Proteomic Dissection of Agonist-Specific TLR-Mediated Inflammatory Responses on Macrophages at Subcellular Resolution Yan Xue,†,‡,# Dong Yun,†,# Alex Esmon,‡ Peng Zou,† Shuai Zuo,† Yanbao Yu,†,‡ Fuchu He,†,§ Pengyuan Yang,† and Xian Chen*,†,‡ Department of Chemistry and Institute of Biomedical Sciences, Fudan University, Shanghai, China, Department of Biochemistry and Biophysics, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, and State Key Laboratory of Proteomics, Beijing Proteome Research Center, Beijing Institute of Radiation Medicine, Beijing, China Received January 11, 2008
Upon stimulation by distinct bacterial/viral products/agonists, APCs including macrophages tend to express particular TLR molecules to coordinate the signaling that ultimately target at chromatin and mediate the activity of downstream transcriptional factors in regulating characteristic sets of gene expression for innate immune response. To investigate largely unknown regulatory mechanism underlying agonist-specific TLR-mediated innate immune responses, at subcellular resolution, we first analyzed Pam3CSK4-induced proteome changes in living macrophages and identified the differentially expressed proteins in the cytosol and chromatin-associated fractions, respectively, by using AACT/ SILAC-based quantitative proteomic approach. In the cytosol fraction, we found that the proteins with notable Pam3CSK4-induced expression changes were primarily involved in post-translational events, energy metabolism, protein transporting, and apoptosis. Among them, a ubiquitous and highly conserved iron-binding protein, Ferritin, was further characterized as a modulator for the expression of a TLR2-specific cytokine IL-10 in murine macrophage cells by using small-interfering RNA (siRNA). Interestingly, we simultaneously identified multiple apoptosis-related proteins showing opposite trend in their regulated expressions, which clearly indicated the existence of systems regulation in differentially modulating the signal for the cross-road balance between protecting cell from apoptosis and the apoptosis of infected cells. For those regulated proteins identified in the nuclear fraction, we integrated bioinformatics to find the interactions of certain chromatin-associated proteins, which suggested their interconnected involvements in proteasome-ubiquitin pathway, DNA replication, and post-translational activity upon Pam3CSK4 stimulation. Certain regulated proteins in our quantitative proteomic data set showed the similar trend of up-regulation in both Pam3CSK4- and LPS-stimulated macrophages (Nature 2007, 447, 972), suggesting their belonging to the recently identified class of pro-inflammatory genes. The regulatory discrepancy between both data sets for other set of genes indicated their agonist-specific nature in innate immune responses. Keywords: Toll-like receptor • innate immune responses • subcellular proteomics • macrophages • smallinterfering RNA (siRNA) • cytosol • chromatin-associated proteins • regulatory proteins • pathogenassociated molecular patterns (PAMPs) • apoptosis • proteasome-ubiquitin pathway
Introduction Different Toll-like receptors (TLRs) or their combinations recognizing pathogen-associated molecular patterns (PAMPs) play a crucial role in mediating specific innate immune response against invading microorganisms.1–6 In the TLR family, TLR2 recognizes the largest number of different ligands/ agonists identified to date including lipoproteins/lipopeptides (LP), peptidoglycan, glycophosphatidylinositol (GPI), zymosan * To whom correspondence should be addressed. Professor Xian Chen. E-mail:
[email protected]. Fax: 919-966-2852. † Fudan University. ‡ University of North Carolina at Chapel Hill. # Equal contribution to this work. § Beijing Institute of Radiation Medicine.
3180 Journal of Proteome Research 2008, 7, 3180–3193 Published on Web 06/24/2008
and certain LPS, and so forth.1,7 Abnormal expression of TLR2 has been correlated with a variety of pathological conditions, including glomerulonephritis, atherosclerosis, meningitis, allergic airway inflammation and acne.8–11 It has become evident that recognition of many known TLR2 ligands is mediated by heterodimers of TLR2/TLR1 or TLR2/TLR6, giving rise to stimulus/agonist-specific immune responses. Among them, Pam3CysSerLys4 (Pam3CSK4) is a potent activator of TLR2/ TLR1 which represents the N-terminal part of bacterial LP.12,13 It was characterized by the lipoamino acid N-acyl-S-diacylglyceryl cysteine where the lipid moiety of the LP and the peptide are essential for cell activation.14 This synthetic LP provides an important tool for studying the TLR2-specific immune recognition mechanism because it is free of other contaminat10.1021/pr800021a CCC: $40.75
2008 American Chemical Society
Agonist-Specific TLR-Mediated Inflammatory Responses
research articles
ing bacteria components. Previous studies indicate that synthetic Pam3CSK4 can effectively stimulate monocytes or macrophages, inducing the translocation of nuclear factor κB (NFκB) and the release of IL-1, IL-6, TNF-R and reactive oxygen species.15,16 However, little is documented about the global view of agonist-specific TLR-mediated protein expression changes that reflect the mechanisms underlying the regulatory control of proper production of defensive cytokines.
chromatin-associated compartment. This subcellular quantitative proteomics analysis can provide precise information not only about the agonist-induced expression changes of proteins, but also their corresponding locations as well as proteins translocation changes between cytosol and nuclei. In addition, from a technical point of view, the subcellular fractionation greatly reduces MS spectral complexity and increases the number and accuracy in the identification of proteins. Further, we integrated bioinformatics to examine the agonist-induced interconnected relationships in our subcellular quantitative proteomic data set in the context of functional category and pathways. Their involvements in regulatory mechanisms will be discussed accordingly.
It is of interest to note that nearly each signaling in response to the exterior stimulation can ultimately affect the activity of particular transcriptional factor(s) at the chromatin, the site of DNA replication and gene transcription.17 Many growth and differentiation signals ultimately target the chromatin to induce alteration in gene expression or DNA replication.18 The association of cellular proteins with chromatin can elicit diverse responses from the cell including transcriptional activation or repression, initiation or inhibition of DNA replication, or chromatin remodeling.19–24 Therefore, in addition to measure the cytosol proteomic changes, a systematic and precise proteomics study on regulatory chromatin proteins will be great help to further understand the biological processes of agonistinduced TLR-mediated innate immune response. To quantitatively measure the proteome variations resulting from various stresses on cells, different isotope labeling techniques have been developed for improving the sensitivity, accuracy, and high-throughput of mass spectrometry (MS)based comparative analysis.25,26 Representative methods include isotope-coded affinity tagging (ICAT), 18O-labeling, and uniform or amino acid-specific metabolic labeling.27–31 Basically, to substitute 2DE-based quantitative measurements of protein abundance changes which were proven to be lesssensitive, laborious, and low-reproducible, stable isotopes either enriched in chemical reagents or at particular amino acids can be utilized as the ‘in-spectra’ quantitative markers to assist MS in the analysis of protein expression changes on a large scale. Recently, an in vitro labeling technique named iTRAQ, which tags the primary amine group in peptide digests, has recently been introduced for quantitative analysis of proteomic changes in comparing up to four different cell populations.32,33 However, this peptide-based labeling strategy requests high consistency in preparation of individual samples involving subcellular fractionation, protein separation and proteolysis, accurate equal mixing of proteolytic digests, and so forth. In addition, LC-unsolved/coeluting peptide signals with different sequences can lead to overlapping signals in the low-mass reporter region on which the quantitative measurements are based. Metabolic labeling strategies such as amino acid coded mass-tagging (AACT)30 or SILAC naturally introduce the tags for quantitative measurements in MS spectra through cell culture. Importantly, the proteins extracted from different cell pools can be processed in a single experiment format after equal mixing of cell numbers or protein concentration; thus, the procedure minimizes the experimental variations, contaminations, and artifacts originated from separating sample processing.30,34,35 For our research purpose, the AACT/SILACbased strategy is particularly useful for subcellular proteome analysis due to its consistency through the course of subcellular fractionation on the paired cell population, for example, unstimulated versus stimulated macrophages in our case. Thereby, we applied AACT-based LC-MS/MS quantitative proteomic method to systematically analyze the Pam3CSK4induced differential protein expression profile in stimulated living macrophages at subcellular resolution, especially in the
Experimental Procedures Chemicals and Antibodies. The deuterium-labeled amino acids such as arginine-d7 and lysine-d4 were obtained from Cambridge Isotope (Andover, MA). All components of cell culture media as well as CHCA and protease inhibitor cocktail were obtained from Sigma (St, Louis, MO) except for fetal bovine serum (FBS) which was obtained from PAA Laboratories GmbH (Linz, Austria). The dialyzed FBS was purchased from Invitrogen. Pam3CSK4 was purchased from Invivogen (San Diego, CA); Trypsin was purchased from Promega (Madison, WI). All the chemicals were sequence- or HPLC-grade unless specifically mentioned. The antibodies against HSP70 and RS6 were obtained from Cell Signaling Technology (#4872, #2217); antibodies to HSP27, hnRNP A2/B1 and actin were purchased from Santa Cruz Biotechnology (sc-13132, sc-10035, sc-8432); HSP90 beta was purchased from Lab Vision Corporation (# RB-118-P0); GAPDH antibody was purchased from ChangChen Bio-tech. (KC-5G4); 14-3-3 epsilon antibody was purchased from Chemicon International, Inc. (AB1671); NF-κB p100 was purchased from Satbio, Inc. (#21016); FHC antibody was purchased from Abcam (ab16875). Cell Lines and Stimulation. The macrophage cell line AMJ2C8 (from ATCC, CRL-2455) was cultured in high-glucose DMEM supplemented with 10% FBS, 100 units/mL penicillin and 100 µg/mL streptomycin, and incubated at 37 °C in a 5% CO2 atmosphere. Labeled AMJ2-C8 cells were grown in arginine-d7 and lysine-d4 labeled medium, in which only the labeled arginine and labeled lysine replaced the normal components. Unlabeled AMJ2-C8 cells were stimulated by Pam3CSK4 (100 ng/mL) for 6 h, while the labeled cells were unstimulated for the same period of time. Cells were harvested and washed with ice-cold PBS twice. The soluble proteins and chromatinassociated proteins were prepared as previously described.36,37 Briefly, similar amount of the stimulated and unstimulated cells were suspended at 4 × 107 cells/mL in buffer A (10 mM Hepes at pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10% glycerol, and protease inhibitors). Triton X-100 was added to 0.1%, the cells were mixed and incubated on ice for 8 min, and the nuclei (P1 fraction) was collected by centrifugation (5 min, 1300g, 4 °C). The supernatant (S1 fraction) was isolated and the P1 nuclei were washed once in buffer A and centrifuged again. The obtained precipitation or P2 fraction was lysed for 30 min in buffer B (3 mM EDTA, 0.2 mM EGTA, and protease inhibitors) on ice. Insoluble chromatin (fraction P3) and the soluble material (fraction S3) were separated by centrifugation at 1700g for 5 min at 4 °C. The chromatin-associated proteins were precipitated via acetone/TCA and resolved in the sample buffer (250 mM Tris-HCl at pH 6.8, 8% (w/v) SDS, 40% (v/v) Journal of Proteome Research • Vol. 7, No. 8, 2008 3181
research articles
Xue et al.
Figure 1. The stepwise extraction procedure and SDS-PAGE gel of S1 and P3 fractions. (a) The extraction procedure and workfow for S1 and P3. The supernatant fraction (S1) and precipitated fraction (P3) were separated and identified by LC-MALDI-TOF/TOF. (b) SDSPAGE gel of S1 and P3. The same amount of Pam3CSK4-stimulated and -unstimulated cells were lysed, mixed at a ratio of 1:1 and separated on 12% polyacrylamide gels stained with CBB-R250. P3 was precipitated with TCA-acetone.
glycerol, 0.01% (w/v) bromphenol blue dye, and 200 mM β-mercaptoethanol). As the procedure in Figure 1a shows, the protein concentration of the S1 and P3 fractions was determined by Bradford Assay and it confirmed the 1:1 ratio of stimulated cells versus unstimulated cells. The fractions S1 and P3 were separated by 12% SDS-PAGE gel and stained using Coomassie Brilliant Blue (CBB) (Figure 1b). In-Gel Trypsin Digestion. The SDS-PAGE bands were cut into 30 sections. The gel pieces were washed once with MilliQ water, and the CBB dye was removed by three rinses in 50% acetonitrile (ACN) and 50 mM ammonium bicarbonate. Gel pieces were dehydrated twice in 100% ACN for 30 min and reconstituted with an in-gel digestion reagent containing 10 ng/µL sequence- grade trypsin overnight at 37 °C. The digested peptides were extracted with 50% ACN and 0.1% TFA from the gel pieces and lyophilized for 6 h. HPLC Separation. Tryptic peptides were resuspended in 10 µL of 0.1% TFA solution and separated using an Agilent 1100 reverse-phase HPLC with a 300 µm × 100 mm C18 column (Grace Vydac, Hesperia, CA). The flow rate was at 4 µL/min, and the mobile phase A was 0.1% formic acid and B was 0.1% formic acid in ACN. The following gradient was used for peptide separation: 0-30% B from 0 to 8 min; 30-75% B from 8 to 55 min; 75-95% B from 55 to 60 min; 95-0% B from 60 to 80 min.The eluted peptides were directly deposited on an Applied Biosystems plate at 30 s intervals in each pore. MALDI-TOF/TOF Analysis. All mass spectra were obtained on a MALDI-TOF/TOF mass spectrometer (4700 proteomics Analyzer, Applied Biosystems, Framingham, MA). The CHCA was used as matrix with a concentration of 3 mg/mL in 50% ACN/0.1% TFA. Before the sample acquisition, six calibrated spots were used for signal and parameter optimization. During MS/MS analysis, air was used as the collision gas and collision voltage was at 1 kV. Spectra were obtained by accumulation of 2000-3000 consecutive laser shots. For one spot, MS/MS 3182
Journal of Proteome Research • Vol. 7, No. 8, 2008
was automatically performed on the seven most intensive peaks observed in the MS scanning. Data Process, Protein Identification, and Quantitative Analysis. Raw data was deisotoped and processed with signal/ noise threshold (S/N > 20 for MS/MS) to obtain peak lists prior to database search by GPS Explorer Software (version 2.0, Applied Biosystems, MA). Peak density in each peak list was restrained to 20 peaks per 200 Da, and the max number of peaks in each peak list was 100. Protein identification was also performed by GPS Explorer (version 2.0, Applied Biosystems, MA), which was integrated with MASCOT search engine (v.1.9.05). All obtained peak lists were searched against Swiss-Prot mouse database (released in 2005.07). Strict parameters were adopted to ensure the confidence of identification as follows: the mass error tolerance was limited within 100 ppm for peptide and 0.6 Da for MS/MS tolerance; monoisotopic search mode; allowing one missed cleavage of trypsin; and arginine-d7 and lysine-d4 were specified as potential modifications. Synchronously, reverse sequence database (RDB) derived from Swiss-Prot mouse database was also searched by MASCOT as a “decoy” database with the same parameters for further investigations on false positive results. The quantitative measurement was accomplished by averaging all the TOF-MS spectra of both light and heavy forms of the AACT-containing peptides and compared their monoisotope peak intensities. The standards of selected peptides for identification and quantification were described previously.38 Briefly, only those peptides containing one arginine, lysine, or both were used to quantify the proteins and obvious pair peaks could be seen in the spectra. The quantitative ratio of stimulation versus unstimulation was obtained by comparing corresponding pair peak intensity. Construction of P3 Proteins Network. The network was drawn with Pajek software (http://vlado.fmf.uni-lj.si/pub/ networks/pajek/) using the following steps. First, we searched
research articles
Agonist-Specific TLR-Mediated Inflammatory Responses the IntACT (http://www.ebi.ac.uk/intact) database with the Swiss-Prot accession numbers and saved the Web search results as text format, then used PERL language to compile the program, utilizing the program to read the saved text files and giving dots and lines to the image definition using protein names as index, and then saved all the results. We converted these saved interaction data into Pajek software format (*.net). Finally, protein-protein interaction network were created by Pajek software. Western Blotting (WB). For the differentially regulated proteins from S1 and P3, we chose some of them related to the inflammation, signal transduction, and metabolism for WB to confirm our results. The procedure consisted of the following: the fraction S1 and P3 proteins (from stimulated and unstimulated cells) were separated by SDS-PAGE, transferred onto PVDF membranes at 15 V for 45 min using semidry transferring device, blocked with 5% nonfat milk in TBS with 0.05% Tween-20 for 1 h at room temperature, washed twice with TBST and then incubated with primary antibodies dissolved in block solution for 1 h. After washing the membranes three times for 10 min in TBST, they were followed by reactions with horseradish peroxidase-linked secondary antibodies for 1 h. After three washes, the proteins were detected by luminol detection reagent and imaged with LAS-3000 machine (FUJIFILM, Japan). Nucleofection of AMJ2 Cells and Nucleic Acid Isolation. AMJ2 murine macrophage cells were grown to 90% confluency in DMEM supplemented with 10% FBS and 1× penicillin/ streptomycin in 12-well plates. Two days prior to nucleofection, macrophage cells were passed to fresh media. Cells were nucleofected with Ferritin Heavy Chain siRNA (mouse specific) from Santa Cruz Biotech, scrambled control siRNA from Invitrogen or the positive control plasmid pmaxGFP from Amaxa using a Nucleofector II device and the nucleofector kit V (Amaxa). We found that AMJ2 cells showed highest positive nucleofection using the program V-001. Cells were harvested, nucleofected and maintained postnucleofection according to the manufacturer’s guidelines (Amaxa, Inc., Gaithersburg, MD). Nucleofection efficiency was established using pmaxGFP. Twenty-four hours postnucleofection, cells were either harvested for RNA isolation or stimulated with TLR2-specific ligand Pam3CSK4 (Invivogen) to a final concentration of 1 µM in solution for 1 or 6 h, after which time the stimulated cells were harvested and RNA was isolated for cDNA synthesis and subsequent gene expression analyses. RNA isolations were performed using the RNeasy kit (Qiagen) according to manufacturer’s guidelines. cDNA synthesis was performed using the SuperScript III first strand synthesis kit (Invitrogen). Quantitative Real-Time PCR. qPCR was performed using an ABI 7300 (Applied Biosystems). All gene expression experiments were done in triplicate. All assays were performed using 2× Taqman Gene Expression Master Mix and mouse-specific Taqman Gene Expression Assays (Applied Biosystems) for the following genes: FHC (to measure the degree of gene knockdown), GAPDH (as a normalizing control), IL-10 and TNFR. Ct values were normalized against GAPDH. Comparisons were made against cells nucleofected with nontargeting scrambled control siRNA duplexes (Invitrogen) and nontransfected controls. For verification of knockdown and percent of gene expression remaining, we used a standard ∆Ct equation. All other expression measurements were made using the pfaffl equation.
Figure 2. The hydrophobicity (HP) and PI distributions of the identified proteins in S1 fraction. Left figure shows the relation between identified protein number and proteins’ HP, most of the proteins are hydrophilic; right figure shows the relation between the protein number and proteins’ PI.
Results and Discussion Subcellular Identification and Quantitative Analysis of the Proteins Differentially Expressed in the Paired Unstimulated versus Pam3CSK4-Stimulated Macrophages. As illustrated in Figure 1a, we stimulated the macrophage cells growing in regular DMEM medium with Pam3CSK4 for 6 h. Because all tryptic peptides have either lysine or arginine at their C terminus, to ensure all detectable peptides containing AACT for quantitative measurements of changes of their abundances, the unstimulated/control cells were maintained in the “heavy” DMEM containing both d7-labeled arginine and d4-labeled lysine. Proteins were extracted following subcellular fractionation in a stepwise manner (Figure 1a). The subcellular fractionation was performed on the 1:1 mixture of stimulated or ‘light’ and unstimulated or ‘heavy’ macrophage cells and two major subcellular fractions, that is, S1 for cytosol and P3 for nuclear fraction, were obtained. To determine the purity of subcellular fractionation, we first examined the identities of the proteins found in S1 and P3 fractions, respectively. As shown in Figure 1b, in the P3 fraction, a majority of the identified proteins were histone (H1, H2A, H2B, H3, H4) or high mobility group (HMG) proteins which were known to be chromatin-associated proteins, while they were absent in the S1 fraction. Soluble proteins such as actin, tubulin, and GAPDH were predominantly detected in the S1 fraction which contained mostly soluble proteins based on the statistic analysis according to the GRAVY value (as hydrophobicity, HP) and PI values of the identified proteins (Figure 2). All of these results support the validity of subcellular fractionation procedure we used. Following the experimental workflow as shown in Figure 1a, a total of 1171 nonredundant proteins were identified according to the original MASCOT score threshold. Noting the great potentiality of false-positive assignment in such unfurtherrestrained data set, the rigorous score threshold from decoy database described above was applied to define the positive matches.39,40 Depending on the proportion of forward and Journal of Proteome Research • Vol. 7, No. 8, 2008 3183
research articles
Figure 3. The quantified proteins were analyzed by their location, known function and biological process involved according to MGA_go-slim rules.
reverse matches, the MASCOT score of 28 and the ranking of the first position in the candidates were assigned as the thresholds of a confident peptide identification with p ) 0.05. The positively identified proteins should contain at least one confident peptide sequence over such threshold. Synchronously, detectable in-spectra AACT quantitative tags were also required. Because of the complexity resulted from the monoisotopic distribution of multiple AACT in one peptide, most of peptides containing more than two arginines or lysines were excluded from AACT-based quantitative measurements. According to these criteria, 540 proteins were identified in high confidence. It should be noticed that most of proteins matched with one peptide in our list were also supported by other MS/MS spectra with relative low scores. While there are some debates in the literature concerning the soundness of such ‘one-hit wonders’,41,42 it suggests that proteins identified here by a single confirmed peptide cannot simply be discarded if the poor-quality MS/MS spectra of related peptides exist. Certainly, if any proteins of interest, for example, potential drug targets or biomarkers, are present as one-hit wonders, the protein’s identity can be further verified by ‘reverse proteomics’ such as Western blotting. Through the subcellular location and functional analysis, as shown in Figure 3, we observe that the identified proteins are widely spread in various subcellular compartments, especially in membrane, cytoplasm, and nucleus, participating in different cell biological processes. Because of the limited sensitivity of mass spectrometry and low abundance of some signal proteins, most of the identified proteins are mainly engaged in the cellular organization and biogenesis or metabolism. For precise identification of differentially expressed proteins, we set the cutoff threshold at 20% changes in intensity ratio of any isotope pair, that is, isotope ratio >1.2 or
