Extracellular Hemoglobin Polarizes the Macrophage Proteome toward

15 Mar 2011 - Figure 9B summarizes the mRNA expression levels of HLA class 1 and 2 ... Suppression of HLA class 2 protein and mRNA by Hb:Hp treatment...
0 downloads 0 Views 4MB Size
ARTICLE pubs.acs.org/jpr

Extracellular Hemoglobin Polarizes the Macrophage Proteome toward Hb-Clearance, Enhanced Antioxidant Capacity and Suppressed HLA Class 2 Expression Theresa Kaempfer,† Elena Duerst,† Peter Gehrig,‡ Bernd Roschitzki,‡ Dorothea Rutishauser,‡ Jonas Grossmann,‡ Gabriele Schoedon,† Florence Vallelian,§ and Dominik J. Schaer*,†,§ †

Internal Medicine, University Hospital, Zurich, Switzerland Functional Genomics Center University and ETH Zurich, Switzerland § Center for Evolutionary Medicine (ZEM), University of Zurich, Switzerland ‡

bS Supporting Information ABSTRACT: Peripheral blood monocytes and macrophages are the only cell population with a proven hemoglobin (Hb) clearance capacity through the CD163 scavenger receptor pathway. Hb detoxification and related adaptive cellular responses are assumed to be essential processes to maintaining tissue homeostasis and promoting wound healing in injured tissues. Using a dual platform mass spectrometry analysis with MALDI-TOF/TOF and LTQ-Orbitrap instruments combined with isobaric tag for relative and absolute quantitation (iTRAQ), we analyzed how Hb exposure could modulate the macrophage phenotype on a proteome level. We identified and relatively quantified 3691 macrophage proteins, representing the largest human macrophage proteome published to date. The Hb polarized macrophage phenotype was characterized by an induced Hb:Hp-CD163-HO1ferritin pathway and enhanced antioxidant enzymes while suppression of HLA class 2 was the most prominent effect. Enhanced Hb clearance and antioxidant capacity together with reduced antigen presentation might therefore be essential qualities of Hb polarized macrophages in wounded tissues and hemorrhage or atherosclerotic plaques. KEYWORDS: monocyte, macrophage, hemoglobin, haptoglobin, HLA class 2, oxidative stress, proteomics

’ INTRODUCTION Hemoglobin (Hb) is sequestered within red blood cells (RBC) under physiologic conditions. Hemoglobin is released from the protected RBC environment only under certain conditions, such as during (i) hemolytic diseases, (ii) blood extravasation in wounded tissues, (iii) hemorrhagic stroke, or (iv) atherosclerotic intraplaque hemorrhage. As a result of heme’s strong reactivity with physiologic oxidants, Hb has been assumed to act as a pro-oxidant and potential pro-inflammatory mediator in extracellular environments.1 Therefore, the clearance of Hb from extracellular sites is essential, in parallel to adaptive cellular responses that are tailored to attenuate the potentially damaging effects of Hb.2 Several mechanisms have evolved to safeguard cells and tissues against systemic or local Hb toxicity. For example, haptoglobin (Hp) is the primary Hb scavenger in plasma that binds free Hb in an irreversible complex.3 Hp slows the release of heme, prevents the formation of covalent protein cross-link products, and protects the structural integrity of globin amino acids, even during very severe oxidative impacts. When bound to Hp, Hb remains sequestered in the intravascular space, and peroxidative processes are confined within the complex, thus shielding the environment from oxidative damage.4 Knowledge remains r 2011 American Chemical Society

limited about the function of Hp in extracellular spaces, such as in a wound where multiple cell types may express and secrete Hp.5 However, it is known that free Hb and the Hb:Hp complex are eventually endocytosed and degraded by circulating blood monocytes and macrophages via the CD163 scavenger receptor.69 Short-term Hb:Hp exposure of CD163þ monocytes induces an antioxidant gene expression profile and enhances the recycling capacity of iron.10,11 Circulating blood monocytes aggregate in the vicinity of injured tissues within a short time of an inflammatory or mechanic impact, and start to differentiate into mature tissue macrophages.12 It is mainly the tissue environment (i.e., secreted cytokines and growth factors) that regulates whether monocytes polarize into either pro-inflammatory phagocytes (classically activated or M1 macrophages) that fight invading pathogens or into a noninflammatory cell type that promotes wound healing (alternatively activated M2 macrophages).1315 Since large amounts of Hb are released into damaged tissues, Hb itself may act as an intrinsic hemoylsis and/or tissue damage signal with the potential to initiate polarization of macrophage phenotypes and associated functions in wounded tissues. Received: December 10, 2010 Published: March 15, 2011 2397

dx.doi.org/10.1021/pr101230y | J. Proteome Res. 2011, 10, 2397–2408

Journal of Proteome Research In this study, we investigated how the differentiation of monocyte derived macrophages is modulated by extracellular Hb:Hp exposure at a proteome level. We adopted a gunshot proteome exploratory strategy to obtain a comprehensive overview of the macrophage proteome, and the quantitative changes in protein abundance resulting from Hb exposure. We combined iTRAQ based protein quantification with the dual-analysis of identical samples by two complementary mass spectrometry platforms (MALDI-TOF/TOF and LCESIMS/MS on a LTQ-Orbitrap). For this reason, we report the most comprehensive macrophage proteome coverage to date. We found that continuous Hb exposure polarized the macrophage proteome toward a phenotype with high Hb clearance and antioxidative capacity that is uniquely characterized by profoundly suppressed HLA class 2 expression.

’ MATERIALS AND METHODS Cell Culture and Experimental Conditions

Human monocytes were prepared from heparinized whole blood (100 U/mL) of different healthy volunteers as described16 and cultured in Iscove’s modified Dulbecco’s medium (IMDM; GIBCO Europe), which was supplemented with 10% autologous human serum (Hb in serum 99%. False discovery rates (FDR) were estimated by searches against a corresponding database containing forward and reversed sequences, according to the method of Kall. Further analysis of proteomics data was performed using MATLAB (version 7.7.0, The Mathworks Inc., Bern, Switzerland). Statistics

Data are expressed as mean ( standard error of mean. Significance was calculated by the t-test or one-way ANOVA. (Graphpad Prism, version 4.0.).

2399

dx.doi.org/10.1021/pr101230y |J. Proteome Res. 2011, 10, 2397–2408

Journal of Proteome Research

ARTICLE

Figure 2. Estimated false discovery rates (FDR) as a function of the protein-score (x-axis) of all proteins identified by (A) MALDI-TOFTOF and (B) LTQ-Orbitrap. The function cfun(x) = a*exp(b*x) þ c*exp(d*x) (red line) was fitted to the data curve by MATLAB. The red dotted line indicates the FDR threshold at 5%.

Flow Cytometric Analysis

Human-blood derived monocytes were prepared from buffy coats of healthy blood donors (Swiss Red Cross, Schlieren, Switzerland) that were seeded into 12-well tissue culture dishes, and stimulated with 2 mg/mL Hb-Hp. At day 7 of culture, cells were harvested by scraping and washed three times with PBS/ 0.1% BSA. Cells were then stained with HLA-DR, CD14, CD86 and CD163 antibodies. Mouse IgG1 κ  APC (BD Pharmingen), Mouse IgG1 κ  PE (BD Pharmingen) and Mouse IgG1 κ  FITC (BD Pharmingen) were used for isotype controls.18 At least 10 000 events per sample were acquired by FACSCanto scan flow cytometer (BD Biosciences, San Jose, CA) using FACS Diva software. Raw data were analyzed using FlowJo 7.2.5 software, and geometric mean fluorescence intensity was normalized to the corresponding isotype control.

’ RESULTS AND DISCUSSION We cultured human peripheral blood derived monocytes in Hb:Hp enriched (2 mg/mL) medium for 10 days to replicate conditions that may occur within damaged tissues. In our experiments, we used Hb:Hp complexes instead of free Hb to avoid uncontrolled oxidative side reactions that may eventually arise in cell culture (Figure 1).19 Proteomic Profiling of Monocyte-Derived Macrophages

To resolve changes in the proteome composition in Hb:Hp exposed macrophages, we analyzed iTRAQ labeled peptides by two complementary mass spectrometers. We expected that the different ionization and detection techniques employed by the two platforms may finally enhance the proteome coverage of our studies. As a first step, to validate the identification of proteins derived from raw spectral data analysis, we estimated theoretical false discovery rates (FDR) as a function of relative Mascot protein ID scores according to the method of Kall,20 and fitted a FDR function cfun(x) = a*exp(b*x) þ c*exp(d*x) to each data set. On the basis of these calculations, we applied a reasonable FDR threshold at 5%, and only identifications above this cutoff point were included for further analysis (Figure 2). Additionally, we only included proteins with at least two unique peptide identifications. In a combined analysis of all the data generated by the two platforms in six biological experiments (three donors, with two

Figure 3. PANTHER GO “cellular compartment” classification of the identified macrophage proteome; 2109 protein IDs (identifier: gene symbol) were classified.

experiments per donor), we were able to identify and provide relative quantification for a total of 3691 macrophage proteins across all conditions, and at the stringent level detailed above. From these 3691 hits, 1449 proteins occurred in both the MALDI-TOF/TOF and Orbitrap data sets. Remarkably, 1047 and 1593 proteins were exclusively identified with only MALDITOF/TOF or LTQ-Orbitrap respectively, indicating that the combined approach significantly enhanced the protein identification yield. To our knowledge, this study presents the most comprehensive proteome coverage of human monocytes or macrophages reported to date.2123 Earlier studies using 2-D gel electrophoresis (2DE) with subsequent LCMS/MS analysis presented only a small fraction of the monocyte proteome, primarily due to the limited resolution of complex proteomes in the gels.24 A recent study by Zhang and co-workers25 identified 1651 unique macrophage proteins that were obtained from shotgun proteomics using two labeling methods (iTRAQ and SILAC) with LCMS/MS analysis. However, we were able to identify and quantify about three times more proteins by using iTRAQ labeling, while avoiding time-consuming subcellular prefractionation through combining two complementary mass spectrometry technologies instead. The identified proteins were compared to a reference list (19911 Homo sapiens genes) to statistically determine the over- and under-representation of the macrophage proteome in 2400

dx.doi.org/10.1021/pr101230y |J. Proteome Res. 2011, 10, 2397–2408

Journal of Proteome Research

ARTICLE

Figure 4. (A) Global representation of the pooled protein IDs from six independent experiments performed with monocytes from three healthy donors and identified by either MALDI-TOF/TOF (n = 2203) or LTQ-Orbitrap (n = 2858). The proteins are sorted along the x-axis by their median ratio (Hb: Hp treatment versus control). Ratio data of the normalized and merged MALDI-TOF/TOF plus LTQ-Orbitrap protein lists are shown in the Masterlist. Red crosses indicate significantly up- or down-regulated proteins (p < 0.005), black crosses indicate unchanged proteins. The red dotted lines illustrate the 1.5-fold deviation threshold. (B) Correlation plot of the significantly regulated (fold change >1.5, p < 0.005) proteins (n = 99) identified by both MALDI-TOF and LTQ Orbitrap. The normalized median values for each protein and the respective linear fit are shown. Only proteins with a FDR < 5% are shown.

relation to the PANTHER “Cellular Compartment” classification categories www.pantherdb.org. All cellular compartments, except nuclei and extracellular proteins, are represented with a slight under-representation of plasma membrane proteins (Table 2 and Figure 3). iTRAQ-based Protein Quantification Reveals a Distinct Macrophage Phenotype Polarization Induced by Hb:Hp Exposure

Protein abundance ratios of all identified proteins were calculated based on the extracted iTRAQ tag signal intensities. Since the linear range of MALDI-TOFTOF and LTQ Orbitrap quantifications are not identical, the range of protein regulations

differed in the two data sets. Therefore, we transformed the protein ratios into log10-values and normalized them according to the median (m): normalized log(10) ratios (y) = [log(10) ratios/(std(m))] (Figure 4). Following normalization, we performed a t-test on the log10(Hb:Hp-treatment/control) ratios against the mean value of the log10(control/control) ratios. A total of 835 proteins, with a p value below 0.005, were considered to be significantly altered by Hb:Hp treatment, when compared to control conditions. After applying a second threshold of >1.5 fold change, 104 and 102 proteins were identified as significantly up- or down-regulated, respectively (Table 1 and Figure 5). Interestingly, of these 206 proteins, only 28 were identified and quantified by MALDI and Orbitrap analysis, suggesting that the 2401

dx.doi.org/10.1021/pr101230y |J. Proteome Res. 2011, 10, 2397–2408

Journal of Proteome Research

ARTICLE

Table 1. Number of Identified Proteins with MALDI-TOF/TOF and LTQ-Orbitrap data set

total no. of identified IDsa

MALDI þ Orbitrapc

1370 (37%)

99

17

11

MALDI

2203 (60%)

365

68

35

2858 (77%)

582

68

86

3691 (100%)

835

104

102

Orbitrap Combined

no. of hits with p < 0.005b

no. of up-regulated proteins (>1.5)

no. of down-regulated proteins (