Proteome signatures of inflammatory activated primary human peripheral blood mononuclear cells.

Proteome profiling is the method of choice to identify marker proteins whose expression may be characteristic for certain diseases. The formation of such marker proteins results from disease-related pathophysiologic processes. In healthy individuals, peripheral blood mononuclear cells (PBMCs) circulate in a quiescent cell state monitoring potential immune-relevant events, but have the competence to respond quickly and efficiently in an inflammatory manner to any invasion of potential pathogens. Activation of these cells is most plausibly accompanied by characteristic proteome alterations. Therefore we investigated untreated and inflammatory activated primary human PBMCs by proteome profiling using a 'top down' 2D-PAGE approach in addition to a 'bottom up' LC-MS/MS-based shotgun approach. Furthermore, we purified primary human T-cells and monocytes and activated them separately. Comparative analysis allowed us to characterize a robust proteome signature including NAMPT and PAI2 which indicates the activation of PBMCs. The T-cell specific inflammation signature included IRF-4, GBP1 and the previously uncharacterized translation product of GBP5; the corresponding monocyte signature included PDCD5, IL1RN and IL1B. The involvement of inflammatory activated PBMCs in certain diseases as well as the responsiveness of these cells to anti-inflammatory drugs may be evaluated by quantification of these marker proteins. This article is part of a Special Issue entitled: Integrated omics.

Introduction

Peripheral blood mononuclear cells (PBMCs) are very important immune players and therefore involved in a large number of diseases. In healthy individuals, these cells circulate in a quiescent state monitoring potential immune-relevant events. Encountering any kind of disease-related abnormalities may elicit strong inflammatory responses in these cells which are targeted at the elimination of potential pathogens but may also cause severe side effects and malicious symptoms. Inflammatory activation engages several biological functions such as leukocyte migration, proliferation of T-cells, interferon response, NF-κB signaling and regulation of cell death. Proteome profiles characteristic for such functional cell states have not yet been comprehensively investigated in inflammatory activated PBMCs.

Proteomics is a powerful screening technology aiming at the high throughput analysis of complex samples mainly using mass spectrometry [1]. Several studies have been performed in order to analyze proteome profiles of quiescent PBMCs, resulting in the classification of proteins according to functional groups and cellular localization [2] or cell type specificity [3]. PBMCs consist of several different cell types and proteome analyses of these cells result in great complexity of data, especially when including the consideration of different functional states. Additionally, the heterogeneity of biological samples and methodological challenges may obstruct straight interpretation of data [4,5].

Indeed, PBMCs consist of different cell types with lymphocytes and monocytes as main constituents and more than two thirds of the lymphocytes represented by T-cells. Furthermore, each of these cells can occur at different functional states, especially when cells from diseased persons are considered. Activated T-cells may account for intrinsic immune response-related tissue damage characteristic for chronic inflammation, autoimmunity and graft versus host disease [6]. Monocytes act as important regulators of T-cell activities. Upon encounter with a potential pathogen, monocytes become activated resulting in differentiation into macrophages. Macrophages present phagocytized and processed material on their cell surface and thus induce the targeted activation of T-cells. Activated T-cells in turn may further activate endothelial cells, fibroblasts and monocytes/macrophages thus maintaining an acute or chronic inflammatory state [7]. Recently, differential PBMC protein expression was shown to support the diagnosis of ulcerative colitis and Crohn's disease [8] as well as of systemic lupus erythematosus [9] or rheumatoid arthritis [10].

While the role of the immune system for several diseases such as pathogen-related disorders is obvious and well understood, the detailed regulatory mechanisms involved in disorders such as chronic inflammatory diseases, atherosclerosis, Alzheimer and cancer are not yet fully understood and therefore in the focus of current research [11-14]. The proteomic characterization of PBMCs and of functional cell states thereof may help to indicate and assess the contribution of these cells to certain diseases.

While current data interpretation strategies for large‐omics data sets rather rely on network analysis [15], we seek a systematic strategy based on the establishment of reference proteome profiles. Especially, we focus on the characterization of proteins specifically expressed in individual cell types at defined functional states. We have already characterized proteins specifically expressed in inflammatory activated as well as tolerogenic primary human dendritic cells [16]. Our research group has also been working with PBMCs focusing on oxidative stress [17] and nutritional intervention [18].

In the present work, in order to generate further reference proteome profiles, we focus on PBMCs which were inflammatory activated in vitro with LPS (lipopolysaccharide) and PHA (phytohaemagglutinin) and subsequently analyzed using 2D-PAGE (‘top down’), as well as an LC–MS/MS based shotgun approach (‘bottom up’) [19]. Proteome profiles were recorded and compared to those of control PBMCs using the Griss Proteomics Database Engine (GPDE), a database specifically engineered for the identification and characterization of marker proteins [20]. In order to relate the observed proteome alterations to the corresponding cell type of origin, we isolated and stimulated T-cells (with PHA) and monocytes (with LPS), the two main cellular constituents of PBMCs, separately before subsequent analysis. We thereby successfully identified proteins that are newly expressed or up-regulated both in T-cells and monocytes. Additionally, we identified proteins specifically induced in T-cells and monocytes, respectively, upon activation. Knowledge of these marker proteins may reveal the involvement of inflammatory-activated T-cells or monocytes in biological samples, which may strongly support the interpretation of more complex clinical proteomics data whenever inflammatory activated PBMCs may be involved.

Material and methods

Blood samples

PBMCs of six individuals were isolated. From each donation we created four aliquots. Two were used for metabolic labeling (untreated and treated) and subsequently analyzed by 2D-PAGE. The other two aliquots were fractionated and further processed for shotgun analysis. Per group, fractionation resulted in six cytoplasmic fractions. Three of them were analyzed twice, resulting in a total of nine shotgun analyses. Five nuclear extracts were successfully isolated and analyzed, while only two secretomes were successfully analyzed. T-cells and monocytes were isolated from four independent blood donations and the corresponding cytoplasmic aliquots were processed for shotgun analysis. In case of two T-cell preparations, aliquots were treated in two different ways (PHA and ionomycin/PMA, respectively). The corresponding PRIDE-experiments are listed in Supplementary Table 2. We have not generated PRIDE-files for the spots identified in 2D-gels.

Isolation and cultivation of PBMCs

PBMCs were isolated from fresh blood (blood samples from volunteers) of healthy donors with written consent of each donor and the approval of the Austrian Ethics committee (no. 297/2011). For the isolation of PBMCs, 50 ml full blood were diluted with RPMI1640 medium (Gibco Ltd., Paisley, Scotland) and supplemented with 2 mM L-glutamine, 100U/ml penicillin, 100 μg/ml streptomycin (Sigma-Aldrich, St. Louis, MO) and 1000U heparin (EBEWE Pharma, Unterach, Austria). 35 ml of the mixture were then carefully overlaid above Ficoll-Paque (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) and centrifuged at 500 g for 30 min at 14 °C. PBMCs were collected from the interphase and were then either re-seeded in diluted autologous plasma or, if used for subsequent cell purification, washed with RPMI Heparin medium and MACS buffer (PBS 1% Human Serum Albumin (Aventis Behring, Vienna, Austria)/5 mM EDTA (Gibco Ltd., Paisley, Scotland) and counted [21,22].

Monocyte and T cell separation

T-cells and monocytes were separated by magnetic sorting using the MACS technique (Miltenyi Biotec, Bergisch Gladbach, Germany), including the use of MACS buffer, Streptavidin MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany), CS columns (Miltenyi Biotec) and the VarioMACS separator. T-cells were obtained by negative selection, which was done by depletion of the PBMCs flowtrough from non-T-cells using an antibody mix containing anti-CD14 (MEM 18) for monocytes, anti-CD16 (3G8) for granulocytes and NK‐cells, anti-CD19 for B cells (BU 12) and anti-CD33 (4D3) for monocytes, thrombocytes and myeloid progenitors, all at concentrations of 10 μg/ml. For monocytes up to 1 × 109 cells were positively enriched by incubating the PBMCs with 15 μg/ml of biotinylated anti-CD14 (VIM13, MEM 18) to label the monocytes [23].

FACS analysis of purified T-cells and monocytes

This method was applied to verify the purity of isolated T-cells and monocytes. The cell suspension (5 × 105 cells/assay) was resuspended in 50 μl Beriglobin (CSL Behring) and kept for 10 min on ice. 20 μl of the following PE or FITC-conjugated mouse antibodies were used at a concentration of 20 μg/ml each: VIAP (clone 2D5), CD3 (UCHT1) and CD4 (vit4)/CD8 (Vit8) provided from the Institute of Immunology from Otto Majdic; CD45 (HI30)/CD14 (TüK4), CD3 (S4.1)/SJ25-C1, CD3 (S4.1)/HLA-DR (Tü36) and CD56 (MEM188) from Caltag. Antibodies were prepared in Micronic-tubes and 50 μl of the cell suspension was added, mixed and incubated for 30 min at 4 °C. Dead cells were labeled by addition of ethidiumbromide or propidiumiodide (1:100). The tubes were kept on ice until they were analyzed by flow-cytometry. Purity of monocytes was found to be above 95%, for T-cells above 98% (see Supplementary Fig. 1). In case of monocytes, we did not collect sufficient cell amounts to allow the analysis of nuclear extracts and secreted protein fractions.

35S-metabolic labeling for measuring of protein synthesis

After isolation PBMCs were reseeded in diluted plasma of the donor in the presence of 35S-labeled methionine and cysteine (Trans35label, Biomedica, MP Biomedicals) for 6 h at 37 °C. The induction of new protein synthesis which is observed by autoradiography is highest within the first few hours after stimulation [24], therefore here we chose 6 h labeling time.

Inflammatory stimulation

For activation of PBMCs for 2D-PAGE, cells were resuspended in blood plasma and treated with 1 μg/ml lipopolysaccharide (LPS, Sigma-Aldrich) [25], inducing an immune response in monocytes via the toll-like pathway (Myd88) [26-28], and 5 μg/ml phytohaemagglutinin (PHA, Biochrom) [29], a lectin acting mitogenic in T-cells causing proliferative cell division [30], for 6 h at 37 °C. For shotgun analysis, PBMCs were diluted in RPMI medium to 1 × 106 cells/ml and treated as above with a combination of LPS and PHA (same concentrations) for 24 h in order to provide the cells sufficient time to cumulate newly synthesized proteins [31]. Isolated monocytes or T-cells were treated with 1 μg/ml LPS or 5 μg/ml PHA for 24 h, respectively. Alternatively T-cells were activated by 1 μM ionomycin/100nM PMA (phorbol 12-myristate 13-acetate) for 24 h [32-34].

Sub-cellular fractionation

The in vitro treated cells were grown in culture medium for 16 h and further in serum-free medium for 8 h (together 24 h) to collect the secretome. To minimize unspecific effects on cells due to the in vitro culturing conditions we obtained controls either by directly processing isolated PBMCs or after 8 h of incubation time.

Secretome protein isolation

After that period of time the supernatant was sterile-filtered through a 0.2 μm filter and precipitated overnight by addition of ethanol tempered to − 20 °C.

Cytoplasmic protein isolation

The isolation of cytoplasmic proteins was performed as described by Gundacker et al. [35]. During all steps samples were kept on ice. Cells were lysed in isotonic lysis buffer (10 mM HEPES/NaOH, pH 7.4, 0.25 M sucrose, 10 mM NaCl, 3 mM MgCl2, 0.5% Triton X-100) supplemented with protease inhibitors (pepstatin, leupeptin and aprotinin, each at 1 μg/ml; 1 mM PMSF) and pressed through a 23 g syringe to induce cell lysis. The cytoplasmic fraction was separated from nuclei by centrifugation at 3500 rpm and 4 °C for 5 min and precipitated overnight by addition of ethanol tempered to − 20 °C.

Nuclear fraction protein isolation

The nuclear pellets swelled up in extraction buffer (500 mM NaCl) for 10 min followed by a 1:10 dilution in NP-40 buffer for 15 min, to reduce the NaCl concentration. The nuclear protein fraction was separated from debris by centrifugation at 3500 rpm and 4 °C for 5 min and precipitated overnight by addition of ethanol tempered to − 20 °C. Afterwards, all protein samples were dissolved in sample buffer (7.5 M urea, 1.5 M thiourea, 4% CHAPS, 0.05% SDS, 100 mM DDT).

2D-PAGE (‘top down’)

Cytoplasmic proteins were loaded by passive rehydration on IPG strips pH 5–8, 17 cm (BioRad, Hercules, CA) at room temperature. Isoelectric focusing (IEF) was performed in a stepwise fashion (1 h 0–500 V linear; 5 h 500 V; 5 h 500–3500 V linear; 12 h 3500 V). After IEF, the strips were equilibrated with 100 mM DTT and 2.5% iodacetamide according to the instructions of the manufacturer (BioRad). For SDS-PAGE the Protean II xi electrophoresis system (BioRad) was used. IPG strips were placed on top of 1.5 mm 12% polyacrylamide slab gels and overlaid with 0.5% low-melting agarose with bromophenol blue. After electrophoresis, gels were stained with a 400 nM solution of Ruthenium II tris (bathophenanthroline disulfonate) (RuBPS) as described before [36]. For this purpose the gels were fixated in 50% methanol/7% acetic acid overnight, next day washed two times for 30 min with 20% methanol, then stained for 6 h with 400 nM RuBPS, and destained overnight in 15% methanol/7% acetic acid. Fluorography scanning was again performed with the FluorImager 595 at a resolution of 100 μm [37]. After scanning the fluorescence of the gels, the gels were dried for subsequent autoradiography. Dried gels were inserted into cassettes including a phosphor screen as detector for ß-radiation of the 35S labeled proteins. These phosphor screens were scanned with the PhosphorImager SI MAC (Molecular Dynamics) with 100 microns. Gels were warped to a reference gel with the TT900 S2S software (version 2006.0.2389, Nonlinear dynamics, Carlsbad, CA) and evaluated with the Progenesis software PG200 (version 2006, Nonlinear) using the “same spot” algorithm. Only protein spots which displayed a more than two-fold increase on average of the corresponding normalized integrated spot intensity were considered as differently regulated and were further analyzed by mass spectrometry.

1D-PAGE for subsequent shotgun analysis (‘bottom up’)

Protein fractions (supernatant, cytoplasm and nuclear extracts) were loaded on 12% polyacrylamid gels, electrophoresis was performed until complete separation of a pre-stained molecular marker (Dual Color, Biorad, Hercules, CA) was visible. After fixation with 50% methanol/10% acetic acid and subsequent silver staining, gel lanes were cut out of the gel and digested with trypsin as described below.

MS-compatible silver staining procedure

SDS-PAGE gels were fixed with 50% methanol, washed and sensitized with 0.02% Na2S2O3. The gels were stained with 0.1% AgNO3 ice cold for 20 min, rinsed with bi-distilled water and subsequently developed with 3% Na2CO3/0.05% formaldehyde as previously described [38].

Protein digestion with trypsin

Spots were cut out from 2D-gels. SDS-gels were cut into slices. After destaining, reduction with DTT and alkylation with iodacetamide, proteins were digested with trypsin (sequencing grade, Roche) at 37 °C overnight as described before [39]. After elution, the peptides were forwarded to LC–MS/MS analysis.

Mass spectrometry analysis

For the identification of isolated 2D spots, the corresponding peptides were loaded on a Zorbax 300SB-C8 (5 μm, 0.3 mm, 5 mm) column and separated by nanoflow LC (1100 Series LC system, Agilent, Palo Alto, CA) with a Zorbax 300SB-C18 (5 μm, 75 mm) column at a flow rate of 250 nl/min using a gradient from 0.2% formic acid and 3% acetonitrile (ACN) to 0.2% formic acid and 45% ACN over 12 min. In case of shotgun analysis, peptides were separated by nanoflow LC (1100 Series LC system, Agilent, Palo Alto, CA) using the HPLC-Chip technology (Agilent) equipped with a 40 nl Zorbax 300SB-C18 trapping column and a 75 μm × 150 mm Zorbax 300SB-C18 separation column at a flow rate of 400 nl/min, using a gradient from 0.2% formic acid and 3% ACN to 0.2% formic acid and 50% ACN for over 60 min. Peptide identification was accomplished by MS/MS fragmentation analysis with an iontrap mass spectrometer (XCT-Ultra, Agilent) equipped with an orthogonal nanospray ion source. The MS/MS data analysis, including peak list-generation and spectrum identification, was done using the Spectrum Mill MS Proteomics Workbench software (Version A.03.03, Agilent) allowing for two missed cleavages and searched against the SwissProt/UniProtKB protein database for human proteins (Version 12/2010 containing 20,328 entries) allowing for precursor mass deviation of 1.5 Da, a product mass tolerance of 0.7 Da and a minimum matched peak intensity (%SPI) of 70%. Due to previous chemical modification, carbamidomethylation of cysteines was set as fixed modification. Oxidation of methionine was the only post-translational modifications considered here. The apparent positive matches found within the search results for peptides having a SpectrumMill peptide score higher than 13 when using the corresponding reversed database compared to the true database were consistenly less than 1% (documented in the PRIDE XML files). Peptides scoring between 9 and 13 were included only if precursor m/z value, retention time and MS2 pattern were found similarly in at least one of our previous experiments and the peptide was thereby scoring above 13. With respect to protein inference, we chose the smallest number of proteins required to explain all observed peptides as described for ProteinProphet [40]. As our protein identification algorithm includes manual selection, we cannot calculate an exact false discovery rate. All identification details including MS2 spectra are fully documented in the PRIDE-XML files available at www.ebi.ac.uk/pride (experiments 22162–22200, 26890–26904, Table S2).

Data evaluation of shotgun analyses

The PRIDE-XML files were loaded into a local version of the GPDE, the software can be downloaded freely from http://sourceforge.net/projects/gpde/. For uploading the files, the parameters “species”, “tissue”, “cell type” and “cell state” as well as “sub-cellular fraction” were set accordingly. Data replicates become assembled as described [20], the emPAI (exponentially modified protein abundance index) values were calculated according to Ishihama et al. [41] using the “Data Analysis” tool. The cell symbols were obtained using the “Protein Expression” tool. In Figs. 3 and 4, snapshots of the database output screens are shown.

Results

2D-PAGE of control and LPS/PHA-treated PBMCs

Primary cells present a significant challenge to proteome research because of their intrinsic heterogeneity, instability and sensitivity to any environmental alteration. Therefore, we applied strict standard operating procedures to minimize the differences between the working procedures. PBMCs from six individual donors were isolated under sterile conditions, transferred back into plasma of the corresponding donor and metabolically labeled for 6 h by the addition of 35S-methionine/cysteine. One aliquot was treated with LPS, an activator of monocytes, and PHA, an activator of T-cells (see Section 2). 2D-PAGE separation of cytoplasmic proteins, fluorescence detection and subsequent autoradiography allowed us to record a marked increase in protein synthesis in the stimulated cells (Fig. 1). The comparison of the activated cells with the untreated controls considering both fluorescence and autoradiographic protein detection identified several proteins to be specifically induced as exemplified in more detail for IFIT-2 (Fig. 2). Selected protein spots of corresponding unlabeled cell preparations were excised, digested with trypsin and analyzed by mass spectrometry. 14 proteins displayed a more than two-fold increase on average of the corresponding normalized integrated spot intensity and were identified using MS as indicated in Fig. 1. Although the autoradiographic spot patterns of the individual donors showed some variations, the inflammation-induced alterations were of very high conformity (Fig. S2).

Fig. 1

Comparison of the cytoplasmic protein fractions from untreated and inflammatory activated PBMCs by fluorescence detection and autoradiography from 2D gels. Cytoplasmic proteins of untreated PBMCs (Con) and, by LPS and PHA, inflammatory activated PBMCs (Act) were separated by 2D-PAGE, stained with the fluorescence dye RuBPS, dried and exposed to phosphor-screens. The first row (Fl) features the fluorescence images, while the second row shows the autoradiography (AR) images. The fluorescence images provide qualitative and quantitative information about the overall protein composition of the PBMCs. The autoradiographs display proteins newly synthesized during the labeling period. UniProtKB/SwissProt accession numbers indicate proteins which were identified as specifically induced or up-regulated upon inflammatory activation.

Fig. 2

Evidence for induced protein expression by 2D-PAGE. The fluorescence pattern of inflammatory activated PBMCs (B) of a representative donor was very similar to the corresponding control (A). The newly detectable spot was identified as IFIT-2 (P09913). The corresponding autoradiograph demonstrated very high 35S incorporation of this protein (D) and absence of detection in the control (C). Image overlays of fluorescence detection of activated (magenta) with the control (green) demonstrate the specificity of this protein expression (E). Image overlay of autoradiography of the activated sample (green) with the corresponding fluorescence detection (magenta) demonstrates the very high labeling of the otherwise hardly detectable spot (F).

Shotgun analysis of control and LPS/PHA-treated PBMCs

Additionally, unlabeled cell aliquots of similar experiments were analyzed using shotgun proteomics. Eight hours after treatment, when inflammation-induced protein synthesis was apparently up-regulated most significantly (Fig. 2), shotgun analysis did not reveal several of the alterations observed in the 2D gels (data not shown). However, after 24 h of treatment, most of the proteins found induced by means of 2D-PAGE were as well identified by shotgun proteomics. Shotgun proteomics requires some threshold protein amounts which may require the accumulation of several hours of successful protein synthesis. As specifically detecting protein synthesis, autoradiography may therefore provide more contrasting results than determination of protein amounts (Fig. 2). Furthermore, we observed that in vitro culturing of PBMCs for several hours without treatment was sufficient to induce some inflammatory activation of cells. The immune cells may easily recognize the environmental change accompanying blood processing and respond to it. Consequently, we processed untreated cells directly after isolation. Then, cytoplasmic, nuclear and secreted protein fractions were separated by SDS-PAGE, digested with trypsin and analyzed by mass spectrometry as described previously [3]. For data analysis only proteins identified with at least two distinct peptides and in at least two independent experiments were considered. Based on these conditions, 1496 proteins were identified in the untreated cells and 1497 proteins in the inflammatory activated cells. 1424 proteins of these were common to both groups (Supplementary Data Table S1). All identification details including MS2 spectra are fully documented in the PRIDE-XML files available at www.ebi.ac.uk/pride (experiments 22162–22200, 26890–26904, Table S2).

2D-PAGE in combination with results from shotgun analysis

18 cytoplasmic and 6 nuclear proteins were considered as specifically expressed upon activation as they were reliably identified by shotgun analysis with three or more distinct peptides in at least three out of the six donors and were not detected in the controls. The 18 cytoplasmic proteins including corresponding gene-ontology terms [42] are listed in Table 1. 13 of these proteins were confirmed to be induced using 2D autoradiographs (Table 1), and only one protein (IFIT2) was found using 2D-PAGE and remained undetectable using shotgun analysis (Fig. 1). One of the newly induced proteins was identified as guanylate-binding protein 5, a UniProtKB entry listed only with “evidenced at transcriptional level”. The corresponding mass spectrometry data are summarized in Fig. 3 and clearly evidence the expression of this protein in activated PBMCs. The specifically induced 6 nuclear proteins were MCM7 (P33993), SATB1 (Q01826), OAS2 (P29728), G3BP1 (Q13283), EIF2AK2 (P19525) and MCM5 (P33992).

Table 1

Grouping of proteins which were specifically expressed in inflammatory activated PBMCs according to their expression in purified and activated T-cells and monocytes. All proteins listed here were found specifically expressed in activated PBMCs with three or more distinct peptides in at least three out of six donors and were not detected in the controls. T-cells and monocytes, the main constituents of PBMCs were purified, activated and analyzed separately. The induced proteins were compared to those observed in activated PBMCs. Proteins identified in both T-cells and monocytes, in only one cell type or in none were assembled into groups. Proteins independently observed to be up-regulated when using 2D-PAGE are indicated by “x” in the column “2D”. A selected “biological process” GO term is listed for each protein.

		2D	GO - biological processes
I) Proteins induced in both activated T-cells and monocytes
P08195	4F2 cell-surface antigen heavy chain	x	Leukocyte migration, cell growth
P05120	Plasminogen activator inhibitor 2 (PAI-2)	x	Anti-apoptosis
P43490	Nicotinamide phosphoribosyltransferase	x	Positive regulation of cell proliferation
P61289	Proteasome activator complex subunit 3	x	Regulation of apoptotic process
II) Proteins induced in activated T-cells only
Q00653	Nuclear factor NF-kappa-B p100 subunit		Proliferation
Q15306	Interferon regulatory factor 4 (IRF-4)	x	T cell activation
P32456	Interferon-induced guanylate-binding protein 2	x	Interferon-gamma-mediated signaling pathway
P12004	Proliferating cell nuclear antigen (PCNA)	x	proliferation
P12268	Inosine-5′-monophosphate dehydrogenase 2	x	GMP biosynthetic process
P80217	Interferon-induced 35 kDa protein (IFP 35)	x	Type I interferon-mediated signaling pathway
Q99873	Protein arginine N-methyltransferase 1		Cell surface receptor linked signaling pathway
Q96PP8	Guanylate-binding protein 5	x	GTP binding
P32455	Interferon-induced guanylate-binding protein 1		Interferon-gamma-mediated signaling pathway
III) Proteins induced in activated monocytes only
O14737	Programmed cell death protein 5		Apoptotic process
P01584	Interleukin-1 beta	x	Positive regulation of T cell proliferation
P18510	Interleukin-1 receptor antagonist protein (IL-1ra)	x	Immune response
IV) Proteins induced in activated PBMCs, but not isolated cells
P20591	Interferon-induced GTP-binding protein Mx1	x	Type I interferon-mediated signaling pathway
P14902	Indoleamine 2,3-dioxygenase 1 (IDO-1)		Tryptophan catabolic process

Fig. 3

Identification of guanylate binding protein 5 (Uniprot Q96PP8). Readouts of the GPDE are shown and suggest safe identification of the indicated protein. Within the amino acid sequence of the protein, the identified peptides are underlined, summing up to sequence coverage of 24.06%. The identified sequences of all present experiments are listed indicating whether the peptide sequence is unique within the human proteome and the number of identifications within the set of data (first peptide: identified in 8 out of nine experiments identifying Q96PP8) as well as amino acid positions within protein sequence, scores and fractions. Below a single MS2 spectrum of the indicated peptide sequence is shown.

To obtain a rough estimate of relative protein abundances, we calculated the average emPAI values for all proteins determined in inflammatory activated PBMCs, over all biological replicates (Table S1). 10 cytoplasmic and 4 nuclear proteins consistently displayed a three-fold or higher increase of the average emPAI (Table 2) compared to controls.

Table 2

Proteins specifically up-regulated in inflammatory activated PBMCs as demonstrated by shotgun proteomics. The assessment of quantitative alterations was based on the calculation of the emPAI (exponentially modified protein abundance index) value for a protein identification in activated versus control PBMCs. We considered the up-regulation of a protein in activated PBMCs as relevant only if the corresponding average emPAI value exceeded three-fold the average value of the untreated PBMCs and was increased in this way in at least three independent experiments. Pep-con and Pep-act, number of distinct peptides identified in untreated or inflammatory activated PBMCs respectively; exp-con and exp-act, number of experiments with a positive protein identification in control or activated PBMCs compared to the total number of the respective experiments; emPAI-con and emPAI-act, average emPAI value determined for a protein identification in the cytoplasmic or nuclear protein fraction of control and activated PBMCs respectively.

		pep-con	exp-con	emPAI-con	pep-act	exp-act	emPAI-act
Cytoplasmic proteins up-regulated
P42224	Signal transducer and activator of transcription 1-alpha/beta	2	3/9	0.044	14	6/9	0.208
P49327	Fatty acid synthase	5	3/9	0.038	28	6/9	0.159
P23381	Tryptophanyl-tRNA synthetase, cytoplasmic	4	8/9	0.121	20	8/9	0.458
P61247	40S ribosomal protein S3a	10	4/9	0.146	13	6/9	0.523
Q14152	Eukaryotic translation initiation factor 3 subunit A	15	4/9	0.031	17	5/9	0.11
P46776	60S ribosomal protein L27a	5	5/9	0.179	6	5/9	0.624
P62263	40S ribosomal protein S14	11	8/9	1.347	13	8/9	4.248
P46781	40S ribosomal protein S9	8	4/9	0.181	13	6/9	0.569
P83731	60S ribosomal protein L24 (Ribosomal protein L30)	5	1/9	0.212	5	7/9	0.664
O00571	ATP-dependent RNA helicase DDX3X	9	1/9	0.049	16	5/9	0.149
Nuclear proteins up-regulated
Q9NR30	Nucleolar RNA helicase 2	6	2/5	0.163	25	5/5	0.777
P10144	Granzyme B	1	2/5	0.145	10	4/5	0.679
Q13765	Nascent polypeptide-associated complex subunit alpha	0	0/5	0	5	5/5	0.648
Q01469	Fatty acid-binding protein, epidermal	0	0/5	0	4	3/5	0.532

Shotgun analysis of isolated T-cells and monocytes

T-cells and monocytes are well known for their distinct functions during inflammation. We therefore tried to relate the activation-induced proteome alterations to the corresponding cell type of origin. T-cells and monocytes were isolated from the PBMC cell mixtures of four individual donors, characterized with respect to cell purity by FACS analysis (Fig. S1), and treated with PHA and LPS for 24 h, respectively. Two T-cell aliquots were treated with Ionomycin/PMA for 24 h as an alternative to activation with PHA, the results were largely similar to those obtained with PHA. 16 of the 18 proteins observed to be induced in activated PBMCs were again identified in these purified and activated cell populations; 4 of them in both, purified and activated T-cells and monocytes, 9 proteins were identified in activated T-cells, 3 proteins in activated monocytes, while 2 proteins, IDO-1 and MX1, remained undetectable in the isolated cells (Table 1). The latter two proteins were thus found induced in the natural cell mixture only, but not in the purified cell populations. IDO1 has been described to be induced in human macrophages and monocyte-derived dendritic cells upon interaction with T-cells [43,44]. MX1 was identified in all 2D-gels and almost all shotgun results from activated PBMCs and also in mature dendritic cells investigated by LC–MS/MS in a previous study [16]. The absence of MX1 expression in the LPS-treated monocytes, which was independently reproduced in the 2D gels thereof (data not shown) may indicate that the isolated cells did not gain the full inflammatory activation state. The full inflammation state was therefore only obtained when the natural cell mixture of PBMCs was present during activation. We interpret this finding as indication for cell cooperation between different PBMCs.

Remarkably, no protein was found induced in the purified cell populations, which has not been identified in the activated PBMCs. These observations lead us to the classification of these 18 proteins into four groups, which we consider as functional signatures (Table 1); group I: signature for activated leukocytes; group II: signature for activated T-cells; group III: signature for activated monocytes; and group IV: signature for the cooperation of activated T-cells and monocytes.

Data interpretation

In this study we used a sub-cellular fractionation approach which was shown to increase the experimental reproducibility of proteome profiles [35]. Additionally, it supports cross-comparisons of protein expression patterns in specific sub-cellular compartments. We have extended our in-house developed data analysis platform, the GPDE [20], to visualize the average protein abundance (calculated by using the average emPAI of biological replicates) between the different cell types. These abundance values are represented as colored cell symbols for a given selection of proteins and cell types. The average emPAI value is translated into a color code with intensities corresponding to the found emPAI values. Cells are symbolized by a small inner circle for the nuclear fraction, an outer circle for the cytoplasm and an outer frame for the secretome (Fig. 4).

Fig. 4

Comparative analysis of protein expression across different cell types and functional cell states. Each cell symbol represents the protein expression of a single protein for a single cell type. Average emPAI values were calculated, increased color intensities correspond to increased emPAI values. All positive identifications were reproduced in at least three different donors, white fields indicate negative finding in six donors. The inner circle represents identification in the nuclear extracts, the outer circle in the cytoplasm and the outer frame in the secreted protein fraction. Five proteins were selected: 1, interleukin-1 beta (P01584); 2, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (P04406); 3, proliferating cell nuclear antigen (P12004); 4, indoleamine 2,3-dioxygenase 1 (IDO-1) (P14902); 5, Nicotinamide phosphoribosyltransferase (NAmPRTase) (P43490). GAPDH may serve as a kind of loading control, the emPAI in the cytoplasm was very similar in all cells presented here. NAmPRTase was found upregulated in T-cells as well as monocytes and monocyte-derived dendritic cells, thus representing a member of the first group of the proteome signature. PCNA was strongly induced in T-cells only, thus identifying activated T-cells (second group). IL-1beta, representative for group 3 was already detectable in untreated PBMCs cultured for eight hours and strongly induced in LPS-treated monocytes and dendritic cells but not in T-cells. IDO-1 was strongly induced in PBMCs (leukocytes) treated with LPS and PHA, but not in activated T-cells, monocytes and dendritic cells. The ladder observation may indicate characteristic cooperation upon inflammatory activation between the cells.

Using this tool the specificity of functional proteome signatures can easily be visualized. Fig. 4 shows a single representative of each of the above described groups as well as a protein not affected by inflammatory activation. The cytoplasmic part of the cell symbol representing GAPDH (Q96PP8) is of similar intensity in all cell types shown in Fig. 4. GAPDH can therefore be regarded as a baseline similar to a loading control in Western experiments. GAPDH was detected in the secretome of inflammatory activated PBMCs but not in the secretome of untreated controls. This clearly indicates the presence of dead cells which have released cytoplasmic proteins into the supernatant. NAMPT (P43490) is shown as a representative of group I: it is undetectable in the untreated PBMCs, T-cells and monocytes but positively identified in all inflammatory activated cells. PCNA (P12004) is shown as a representative of group II: it is up-regulated in activated T-cells but not identified in monocytes and dendritic cells. Remarkably, PCNA was detected in the nuclear fraction of untreated PBMCs which is compatible with the known sub-cellular location of this cyclin [45]. However, the newly induced synthesis of this protein upon induction of cell proliferation needs to take place in the cytoplasm. Indeed, PCNA was detected in the cytoplasmic protein fraction of activated PBMCs. IL-1beta (P01584), representative of group III, was found to be induced in monocytes but not in T-cells. IL-1beta was previously described to be up-regulated in LPS-activated monocytes [46]. IDO1 (P14902) was identified in activated PBMCs only, thus representing group IV.

Discussion

The results of most proteome experiments are long protein lists which may be difficult to interpret. To support biological data interpretation we already classified PBMC-derived proteins according to their cellular origin [3]. We identified proteins specifically occurring in single cell types as well as proteins shared between two or more cell types, and proteins common to many cell types [39]. Analyzing a complex biological sample and identifying therein proteins known to be selectively expressed may thus identify the corresponding cell type in the sample. Proteome signatures specific for cell types in a defined functional state could further support interpretation of complex data. In this study we therefore present reference proteome profiles of inflammatory activated white blood cells, including cell type-specific inflammation signatures.

PBMCs are the immediate players of inflammatory responses and mainly consist of lymphocytes and monocytes [47-49]. B-cells are producers of specific antibodies, while T-cells are important regulators as well as effectors of inflammatory processes. Monocytes are the most important partners of lymphocytes with a complex repertoire of cellular functions comprising phagocytosis, presentation of antigens and paracrine regulation of inflammation. In addition to the analysis of the bulk PBMC mixture we have included the selective analysis of the two most abundant constituents, i.e. purified T-cells and monocytes. B-cells and other leukocyte subtypes which are present in relatively much smaller amounts were not considered here.

For a robust and reliable assessment of function-related proteome alterations we applied two different analysis strategies, a ‘top down’ and ‘bottom up’ approach, in parallel [19]. 2D-PAGE is a well-established ‘top town’ technology providing accurate quantitative protein expression patterns. However, 2D-PAGE is limited with respect to the number of proteins accessible for quantification and a rather lab intensive technique which is hard to automatize [50]. However, by the application of metabolic labeling, a very sensitive measure for the induction of protein synthesis was achieved (Fig. 2). In order to identify more proteins and enable database-supported data interpretation, we complemented this approach with the ‘bottom up’ shotgun approach, our second analysis strategy.

Table 1 presents the newly synthesized proteins identified upon inflammatory activation of PBMCs using LPS and PHA. Corresponding gene ontology terms are also listed in order to give insights into known biological functions of the proteins. As can be seen, all induced proteins relate to known consequences of inflammatory activation such as leukocyte migration (4F2 cell-surface antigen heavy chain) [51], proliferation of T-cells (PCNA) [52], regulation of cell death (Proteasome activator complex subunit 3, Programmed cell death protein 5) [53,54], NF-κB signaling (NF-kappa-B p100 subunit) [55] and interferon response (interferon regulatory factor 4, interferon-induced guanylate-binding protein 2, interferon-induced 35 kDa protein, interferon-induced GTP-binding protein Mx1) [56-59]. As T-cells and monocytes have different biological tasks it is not surprising that these cells also display different responses to activation. T-cells induced proteins related to the induction of proliferation (PCNA) and several interferon-responsive proteins. Monocytes, which have no capability to proliferate, rather expressed proteins regulating cell death (programmed cell death protein 5) as well as proteins acting in a paracrine fashion (IL-1beta). By assembling these proteins according to their expression specificity we obtained inflammatory signatures of T-cells and monocytes. Knowledge of these specifically expressed proteins may reveal the involvement of inflammatory activated T-cells or monocytes in biological samples, which may strongly support the interpretation of complex clinical proteomics data when inflammatory activated PBMCs are involved.

The application of our standard proteome analysis procedure and data processing system enabled us to include previously published data into the present comparative analysis. Such cross-experimental comparisons may further support data validation and interpretation. To give an example: some subtle inflammatory activation of cells was evidenced in the untreated cells kept in culture for 8 h. In vitro cultivation without any treatment apparently induced small amounts of IL-1beta and NAMPT, indicating activation of monocytes (Fig. 4). This is the reason why we used directly processed cells as controls as we were not focusing on the specific effects of the inflammatory agonists but rather on the finally obtained cell states. This observation also demonstrates the great sensitivity of the PBMCs to cell manipulation ex vivo. Here we also present a comparison to previously published proteome profiles of primary monocyte-derived dendritic cells (DCs) [16]. While none of the inflammation signature members were identified in the immature DCs, eight members of the inflammation signatures were also identified as up-regulated in the inflammatory activated DCs. Remarkably, three members of group II, the signature of activated T-cells, were identified in activated DCs. This finding may be somewhat unexpected as dendritic cells are close relatives to monocytes. However, it is in line with the previously described observation that DCs express surface markers not found in monocytes but characteristic for T-cells during maturation in the thymus [60]. Considering activated DCs, which are not members of PBMCs, we found, amongst others, interferon lambda-1 and C-X-C motif chemokine 9 [16]. These proteins were not identified in the activated PBMCs and may thus represent members of an inflammatory signature of DCs.

It is our aim to extend these systematic analyses of functional signatures of different cell types eventually resulting in signatures which are highly specific for each single cell type and cell state. Such knowledge would support a fully automated assessment of biological proteome profiles with respect to the presence of certain cells in defined cell states. Such assessments could greatly support the identification or recognition of pathophysiologic pathways in individual samples. A recent paper presenting proteome profiles of human vulvar cancer samples investigated characteristic features of samples derived from patients suffering from early relapse of disease. Interestingly, the proteins correlating with such unfavorable clinical situation are all contained in our lists of proteins which were found induced or up-regulated in inflammatory activated PBMCs (Tables 1, 2). This finding suggests that the invasion of inflammatory activated leukocytes may have been the characteristic feature of tumor tissue derived from these early relapsing patients. Such recognized pathophysiologic events could then be specifically targeted by pharmacologic means. This is exactly what we intended to achieve with our approach: to provide means for researchers to interpret complex data with respect to functional aspects in order to create clear hypotheses which may then become verified subsequently.

Outlook

The assessment of cell type-specific activation states out of a proteome profile of a complex clinical sample should support different issues: to recognize involved pathologic mechanisms and to assess individual variations thereof. It may also help to monitor drug effects. It can be assumed that drug-induced down-regulation of inflammation will be accompanied by the down-regulation of members of the inflammatory signatures. Monitoring the expression rate of such proteins may therefore provide novel means in order to assess drug effects and drug efficiency in an individualized fashion.

The following are the supplementary data related to this article.

Supplementary figures

Supplementary figures

Supplementary Table S1

Proteins identified in untreated (con) and inflammatory activated (act) PBMCs by shotgun proteomics. Nine independent experiments out of six different donors for each group were pooled. The positive identification rates are listed for each protein for each group (pos-id; e.g. 3/9: three independent identifications out of nine experiments). Proteins were identified with at least two distinct peptides (indicated in the column “dist-pep”) in at least two experiments. Average exponentially modified Protein Abundance Index values (emPAI) were calculated for each protein. Proteins are grouped according to their expression within different cell types as described in the text.

Supplementary Table S2

List of PRIDE XML files submitted to the PRIDE database.