ABSTRACT
Milk fat globules (MFGs) are vesicles released in milk as fat droplets surrounded by the endoplasmic reticulum and apical cell membranes. During formation and apocrine secretion by lactocytes, various amounts of cytoplasmic crescents remain trapped within the released vesicle, making MFGs a natural sampling mechanism of the lactating cell contents. With the aim of investigating the events occurring in the mammary epithelium during bacterial infection, the MFG proteome was characterized by two-dimensional difference gel electrophoresis (2-D DIGE), SDS-PAGE followed by shotgun liquid chromatography-tandem mass spectrometry (GeLC-MS/MS), label-free quantification by the normalized spectral abundance factor (NSAF) approach, Western blotting, and pathway analysis, using sheep naturally infected by Mycoplasma agalactiae. A number of protein classes were found to increase in MFGs upon infection, including proteins involved in inflammation and host defense, cortical cytoskeleton proteins, heat shock proteins, and proteins related to oxidative stress. Conversely, a strikingly lower abundance was observed for proteins devoted to MFG metabolism and secretion. To our knowledge, this is the first report describing proteomic changes occurring in MFGs during sheep infectious mastitis. The results presented here offer new insights into the in vivo response of mammary epithelial cells to bacterial infection and open the way to the discovery of protein biomarkers for monitoring clinical and subclinical mastitis.
INTRODUCTION
Mycoplasma agalactiae is the causal agent of contagious agalactia (CA), one of the most serious infectious diseases affecting small ruminants worldwide (11, 22, 34). The main clinical sign of CA in acutely infected flocks is the alteration of milk consistency in lactating ewes, with a decline and subsequent failure of milk production as a result of interstitial mastitis. Also, arthritis and keratoconjunctivitis can affect about 5 to 10% of infected individuals (4). In areas where CA is endemic, however, subacute and chronic infections are by far the most frequent occurrences (10, 11), making control and eradication of this pathogen extremely difficult. In fact, acute mastitis subsequent to mammary gland infection often progresses to subacute or chronic disease, during the course of which the pathogen is shed in milk for extended periods; after clinical resolution of the disease, mycoplasmas continue to be shed and animals become asymptomatic carriers (4). Eradication plans have been in place for decades in several countries, but the disease persists in many areas, where it is still responsible for important economic losses.
Although CA is widely described clinically, the molecular pathogenesis of CA is not well understood, and the host-pathogen interplay during natural mycoplasma infection has yet to be elucidated. In general, M. agalactiae affects the mammary glands, joints, eyes, ears, and respiratory tract, causing inflammation with different degrees of severity. The microorganism is believed to be unable to penetrate cells, although it is known to adhere tightly to the colonized epithelium (34, 37). Despite recent in-depth genomic and proteomic analyses (7, 27, 40), which have led to fundamental insights into its genomic and proteomic composition, very few proteins of M. agalactiae have been proved to be virulence determinants. These include a family of variable surface proteins, named Vpma proteins (16), the immunodominant adhesin P40 (14), and P48, a homologue of the macrophage-activating lipopeptide (MALP) of M. fermentans (36).
The lack of knowledge concerning virulence mechanisms might also be due to the scarcity of data on alterations occurring in the host cell counterpart. To date, there are no proteomic or genomic studies aimed at investigating the response of sheep tissues to M. agalactiae, and expression levels of low-abundance milk proteins have seldom been assessed during mastitis. The milk proteome has been investigated in naturally and experimentally infected cows with signs of mastitis (5, 6, 19, 41); nevertheless, the response of the mammary gland to natural mycoplasma infections has never been subjected to a dedicated study. To the best of our knowledge, just one study evaluated the local immune response of the goat mammary gland in an experimental M. agalactiae infection (9).
Milk fat globules (MFGs) are released from the lactating cell as a result of an apocrine secretion mechanism leading to the formation of fat droplets surrounded by the endoplasmic reticulum (ER) membrane and by the apical cell membrane on the external surface (17). Cytoplasmic crescents are often trapped between these membrane layers (17), making MFGs a natural mechanism for sampling the lactating cell in vivo. In this way, the molecular content of MFGs might be exploited to study the biology of the lactating cell and to evaluate the alterations occurring in vivo under pathological conditions. Indeed, several proteins involved in host defense have been identified in bovine milk during natural and experimentally induced mastitis, and some of these have been associated with MFGs (19, 41, 42). Moreover, MFGs display intriguing similarities with exosomes, small secretory vesicles released by several tissues and involved in manifold functions, including immunomodulatory activity (33, 39, 49). As a further advantage, the investigation of a purified subproteome directly derived from the lactating cell reduces sample complexity and uncovers the deep proteome represented by low-abundance proteins, overcoming the problems generated by the massive amounts of caseins and whey proteins present in milk.
Recently, we accomplished an in-depth proteomic characterization of sheep MFG proteins (MFGPs) under physiological conditions, revealing a complex but highly reproducible protein profile for midlactation ewes (31). Here we comparatively investigated the proteomic profiles of MFGPs in sheep milk samples collected from CA-affected and CA-free flocks. Protein expression profiles were evaluated by means of two-dimensional difference gel electrophoresis (2-D DIGE) and SDS-PAGE followed by shotgun liquid chromatography-tandem mass spectrometry (GeLC-MS/MS), and protein abundance data were compared to the healthy sheep MFGP profile. Differentially expressed proteins were identified and characterized by gene ontology (GO) and pathway analyses.
MATERIALS AND METHODS
Animals and samples.For this study, frozen 50-ml milk samples belonging to a total of 7 CA-affected flocks were retrieved retrospectively from the Istituto Zooprofilattico Sperimentale della Sardegna (IZS; a public animal health institution), with samples classified as positive for M. agalactiae culture or as microbiologically negative (n = 12 and n = 15, respectively; total number of samples = 27), and were used for proteomic investigations. Samples from an M. agalactiae-free flock were also collected as negative controls (n = 3; total number of samples = 30). All milk samples were retrieved among those collected by IZS as part of the control and eradication program established by the Sardinian Regional Government. According to this program, sheep belonging to flocks where CA outbreaks have occurred or are suspected are subjected to clinical examination by the IZS veterinary personnel, and 50 ml of mixed milk from both half-udders is collected as described previously (24). Samples are refrigerated and subjected to microbiological examination at the IZS laboratory within 24 h of collection (24). For microbiological cultures, 10 μl of milk is seeded in 5% sheep blood agar and incubated at 37°C for 24 to 48 h. For mycoplasma culture, 10 μl of milk is seeded in Hayflick agar and incubated in a wet chamber for up to 7 days. Upon growth of fried-egg colonies, isolates are cloned and identified by PCR (45). For this study, only milk samples which were positive for M. agalactiae culture and microbiologically negative for other microbial pathogens were selected and retrieved. Once transported to the proteomics laboratory, all milk samples (n = 30) were thawed and tested by PCR for the presence of M. agalactiae DNA (7, 45).
Extraction of sheep MFGPs.MFGPs were extracted as described previously (31) from the same sheep milk samples examined by bacterial culture and subjected to whole-milk PCR for detection of M. agalactiae. Briefly, milk samples were centrifuged to obtain the cream fraction containing MFGs, and the cream was washed twice in phosphate-buffered saline and once in triple-distilled water. The fat globules were then crystallized at 4°C overnight, homogenized in triple-distilled water with a TissueLyser mechanical homogenizer (Qiagen, Hilden, Germany), and then warmed to melt the fat. After centrifugation, the pellet was resuspended in water and the protein concentration was determined with a 2-D Quant kit (GE Healthcare, Uppsala, Sweden). Samples were stored at −20°C prior to analysis.
SDS-PAGE and Western immunoblotting.MFGP samples (n = 21) were resuspended in Laemmli buffer (21), boiled, loaded into precast TGX acrylamide gels (Bio-Rad Laboratories, Hercules, CA), subjected to electrophoresis (1), and stained with Coomassie blue (8). Western immunoblotting was performed as described previously (1), using the following antisera: monoclonal anti-actin (clone AC-40) antibody, monoclonal anti-α-tubulin (clone B-5-1-2) antibody, anti-S100A9 rabbit antibody, anti-cathelicidin rabbit antibody, monoclonal anti-S100A11 (clone 2F4) antibody (Sigma-Aldrich, St. Louis, MO), and a rabbit hyperimmune serum raised against recombinant M. agalactiae P48 (M. agalactiae rP48) (36).
2-D DIGE.Sixty-microgram protein samples extracted from triplicate samples of mycoplasma-negative milk samples (uninfected) and mycoplasma-positive milk samples (infected) were labeled with 400 pmol N-hydroxysuccinimidyl ester of cyanine dyes Cy3 and Cy2, respectively (GE Healthcare), as indicated by the manufacturer. The Cy5 cyanine dye was used to label a pooled sample comprising equal amounts of each of the specimens in the study (uninfected and infected), which served as the internal pooled standard. The labeled protein samples and the internal pooled standard were mixed in suitable combinations, brought to the final rehydration volume with IPG buffer (GE Healthcare) and Destreak rehydration solution (GE Healthcare), and applied to 24-cm IPG strips (pH 3 to 11, nonlinear; GE Healthcare) by passive rehydration overnight at room temperature (see Table S1 in the supplemental material). Rehydrated strips were then run together in an IPGphor device equipped with an Ettan IPGphor 3 loading manifold (GE Healthcare) at 20°C for a total of about 90,000 V-h. After isoelectric focusing (IEF), the strips were equilibrated (6), subjected to second-dimension SDS-PAGE, and digitalized as described previously (44). The images were analyzed with a DeCyder batch processor and differential in-gel analysis (DIA) modules (GE Healthcare). Statistical analysis of protein level changes was performed with the DeCyder BVA (biological variation analysis; v.6.5) module. The results related to uninfected and infected samples were compared and statistically evaluated by one-way analysis of variance (ANOVA) with the DeCyder BVA module, applying the false discovery rate (FDR) to minimize the number of false-positive results. Protein spots with statistically significant variation (P ≤ 0.05), showing a difference in volume of >2-fold, were selected as differentially expressed. Cluster analysis and visualizations were performed using the DeCyder EDA (extended data analysis) module.
MALDI-TOF MS.For protein identification, preparative 2-D PAGE gels were set up by overnight rehydration loading of 300 μg of protein extract into pH 3 to 11 NL 24-cm IPG strips. Strips were then focused and subjected to second-dimension electrophoresis as described above. All blue molecular weight markers (Bio-Rad Laboratories) were also loaded on an electrode wick and run together with the isoelectrofocused strips. After electrophoresis, the gel slab was subjected to colloidal Coomassie staining (8). Visible protein spots of interest were excised manually from the gels, and mass spectra were recorded on a matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) instrument (MALDI micro MX; Waters, Manchester, United Kingdom) as described previously (2, 7, 44). Raw data, reported as monoisotopic masses, were then introduced into our in-house Mascot peptide mass fingerprinting software (version 2.2; Matrix Science, Boston, MA) and used for protein identification.
GeLC-MS/MS and pathway analysis.Proteins (30 μg) were resolved by SDS-PAGE. After staining by colloidal Coomassie blue (8), the gel lane was cut into 30 slices and subjected to in-gel tryptic digestion as described previously (1). Briefly, the gel slices were destained, washed, reduced, alkylated, and digested with trypsin; peptides were extracted, dried, and resuspended as described previously (1, 7, 31, 43). LC-MS/MS analyses were performed on a quadrapole TOF (Q-TOF) hybrid mass spectrometer equipped with a nano-lock Z-spray source and coupled online with a CapLC capillary chromatography system (Waters) as described previously (31, 43). In this case, protein identification was carried out in the NCBI database, using the Mammalia (mammals) and Mycoplasma taxonomies with the following search parameters: peptide tolerance, 30 ppm; MS/MS tolerance, 0.4 Da; charge state, +2 and +3; enzyme, trypsin; and up to two missed cleavages. For false-positive analysis, a decoy search was performed automatically by choosing the decoy checkbox on the search form. Mascot results were parsed using the IRMa toolbox (version 1.26.1) (13) with the following inclusion criteria in order to keep an FDR value of ≤2% for each analysis: protein significance threshold, P value of <0.01; peptide rank, 1; and peptide ion score cutoff, 20. Unique peptides (UP), spectral counts (SpC), and exponentially modified protein abundance index (emPAI) values were reported as calculated by IRMa (13). Skin keratins were excluded from the final protein list. Proteins with similar peptides which could not be differentiated based on MS/MS analysis alone were grouped (38). SpC values were used as a semiquantitative parameter for estimating protein abundance and comparing the expression of the same protein among different samples, as described previously (31, 43). The SpC log ratio (RSC) and normalized spectral abundance factor (NSAF) were calculated according to the methods of Old et al. (28) and Zybailov et al. (54), respectively, considering for each identified protein the average SpC value among biological replicates. NSAF values were used to compare expression of diverse proteins among samples. NSAF was calculated as described previously (31, 53).
GO assignments were carried out using online DAVID bioinformatic resources (version 6.7) (12, 18). NSAF comparisons among classes were plotted using Microsoft Excel, and the statistical significance of differences in protein expression levels among the groups compared during the study was assessed by a two-tailed t test with a 95% confidence level. The beta-binomial test was performed to identify differentially expressed proteins according to the method of Pham et al. (30), using the software kindly provided by the authors. With the aim of increasing the stringency of this analysis, proteins identified with less than one SpC in at least one replicate or with fewer than two SpCs in more than one replicate of the subgroup in which they were more expressed were excluded from the list. The list of protein identifications (IDs) with P values of <0.05, together with their respective RSC values, was imported into the online software package IPA (version 8.7; Ingenuity Systems, Redwood City, CA), and network analyses were performed with thresholds of 1.5 for RSC and 0.05 for P value. Sheep UniProt IDs were replaced with the UniProt IDs for the closest human protein equivalents in order to enable the best exploitation of the knowledge-based IPA software (version 8.8; updated 13 November 2010). To determine the biological processes, functions, pathways, and molecular networks most significantly altered during infection by M. agalactiae, both over- and underrepresented proteins were defined as value parameters, all identifier types and data sources were selected in order to access all available information in the IPA database, and both direct and indirect relationships were considered.
RESULTS
Alterations in total protein profile of MFGPs in sheep infected by M. agalactiae.In a previous study, the proteomic profile of sheep MFGPs was characterized under physiological conditions, revealing a high interindividual reproducibility of the MFGP expression profile for healthy sheep of the dairy breed Sarda (31). In order to evaluate possible changes in MFGPs during mastitis, 27 sheep milk samples from 7 different flocks affected by CA were retrieved from IZS, a public animal health institution, and 3 sheep milk samples were collected as controls from a certified CA-free herd. At the beginning of the study, all samples were subjected to whole-milk PCR to detect the presence of M. agalactiae DNA and to extraction of MFGPs, which were obtained successfully from 21 samples (Table 1). MFGP expression profiles were initially evaluated in representative milk samples from sheep with M. agalactiae-positive cultures, from sheep with M. agalactiae-negative cultures but belonging to the same CA-affected flock, and from sheep belonging to a CA-free flock (Fig. 1, left panel). Dramatic alterations were evident in the SDS-PAGE profiles of MFGPs from sheep with M. agalactiae-positive cultures (lanes 1 and 2) compared to those for MFGPs from bacteriologically negative sheep from the same flock (lanes 3 and 4). However, slight differences could also be observed in the latter samples compared to MFGPs of sheep from the CA-free flock (lanes 5 and 6). In order to investigate the presence of mycoplasma proteins in culturally negative animals, the same samples were subjected to Western immunoblotting (WB) with antibodies directed against P48, an immunodominant protein of M. agalactiae (36). In fact, since MFG membranes (MFGMs) are derived directly from the lactocyte apical membrane, bacterial proteins could remain associated with this cellular fraction and be shed in milk during chronic infections. Should this be true, bacteria or proteins adhered to MFGMs would be enriched and their detection by means of immunological techniques would be enhanced. The results are shown in Fig. 1 (right panel). As expected, samples 1 and 2 (culturally positive samples) showed clearly positive signals, while samples 5 and 6 (samples from the M. agalactiae-free flock) were negative. However, samples 3 and 4, obtained from sheep which were culturally negative but belonged to the flock infected by M. agalactiae, showed a clearly visible band (sample 4) or a faint but visible signal (sample 3), highlighting the presence of bacterial proteins in these culturally negative samples. Moreover, this result indicates the ability of MFGs to enrich diagnostically relevant proteins of the infecting bacterial pathogen, since milk samples from the same animals produced a negative result when tested with the same antibody (data not shown).
Summary of sheep milk samples used in this worka
SDS-PAGE of sheep MFGPs and Western blotting results obtained for the immunodominant M. agalactiae antigen P48. (Left) Total protein profiles of MFGPs extracted from sheep culturally positive for M. agalactiae (lanes 1 and 2), from culturally negative sheep from the same flock (lanes 3 and 4), and from sheep belonging to a CA-negative flock (lanes 5 and 6). Total M. agalactiae proteins were loaded in lane 7 as a control. (Right) Western immunoblotting with antibodies directed against the immunodominant M. agalactiae lipoprotein rP48. M, molecular weight markers.
2-D DIGE characterization of sheep MFGPs during M. agalactiae infection.In view of the preliminary observations by SDS-PAGE, a 2-D DIGE study was performed with the aim of characterizing, both qualitatively and quantitatively, the differences in MFGPs of infected and uninfected animals. Therefore, three samples from the CA-free flock and three samples from the CA-affected flock, one of which was negative for bacterial culture but showed WB positivity for P48 (C−/WB+), were subjected to a 2-D DIGE study. Figure 2 summarizes the obtained results. A representative 2-D DIGE image comparing signals generated by a negative sample (C−/WB−) (green signal) and by a culturally M. agalactiae-positive sample (C+/WB+) (blue signal) is shown, together with the six individual 2-D maps used for image analysis and statistical analysis.
2-D DIGE of MFGPs from M. agalactiae-infected and uninfected sheep. (Top) Overlay image of MFGPs extracted from representative infected (blue) and uninfected (green) milk samples (samples A and D, respectively). Spots indicate proteins with statistically significant differences in amount among all samples examined. Identities of differentially represented proteins are reported in Table 1. (Bottom) Single-channel images of MFGPs extracted from culturally positive sheep milk samples (A and B), from a milk sample with culture negativity and Western immunoblotting positivity for M. agalactiae (C), and from sheep milk samples from a CA-free flock (D to F).
Analysis of the 2-D DIGE images with DeCyder software (GE Healthcare) revealed statistically significant differences in abundance for numerous spots, 28 of which were successfully identified by MS (Fig. 2; Table 2). The proteins found to be increased in positive samples were involved mostly in membrane and vesicular trafficking (30%), such as annexins, actin, and myosin; in immune function, inflammation, and host defense (30%), such as S100 proteins, cathelicidins, and antimicrobial peptides; in protein synthesis and folding (25%), such as heat shock proteins (HSPs); and in enzymatic activity (15%), such as the mitochondrial superoxide dismutase and protein disulfide isomerase. In contrast, proteins found to be decreased were involved almost exclusively in fat transport/metabolism and in MFG secretion, including butyrophilin, lactadherin, adipophilin, and xanthine dehydrogenase/oxidase, with only a minimal fraction being represented by milk proteins.
Protein spots showing statistically significant differences in protein abundance among MFGP samples examined by 2-D DIGE
Multivariate analysis based on principle component analysis (PCA) was performed on expression data to assess global changes in the MFGP profiles for infected sheep (C+/WB+ and C−/WB+) compared to uninfected sheep (C−/WB−). PCA allows for grouping of samples with overall similar expression characteristics and for identification of proteins which are responsible for the differences between groups. PCA with the differentially abundant proteins revealed a clear separation of positive and negative samples into two groups (Fig. 3, top panel, red and purple groups) in the score plot. Interestingly, the sample with culture negativity and WB positivity (C−/WB+) clustered separately from the other sample groups (Fig. 3, top panel, green group). A heat map was also generated in order to compare protein expression patterns within the three classes: uninfected (C−/WB−), infected (C+/WB+), and infected but culture negative (C−/WB+). The results obtained, illustrated in Fig. 3, bottom panel, further highlight the differences in expression existing between infected and noninfected individuals and the intermediate status of the C−/WB+ sample.
Statistical analysis of 2-D DIGE results. A score plot (upper diagram) and heat map (lower diagram) obtained upon comparison of MFGP samples from M. agalactiae-infected sheep (A, B, and C) with samples from CA-free animals (D, E, and F) are shown. In the heat map, each colored cell represents the protein abundance value for a single sample. Increasingly positive values are indicated by reds of increasing intensity, and increasingly negative values are indicated by greens of increasing intensity. Cells with a value of 0 are colored black.
GeLC-MS/MS analysis of sheep MFGPs during M. agalactiae infection.Proteome coverage was subsequently increased by means of GeLC-MS/MS analysis. This approach overcomes some of the limitations suffered by 2-D PAGE in analysis of membrane proteins and liposoluble proteins. Moreover, by means of recent advancements in label-free quantification, the ability to perform a relative evaluation of protein expression among samples is maintained (30, 52). To this extent, MFGPs extracted from the two M. agalactiae-positive samples examined by 2-D DIGE were also subjected to GeLC-MS/MS analysis. In total, 185 unique proteins were identified, subjected to label-free relative quantification of protein expression by means of the spectral counting approach, and normalized by means of the NSAF (28, 54) as described previously (31, 43) (see Table S2 in the supplemental material). A protein ID and abundance database was generated for healthy sheep MFGs (31). Building on this existing database, a beta-binomial test was applied to spectral counting data with the purpose of identifying proteins with statistically significant differences in abundance between animals infected by M. agalactiae and animals from a CA-free flock. Table 3 reports identities, RSC values, SpC values, and P values for the 68 statistically significant differentially expressed proteins (statistically significant proteins are plotted according to RSC in Fig. S3 in the supplemental material, and the complete list of statistical results is reported in Table S4 in the supplemental material). A classification based on cellular localization and biological function was then performed by DAVID on all differential MFGPs, using the normalized protein abundance values to allow for comparison among localization and functional classes. As shown in Fig. 4, marked changes were observed both in terms of cellular localization (Fig. 4A) and in terms of biological function (Fig. 4B). In particular, membrane proteins decreased, while secreted, mitochondrial, and lysosomal proteins increased significantly in positive samples. In terms of biological function, membrane and vesicular trafficking proteins decreased, while enzymes and proteins with immune functions increased in positive samples.
Proteins showing statistically significant differences in abundance among MFGP samples examined by GeLC-MS/MS
Normalized spectral abundance of MFGPs in M. agalactiae-infected (white) and uninfected (black) sheep. Proteins were categorized by DAVID according to cellular localization (A) and function (B). Asterisks indicate statistically significant differences between the two groups according to a two-tailed t test with a 95% confidence level.
The GeLC-MS/MS approach was then applied to the C−/WB+ sample (i.e., the third sample analyzed by 2-D DIGE) in order to evaluate the extent of its protein expression variation and to relate it to truly positive and truly negative samples. Interestingly, the MFGP expression profile for this sheep showed alterations that followed a pattern similar to that for MFGPs of culturally positive sheep, although with a lower intensity (see Fig. S5 in the supplemental material). Notably, however, for the C−/WB+ sheep, several proteins involved in host defense, immune function, and inflammation, especially S100A9, cathelicidin, and lactoferrin, reached high RSC values, as shown in Fig. 5. This might indicate that in such subjects, the expression levels of proteins involved in lactation and in the mechanisms governing milk fat globule secretion are altered only slightly, but the inflammation and host defense mechanisms are activated and clearly detectable at the proteomic level.
Comparison of RSC values of selected MFGPs. Bars indicate the protein levels in C+/WB+ sheep (black) and C−/WB+ sheep (gray) compared to MFGP levels in uninfected sheep.
In order to investigate the presence of Mycoplasma proteins, the Mycoplasma taxonomy was also searched. As a result, using the same stringency parameters used for the mammalian taxonomy, 15 protein identities were statistically supported by the software. However, a univocal attribution could not be made for many of these identifications, due to sequence homology between the sheep and mycoplasma peptides detected. Only two proteins showed sequence differences enabling univocal attribution to mycoplasmas: lactate dehydrogenase, a prominent mycoplasma immunogen, and AvgC, a variable surface protein involved in immunological escape. The missed identification by GeLC-MS/MS of relevant M. agalactiae proteins is not surprising and may be due to several factors dependent on the sensitivity of the technique and on the proteomic approach, such as identity of the peptide sequences found by MS/MS for bacterial and host homologues, which impairs species attribution, as well as colocalization of bacterial proteins of the same molecular weight among highly abundant proteins during electrophoresis, as might be the case for P48 and butyrophilin, which hampers identification of the less abundant protein species upon analysis by GeLC-MS/MS due to instrument sensitivity constraints.
Identification and pathway analysis of MFGPs showing changes in abundance upon infection.All proteins showing statistically significant differences in the protein profiles of positive and negative MFG samples and their respective fold change values were subjected to pathway analysis using IPA software, with the aim of elucidating the main molecular interactions and biological connections and representing them by networks. Since the IPA database builds on the literature generated for humans and rodents, the UniProt codes for sheep proteins were replaced with the UniProt codes for the closest human protein equivalents for the purpose of this analysis.
When all statistically significant differentially abundant proteins were subjected to pathway analysis, the network scoring the highest significance value was cellular movement, hematological system development and function, and immune cell trafficking (score of 64). The diseases and disorders showing the strongest statistically significant associations with the differentially expressed proteins were respiratory disease (P values of 1.97E−10 to 3.84E−03), infectious disease (P values of 7.55E−10 to 5.20E−03), and the inflammatory response (P values of 1.01E−08 to 7.71E−03), consistent with the pathogen under examination. The most statistically significant molecular and cellular functions being altered were cellular movement (P values of 9.74E−11 to 7.71E−03), cell death (P values of 3.04E−10 to 7.71E−03), and free radical scavenging (P values of 1.50E−07 to 7.18E−03) (the complete analysis summary is reported in Table S6 in the supplemental material). In order to highlight the different events taking place in infected MFGs, a pathway analysis was then performed by separately examining proteins significantly increased or decreased in abundance above the 1.5-fold change threshold. Increased proteins were associated mostly with the inflammatory response (P values of 1.56E−10 to 2.06E−02), infectious disease (P values of 2.18E−08 to 2.06E−02), and respiratory disease (P values of 2.18E−08 to 9.27E−03), while decreased proteins were implicated mainly in lipid metabolism (P values of 6.18E−07 to 2.38E−02), molecular transport (P values of 6.18E−07 to 2.38E−02), and small-molecule biochemistry (P values of 6.18E−07 to 2.38E−02). Figure 6 depicts the networks with the highest scores for all differentially abundant proteins (Fig. 6A) and for increased (Fig. 6B) and decreased (Fig. 6C) proteins. (The complete analysis summaries for increased and decreased proteins are reported in Tables S7 and S8 in the supplemental material).
Results of Ingenuity pathway analysis. The highest-scoring networks for all differentially represented proteins (A) and for proteins overrepresented (B) and underrepresented (C) in infected sheep are illustrated. (A and B) Results for the highest-scoring network, i.e., cellular movement, hematological system development and function, and immune cell trafficking, are illustrated. (C) Results for the highest-scoring network, i.e., lipid metabolism, molecular transport, and small-molecule biochemistry, are illustrated. Red, overrepresented proteins; green, underrepresented proteins; white, proteins indicated by IPA as significantly associated with the reported network; continuous line, direct relationship; dotted line, indirect relationship. Color intensity represents the extent of differential protein abundance.
Western immunoblotting validation of findings for differentially abundant MFGPs.Since cytoskeletal and host defense proteins were among the most intensely increased MFGPs, as indicated also by the GO and IPA analyses, these were evaluated by WB of a larger number of samples (n = 21) with the purpose of validating proteomic results as well as investigating their suitability as indicators of M. agalactiae colonization and host response. Commercial antibodies against actin, tubulin, S100A9, cathelicidin, and S100A11 were evaluated for reactivity with MFG protein extracts, and the results were compared to those for PCR, culture, and positivity for the M. agalactiae lipoprotein P48. The results are summarized in Table 1 and illustrated in Fig. 7.
Composite image summarizing results obtained for all milk samples. Results were generated by culture, PCR, and Western immunoblotting for selected cytoskeletal and host defense MFGPs for all samples included in this work. Samples subjected to 2-D DIGE and GeLC-MS/MS are enclosed in the left section of the figure. Sample ID numbers are as indicated in Table 1.
MFGP samples positive for M. agalactiae, by either culture or PCR, displayed strong signals for the host defense proteins tested by WB, especially S100A9, whose signal was clearly detectable in all 16 PCR-positive samples tested with this antibody. Antibodies against cathelicidin and S100A11 produced clear positive bands for only a subset of M. agalactiae-positive samples. On the other hand, negative samples did not display any detectable signal for any of the host defense proteins revealed to be expressed upon M. agalactiae infection, while the same showed a slight positivity for cortical cytoskeleton proteins. Therefore, validation with a larger number of samples from different CA-affected flocks confirmed the observations obtained with the proteomic and data analysis approaches and further highlighted the significant and specific increase of the host defense protein S100A9 in M. agalactiae-positive samples. However, it should be considered that the apparently smaller amounts of cathelicidin and S100A11 than of S100A9 seen by Western immunoblotting, although consistent with the observations made by 2-D DIGE and GeLC-MS/MS, might be due to differences in specificity of the commercially available antibodies used in this work, which are not specific for sheep proteins. In this respect, this should be considered only a validation of their increased abundance in affected animals from different flocks, while quantitative conclusions should be drawn only by means of quantitative proteomic approaches or alternative quantitative assays.
DISCUSSION
This report presents a comprehensive evaluation of proteomic changes occurring in MFGs of sheep infected by M. agalactiae, providing new information on the in vivo events taking place in the lactating mammary epithelium during a bacterial infection. Specifically, a combined approach based on 2-D DIGE, GeLC-MS/MS, and pathway analysis was applied to study the protein profile of MFGs produced by lactating epithelial cells during a natural M. agalactiae infection. Taken together, the different approaches converged on consistent results. Increased levels of proteins involved in inflammation and immune defense (such as antimicrobial proteins and peptides), in folding (such as HSPs), and in the cortical cytoskeleton were detected in infected sheep MFGs. Increased amounts of proteins involved in oxidative stress, such as the mitochondrial superoxide dismutase, were also observed, consistent with the findings of other researchers upon transcriptome studies of the bovine mammary gland infected with Escherichia coli (26). On the other hand, there was an evident and significant decrease in abundance of proteins devoted to the physiological functions of MFGs, i.e., lipid synthesis, transport, and secretion. In MFGs from infected animals, S100A9 (calgranulin B) was the protein showing the most upregulation, with a striking average positive/negative ratio of 61.46 upon 2-D DIGE analysis, followed by S100A11 (calgizzarin) and cathelicidin-1, proteins involved in inflammation/host defense, as also confirmed by GeLC-MS/MS and immunoblotting. With all the techniques applied in this study, S100 proteins and cathelicidins were never detected in MFGs from healthy, M. agalactiae-free sheep, demonstrating their specific and significant increase during infection. Nevertheless, an abundance of these proteins, together with other host defense proteins, was clearly detected in MFG extracts obtained from sheep infected by M. agalactiae but negative by milk culture, i.e., with lower bacterial loads; in fact, S100A9, S100A11, cathelicidin, and the myeloid antimicrobial peptide were overrepresented, while proteins involved in milk fat secretion/metabolism were slightly underrepresented.
S100 proteins and cathelicidins are known to possess proinflammatory functions and direct antimicrobial effects (5, 23), and their upregulation is increasingly reported for microbial infections. Recent proteomic and transcriptomic studies reported a significant increase of S100 proteins and cathelicidins in milk during natural and experimental bovine mastitis (19, 26), although their cellular source was not investigated. However, an increase in mRNAs for host defense, cytoskeletal, and heat shock proteins was recently demonstrated for the bovine mammary epithelial cell line MAC-T infected with different bacterial pathogens, suggesting an in vivo role in their production for mammary epithelial cells. High levels of S100 proteins and cathelicidins have been reported for several epithelial tissues, mostly associated with both acute and chronic inflammatory and infectious conditions, such as genitourinary infections, arthritis, psoriasis, and degenerative disorders (15, 29, 35, 50–52), further supporting the hypothesis that these high levels might also occur in the mammary epithelium. Yet the contribution of phagocytic cells cannot be ruled out completely, since phagocytic cells with engulfed MFGs, possibly opsonizing bacteria or bacterial proteins, might hypothetically contaminate the fat ring preparation due to their lower density, despite repeated and extensive washing and centrifugation.
Interesting insights into the role played by the proinflammatory S100 proteins in the response of epithelial cells to infection were recently provided by a study on Candida infection of mouse vaginal epithelia (51). The authors of that study detected an increase of both S100 mRNAs and proteins in infected epithelial cells, leading to recruitment of polymorphonuclear neutrophils (PMNs) in the vagina, with the extent depending on the amount of proinflammatory mediators produced by epithelial cells. Similarly, since the mammary gland is exposed to a variety of microbial pathogens in the environment, a first line of defense is built up by lactocytes. The results obtained here indicate that exploitation of MFGs as surrogates of the lactocyte cytoplasm has the potential to provide clues to the extent of the mammary gland response to infection in vivo and to elucidate the role played by lactocytes in the establishment of acute or chronic infections by M. agalactiae.
Indications of the pathways involved in mastitis were provided by the IPA analysis performed on all MFGPs undergoing significant changes in abundance upon infection by M. agalactiae. In particular, in the category of diseases and functions, IPA highlighted an increase in proteins associated with respiratory disease, infectious disease, and the inflammatory response. In evaluating the pathway analysis data, it must be taken into account that IPA builds on the literature findings for humans and rodents; therefore, the results generated upon IPA analysis of our protein expression database are biased by findings on human infections. Notably, clinically significant human mycoplasma infections occur mainly in the airways (e.g., M. pneumoniae) and the genitourinary tract (e.g., M. hominis and M. genitalium). This could explain in part why IPA analysis of our data generated a significant score for respiratory disease in the category of diseases and disorders and might underline similarities in the host responses elicited by mycoplasmas upon epithelial tissue infection in different hosts, as well as pointing out common traits in the innate immune response mechanisms elicited in different secretory epithelia. It is intriguing that the respiratory epithelium (20, 32) and many other secretory epithelia are increasingly being reported to produce exosomes involved in innate defense and immunomodulation (33, 49), such as biliary duct cells (47), intestinal epithelial cells (48), and most interestingly, the human mammary gland (3). Since sheep MFGs present many of the physicochemical attributes and proteomic signatures of exosomes (25, 31), functions associated with these secretory vesicles might likely be observed in MFGs during infection of the mammary gland.
Finally, important insights into disease diagnosis and control can be drawn from this study. Currently, the cultural and immunological tools available for detection of M. agalactiae provide a rapid diagnosis of disease but may not be very sensitive for use with chronically affected herds and flocks. The limitations of microbial culture, which currently remains the most widespread method for assessing the presence of M. agalactiae in milk, clearly emerged in this work. In fact, despite positivity by milk PCR and evident macroscopic alterations which impaired separation of the fat ring, several samples were negative for M. agalactiae culture (Table 1). This is a known problem, likely due to unreported administration of antimicrobial agents, to acidification of milk causing a reduced viability of mycoplasmas, or to the massive presence of other bacteria that may hinder the growth of mycoplasmas in selective agar plates (46). Another reason for culture negativity can be a very low bacterial load, as might have occurred in this work for several milk samples which were negative for mycoplasma culture, had a normal appearance, and formed an apparently normal fat ring but were positive by milk PCR and by Western immunoblotting for P48 and S100A9. This occurrence might be of particular relevance, as it is the typical condition encountered during more insidious chronic, subclinical infections or at the first stages of infection. Here we observed that some host proteins are strongly overrepresented upon infection of the mammary gland and behave as diagnostic antigens with a level of sensitivity comparable to that of milk PCR. Future studies might enable the exploitation of these proteins as sensitive and easily detectable markers of mammary infection and inflammation. Moreover, should their increase also be demonstrated upon infection by other microbial pathogens, as suggested by studies of cow mastitis, these proteins may also possess future potential as pathogen-independent indicators of sheep mammary infection, providing a simple, sensitive, and comprehensive tool for rapidly detecting chronic infections in asymptomatic carrier animals and for preventing their spread to the whole flock or herd.
In summary, this is the first report detailing the proteomic changes occurring in MFGs of sheep naturally infected by a bacterial pathogen. The results reported here open the way to elucidation of the molecular events taking place in the infected mammary epithelium in vivo, offer valuable insights for understanding the changes induced by M. agalactiae in its natural host, and possess significant potential for the development of tools enabling diagnosis and control of chronic, subclinical infections of the mammary gland in sheep.
ACKNOWLEDGMENTS
We thank Roberto Tonelli for his help with the statistical processing of data.
This work was supported by funding from the Regione Autonoma della Sardegna—Progetto Cluster Proteomica.
FOOTNOTES
- Received 11 January 2011.
- Returned for modification 26 February 2011.
- Accepted 1 June 2011.
- Accepted manuscript posted online 20 June 2011.
↵‡ Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00040-11.
- Copyright © 2011, American Society for Microbiology. All Rights Reserved.