Lipopolysaccharide priming enhances expression of effectors of immune defence while decreasing expression of pro-inflammatory cytokines in mammary epithelia cells from cows

Profiling the LPS-mediated modulation of the global transcriptome in MEC

Cultures of primary epithelial cells (pbMEC) were used to globally profile the effects of LPS on the transcriptome. The cells were stimulated (primed) for 12 h with a low concentration of LPS (100 ng/ml) and their RNA was collected after either a short (18 h) or an extended (48 h) waiting period (Figure 1). Respective controls of untreated cultures were included. For monitoring the effect of LPS priming upon the immune responsiveness of those cells, we challenged the cultures for 6 h with 10⁷/ml of E. coli particles either 12 h or 42 h after the LPS pre-treatment. Thus, cells were either harvested 30 h (short waiting experiment) or 60 h (long waiting experiment) after start. Adequate control cultures of non-primed cells challenged with the same E. coli stimulus were also analysed. The entire experiment therefore consisted of eight different conditions. It was replicated with pbMEC cultures obtained from three different cows. Thus in all 24 RNA samples were collected and individually analysed with GeneChip bovine genome arrays (Affymetrix).

LPS priming induced changes level off after time

The correlation analysis of the microarrays (Figure 2) revealed that the three biological replicates clearly clustered together in any of the respective treatment groups if the cells were collected 18 h after the LPS treatment. This was no longer observed if the cells were analysed after extending the waiting period to 48 h post priming. Hence, the LPS effect vanishes with time.

Challenging the cultures 48 h post-LPS-priming modulated the abundance of only 15 DET including 13 DEG compared to the controls. Considering the modulation of the immune responsiveness through LPS priming we found that this treatment regulated only six genes differently, if the E. coli challenge was applied 48 h after the end of the LPS priming (Table 1). Therefore, we focused our analysis on data obtained from the short waiting experiments. We identified the human orthologs of those bovine DEGs and only considered the human identifiers for the subsequent analyses, since more comprehensive knowledge is available about the human factors regarding their functions and regulatory relationships.

A complete list of significantly regulated genes and all statistically relevant parameters is given in the Additional material (Additional file 1, Table S1 for the comparisons primed vs. control and Additional file 2, Table S2 for induced post priming vs. induced).

Priming of pbMEC induced the expression of genes involved in immune-protective mechanisms

Priming decreased the expression of only ten genes. Overall, the magnitude of downregulation was very low. The mRNA concentration of the most downregulated gene ELTD1 (EGF, latrophilin and seven transmembrane domain containing 1) was in primed cultures only 2/3^rd of that found in the control cells (Table 2). This gene encodes a G-protein coupled receptor with unknown function. Three other downregulated genes are known to increase "Apoptosis of Cell Lines" (tribbles homolog 3, Ras association domain family 4, and putative lymphocyte G0/G1 switch gene 2; Table 2).

Only nine genes were found to be differentially expressed relative to the control if the waiting period after the LPS stimulus was extended to 48 h. Priming enhanced their expression. Three of those genes were also found to be regulated after the short waiting period. The latter genes encode CCL5, LTF, and transglutaminase 3 (TGM3). The extent of priming-induced upregulation of CCL5 and LTF expression was weaker in the long waiting experiment (36- and 2-fold, respectively; Additional file 1, Table S1B). However, it was stronger for TGM3 (7-fold after long- and 2-fold after short waiting period; Additional file 1, Table S1B). The mRNAs encoding interleukin IL-6 and the chemokine CXCL6 were upregulated by priming only after the long waiting period (6- and 4-fold; Additional file 1, Table S1B).

Priming enhanced the E. coli mediated induction of immune-protective molecules, but decreased expression of pro-inflammatory and cell death associated factors

We have previously reported that an E. coli challenge of naïve pbMEC regulated the expression of > 300 genes [22]. Given this background knowledge, we analyse in this study only the LPS priming-mediated modulation of the E. coli-elicited response.

We observed for 226 genes a priming related modulation of their expression subsequent to an E. coli challenge (Table 1; columns I.p.P. vs. I.). Expression of 38 genes was higher while that of 188 genes was lower than in non-primed but challenged cells (Figure 3A). We first quantitatively differentiate the impact of the LPS priming upon their subsequent regulation by the E. coli challenge. The biological functions of the factors encoded by the DEG were determined using the Ingenuity software. In a second step we identified the molecular relationships of all 226 DEG by an IPA network analysis.

The LPS priming mediated quantitative modulation of mRNA levels eventually affects all different regulatory levels controlling mRNA abundance. Accordingly, we sort the genes into six different regulatory groups.

Identification of central factors regulating the different response to E. coli after priming

The expression related interdependence of those 226 priming mediated DEGs was analysed with the IPA software in order to uncover regulatory key factors. A network emerged comprising 78 of those DEGs. It is associated with the IPA functions "Inflammatory Response", "Antimicrobial Response", and "Inflammatory Disease" (Figure 3B). The central node molecules in this network are the cytokine IL-1B and the transcription factor interferon regulatory factor 7 (IRF7). IL-1B is known as a central regulator of immune response and the IPA data base indicates that it may affect the expression of 44 from among our 226 DEGs, including IRF7. This factor is activated through the TRIF-factor dependent downstream signalling of TLR4 and of the endosomal TLR receptors (TLR3, 7, 8, 9) [26]. IRF7 is known to directly influence the expression of 34 DEGs. Other factors occupying node positions in this network are known as DEAD (Asp-Glu-Ala-Asp) box polypeptide 58 (DDX58; linked to 14 downstream factors), IRF1 (linked to 13 downstream factors), and TNFSF10 (linked to 12 downstream factors). The expression levels of all these factors were reduced by the LPS priming prior to the pathogen challenge.

Validation of the induction post-priming experiment by RT-qPCR

Our microarray analysis had shown that priming prior to a pathogen challenge enhances the expression of bactericidal and immune-protective factors while attenuating cytokine and chemokine expression. We used RT-qPCR to validate from the same RNA samples as used for the microarray analysis the differential expression of representative candidate genes (Figure 4; Additional file 3, Table S3 and Additional file 4, Table S4). We included some other key immune regulatory factors (TNF-α, IL6, CXCL8 [also known as IL8]) factors revealing a statistically insignificant modulation in the microarray experiments under the stringent conditions applied to that analysis. We found a good correlation (r, 0.91; 14 genes) between changes measured by RT-qPCR and the microarray data (Additional file 5, Table S5). The RT-qPCR data show that priming quenched the pathogen induced expression of the cytokines IL1B, TNF-α, and IL6 by ~30-50%. The mRNA concentrations of these factors are known to eventually peak in MEC within 1 h after an E. coli stimulation and are subsequently massively downregulated to lower levels which are sustained from 6 h to 24 h post stimulation [22, 27]. Thus, the only moderate modulation after 6 h of pathogen stimulation as observed here is conceivably a conservative estimate of the actual priming-related changes.

IL-1B, TNF-α, and IL-6 are often called "master" cytokines, owing to their secretion and perception by many different cell types and their key role in orchestrating the synthesis of other cytokines and chemokines. TNF-α may eventually influence the expression of 60 of those 226 priming-mediated DEGs while IL-6 might affect the expression of another 27 DEGs (Table 3; Additional file 1, Table S1 and Additional file 2, Table S2). Similarly reduced was the IL-1/IRF7 regulated induction of the factors NOS2 (bactericidal and immune regulatory), IL15, MX2, and RTP4. On the other hand, the RT-qPCR measurements very clearly confirmed that priming alone significantly enhances the expression of the key chemokines (CCL5, IL8), of anti-microbial factors (LAP, SLPI) and also of molecules which are known to promote cell and tissue protection in the course of an inflammation (SLPI, TGM3, SAA3, LTF). The priming-mediated enhanced expression of LAP, SLPI, and TGM3 after pathogen stimulation was also confirmed.

Priming enforces sentinel functions of the MEC

A mild LPS priming (1 μg/udder quarter, for 12 h) was found to significantly increase in vivo the number of somatic cells in udders [23]. Thus, it was not surprising to find in our in vitro pbMEC model that LPS priming enhanced quite strongly the level of mRNAs encoding the chemokines CCL5 (also known as RANTES) and CXCL8. Both factors are known to trigger diapedesis of mononuclear cells, T-cells and macrophages (CCL5 [29]) and polymorph nuclear granulocytes (PMN; CXCL8 [30]) through the endothelia of the blood vessels recruiting them into the inflamed sites. Elevated levels of both factors enhance the abundance of professional sentinel cells (e.g. macrophages) in the udder and conceivably improve the competence of the gland to counter act recurrent infections. It is known from macrophages that LPS priming enhances CCL5 expression in response to a second LPS stimulus [11].

In contrast to observations made on endotoxin tolerant macrophages [14, 15], we found that priming increased the abundances of a variety of mRNA moieties encoding peptidases for antigen processing (ERAP2, CTTS) and MHC-II factors (bovine orthologs of HLA-DQA1, -DRA, -DQB1) in MEC. While MEC are generally not considered as professional antigen-presenting cells (APCs) it is known that alveolar epithelial cells express MHC-II factors and are relevant APCs in the lung [31]. Thus, enhanced expression of these factors in LPS-primed MEC suggests an improved readiness of the primed cells for antigen presentation and thus improving their sentinel competence.

Priming improves protection against pathogens and tissue damage

Priming increased the levels of mRNAs encoding the bactericidal β-defensin lingual antimicrobial peptide (LAP). It was shown in macrophages that ET remodels the chromatin at the promoters of such "not LPS tolerant" effector genes of immune defence through differential histone modification making them more quickly responsive to a second stimulus [11]. Epigenetic mechanism might also operate during ET at the LAP promoter of the MEC, since mastitis induced expression of the LAP gene requires chromatin decompaction at the promoter [32].

Focussing on LAP as an example for a β-defensin encoding gene serves only as a paradigm for the regulation of this class of molecules, which are so abundantly encoded in the bovine genome. More than 100 highly related copies of β-defensin-encoding genes were uncovered in the analysis of the entire bovine genome [33]. The infection induced expression of LAP and other β-defensins in the udder has previously been reported [18, 19, 34]. The wealth of these β-defensin-encoding genes could not be analysed with the tools applied in this study, but LPS priming-enhanced abundance of the LAP-encoding RNA might be indicative for the regulation of other members of this gene family. Hence, LPS priming supports protection against colonization of the udder by bacterial pathogens. This was indeed found to be the case in vivo[10, 23].

Priming increased also the mRNA level of the defensin like factor secretory leukocyte peptidase inhibitor (SLPI), known as contributing to the defence against bacteria, fungi and retroviruses in epithelial tissues [35]. This multifunctional serine protease inhibitor protects epithelial tissue during inflammation from the attack by endogenous proteolytic enzymes. Moreover, SLPI is known to bind and synergize with proepithelin. This growth factor promotes proliferation of epithelial cells and suppresses activation of neutrophils. SLPI abrogates proteolytic cleavage of proepithelin into the inflammation sustaining factor epithelin [36]. Lactotransferrin (LTF) is another multifunctional factor revealing LPS priming-related upregulated mRNA levels. This factor has bacteriostatic, bactericidal and some immune modulating properties, but also the capability to bind LPS (reviewed in [37]). It was shown that LTF pre-treatment protects mice from pathophysiological effects of LPS and enhances the survival after an LPS injection [38]. Priming enhanced the levels of the mRNA encoding the complement factor CFB. This protein activates the alternative complement pathway thereby strengthening unspecific immune defence mechanisms. Expression of this factor in the udder is also strongly induced during E. coli-elicited mastitis [17].

Priming increased the mRNA abundances of several factors protecting against membrane damage associated with the so called "bystander killing" effect of activated immune defence. These include the decay accelerating factor for complement CD55 (also known as DAF) protecting membranes of the host cells against misguided attacks through complement factors and also the glycoprotein mucin 1 (MUC1). This mucin is a key component of the apical (luminal) membrane of the MEC [39] forming a physical barrier against microbial attacks [40]. Increased expression of the transglutaminases TGM1 and TGM3 indicates stabilization of tissues and cells by protein cross-linking [41]. These factors contribute also to the formation of tight junctions between mammary epithelial cells and are known to promote cross linking of proteins in the extracellular matrix thereby initiating wound healing.

Priming reduces expression of master cytokines and of potentially harmful factors

It is known from macrophages that endotoxin tolerance (ET) down-regulates the expression of the inflammatory master cytokines tumor necrosis factor α (TNF-α), interleukin 1 (IL-1) and IL-6 (reviewed in [13]). They all belong to the class of "LPS-tolerant" genes which become refractory in macrophages to repeated LPS stimulation [11]. We now made a similar observation in the MEC. LPS priming moderately, yet significantly reduces their mRNA levels after a subsequent E. coli challenge in these cells. These three factors together orchestrate and dominate many aspects of the local as well as systemic immune response [42, 43]. While their adequate induction through invading pathogens is indispensable for mounting a productive immune defence in the host, their overshooting expression may be detrimental. Confining their expression is therefore an overarching beneficial effect of ET and contributes to prevent immune-pathology. Our data suggest that this principle applies also to the regulation of the immune defence in the udder mediated through MEC. Moreover, we validated reduced and confined expression for some of their secondary response factors during the subsequent pathogen challenge. These factors include the nitric oxide synthase NOS2. This enzyme produces nitric oxide radicals, which are bactericidal but also harmful to the host cells and tissues as well. Similarly, expression of the metallopeptidases (MMP9, MMP13) was found to be confined by LPS priming. These proteases may disintegrate the extra-cellular matrix and connectives. LPS priming appears to prohibit their exuberant production and thus reduces activity of these potentially harmful factors.

Expression of many factors associated with cell death was found to be limited by LPS priming. These factors include the caspases CASP4 and -7, DAXX, CFLAR to name only some of them. Reduced and limited induction of these factors is conceivably a consequence of reduced TNF-α induction and indicates improved preservation of cell- and tissue-integrity.

Reduced and dampened induction of the TLR/NF-κB axis of pathogen signal transduction appears to be crucial for the ET mediated modulation of the immune response in the MEC. Not only was the pathogen stimulated expression of several TLRs reduced (TLR2, -3,-4) but also were factors induced by the LPS priming which are known to interfere with NF-κB activation, including the coronin CORO1A and SPLI. CORO1A is known to suppress TLR2, TLR3, TLR4 and TNF-α mediated NF-κB activation as well as IFN-β promoter activation [44]. Indeed, we found that the mRNA expression of various type I interferon response factors was decreased by priming (e.g. myxovirus resistance 2 [MX2], receptor transporter protein 4 [RTP4], 2'-5'-oligoadenylate synthetase 1 [OAS1], interferon stimulated exonuclease gene 20 kDa [ISG20]). The factor SLPI has not only a broad spectrum of antimicrobial activities but also anti-inflammatory/immunomodulatory functions [45]. SLPI interferes also with TLR-4 mediated NF-κB activation by inhibiting the interaction between CD14 and LPS [46] thus quenching the production of inflammatory factors like TNF-α, NOS2, COX2, and MMPs [47, 48].

Cell culture

Primary cultures of MEC (pbMEC) were obtained from udders of three healthy, pregnant (day 130 of gestation) cows in the mid of their first lactation. Cell preparation and their culture on collagen coated dishes were as previously described [17]. The pbMEC growth medium (GM) was RPMI 1640 (Biochrom AG; Berlin, Germany) supplemented with prolactin, hydrocortisone, insulin and 10% FCS as described [18]. Fibroblasts were selectively removed by repeated trypsinization. We number throughout the cultures derived from those three individual cows as 1, 2 or 3.

E. coli 1303 and LPS preparation

Heat killed (60°C, 30 min) particles of the mastitis causing E. coli strain 1303 [19] have been prepared as described [17]. LPS was prepared from this strain by butanol extraction and hydrophobic interaction chromatography [49] followed by a purification with triethylamine and deoxycholate [50] (kindly provided by Sonja von Aulock, University of Konstanz, Germany). It was dissolved in GM (1 mg/ml) by ultra-sonic dispersion (2 min) and aliquots were stored at -20°C.

LPS priming and challenge with heat inactivated E. coli particles

Figure 1 illustrates the experimental settings. Each experiment was performed in triplicate using separate pbMEC cultures (80% confluence) prepared from the udders of those three different cows. Controls (C.) were cultured for 12 h in GM, washed 3× with PBS and cultivated for additional 18 h (short waiting experiment) or 48 h (long waiting experiment). For the priming experiment (P.) the pbMEC cultures were stimulated ("primed") at time 0 h with 100 ng/ml LPS, washed three times with PBS and cultivated for another 18 h or 48 h. For the induction trial (I.) cells were challenged with 10⁷/ml particles of heat killed E. coli₁₃₀₃ for 6 h. The cultures were handled exactly like the control group (C.) prior to the E. coli challenge. For the induction post priming experiment (I.p.P.) cultures were primed for 12 h with LPS, just as described for the priming group (P.). After 3× washing with PBS and waiting for 12 h or 42 h in GM the I.p.P. cultures were also challenged with 10⁷/ml particles of E. coli₁₃₀₃. Cells were harvested at time 30 h (short waiting experiments) or 60 h (long waiting experiments) and total RNA was prepared.

Microarray hybridisation

Total RNA for the microarray analysis was extracted with the RNeasy kit exactly as described by the manufacturer (Qiagen, Hilden, Germany). RNA processing, labelling and hybridization used the respective reagent kits from Affymetrix and was exactly done as previously described [22]. Briefly, 5 μg total RNA of each sample was used for cRNA preparation and labelling with the Affymetrix GeneChip^® Expression 3' Amplification One-Cycle Target Labelling Kit (Affymetrix, St. Clara, USA). The fragmented cRNA was hybridized for 16 hours at 45°C to GeneChip^® Bovine Genome (Affymetrix). The Microarrays were scanned at 1.56 micron resolution using the GeneChip Scanner 3000 7G (Affymetrix). The microarray data sets were submitted to the Gene Expression Omnibus (GEO) database (accession no. GSE32186).

Microarray analysis

The microarray data were analysed using the Biometric Research Branch (BRB) array tools version 4.1 [http://linus.nci.nih.gov/BRB-ArrayTools.html]. Background correction and normalization of the expression values was performed using the GC Robust Multi-array Average (GCRMA) algorithm [51]. Transcripts were defined as significant differentially expressed (DET) among Control (C.) and Priming (P.) as well as among Induction (I.) and Induction post Priming (I.p.P.) groups if their fold change exceeded > 1.5 and the p-value of the univariate t-test between values paired according the pbMEC preparation was < 0.005. False discovery rates (FDR) values were calculated and are listed in the additional files (Additional file 1, Table S1 and Additional file 2, Table S2). However, this parameter was not applied as a cut off criterion since it would have been too stringent for some comparisons, resulting in no significantly regulated genes at all. This applies in particular to the comparison between primed vs. control cultures. Human orthologs of differentially expressed genes (DEG) were determined as described previously [22, 52]. Gene ontology and network analysis was performed using the Ingenuity Pathways Analysis (IPA 9.0) software (Ingenuity Systems, Redwood City, CA). This program identifies the biological functions that were most significant to a data set using Fisher's exact test. The calculated p-value specifies the probability that each biological function assigned to a gene list is solely due to chance. The heatmap was generated using the software tool matrix2png [53].

Quantitative real-time PCR (RT-qPCR)

Total RNA extraction with Trizol (Invitrogen, Karlsruhe, Germany), cDNA preparation with Superscript II (Invitrogen) and quantification of mRNA copy numbers with the LightCycler instrument and the SYBR Green plus reagent kit (both from Roche, Basel, Switzerland) was as described [54]. Quality of the PCR products obtained in those LightCycler runs was validated by subcloning the amplicons into pGEM-Teasy (Promega) and sequencing. Relative copy numbers were titrated against external standards consisting of a dilution series (10⁶ to 10 copies) of those sequenced plasmids. Sequences of the oligonucleotide primers used are given in Additional file 6, Table S6. Significance of differences was assessed with the t-test of paired values. Spearman's Rank Correlation was used to compare microarray and RT-qPCR measurements using the SigmaStat 3.5 software (Systat Software Inc., San Jose, California, USA).