staphylococcus aureus mastitis is contagious and does not respond well to classical antibiotic therapies (12)(13)(14). This Gram-positive pathogen is characterized by biofilm formation and intra-epithelial localization, which hinder access of therapeutic agents (11). Consequently, S. aureus udder infections have led to the overuse of antibiotics that increases drug resistance (51)(52)(53). Therefore, either novel vaccination or other innovative preventive measures with negligible side-effects are highly anticipated as they have remained unsuccessful in Grampositive mastitis (7). The hosts' recognition of pathogen-associated factors and its innate immune response determines the outcome of bovine mastitis (54,55). The persistent character of S. aureus udder infections is believed to result from a limited mammary cytokine production compared to the "cytokine storm" raised against E. coli (20,21). This implies that E. coli antigens are either more easily recognized by mammary PRR or that Gram-negative virulence factors have a higher potency to trigger a strong local immune response. In this study, the dose-dependent stimulation with LPS versus LTA allowed controlled investigation of the potency and differential activation of PRRs through monitoring of local IL-8 production and immune cell infiltration at 24 h p.i. Histology demonstrated that the highest inoculum dose of LPS and LTA initiated innate immune cell recruitment. However, the innate immune response was significantly faster and higher stimulated by LPS compared to LTA.
It has been suggested that LPS may influence the course of a subsequent bacterial udder infection (56,57). More specifically, injection of LPS in the udder reduced the severity of a subsequent experimental E. coli mastitis (57). Whether preconditioning enhances the pathogen recognition or the host immune response is not clear. Staphylococcus aureus stimulates TLR2 but lacks efficient activation of TLR4 compared to E. coli (58,59). Similarly, as TLR4 activation by LPS is protective against E. coli mastitis, this study aimed to boost TLR2-induced immune response of the murine mammary microenvironment by LTA to protect the host against S. aureus mastitis. On the other hand, we also explored the possibility that we should compensate for the lack of TLR4 signaling during S. aureus mastitis by preconditioning with LPS. Our results confirmed the observations previously suggested for an in vitro S. aureus infection (60,61): preconditioning the host significantly reduced bacterial loads at 24 h p.i. We also described the participation of local chemo-or cytokines. Although several major local pro-inflammatory mediators decreased, the host was still able to recruit neutrophils to the lumen of LPS or LTA preconditioned alveoli. The lower local IL-8 levels are, therefore, likely an indication of abolished further neutrophil recruitment rather than of their initial LPS-or LTA-induced recruitment. In addition, local MCP-1 and IL-17A levels are signature cytokines for macrophage recruitment and T helper cell activation, indicating that the immunomodulatory response is not limited to neutrophils. On the one hand, LTA but not LPS preconditioning additionally lowered local BAFF levels, a cytokine necessary for proliferation and differentiation of B cells. On the other hand, LPS but not LTA preconditioning elevated local RANTES/CCL5 levels, in line with the in vitro observations of cytokine networks after recurrent exposure of bovine MECs to LPS (56). Overall, our different preconditioning protocols against S. aureus mastitis together with the reported sterile mammary inflammation data indicate that the two key Gram-negative and -positive virulence factors have unique traits and differentially immunomodulate the mouse mammary gland.
Mammary sentinel cell types-each featuring a distinct immune competence-govern the host response to virulence factors in the mammary gland (16). Some mammary epithelial tissue exposure experiments have described an effect of LPS and LTA on the innate immune response, but how these key virulence factors modulate the mammary gland microenvironment has been scarcely studied in vitro and especially in vivo (16,27,32). The validated murine intramammary inoculation model previously demonstrated that local macrophages are of key importance for the LPS-TLR4 axis-mediated neutrophil recruitment into the alveolar lumen, while this is far less the case for TLR2-mediated signaling (27,62). In general, macrophages can be programmed to different priming and tolerance states by varying dosages of LPS (63). Partly corroborating the observations from our in vivo sterile mammary inflammation data, we demonstrated that macrophages alone affect the mammary environment albeit to a limited degree. Indeed, in vitro preconditioning of alveolar macrophages with either LPS or LTA, and their transfer into sterile murine mammary glands did not induce a neutrophil influx. Moreover, upon S. aureus infection, activation of these differently preconditioned macrophages reduced the bacterial loads only to a limited extend compared to sham-exposed macrophages. Of relevance, the different TLR trigger changed the response of macrophages as LTA-dependent macrophage signaling was more efficient in reducing the inflammatory signaling compared to the LPS-dependent macrophage signaling. Injection of macrophages treated with LPS or LTA into the gland prior to infection attenuates the S. aureus-induced cytokine release when compared to mice receiving PBS-treated macrophages. However, the pretreated transferred macrophages are not sufficient to reduce the bacterial burden, suggesting that LPS and LTA have to act directly on cells in the mammary gland. MECs have been proposed as special non-professional sentinel cells that are responsible for macrophage-independent local signaling (64). To investigate the role of the mammary epithelium in the LPS-and LTA-mediated immunomodulatory response against S. aureus, macrophages were depleted in our preconditioning mastitis model by clodronate-containing liposomes. This local treatment did not change the antibacterial host response, but strongly immunomodulated a subsequent mammary S. aureus infection. Consequently, the antibacterial response after preconditioning with either LPS or LTA likely originated from macrophage-independent signaling by the mammary epithelium. Moreover, these data extend those obtained with the preconditioned macrophages. Indeed, macrophages again immunomodulated the mammary compartment during preconditioning as clodronate depletion influenced the local cytokine profile. Depletion of local macrophages was able to further distinguish the differential LPS versus LTA immunomodulatory signaling prior to a S. aureus infection. More specifically, our data showed that macrophages inhibited BAFF levels upon LPS preconditioning, while inhibiting IL-17A upon LTA preconditioning. In addition, macrophages had a pro-inflammatory effect by induction of RANTES/CCL5 upon LTA preconditioning prior to S. aureus infection. However, based on the methods used in this study, it is not possible to verify whether the changes in cytokine, chemokine, and antimicrobial peptide levels following clodronate pretreatment and subsequent LPS/LTA preconditioning followed by a S. aureus infection are either due to an enhanced protein production by the mammary epithelia or the consequence of a reduced uptake by the macrophages after their depletion. Gene expression profiles on isolated mammary cell types would be of added value to distinguish between both phenomena, albeit these were not assessed in this study. In essence, mammary macrophages support the MECs to respond adequately to a local infection. This corroborates the key observations that alveolar macrophages aid the lung epithelia in tissue homeostasis (65) and local immune response (66).
Complementing the local cytokine profiles, we at first evaluated LCN2 and CHI3L1 as potential in vivo and in vitro innate immunity modulators in the context of mastitis. Both LCN2 and CHI3L1 are stored in granules from mature human neutrophils (67)(68)(69). In line with our demonstration of neutrophils on histology, increasing doses of LPS or LTA in the mammary gland comparably induced local LCN2 and CHI3L1 levels. However, no difference was seen at 24 h p.i. for either LPS or LTA at the highest inoculum dose (10,000 EU) as expected based on the differential local IL-8 patterns. Moreover, during S. aureus-induced mastitis, a salient finding was that neutrophil recruitment only caused a limited induction of local CHI3L1 and even no increase in mammary LCN2 compared to sham inoculation. These unexpected data strongly indicated that S. aureus, in marked contrast to its key virulence factor LTA, did not fully trigger this innate immune signaling even if the basic mastitis characteristic, i.e., neutrophil influx was present. Moreover, our data indicated that neutrophils can be ruled out as the (only) source of LCN2 and CHI3L1, as a strong increase of both innate immune signals was absent in experimental S. aureus murine mastitis. Nonetheless, we showed a macrophage-dependent induction of both LCN2 and CHI3L1 in the mammary compartment found to be typically triggered by LPS and less by LTA. Indeed, the latter virulence factor had a remarkably limited influence on the local cytokine profile. By contrast, clodronate treatment demonstrated that the macrophage-independent triggering by LPS was characterized by a strong induction of local CHI3L1 but not of LCN2. Again in marked contrast, LTA now also had a major influence, however, selectively on LCN2 and not on CHI3L1. These important findings strongly indicate that the LPS-induced upregulation of CHI3L1 in the mammary gland is mainly mediated by local macrophages, while the LTA-induced upregulation of LCN2 is inhibited by macrophages. Based on the local cytokine patterns combined with the LCN2 and CHI3L1 levels, a complex interplay between macrophages and the MECs occurs which is of critical importance to effectively induce antibacterial and anti-inflammatory processes and which encompasses either LCN2 or CHI3L1 induction.
In conclusion, extrapolation of our preclinical data allows us to suggest that during a bovine S. aureus-induced mastitis, the effective activation of host innate immunity in the murine mammary gland is lacking. We unequivocally demonstrated that activation of the complete mammary gland microenvironment is required to effectively clear this Gram-positive infection and dampen the associated local pro-inflammatory cytokine response. Our current data also underscore the added value of the novel sentinel markers such as LCN2 and CHI3L1 that can be induced by different mammary cell types and may aid in the evaluation of preconditioning protocols.

eThics sTaTeMenT
This study was carried out in accordance with the FELASA guidelines and recommendations. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Ghent University.
aUThOr cOnTriBUTiOns KB, JS, and KD were essential for the conception and design of the work; the acquisition, analysis, and interpretation of data for the work; and drafted the work. TB and EM revised it critically for important intellectual content. KB, JS, KD, TB, and EM gave their final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.