Differential protein expression in chicken macrophages and heterophils in vivo following infection with Salmonella Enteritidis

In this study we compared the proteomes of macrophages and heterophils isolated from the spleen 4 days after intravenous infection of chickens with Salmonella Enteritidis. Heterophils were characterized by expression of MMP9, MRP126, LECT2, CATHL1, CATHL2, CATHL3, LYG2, LYZ and RSFR. Macrophages specifically expressed receptor proteins, e.g. MRC1L, LRP1, LGALS1, LRPAP1 and a DMBT1L. Following infection, heterophils decreased ALB and FN1, and released MMP9 to enable their translocation to the site of infection. In addition, the endoplasmic reticulum proteins increased in heterophils which resulted in the release of granular proteins. Since transcription of genes encoding granular proteins did not decrease, these genes remained continuously transcribed and translated even after initial degranulation. Macrophages increased amounts of fatty acid elongation pathway proteins, lysosomal and phagosomal proteins. Macrophages were less responsive to acute infection than heterophils and an increase in proteins like CATHL1, CATHL2, RSFR, LECT2 and GAL1 in the absence of any change in their expression at RNA level could even be explained by capturing these proteins from the external environment into which these could have been released by heterophils. Electronic supplementary material The online version of this article (doi:10.1186/s13567-017-0439-0) contains supplementary material, which is available to authorized users.


Introduction
Macrophages and heterophils represent professional phagocytes acting as effectors and modulators of innate immunity as well as orchestrators of adaptive immunity [1]. Heterophils, the avian counterparts of mammalian neutrophils, belong among the first responders to bacterial infections and sensing of pathogen associated molecular patterns (PAMPs) stimulates heterophils for phagocytosis as well as release of bactericidal proteins stored in heterophil granules into the extracellular environment [2]. In agreement with their general function in host protection against pathogens, heterophils play a crucial role in the protection of chickens against Salmonella infection and chickens with heterophil depletion are not protected against colonization of systemic sites [3][4][5]. However, although there are several reports on specific heterophil functions during infection of chickens with Salmonella enterica, their genome-wide response to infection has not been characterized so far.
Macrophages are professional phagocytes responsible for the destruction and clearance of pathogens. When activated, macrophages increase their antibacterial activity by the expression of antimicrobial peptides like cathepsins B, C, D and S, avidin, ferritin or ovotransferrin [6], and production of NO radicals from arginine by inducible NO synthase. The antimicrobial proteins expressed by macrophages are commonly produced also by heterophils though it is not known to what extent these may differ in their immediate availability and total amount produced by both cell types. Macrophages can also regulate the immune response by the expression of cytokines e.g. IL1β, IL6, IL8, IL18 or LITAF [7] and are capable of antigen presentation [8][9][10]. However, similar to heterophils, an unbiased report on total proteome expressed by chicken macrophages is absent.
In our previous study we showed that heterophils and macrophages increase in the spleen of chickens when intravenously infected with Salmonella Enteritidis (S. Enteritidis) [7]. Next we characterized the gene expression at the tissue level in the whole spleen and expression of selected transcripts was tested in sorted leukocyte subpopulations [6]. However, none of this provided general data on the protein expression in chicken heterophils and macrophages. Although intravenous infection of chickens only partially represents specific Salmonella-chicken interactions which are mixed up with a general response to bacteremia caused by Gram negative bacterium, this way of infection represents a model for the understanding heterophil and macrophage functions during early response to infection. In the current study we therefore isolated heterophils and macrophages from chicken spleens by fluorescence-activated cell sorting (FACS), purified proteins from these cells and identified them by mass spectrometry. This allowed us to (1) characterize the total proteome of heterophils and macrophages, (2) define proteins which exhibited differential abundance in chicken heterophils compared to macrophages and (3) identify proteins that changed in abundance following the intravenous infection with S. Enteritidis in either of these populations. Since we also included a group of chickens which was vaccinated prior to challenge, we also addressed whether there are any proteins specifically expressed by the macrophages or heterophils from the vaccinated chickens. Using this approach we identified over one hundred proteins characteristic of either chicken heterophils or macrophages which allowed us to further refine their function in chickens.

Ethics statement
The handling of animals in this study was performed in accordance with current Czech legislation (Animal protection and welfare Act No. 246/1992 Coll. of the Government of the Czech Republic). The specific experiments were approved by the Ethics Committee of the Veterinary Research Institute (permit number 5/2013) followed by the Committee for Animal Welfare of the Ministry of Agriculture of the Czech Republic (permit number MZe 1480).

Bacterial strains and chicken line
Newly hatched ISA Brown chickens from an egg laying line (Hendrix Genetics, Netherlands) were used in this study. Chickens were reared in perforated plastic boxes with free access to water and feed and each experimental or control group was kept in a separate room. The chickens were vaccinated with S. Enteritidis mutant completely lacking Salmonella pathogenicity island 1 (SPI-1) constructed as described earlier [11] and infected with isogenic wild type S. Enteritidis 147 spontaneously resistant to nalidixic acid. The strains were grown in LB broth at 37 °C for 18 h followed by pelleting bacteria at 10 000 × g for 1 min and re-suspending the pellet in the same volume of PBS as was the original volume of LB broth.

Experimental infection
There were 3 groups of chickens. Six chickens from the control group were sacrificed on day 48 of life. An additional 6 chickens (group 2) were infected intravenously with 10 7 CFU of wild type S. Enteritidis in 0.1 mL PBS on day 44 of life. The last 6 chickens (group 3) were orally vaccinated on day 1, revaccinated on day 21 of life with 10 7 CFU of S. Enteritidis SPI-1 mutant in 0.1 mL of inoculum and challenged intravenously with 10 7 CFU of wild type S. Enteritidis on day 44 of life. Intravenous mode of infection was used mainly to stimulate macrophage and heterophil response rather than to model natural infection of chickens with S. Enteritidis. All chickens in groups 2 and 3 were sacrificed 4 days post infection, i.e. when aged 48 days. The spleens from the chickens from all three groups were collected into PBS during necropsy. To confirm S. Enteritidis infection, approximately 0.5 g of liver tissue was homogenised in 5 mL of peptone water, tenfold serially diluted and plated in XLD agar, as described previously [11].

Collecting heterophil and macrophage subpopulations by flow cytometry
The cell suspensions were prepared by pressing the spleen tissue through a fine nylon mesh followed by 2 washes with 30 mL of cold PBS. After the last washing step, the splenic leukocytes were re-suspended in 1 mL of PBS and used for surface marker staining.
In total 10 8 of cells were incubated for 20 min with antimonocyte/macrophage:FITC (clone KUL01 from Southern Biotech) and CD45:APC (clone LT40 from Southern Biotech), followed by wash with PBS. Monocytes/macrophages (CD45+KUL01+) and heterophils (identified based on FSC/SSC characteristics within CD45+ cells) were sorted using a FACSFusion flow cytometer operated by FACSDiva software (BD Biosciences). Only for simplicity, the monocytes/macrophages population will be called as "macrophage (Ma)" in the rest of this paper. Sorted cells were collected in PBS and immediately processed as described below. A small aliquot from each sample was subjected to immediate purity analysis. The purity of macrophages was 88.6 ± 5.3% and of heterophils 88.1 ± 4.2% when counting cell of expected staining, and FSC and SSC parameters out of all particles. When we gated at the area with live cells, the purity of macrophages and heterophils was between 97 and 98%.
Majority of contaminants therefore represented cellular debris and only around 2.5% of contaminants were formed by non-target cells.
Protein and RNA isolation from sorted cells, reverse transcription of mRNA and quantitative real time PCR (qPCR) Sorted leukocyte subpopulations were lysed in 500 µL of Tri Reagent (MRC) for parallel isolation of RNA and proteins. Upon addition of 4-bromoanisole and 15 min centrifugation at 14 000 × g, proteins were precipitated with acetone from the lower organic phase. RNA present in upper aqueous phase was further purified using RNeasy purification columns according to the instructions of the manufacturer (Qiagen). The concentration of RNA was determined spectrophotometrically (Nanodrop, Thermo Scientific) and 1 µg of RNA was immediately reverse transcribed into cDNA using MuMLV reverse transcriptase (Invitrogen) and oligo dT primers. After reverse transcription, the cDNA was diluted 10 times with sterile water and stored at −20 °C prior qPCR. qPCR was performed in 3 µL volumes in 384-well microplates using QuantiTect SYBR Green PCR Master Mix (Qiagen) and a Nanodrop pipetting station from Innovadyne for PCR mix dispensing following MIQE recommendations [12]. Amplification of PCR products and signal detection were performed using a LightCycler II (Roche) with an initial denaturation at 95 °C for 15 min followed by 40 cycles of 95 °C for 20 s, 60 °C for 30 s and 72 °C for 30 s, followed by the determination of melting temperature of resulting PCR products to exclude false positive amplification. Each sample was subjected to qPCR in duplicate and the mean values of the Cq values of genes of interest were normalized (ΔCt) to an average Cq value of three reference genes (GAPDH, TBP and UB). The relative expression of each gene of interest was finally calculated as 2 −ΔCq . Statistical analysis using a two sample t test for means equality was performed when comparing levels of mRNA expression between chicken groups and results with p value ≤ 0.05 were considered as significantly different in expression. Sequence of reference genes GAPDH, TBP and UB have been published elsewhere [13,14]. Sequences of all newly designed primers used in this study including their location within different exons and sizes of PCR products are listed in Additional file 1.

Sample preparation for LC-MS/MS analysis
Precipitated proteins were washed with acetone and dried. The pellets were dissolved in 300 µL of 8 M urea and processed by the filter aided sample preparation method [15] using Vivacon 10 kDa MWCO filter (Sartorius Stedim Biotech). Proteins were washed twice with 100 µL of 8 M urea and reduced by 100 µL of 10 mM DTT. After reduction, proteins were incubated with 100 µL of 50 mM IAA and washed twice with 100 µL of 25 mM TEAB. Trypsin (Promega) was used at 1:50 ratio (w/w) and the digestion proceeded for 16 h at 30 °C.
For comparative analysis, peptide concentration was determined spectrophotometrically (Nanodrop, Thermo Scientific) and samples from the same group of chickens were pooled. Pooled samples were then labelled using the stable isotope dimethyl labelling protocol as described previously [16]. Labeled samples were mixed and 3 subfractions were prepared using Oasis MCX Extraction Cartridges (Waters). The samples were desalted on SPE C18 Extraction Cartridges (Empore) and concentrated in a SpeedVac (Thermo Scientific) prior to LC-MS/MS.
High resolution (30 000 FWHM at 400 m/z) MS spectra were acquired for the 390-1700 m/z interval in an Orbitrap analyser with an AGC target value of 1 × 10 6 ions and maximal injection time of 100 ms. Low resolution MS/MS spectra were acquired in Linear Ion Trap in a data-dependent manner and the top 10 precursors exceeding a threshold of 10 000 counts and having a charge state of +2 or +3 were isolated within a 2 Da window and fragmented using CID.

Data processing, protein identification and quantification
Raw data were analysed using the Proteome Discoverer (v.1.4). MS/MS spectra identification was performed by SEQUEST using the Gallus gallus protein sequences obtained from Uniprot database. Precursor and fragment mass tolerance were 10 ppm and 0.6 Da, respectively. Carbamidomethylation (C) and oxidation (M) were set as static and dynamic modifications, respectively. Dimethylation (N-term and K) was set as static modification in the label-based analysis. Only peptides with a false discovery rate FDR ≤ 5% were used for protein identification.
Spectral counting, the protocol in which abundance of a protein is expressed as the total number of tandem mass spectra matching its peptides (peptide spectrum matches, PSM), was used for comparative label-free analysis of heterophil and macrophage proteomes [17]. For a general comparison of protein abundance between heterophils and macrophages, PSMs belonging to a particular protein from all three groups of chickens, i.e. 18 samples, were summed up. The identification of at least two distinct peptides belonging to the particular protein and the threshold of at least 5 PSMs in at least one sample was required for its reliable identification [18,19]. All data were normalized to the total number of PSMs in individual samples. Statistical analysis using a t test was performed and the proteins with p value ≤ 0.05 and with at least four fold differences in its amounts were considered as significantly different in their abundance between the subpopulations.
In the label-based quantification, only unique peptide sequences with at least 20 PSMs were considered for peptide ratio calculations. Subsequent analysis of labelbased data was performed in R (https://www.R-project. org). For each protein, its individual peptide ratios were log 2 transformed, mean values were calculated and tested with a one sample t test. Benjamini-Hochberg correction for multiple testing was then applied to the obtained p values. Only proteins having ≥ twofold change and adjusted p value ≤ 0.05 were considered as being significantly different in abundance.

Bioinformatic analysis
Protein interaction networks were built using the online database resource Search Tool for the Retrieval of Interacting Genes (STRING). Proteins were further analyzed using Gene Ontology (GO) database and the Kyoto Encyclopedia of Genes and Genomes (KEGG) for their classification into specific pathways. PCA plots were calculated and created in R (https://www.R-project.org).

S. Enteritidis infection
Intravenous S. Enteritidis infection resulted in a high colonization of systemic sites. Average log 10 S. Enteritidis counts were 5.03 ± 0.54 and 3.06 ± 0.99 CFU/g of liver in the infected chickens and the vaccinated and infected chickens, respectively. Despite this, no fatalities were observed among infected chickens. No S. Enteritidis was detected in any of the control non-infected chickens.

Identification of heterophil and macrophage specific proteins
Proteins specific for chicken heterophils or macrophages were determined irrespective whether these were obtained from the infected or non-infected chickens.
Altogether, 858 proteins from heterophils and 1032 proteins from macrophages were detected. Out of these, 654 proteins were expressed both in heterophils and macrophages. Two-hundred and eight proteins were detected in macrophages only and an additional 126 proteins were 4 times or more abundant in macrophages than in heterophils. On the other hand, 34 proteins were detected in heterophils only and an additional 44 proteins were 4 times or more abundant in heterophils than in macrophages (Additional file 2).

Proteins characteristic for heterophils
Out of 78 proteins characteristic for heterophils (Additional file 2), 20 with the highest PSM difference between heterophils and macrophages are listed in Table 1. These included MRP126, LECT2, CATHL1, CATHL2, CATHL3, LYG2, LYZ and RSFR proteins, all with antibacterial functions. STOM and RAB27A proteins controlling storage and release of granular proteins in neutrophils also belonged among the characteristic and highly expressed proteins in heterophils. Two serine protease inhibitors, SERPINB10 and SERPINB1, were also found among the 20 most characteristic heterophil proteins (Table 1). Only a single KEGG pathway was specifically enriched in heterophils and this was the starch and sucrose metabolism pathway comprising PYGL, PGM1 and PGM2 proteins (p = 1.7E−4). Despite the KEGG pathway designation, all these proteins represent enzymes involved in glycogen metabolism [20].

Heterophil proteins responding to in vivo infection with S. Enteritidis
Altogether, 153 proteins were present in different abundance in the heterophils before and after S. Enteritidis infection. Of these, 109 proteins increased and 44 proteins decreased in abundance (Additional files 3 and 4 for all quantified heterophil proteins). Proteins belonging to 2 KEGG categories were enriched in heterophils following S. Enteritidis infection. These included the category translation with 39 proteins (p = 2.58E−62) and protein processing in endoplasmic reticulum (12 proteins, p = 1.74E−11). Twenty proteins with the highest increase in abundance, except for those belonging to the category translation, are listed in Table 3. Among others, these included AVD, F13A, ANXA2, ANXA7 or CTSC.
Forty-four proteins decreased in abundance in heterophils following S. Enteritidis infection and 20 of these with the highest decrease are listed in Table 4. Proteins with decreased abundance were those found in heterophil granules such as MPO, LYZ, LYG2, CTSG, CTSL1, CATHL1, CATHL2, RSFR, MMP9 and LECT2. Another set of proteins which decreased in heterophils following S. Enteritidis infection included ALB, FN1 and OTFB (Table 4).

RNA expression
Finally we verified the expression of 37 genes coding for selected proteins listed in Tables 1, 2, 3, 4 and 5. Expression of 4 genes, LRP1, MPO, PPIB and TUBA3A was too low and these genes were excluded from further consideration (Additional file 7).
Six genes (LGALS1, MRC1L, GDA, MECR, DMBT1, LRPAP1) out of 7 proteins selected as specific for macrophages were transcribed in macrophages at a higher level than in heterophils. Only HSP70 was transcribed in macrophages and heterophils at the same level though it was present in higher abundance at the protein level in macrophages. Nine genes (MRP126, OTFB, LYG2, LYZ, SERPINB1, CATHL1, CATHL2, MMP9, LECT2) out of 14 heterophil specific proteins were transcribed in heterophils at a higher level than in macrophages. Two genes of this group (GPX, CTSG) were transcribed in heterophils and macrophages at the same level and the remaining 2 genes (RSFR, LTA4H) were transcribed at a higher level in macrophages though protein mass spectrometry indicated their higher abundance in heterophils. Expression of 11 proteins which increased in abundance in macrophages following infection of chickens with S. Enteritidis was also tested at the RNA level. Except for MRP126, 10 of these (MECR, CTSC, ERAP1, RSFR, SOD1, CALR, CATHL1, CATHL2, LECT2, GAL1) did not exhibit any difference at the transcriptional level. Similar to the results of protein mass spectrometry, RNA levels of the tested genes in the heterophils or macrophages from the vaccinated chickens were in between the expression in non-infected chickens and chickens infected without previous vaccination. Only 3 genes in heterophils did not follow this scheme and CATHL1, CATHL2 and LECT2 were expressed in heterophils from the vaccinated chickens at significantly higher level than in the heterophils from infected chickens.

Discussion
Until now, chicken heterophils and macrophages have been characterized only by their specific characteristics like cytokine signaling or production of antimicrobial peptides [2,6,7,24,25] and an unbiased report characterizing their total proteome, before and after infection, has been missing. In the current study we therefore isolated proteins from heterophils and macrophages and quantified their abundance before and after infection with S. Enteritidis by mass spectrometry. We have to remind that mass spectrometry provides reliable data for approximately 800 the most abundant proteins. The lowly represented proteins, despite their potential specificity or responsiveness to infection, could not be therefore detected.
Chicken macrophages differed from heterophils in 3 specific features. First, macrophages specifically expressed receptors such as MRC1L, LRP1, LGALS1, LRPAP1 and DMBT1L. Second, macrophages exhibited higher mitochondrial activity including fatty acid degradation, TCA cycle and oxidative phosphorylation. And third, macrophages specifically expressed enzymes involved in arginine and proline metabolism (Figure 1). Receptors specifically expressed by macrophages indicate their potential to sense signals from the external environment which allows them to modulate immune response [6,7] including their own polarization [26,27]. The dependency of macrophages on oxidative phosphorylation and mitochondria functions was already described for human macrophages and neutrophils [28]. Macrophages were also enriched in arginine and proline metabolism since one of their bactericidal activities is the production of NO radicals by iNOS and arginine [29]. Following infection with S. Enteritidis, macrophages increased the expression of lysosomal and phagosomal proteins what could be associated not only with S. Enteritidis inactivation but also with macrophage ability of antigen presentation. Heterophils specifically expressed granular proteins MPO, LYZ, LYG2, RSFR, LECT2, CATHL1, CATHL2, CTSL1, CTSG, OTFB, SERPINB1 and MMP9, and endoplasmic reticulum proteins SSR1, PDIA4, PDIA6, PPIB, BiP, HSP90B1 and CANX. The latter group of proteins is activated when lumenal conditions in endoplasmic reticulum are altered or chaperone capacity is overwhelmed by unfolded or misfolded proteins [30]. Induction of an unfolded protein response leads to neutrophil degranulation in mice [31] and based on our results, a similar response can be predicted also in chicken heterophils.
Granular proteins decreased in heterophils in response to infection. Since transcription of genes encoding these proteins did not change and the number of ribosomal proteins increased, these genes must have remained continuously transcribed and translated even after initial degranulation [24,[32][33][34][35]. However, not all proteins that decreased in heterophils following S. Enteritidis infection were assigned to pathogen inactivation. Matrix metalloproteinase MMP9 is used for degradation of the extracellular matrix to enable leukocyte infiltration to the site of inflammation [36], and ALB and FN1, are found at the surface of granulocytes and inhibit their migration [37,38]. The decrease of ALB and FN1 together with the degradation of extracellular matrix by MMP9 leads to heterophil translocation from the blood circulation to the site of inflammation.
Comparing expression at the protein and RNA levels provided several unexpected results. Changes in expression at the RNA level in response to infection were more pronounced in heterophils than in macrophages. We can exclude any technical issues in macrophage gene expression analysis since there were at least 3 genes inducible at the RNA level also in macrophages (AVD, MRP126 and F13A). Unlike macrophages, there were also greater differences in the expression profiles of heterophils obtained from vaccinated chickens in comparison to those obtained from naive but infected animals and an increase in CATHL2 and LECT2 in the heterophils from the vaccinated chickens following S. Enteritidis challenge appeared as a specific positive marker of vaccination. Despite this, expression in heterophils and macrophages in naive but infected chickens tended to approach a similar expression profile ( Figure 2). In this study we characterized protein expression in chicken heterophils and macrophages in response to intravenous infection with S. Enteritidis. Heterophils decreased ALB and FN1, and released MMP9 to enable their translocation to the site of infection. Secondly the endoplasmic reticulum proteins increased in heterophils which resulted in the release of granular proteins. On the other hand, macrophages were less responsive to acute infection and an increase in proteins like CATHL1, CATHL2, RSFR, LECT2 and GAL1 in the absence of any Figure 1 The most characteristic proteins and their functions in chicken heterophils and macrophages. Heterophils express MMP9, MRP126, LECT2, CATHL1, CATHL2, CATHL3, LYG2, LYZ and RSFR proteins. Following S. Enteritidis infection, heterophils decreased fibrinogen FN1 and albumin ALB, and increased ribosomal proteins. In addition, endoplasmic reticulum proteins are activated which results in the release of granular proteins. Heterophils expressed glycogen (Gly) metabolism pathway which allows for rapid glucose (Glu) availability and anaerobic ATP generation via glycolysis while macrophages increased mitochondrial activity. Macrophages expressed receptor proteins MRC1, LGALS1, LRPAP1 and DMBT1L, mitochondria-localized proteins and arginine metabolism proteins. Following infection with S. Enteritidis, macrophages increased the expression of lysosomal and phagosomal proteins (CTSB, CTSC, RAB7A, CATHL1, RSFR, GAL1, SOD1).

Figure 2
PCA cluster analysis of chicken heterophils and macrophages using expression data from qPCR. Each spot represents heterophils (circles) or macrophages (triangles) isolated from non-infected (green color), infected (red color), and vaccinated and infected chickens (blue color), 6 chickens per group. Heterophils from vaccinated chickens responded to infection more than macrophages from the same chicken. Transcription of heterophils and macrophages from naive but infected chickens approached the same profile.