Article pubs.acs.org/jpr
Establishing a Proteomics-Based Monocyte Assay To Assess Differential Innate Immune Activation Responses Nataliya K. Tarasova,† A. Jimmy Ytterberg,†,‡ Karin Lundberg,‡ Xing-Mei Zhang,§,∥ Robert A. Harris,§,∥,⊥ and Roman A. Zubarev*,†,⊥ †
Department of Medical Biochemistry and Biophysics, ‡Department of Medicine, Solna, and §Department of Clinical Neuroscience, Karolinska Institutet, SE 17177 Stockholm, Sweden ∥ Department of Clinical Neuroscience, Centre for Molecular Medicine, Karolinska Hospital, Karolinska Institutet, SE 17176 Stockholm, Sweden S Supporting Information *
ABSTRACT: Innate immune cells are complex systems that can be simultaneously activated in a variety of ways. Common methods currently used to estimate the response of innate immune cells to stimuli are usually biased toward a single mode of activation. The aim of this study was to assess the possibility of designing an assay based on unbiased proteome analysis that would be capable of predicting the complex response of the innate immune system to various challenges. Monocytes were used as representative cells of the innate immune system. The underlying hypothesis was that their proteome response to different activating molecules would reflect the immunogenicity of these molecules. To identify the main modes of response, we treated the human monocytic THP-1 cell line with nine different stimuli. Differentiation and activation were determined to be the two major modes of monocyte response, with PMA causing the strongest differentiation and Pam3CSK4 causing the strongest proinflammatory activation. The established assay was applied to characterize the monocyte response to epidermal growth factor peptide containing isoaspartate, which induced differentiation but not proinflammatory activation. Because of its versatility, robustness, and specificity, this new assay is likely to find a niche among the more established immunological methods. KEYWORDS: label-free proteomics, THP-1 cell line, TLR ligands, activation, differentiation
■
INTRODUCTION Every day the human organism is exposed to thousands of different substances and microorganisms as well as to variations in environmental conditions. The organism adapts to some conditions and foreign entities while removing anything that appears dangerous. The question is how the human organism decides which of these substances, particles, and microorganisms are dangerous and have to be removed. The system most likely to be the first to detect and respond to a foreign object is the immune system, whose main function is defense of the organism.1 It is a complicated and finely tuned system consisting of a network of biological entities, including molecules, molecular complexes, and cells that interact with each other during complex processes. The immune system can be generally divided into “innate” and “adaptive” arms.1 The innate immune system provides immediate host defense. It includes granulocytes, monocytes, macrophages, dendritic cells, complement proteins, cytokines, and acute phase proteins.1 These entities, upon detecting danger, are able to initiate a system-wide response such as inflammation.2 Altogether this implies that the response of the innate immune cells to a © XXXX American Chemical Society
particular compound or particle may in many cases predict the response of immune system and even the whole organism. Innate immune cells are known to recognize and respond to a large number of molecular patterns that can be generally divided into two groups: PAMPs (pathogen associated molecular patterns) and DAMPs (danger/damage associated molecular patterns). PAMPs and DAMPs are recognized by pattern recognition receptors (PRRs) on the surface of or inside cells.3 Toll-like receptors (TLRs) represent one of the PRR families, detecting molecular signatures that are conserved among a large group of pathogenic microorganisms.4−6 A total of 10 TLR members have been identified in humans (TLR1TLR10) and 12 in mice (TLR1-TLR9, TLR11-TLR13).7 TLRs are located both on the cell surface (TLR1, TLR2, and TLR4TLR6) and inside of cells.8 Each TLR can recognize a particular class of PAMPs (Table 1). Ligand binding to TLRs initiates one of the two major signaling cascades depending on the TollInterleukin 1 Receptor (TIR) domain-containing adaptors Received: May 9, 2016
A
DOI: 10.1021/acs.jproteome.6b00422 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
multivariate analysis. A human monocytic cell line THP-127 (THP-1 cells) was employed to establish the assay. This cell line is one of the most common and well-suited models for monocytes and monocyte-macrophage differentiation.28−31 The most important applications of the assay are projected to be the situations in which the monocyte response is not very strong and conventional methods are insufficiently sensitive enough to detect it. This could be the case when the stimulus concentration is low or when the molecule has borderline immunogenicity. The assay could also be employed for molecules that are hypothesized to trigger autoimmune or neurodegenerative diseases to test their initiation of innate immune cascades. A traditional view is that dysregulation of the adaptive immune system is the main cause of autoimmune diseases, while that of the innate immune system is the main cause of chronic inflammation;32−34 however, it has recently been reported that the innate immune system is also able to facilitate autoimmune diseases.32,33,35,36 The agents that can potentially trigger or enhance the innate immune system response include damaged or modified proteins. For instance, it is believed that citrullination (deimidation of arginine) is involved in triggering rheumatoid arthritis.37−39 Herein, we tested our monocyte assay with another posttranslational modification (PTM). The formation of isoaspartyl peptide bonds (IsoD) is one of the most common forms of nonenzymatic PTM of peptides and proteins under physiological conditions. IsoD build-up can decrease the biological activity of a protein pharmaceutical, alter its susceptibility to proteolytic degradation, and elicit autoimmunity.40 We recently demonstrated the link between the IsoD levels and Alzheimer’s disease (AD).41 In the current study we applied the developed assay to evaluate whether IsoD-containing peptides could trigger a response in monocytes. The presence of a monocytic response would be consistent with the inflammation proposed to be associated with AD.42
Table 1. Human Toll-Like Receptors (TLRs) and Their Ligands TLR
ligand(s)
TLR1 TLR2
triacyl lipopeptides zymosan, peptidoglycan, lipopeptides, lipoteichoic acid, lipoarabinomannan, GPI anchors, phenol-soluble modulin, glycolipids double-stranded RNA (dsRNA) lipopolysaccharide, taxol, RSV fusion protein, MMTV envelope protein, endogenous ligands (HSPs, fibronectin, hyaluronic acid) flagellin diacyl lipopeptides single-stranded RNA (ssRNA), imidazoquinolines single-stranded RNA (ssRNA), imidazoquinolines CpG DNA unknown
TLR3 TLR4 TLR5 TLR6 TLR7 TLR8 TLR9 TLR10
MyD88 or TRIF. Via all TLRs, except for TLR3, MyD88 mediates the activation of NF-κB, leading to the induction of inflammatory cytokine genes.9 TRIF mediates TLR3- and TLR4-dependent activation of IRF3 and NF-κB, inducing IFNβ production.7 Engulfed foreign molecular patterns (antigens) can subsequently be digested in lysosomes and presented on the surface to cells of the adaptive immune system.10,11 The cells that are capable of presenting antigens on their surface are termed antigen-presenting cells (APCs). Classical APCs include macrophages, dendritic cells, Langerhans cells (dendritic cells of the epidermis),12 and B lymphocytes. Macrophages located in tissues are the first cells responding to the invading foreign entities or to damage.13 These resident macrophages originate from monocytes in the yolk sac during embryonic development and self-replenish in the tissue.14 The main function of resident macrophages is to maintain homeostasis. In contrast, tissue clearance and inflammatory responses are mainly performed by macrophages derived from blood monocytes that upon stimulation leave the bloodstream, migrate into tissue, and differentiate into macrophages.14 Monocytes thus act as “immunological orchestrators”,15 and monocyte responses to different stimuli can in many cases reflect the innate immune responses and even predict the response of the whole organism. The common methods used to characterize monocyte responses include ELISA,16,17 PCR,18,19 and flow cytometry.20,21 These methods are limited in scope of analysis and in general are biased toward known response types and markers of cell activation: cytokine or ROS production, changes in surface markers, or the presence of particular proteins in the cell; however, monocytes, as well as other innate immune cells, are capable of inducing multiple types of activation simultaneously. For instance, the major component of Gram-negative bacterial membrane, lipopolysaccharide (LPS), induces ROS and NO production, cytokine release, and upregulation of number of surface markers and receptors (e.g., CD14, ICAM-1, CD11b, and TLR4) in monocytes.22−26 An unbiased proteomics analysis assessing the abundance change of thousands of proteins may therefore have a better chance to identify and quantify the response of monocytes to a variety of stimuli. Our main hypothesis herein was that the proteome responses of monocytes to different activating molecules would reflect the relative immunogenicities of these molecules. We tested the above hypothesis by creating a prototype of a proteomics-based monocyte assay. To assess the complex modality of the innate immune response, we employed
■
METHODS
Cell Culture, Treatment, and Counting
The human monocytic cell line THP-1 (ATCC) was cultured in RPMI-1640 medium supplemented with 10% heatinactivated fetal bovine serum (Biochrome, Merck Millipore), 0.05 mM mercaptoethanol, 2 mM L-glutamine, and 100 U/mL penicillin−streptomycin mixture. Treatments and their concentrations are described in Table 2. Details of the experiments performed with the cells are summarized in Table S-1, including time of incubation, number of replicates, and initial number of cells used. All experiments were performed in 24well plates (Sarstedt). Nonadherent THP-1 cells were used at the initiation of all experiments. TLR ligands from the Human TLR1-9 Agonist kit (InvivoGen) were diluted following the manufacturers’ instructions. The more detailed description of these TLR ligands is available in Table S-1. LPS (Sigma, L2654−1MG) was from E. coli. Phorbol 12-myristate 13acetate (PMA; Sigma) was diluted with dimethyl sulfoxide (DMSO); the final concentration of DMSO in culture was 0.005%. Synthetic EGF peptides (Peptide 2.0) were diluted in DMSO; the final concentration of DMSO in the culture was 0.15%. The cells in suspension were collected first. The cells remaining adherent were incubated for 5 min with 0.05% trypsin mixed with EDTA and phenol red (Gibco) before B
DOI: 10.1021/acs.jproteome.6b00422 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
min. Mass spectra were acquired with a resolution of R = 70 000, followed by up to 10 consecutive data-dependent MS/ MS spectra taken using higher-energy dissociation (HCD) with the collisional energy set at 25 units. Samples were analyzed in a randomized order.
Table 2. Treatments and Their Concentrations treatment LPS CpG DNA (ODN 2006) HKLM flagellin (FLA) zymosan FSL-1 Pam3CSK4 ssRNA (ssRNA40) PMA (TPA) N- peptide from EGF D- peptide from EGF IsoD- peptide from EGF
description/sequence
conc. (μg/mL)
lipopolysaccharide 5′-tcgtcgttttgtcgttttgtcgtt-3′
0.02; 0.1; 0.5 0.01
heat killed Listeria monocytogenes principal component of bacterial flagella glucan with repeating glucose units connected by β-1,3-glycosidic linkages (palmitoyl)2-CGDPKHPKSF (palmitoyl)3-CSK4 5′-GCCCGUCUGUUGUGUGACUC-3′
108 cells/mL 1 10
phorbol 12-myristate 13-acetate (12-Otetradecanoylphorbol-13-acetate) NSDSEGPLSHDGYGLHDGV
0.01
DSDSEGPLSHDGYGLHDGV
0.15
IsoDSDSEGPLSHDGYGLHDGV
0.15
Protein Identification and Quantification
MS/MS data were extracted and processed according to a previously described protocol43,44 and searched against the concatenated version of the UniProtKB/Swiss-Prot database (release 2014_01; 20 279 human sequences or release 2012_06; 20 257 human sequences) using the Mascot search engine v. 2.3 or v. 2.4 (Matrix Science, U.K.; www. matrixscience.com). The following parameters were used: trypsin digestion with a maximum of two missed cleavages; carbamidomethylation (C) as a fixed modification; pyroglutamate (Q) and oxidation (M) as variable modifications; and a precursor mass tolerance of 10 ppm and a fragment mass tolerance of 0.1 Da. The list of identified proteins was filtered using 1% false discovery rate (FDR) and at least two peptides per protein as limiting parameters. Label-free quantification was performed using the program Quanti that compensates in silico for electrospray current fluctuations.43 Further details and results of the analyses are provided in the Supporting Information (Tables S-2−S-4). In total, between 2000 and 3007 proteins were quantified in each experiment with at least two unique peptides per protein.
1 0.1; 0.5; 2.5 1
0.15
collection. Collected cells were counted using a TC10 cell counter (BioRad) with trypan blue staining. After centrifugation (6 min; 1000g), the cell pellets were washed with PBS and lysed in a lysis buffer (8 M urea in 100 mM ammonium bicarbonate) by sonication on ice (21 s; in three repeats with breaks in between). Cytokine Measurement with Luminex
Bioinformatics and Statistical Analysis
Concentrations of soluble cytokines in media were measured using a Bio-Plex Pro Human Cytokine 10-plex Assay kit (Luminex), following the manufacturers’ instructions.
Protein abundances were normalized with the assumption that equal amounts of protein digests were injected for each sample. Proteins that likely originated from the media (serum albumin and hemoglobin) or sample handling (keratins) were excluded from the results. Log-transformed abundance values were used for further analysis. Principal component analysis (PCA) and Orthogonal Projections to Latent Structures (OPLS) were performed using Simca software version 14.0 (Umetrics). Unpaired Student’s t test with equal or unequal variance (depending on the result of Excel F-test) was applied to calculate the p values. FDR (Benjamini−Hochberg)45 adjusted (q values) or Bonferroni adjusted p values (E values) with a threshold of 5% were used to identify significantly up-/downregulated proteins and pathways. Pathway analysis was conducted using String version 10 (http://string-db.org/). The q values for Gene Ontology (GO) terms enrichment were calculated against the data set with all identified proteins. All error bars in the figures represent sample standard errors of the mean. Fold change was calculated as a ratio between the average relative abundances of a protein in treated cells and corresponding control cells. Quality control of proteomics data was performed in several ways: (1) The retention times of peptides were compared for consistency among different samples within the same experiment. Three samples were excluded from the third experiment. (2) Whether protein abundances correlated with the order of injection was assessed. No protein with a correlation better than R2 > 0.80 was identified. (3) PCA plots were analyzed for the presence of outliers. As a result, one sample was excluded from the first experiment and two samples from the third experiment. (4) Proteins quantified in
