Early Phosphoproteomic Changes in the Mouse Spleen During Deoxynivalenol-Induced Ribotoxic Stress

Abstract

The trichothecene mycotoxin deoxynivalenol (DON) targets the innate immune system and is of public health significance because of its frequent presence in human and animal food. DON-induced proinflammatory gene expression and apoptosis in the lymphoid tissue have been associated with a ribotoxic stress response (RSR) that involves rapid phosphorylation of mitogen-activated protein kinases (MAPKs). To better understand the relationship between protein phosphorylation and DON’s immunotoxic effects, stable isotope dimethyl labeling–based proteomics in conjunction with titanium dioxide chromatography was employed to quantitatively profile the immediate (≤ 30min) phosphoproteome changes in the spleens of mice orally exposed to 5mg/kg body weight DON. A total of 90 phosphoproteins indicative of novel phosphorylation events were significantly modulated by DON. In addition to critical branches and scaffolds of MAPK signaling being affected, DON exposure also altered phosphorylation of proteins that mediate phosphatidylinositol 3-kinase/AKT pathways. Gene ontology analysis revealed that DON exposure affected biological processes such as cytoskeleton organization, regulation of apoptosis, and lymphocyte activation and development, which likely contribute to immune dysregulation associated with DON-induced RSR. Consistent with these findings, DON impacted phosphorylation of proteins within diverse immune cell populations, including monocytes, macrophages, T cells, B cells, dendritic cells, and mast cells. Fuzzy c-means clustering analysis further indicated that DON evoked several distinctive temporal profiles of regulated phosphopeptides. Overall, the findings from this investigation can serve as a template for future focused exploration and modeling of cellular responses associated with the immunotoxicity evoked by DON and other ribotoxins.

The trichothecene deoxynivalenol (DON), a ribotoxic mycotoxin produced by toxigenic Fusarium species, commonly contaminates cereal-based foods and has the potential to adversely affect human and animal health (Pestka, 2010a). A primary target of this mycotoxin is the innate immune system, as has been demonstrated in vivo in mouse systemic and mucosal immune organs (Zhou et al., 1999, 2003a), as well as in vitro in primary and cloned cell cultures derived from mice and humans (Islam et al., 2006; Zhou et al., 2003b, 2005). In the mouse, low-dose DON exposures (1–5mg/kg body weight [bw]) are immunostimulatory as evidenced by robust upregulation of mRNAs and proteins for a diverse array of cytokine, chemokine, and other inflammation-related genes in the spleen, such as tumor necrosis factor-α, interleukin (IL)-1β, and IL-6 (Amuzie et al., 2008; Kinser et al., 2004). High-dose DON exposures (10–25mg/kg bw) evoke apoptosis in the spleen, which could contribute to immunosuppression (Islam et al., 2003; Zhou et al., 1999, 2000).

DON-induced proinflammatory gene activation and apoptosis are mediated by the activation of mitogen-activated protein kinases (MAPKs) via a process known as the ribotoxic stress response (RSR) (Pestka, 2010a). RSR can be induced by other ribotoxins that interfere with translation, including chemicals (e.g., emetine and anisomycin) and proteins (e.g., ricin and Shiga toxin) produced by bacteria, fungi, or plants (Iordanov et al., 1997; Laskin et al., 2002).

Besides MAPKs and their substrates, DON-induced RSR involves the rapid and transient activation of at least two upstream ribosome-associated kinases in vitro, double-stranded RNA-dependent protein kinase (PKR) and hematopoietic cell kinase (Hck), suggesting the pivotal role of protein phosphorylation in DON-induced RSR (Bae and Pestka, 2010; Zhou et al., 2003b, 2005). Phosphoproteomic changes in cloned immune cell lines, including mouse macrophage (Pan et al., 2013), and human T and B cell lines (Nogueira da Costa et al., 2011) have been performed to model molecular mechanisms of DON-induced RSR in the responsive cell populations and to identify biomarkers of effect for DON. However, DON’s effects of protein phosphorylation during RSR have not yet been comprehensively characterized within the immune system of an intact animal.

Resolving the complexity of DON-induced RSR in vivo requires an integrative approach to map the signaling and molecular events occurring at the cellular and subcellular levels. Due to the naturally low stoichiometry of phosphorylation, assessment of its dynamics requires specific affinity enrichment for identification and metabolic or chemical labeling in conjunction with high-accuracy proteomics for relative quantification (Mann et al., 2002). Although in vitro metabolic labeling using stable isotope labeling of amino acids in cell culture (SILAC) is commonly used for in vitro proteomic studies and has been recently successfully employed to track DON-induced phosphorylation changes in macrophages (Pan et al., 2013), this approach is not readily amenable to animal studies. Two alternative strategies for in vivo studies are chemical labeling by isobaric tag for relative and absolute quantitation (iTRAQ) and by stable isotope dimethyl labeling (Kovanich et al., 2012). The latter method entails rapid and complete reductive amination of peptides, which makes it a relatively simple and economical strategy for large-scale quantitative analyses. It involves chemically labeling primary amine groups (lysine and amino termini) of peptides with different isotopically labeled formaldehyde through reductive amination (Kovanich et al., 2012). Because this approach has been used for comprehensive quantitative proteome analyses for assessing protein expression and posttranslational modifications, such as phosphorylation, glycosylation, and proteolysis (Kovanich et al., 2012), it should be ideal for the measurement of DON-induced phosphoproteome changes in vivo.

The overall goal of this investigation was to identify early protein phosphorylation changes in the immune system of DON-exposed mice and relate these to the toxin’s downstream immunomodulatory effects. Specifically, we identified and quantitated phosphoproteomic changes in the spleens of mice orally exposed to a dose of DON (5mg/kg bw). This dose of DON is known to be immunostimulatory, eliciting robust proinflammatory gene expression and MAPK activation in the spleen (Amuzie et al., 2008; Zhou et al., 1997, 2003a). We employed a 30-min time window to view events in the spleen during DON’s fast pharmacokinetic turnover prior to proinflammatory gene activation (1–3h) and apoptosis (≥ 6h). Stable isotope dimethyl labeling in conjunction with titanium dioxide (TiO₂) chromatography and liquid chromatography-mass spectrometry (LC-MS)/MS was utilized to profile phosphoproteomic changes (Fig. 1). The results revealed the involvement of an extremely complex cell signaling network in vivo that was consistent with the induction of resultant immune cell activation and apoptosis by DON.

Fig. 1.

Experimental design for stable isotope dimethyl labeling–based quantitative phosphoproteomic analysis of DON-induced RSR in the spleen. Mice were orally gavaged with vehicle or 5mg/kg bw DON for different time periods in two independent groups. After nuclear and cytoplasmic proteins were extracted, equal amount of proteins from different treatment groups were trypsin digested, dimethyl labeled, and equally mixed. Phosphopeptides were enriched by titanium dioxide chromatography and analyzed by high-resolution mass spectrometry. Each set was repeated in three independent experiments (n = 3).

MATERIALS AND METHODS

Experimental design.

All animal studies followed NIH guidelines and were approved by the Michigan State University Committee on Animal Care. Eight- to 10-week-old male B₆C₃F₁ mice (Charles River, Portage, MI) were used in the study. Prior to the experiment, animals were held for a 1-week acclimation period in a room with a 12-h light/dark cycle, 70% humidity, and a temperature of 22–23°C.

DON (> 98% purity by elemental analysis) was obtained from Dr Tony Durst (University of Ottawa). An initial pilot study was conducted to measure DON tissue levels in the spleen of mice (n = 6) at 0, 5, 15, and 30min after exposure to the toxin at 5mg/kg bw. Mice were sacrificed by cervical dislocation and spleens removed. Spleens (≤ 200mg) were homogenized in PBS (1:10 [wt/vol]), and the homogenates were centrifuged at 15,000 × g for 10min. The supernatant fraction first heated to 100°C for 5min and then centrifuged at 15,000 × g for 10min. DON concentrations in the supernatant were measured using a Veratox High Sensitivity (HS) ELISA (Neogen, Lansing, MI) (Amuzie et al., 2008).

For the proteomic study, two groups of mice were used (Fig. 1). This first group included mice that were (1) untreated, (2) treated with vehicle and held for 15min, or (3) treated with 5mg/kg bw DON via oral gavage and held for 30min. The second group of mice were orally gavaged with 5mg/kg DON and held for (1) 5min, (2) 15min, or (3) 30min. The group treated with vehicle and held for 15min served as a representative of vehicle treatment to discern the effect of animal handling and the vehicle used. Mice were sacrificed by cervical dislocation, and spleens were removed for extraction of proteins to be used in Western blot and proteomic analyses. Each group was repeated in three independent experiments (n = 3).

Western blot.

Nuclear and cytoplasmic fractions were extracted from spleen using a Nuclear Extract Kit (Active Motif, Carlsbad, CA). Fractions from three treatment types within a group were pooled equally based on protein content as determined by BCA Protein Assay (Pierce, Rockford, IL). Immunoblotting analysis was performed as described (Bae and Pestka, 2008). Antibodies against the following proteins were used: p38, phospho-p38 MAPK (Thr180/Tyr182), p42/p44 MAP kinase, phospho-p42/p44 MAP kinase (Thr202/Tyr204), SAPK/JNK, phospho-SAPK/JNK (Thr183/Tyr185), beta-actin, stathmin, phospho-stathmin (Ser38), filamin A (FLNA), phospho-filamin A (Ser2152), myosin 9, phospho-myosin 9 (Ser1943), lamin A (Cell Signaling, Danvers, MA), and phospho-lamin A (Ser392) (Abcam, Cambridge, MA). Blots were scanned on the Odyssey IR imager (LICOR, Lincoln, NE), and quantification was performed using LICOR software v.3.0.

Phosphopeptide measurement.

Protein phosphorylation changes in the spleens were analyzed by proteomics analysis using stable isotope dimethyl labeling coupled with TiO₂ chromatography for phosphopeptide enrichment and high-accuracy mass spectrometric characterization (Fig. 1). Cytoplasmic (3.3mg) and nuclear (500 µg) protein were digested with trypsin using the FASP protocol (Wisniewski et al., 2009) using spin ultrafiltration units of nominal molecular weight cut of 30,000 (Millipore, Billerica, MA). Employing a triplex dimethyl labeling strategy, peptides originating from nuclear and cytoplasmic proteins of three treatment types within a group were labeled by the on column method with light, intermediate, or heavy dimethyl reagents as described previously (Boersema et al., 2009). The defined mass increments introduced among the three groups resulted in characteristic peptide triplets that enabled relative quantification of peptide abundance. Eluted labeled peptides within each set were mixed equally and further enriched for phosphopeptide with Titansphere PHOS-TiO Kit (GL Sciences, Torrance, CA) following manufacturer’s instructions. Enriched peptides were desalted using 100mg reverse-phase tC18 SepPak solid-phase extraction cartridges (Waters, Milford, MA). The phosphoproteomic study comprised three independent biological replicates to account for the biological and technical variability.

Peptides were resuspended in 2% acetonitrile (ACN)/0.1% trifluoroacetic acid (vol/vol) to final concentration of 20 µg/µl. From this, 10 µl (200 µg) was loaded for 5min onto a Waters Symmetry C18 peptide trap (5 µm, 180 µm × 20mm) using a Waters nanoAcquity Sample Manager (www.waters.com). The column was washed at 4 µl/min with 2% ACN/0.1% formic acid (vol/vol). Bound peptides were eluted using a Waters nanoAcquity UPLC (Buffer A = 99.9% Water/0.1% formic acid, Buffer B = 99.9% ACN/0.1% formic acid [vol/vol]) onto a Michrom MAGIC C18AQ column (3 µ, 200 A, 100U × 150mm, www.michrom.com) and eluted over 240min with a gradient of 5–30% B in 225min, ramping to 90% B at 227min and back to 5% B at 228.1min at a constant flow rate of 0.8 µl/min. Eluted peptides were sprayed into a ThermoFisher LTQ-FT Ultra mass spectrometer (www.thermo.com) using a Michrom ADVANCE nanospray source. Survey scans were taken in the Fourier Transform (FT) mass spectrometer (25,000 resolution determined at m/z 400), and the top 10 ions in each survey scan were then subjected to automatic low-energy collision-induced dissociation in the Linear trap quadrupole (LTQ).

Data analysis.

Peak list generation, protein quantitation based upon dimethyl labeling, extracted ion chromatograms (XICs), and estimation of false discovery rate (FDR) were all performed using MaxQuant (Cox and Mann, 2008), v1.2.2.5. MS/MS spectra were searched against the IPI mouse database, v3.72, appended with common environmental contaminants using the Andromeda searching algorithm (Cox et al., 2011). Further statistical analysis of the dimethyl-labeled protein and peptide ratio significance was done using Perseus, v1.2.0.17 (www.maxquant.org). MaxQuant parameters were protein, peptide, and modification site maximum FDR = 0.01; minimum number of peptides = 1; minimum peptide length = 6; and minimum ratio count = 1 protein quantitation was done using all modified and unmodified razor and unique peptides. Andromeda parameters were triplex dimethyl labeling: light condition (DimethylLys0, DimethylNter0), medium condition (DimethylLys4, DimethylNter4), and heavy condition (DimethylLys8, DimethylNter8); fixed modification of Carbamidomethylation (C), variable modifications of Oxidation (M), Phosphorylation (STY) and Acetyl (Protein N-term); maximum number of modifications per peptide = 5; enzyme trypsin max missed cleavage = 2; parent ion tolerance = 6 ppm; and fragment ion tolerance = 0.6Da reverse database search was included. Given that each group of mice was independently dosed and sacrificed, and proteins were independently extracted, digested, and labeled, each of the treatment groups was considered independent and was compared among each other as independent groups. For each identified dimethyl triplet, MaxQuant calculated the three XIC values, and XICs for the light and medium member of the triplet were normalized with respect to the heavy common 30-min time point of DON treatment, which was scaled to 1. A significance value, Significance B, was calculated for each dimethyl relative quantification ratio and corrected by the method of Benjamini and Hochberg using the statistical program Perseus in the MaxQuant environment (Cox and Mann, 2008). Significance B values of a phosphopeptide in the two groups are calculated independently. Significant protein ratio cutoff was set as at least one set of time course has a Significance B value ≤ 0.05 as calculated by MaxQuant (Cox et al., 2009). Phosphosites were mapped with PhosphoSitePlus (Hornbeck et al., 2012).

DAVID analysis.

Analyses for gene ontology (GO) annotation terms were performed with the functional annotation tool DAVID (http://david.abcc.ncifcrf.gov/). Uniprot accession identifiers of significantly regulated phosphoproteins were submitted for analysis of GO biological processes and KEGG pathways, with a maximal DAVID EASE score of 0.05 for categories represented by at least two proteins.

Fuzzy c-means clustering.

Fuzzy c-means (FCM) clustering (Olsen et al., 2006) in MATLAB (MathWorks) was used to search for patterns in the time profile patterns among all identified phosphopeptides. To robustly represent the ratio of a peptide being quantified multiple times, the median values of the three biological replicates were used for clustering. The phosphoproteomic ratios of peptides quantified over all four time points were included, natural log transformed, and normalized. From the 20 clusters, we selected 6 based on the average probability of each cluster to represent early, intermediate, and late responders with phosphorylation levels up- or downregulated.

Statistics.

Data were analyzed using SigmaStat for Windows (Jandel Scientific, San Rafael, CA). A Kruskal-Wallis one-way ANOVA was performed. Data sets were significantly different when p < 0.05.

RESULTS AND DISCUSSION

Relation of DON Distribution to RSR Onset in the Spleen

Following oral administration, DON was detectable in the spleen within 5min and reached peak concentration within 15min (Fig. 2A). The toxin was rapidly cleared to about half the peak concentration around 30min, which is consistent with previous observations (Pestka et al., 2008). DON-induced phosphorylation of MAPKs, including p38, extracellular signal- regulated protein kinase (ERK) 1/2, and c-Jun N-terminal kinase (JNK) 2, was evident at 15min (Fig. 2B). These results confirm that the time window used in this investigation included both peak DON concentration and the initiation of RSR.

Fig. 2.

Relationship between DON tissue concentrations and RSR onset in the spleen. (A) DON concentrations in the spleen after oral exposure to 5mg/kg bw DON. Data are presented as mean ± SEM. (n = 6). Different letters indicate a statistically significant difference between time points (p < 0.05). (B) Representative Western blot of p38, ERK, and JNK and their phosphorylation status in the spleen of mice exposed to DON using total and phospho-specific antibodies, with beta-actin as internal control.

DON-Induced Changes in Splenic Phosphoproteomic Profile

We reproducibly identified 500 and 516 unique phosphopeptides corresponding to 226 and 302 proteins in the nuclear and cytoplasmic fractions (Supplementary dataset 1), respectively, at an accepted FDR of 1%. Approximately 22% of the total phosphopeptides were identified at levels that were considered statistically significant compared with vehicle controls (Significance B with a Benjamini-Hochberg corrected FDR < 5%) (Figs. 3A and B). A total of 90 proteins possessed significantly altered phosphorylation status, with 43 in the nucleus (Supplementary table 1) and 53 in the cytoplasm (Supplementary table 2). More than 96% of identified peptides harbored at least one phosphosite, indicating that there was highly efficient phosphopeptide enrichment by TiO₂ chromatography. The distribution of all phosphorylated serine, threonine, and tyrosine sites identified was 83.0, 14.8, and 2.1%, respectively (Figs. 3C and D).

Fig. 3.

Summary of protein phosphorylation changes during early DON-induced RSR in the mouse spleen. Numbers of significantly regulated (A) phosphopeptides and (B) phosphoproteins (grey) among all unique proteins and peptides identified (grey + black). Significance was determined by Significance B values corrected for FDR ≤ 0.05. Overall distribution of phospho-serine, threonine, and tyrosine detected in the study (C) and significantly regulated by DON (D) in nuclear and cytoplasmic fractions combined.

The potential for handling stress and vehicle-induced changes in phosphorylation was evaluated by comparing no treatment and vehicle groups. There were changes in 15 proteins that were potentially animal-handling-stress- and vehicle-induced ones (Supplementary table 4), and these were excluded for functional analysis. Overall, extensive phosphoproteome changes occurred in the spleen following DON exposure, which encompass known and yet-to-be functionally characterized phosphosites.

Confirmation of Proteomic Changes

To verify the quantification of phosphorylation in stable isotope dimethyl labeling–based proteomics, the phosphorylation kinetics of four representative proteins, stathmin 1 (STMN1), FLNA, myosin 9 (MYH9), and lamin A (LMNA) were measured by Western blotting using phosphosite-specific antibodies. Detection of relative up- and downregulation of protein phosphorylation by proteomics and immunoblot was in agreement (Fig. 4), supporting the reliability of the proteomics approach in measuring the dynamic phosphorylation during DON-induced RSR.

Fig. 4.

Confirmation of phosphoproteome quantification. Phosphorylation levels of STMN1, FLNA, MYH9, and LMNA were measured using Western blot by normalizing the signaling intensity of the bands from phosphosite-specific antibody with those from total antibody. Normalized intensities (square, shown as mean ± SEM) were then correlated with the phosphoproteomic ratios (circle, shown as median).

Biological Analysis of Altered Phosphoproteins

To discern the potential impact of DON-induced phosphory lation, changes in the prevalence of differentially phosphorylated proteins were related to specific biological processes using GO annotation in DAVID. GO analysis indicated that 14% of all significantly regulated phosphoproteins were associated with intracellular signaling cascades (Fig. 5). Thus, DON-induced RSR in the spleen was extraordinarily complex involving the interrelated MAPK and phosphatidylinositol 3-kinase (PI3K)/AKT pathways (Table 1) that linked early cellular events to later proinflammatory gene activation, apoptosis, and other immunotoxic effects of DON. In addition, DON exposure affected other biological processes such as cytoskeleton organization, regulation of apoptosis, and lymphocyte activation and development (Fig. 5).

Fig. 5.

GO associated with significantly regulated phosphoproteins as determined by DAVID biological process terms (DAVID EASE score < 0.05). Data are presented as a histogram of the relevant biological processes identified and shown as a percentage of the total identified proteins that fall within each category.

Table 1.

Splenic Phosphosites Associated With Proteins Involved in MAPK and PI3K/AKT Pathways in the Mouse Spleen

Protein name	Protein description	Phosphosite
MAPK pathway
PEA15	Astrocytic phosphoprotein PEA15	S116
STMN1	Stathmin	S38
MNK2	MAP kinase-interacting serine/threonine-protein kinase 2	S270, T272
FLNA	Filamin A	S2152
PI3K/AKT pathway
PI4K	Phosphatidylinositol 4-kinase	T292
PIK3CG	Phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit gamma	S257, S262, S264
NHERF1	Sodium-hydrogen exchanger regulatory factor 1	S275
PKD2	Protein kinase D2	S197
CTNNB1	β-catenin	T556
DNMT1	DNA methyltransferase 1	S125
PML	Promyelocytic leukemia	T527
		S528

It is important to note that phosphorylation changes described in this study represent the average among many cell types in the spleen. However, the phosphorylation of proteins could be differentially regulated within individual cells. Some of the phosphorylation changes might cross-compete and therefore have little or no impact on cell functions. In addition, it is possible that a temporal gap exists in the onset of DON’s effects in cell signaling, proinflammatory gene activation, and apoptosis. Consequently, it is important to meld the in vivo findings herein with prior in vitro and in vivo studies to dissect the differential effects of DON within various cell populations and the spleen as a whole.

Integration of Phosphoproteomic Data With Prior Research: MAPK Signaling

The MAPK pathway is central to DON-induced RSR, and the phosphorylation of p38, ERK1/2, and JNK1/2 has been demonstrated in the mouse spleen here (Fig. 2B) and previously (Zhou et al., 2003a). Consistent with these findings, phosphoproteomic analysis identified kinases that could function as upstream and downstream mediators of the activated MAPKs (Fig. 6A). These included astrocytic phosphoprotein PEA15 (upstream of ERK), STMN1 and MAP kinase-interacting serine/threonine-protein kinase 2 (MNK2) (downstream of ERK), and FLNA, which is a scaffold in MAPK signaling.

Fig. 6.

MAPK (A) and PI3K/AKT (B) signaling pathways linked to DON-induced RSR in the spleen.

The upstream mediator PEA15 binds to ERK1/2 and inhibits its nuclear localization, thus blocking cell proliferation (Krueger et al., 2005). Such binding and inhibitory effects are abolished when PEA15 is phosphorylated at Ser116, which was observed at 15min during DON-induced RSR. The downstream mediator STMN1 is a small microtubule binding protein that regulates the rates of polymerization and disassembly of microtubules (Nakamura et al., 2006). ERK-driven phosphorylation of STMN1 at four serine residues including Ser38, which is significantly regulated in DON-induced RSR, leads to cell cycle arrest and apoptosis.

Dual phosphorylation of another downstream mediator, MNK2, at previously unreported sites—Ser270 and Tyr272—was decreased at 30min. With ERK and serine/threonine- protein phosphatase 2A (PP2A) controlling its phosphorylation status, MNK1/2 could also regulate the release of arachidonic acid and proinflammatory gene activation by phosphorylating cytosolic phospholipase A2 (cPLA2) (Hefner et al., 2000), which was recently shown to be phosphorylated in DON-treated RAW 264.7 cells (Pan et al., 2013).

Phosphorylation of a scaffold protein that organizes MAPK signaling, FLNA, was also altered during DON-induced RSR. FLNA integrates the JNK pathway via interaction with MKK4, the p38 pathway via interaction with TAK1 and p38, and the ERK pathway via interaction with MAPK6, and transcription factors c-JUN, p65, and STAT1 (Bandyopadhyay et al., 2010; Morrison and Davis, 2003). Over the 30min of DON treatment period, FLNA phosphorylation at Ser2152 gradually increased, which could protect it against proteolytic cleavage by calpain (Chen and Stracher, 1989), thereby strengthening the structural framework and enhancing MAPK signaling in the RSR. Accordingly, phosphoproteins affected by DON exposure might not only be involved in directly mediating the MAPK signaling, but also in the regulation of MAPK signaling organization.

Integration of Phosphoproteomic Data With Prior Studies: PI3K/AKT Signaling

The PI3K/AKT pathway was also potentially targeted in DON-induced RSR (Fig. 6B), with affected proteins being involved in the regulation of phosphoinositide turnover (via phosphatidylinositol 4-kinase [PI4K], PI3K, and NHERF1), as well as transcriptional regulation (via DNMT1, PML, and PKD2).

Phosphoinositide turnover involves a series of reversible lipid phosphorylation reactions. The catalytic domain of PI4K had decreased phosphorylation at T292 in the PI3K/PI4K domain at 30min after DON treatment. This kinase mediates the production of P(4,5)P2, which could be further phosphorylated to P(3,4,5)P3 (PIP3) by PI3K, or hydrolyzed to inositol triphosphate and diacylglycerol by phospholipase C. All of these lipids are important secondary messengers in cellular signaling.

PI3K, which positively catalyzes phosphorylation of second messenger PIP3, was differentially phosphorylated by DON in the spleen. PIP3 is negatively regulated by two lipid phosphatases, namely SH2-containing inositol-5′-phosphatase (SHIP) that was shown to be affected by DON in the RAW 264.7 (Pan et al., 2013) and phosphatase and tensin homolog (PTEN). Triple phosphorylation of the catalytic subunit of PI3K (PIK3CG) at novel phosphosites Ser257, Ser272, Ser274 within the RAS-binding domain was increased at 15min but rapidly decreased at 30min, which might indicate a transient modulation of PI3K interaction with and activation by RAS.

The sodium-hydrogen exchanger regulatory factor 1 (NHERF1) is known to recruit PTEN and restrict PI3K activation (Takahashi et al., 2006). Dephosphorylation of NHERF1 at Ser275 occurred at 30min in DON-induced RSR, which could weaken the association of NHERF1 with PTEN, and facilitate the activation of PI3K pathway (Voltz et al., 2007).

Transcription is potentially regulated during DON-induced RSR via transcription factor and epigenetic mechanisms. AKT1 is a kinase activated by PIP3 that was previously demonstrated to be rapidly phosphorylated and activated by DON in RAW 264.7 cells and peritoneal macrophages (Shi and Pestka, 2009; Zhou et al., 2005). This results in phosphorylation and inhibition of glycogen synthase kinase-3 beta, which phosphorylates β-catenin (CTNNB1) and targets it for degradation. As shown here, DON induced CTNNB1 dephosphorylation at 15min. This protein has been previously shown to translocate to the nucleus, where it acts as a coactivator for transcription factors of the TCF/LEF family, c-JUN, c-MYC, and n-MYC (Staal et al., 2008). AKT1 is also a kinase for DNA methyltransferase 1 (DNMT1) (Estève et al., 2011), which was phosphorylated here at Ser125. The possible effect of DON on DNA methylation has been previously suggested in RAW 264.7 macrophages (Pan et al., 2013) and Caco-2 intestinal epithelial model (Kouadio et al., 2007).

Also significantly altered in both mouse spleen and RAW 264.7 was promyelocytic leukemia (PML), a protein that induces gene hypermethylation and silencing by recruiting DNMTs to target promoters (Di Croce et al., 2002). Decreased phosphorylation of PML at Ser528 at 30min stabilizes it from being targeted for degradation via the ubiquitination pathway (Yuan et al., 2011). Notably protein kinase D2 (PKD2) was significantly phosphorylated at 30min in DON-induced RSR. Activated PKD2 could translocate to the nucleus where it phosphorylates and inhibits Class IIa HDACs (Vega et al., 2004). Similarly, DON was previously shown to modulate the phosphorylation of HDAC1 and HDAC2 in RAW 264.7 (Pan et al., 2013). Overall, phosphoproteins impacted during DON-induced RSR likely affect phosphoinositide signaling and transcriptional regulation via the PI3K/AKT pathway.

Therefore, the MAPK pathway affected could contribute to the proinflammatory response by further activation of the NF-κB (Zhou et al., 2003a), and the PI3K/AKT pathway could also regulate a multitude of transcription factors, including NF-κB, AP-1, NFAT, and CREB (Hazeki et al., 2007), and contribute to the immunostimulation caused by DON.

Integration of Phosphoproteomic Data With Prior Research—Leukocyte Targets of DON

Phosphorylation of proteins involved in lymphocyte activation and development was significantly altered, suggesting that diverse immune cell populations, including monocytes, macrophages, T cells, B cells, dendritic cells, and mast cells, could be impacted by DON exposure (Fig. 7). Two of the identified proteins, lymphocyte cytosolic protein 1 (LCP1) and myosin, heavy polypeptide 9 (MYH9), are involved in the activation and development of monocytes and T cells. LCP1 is an actin-binding protein expressed exclusively in hemopoietic cell lineages. Phosphorylation of LCP1 at Ser5, as observed in DON-induced RSR at 5min, plays a role in the activation of T cells in response to costimulation through T cell receptor (TCR)/CD3 and CD2 or CD28 (Wabnitz et al., 2007). This phosphosite has also been shown to be phosphorylated in response to lipopolysaccharide (Shinomiya et al., 1995). MYH9 is a myosin protein involved in cell adhesion and is required for the establishment of T-cell polarity and monocyte differentiation (Jacobelli et al., 2004). It is phosphorylated at Ser1943 transiently at 5min in DON-induced RSR. Apoptotic chromatin condensation inducer 1 (ACIN1) plays a part in positive regulation of monocyte differentiation (Sordet et al., 2002) and erythrocyte differentiation (Zermati et al., 2001).

Fig. 7.

Phosphorylation changes of proteins associated with lymphocyte activation and development during DON-induced RSR in the spleen. The heat map depicts the log-2 transformed relative abundance of the protein phosphorylation.

T cells and B cells have been implicated previously in DON-induced RSR. It has been proposed that, based on transcriptomic studies in the mouse thymus (van Kol et al., 2011) and Jurkat human T cells (Katika et al., 2012), DON induces cellular events that contribute to T-cell activation. Furthermore, B-cell activation has been implicated in mediating DON-induced IgA nephropathy (Pestka and Dong, 1994). The phosphoproteomic results suggested that early signaling is likely to precede these events. Relative to T-cell activation, minor histocompatibility antigen HA-1 (HMHA1) was differentially phosphorylated significantly in DON-induced RSR. When complexed with major histocompatibility complex, this protein is known to drive an immune response following recognition by TCR complex on T cells (Lin and Weiss, 2001). Finally, SH2 domain containing 3C (SH2D3C), which exhibited significantly altered phosphorylation status here, enhances T-cell activation by interacting with tyrosine kinase PYK2 (Sakakibara et al., 2003).

Among the shared components of T-cell and B-cell activation, CD45 is a surface antigen and tyrosine phosphatase. It enhances the activation of T and B cells by dephosphorylating the C-terminal inhibitory tyrosine and activating the SRC proteins LCK and FYN (Lin and Weiss, 2001). Phosphorylation of CD45 at Ser962 within its tyrosine-protein phosphatase domain 2 was significantly increased at 30min, which might suggest increased CD45 phosphatase activity and enhanced T-cell and B-cell activation. T-cell and B-cell activation both involve the phosphorylation of caspase recruitment domain family member 11 (CARD11). This protein, transiently phosphorylated at 15min in DON-induced RSR, could contribute to the activation of NF-κB (Sagaert et al., 2010), which has been demonstrated in RAW 264.7 macrophages under DON-induced ribotoxic stress (Pan et al., 2013; Wong et al., 2002).

In addition to its involvement in PI3K/AKT signaling as discussed above, CTNNB1 is a regulator for T cell and dendritic cell function. It is a critical factor in CD8-positive T-cell differentiation and memory development (Gattinoni et al., 2009). In intestinal dendritic cells, CTNNB1 is required for the expression of anti-inflammatory mediators such as retinoic acid–metabolizing enzymes, IL-10, and transforming growth factor-beta (Manicassamy et al., 2010). Unc-13 homolog D (UNC13D), another protein with significantly regulated phosphorylation status, plays a role in cytotoxic granule exocytosis in lymphocytes and is required for granule maturation, granule docking, and priming at the immunologic synapse (Ménager et al., 2007). UNC13D also regulates Ca²⁺-dependent secretory lysosome exocytosis in mast cells (Neeft et al., 2005). In summary, DON-induced RSR in the spleen entails activation and regulation of multiple leukocyte targets including monocytes, macrophages, T cells, B cells, dendritic cells, and mast cells.

Large-scale phosphoproteomic studies have been employed previously to elucidate molecular mechanisms of DON-induced RSR in cloned immune cell lines, including mouse macrophage (Pan et al., 2013), and human T and B cell lines (Nogueira da Costa et al., 2011), as well as to identify biomarkers of effect for DON. Although valuable for mechanistic exploration of DON-induced RSR in specific immune cell types, it is recognized that alterations in signaling pathways and biological processes derived from these in vitro models cannot always be extrapolated to in vivo systems, due to the complexity relative in (1) diversity of cell types, (2) pharmacokinetic distribution of DON, and (3) intercellular communication in the immune system of an intact animal. Nevertheless, among the early phosphoproteome alterations identified in vivo, a total of 30 significantly regulated phosphoproteins (Supplementary table 3), including STMN1, LMNA, PML, and BCLAF1, overlap with those previously detected in vitro in DON-exposed RAW 264.7 macrophages, suggesting that the innate immune system is a critical target.

Overall, these findings are consistent with diverse immune cell populations, including monocytes, macrophages, T cells, B cells, dendritic cells, and mast cells, being impacted by DON exposure and ultimately mediating the immunomodulatory effects of this toxin.

Integration of Phosphoproteomic Data With Prior Research—Regulation of Apoptosis

DON-induced apoptosis has been demonstrated both in vivo and in vitro. In vivo administration of high doses of DON to mice causes apoptosis in spleen, thymus, Peyer’s patches, and bone marrow (Zhou et al., 1999). Similarly, high concentrations of DON can evoke rapid onset of apoptosis in vitro, such as in RAW 264.7 murine macrophages, T cells, and B cells (Pestka et al., 1994; Zhou et al., 2005) potentially leading to suppression of immune function. DON induces apoptosis by both intrinsic and extrinsic pathways (He et al., 2012; Zhou et al., 2005). As shown here, prior to apoptosis onset, regulators of both survival and apoptotic pathways are differentially phosphorylated in the spleen of mice treated with DON (Fig. 8).

Fig. 8.

Phosphorylation changes of proteins associated with regulation of apoptosis during DON-induced RSR in the spleen. The heat map depicts the log-2 transformed relative abundance of the protein phosphorylation.

One protein with altered phosphorylation status during DON-induced RSR was Bcl-2-associated transcription factor 1 (BCLAF1), a Bcl-2 family protein, which is proapoptotic (Kasof et al., 1999) and encodes a transcriptional repressor that interacts with several members of the Bcl-2 family of proteins. The Bcl-2 family includes both proapoptotic and antiapoptotic regulators that regulate the intrinsic apoptosis pathway. CD45 mediates apoptosis via both the intrinsic and extrinsic pathways and is involved in the DNA fragmentation and DNA condensation process, playing a role in the proapoptotic pathway activation (Dupéré-Minier et al., 2010). Another protein impacted in DON-induced RSR also identified in the phosphoproteomic study, PML, mediates apoptosis in a p53-dependent manner by stabilizing p53 and phosphorylation/activation of p53 at Ser20 (Guo et al., 2000). p53 phosphorylation and binding activity are induced by DON in RAW 264.7 macrophages (Zhou et al., 2005). Alternatively, PML could stimulate apoptosis in a p53-independent manner via the ATM and checkpoint kinase-2 (CHEK2) in U937 human monocytes, in which overexpression of PML led to apoptosis (Yang et al., 2002).

Multiple caspase substrates showed altered phosphorylation during DON-induced RSR, including ACIN1, DNA fragmen tation factor alpha subunit (DFFA), LMNA, and lamin B1 (LMNB1). ACIN1 induces apoptotic chromatin condensation after activation by caspase-3 (Sahara et al., 1999). DNA fragmentation factor (DFF) is composed of heterodimer of DFFA (significantly regulated during DON-induced RSR) and DFFB subunits and is one of the major endonucleases responsible for internucleosomal DNA cleavage during apoptosis. Upon cleavage of DFFA by caspase-3, DFFB triggers DNA fragmentation and chromatin condensation (Zhang et al., 1999). In addition, caspase 6-dependent degradation of lamins, including LMNA and LMNB1, facilitates the nuclear events of apoptosis by inducing nuclear breakdown as a prelude to nuclear destruction (Burke, 2001).

In addition to proapoptotic proteins, regulators of antiapoptotic pathways were also impacted during early DON-induced RSR. CARD11 might contribute to NF-κB activation, which could further activate members of the inhibitor of apoptosis protein (IAP) family (Ryan et al., 2000). PEA15 contains death effecter domain (DED) and binds other DED-containing proteins, preventing the formation of the death-inducing signaling complex and inhibiting activation of the caspase cascade (Trencia et al., 2004). Finally, the PI3K/AKT activated pathway during DON-induced RSR is one of the strongest intracellular prosurvival signaling systems. It can phosphorylate and inactivate BAD (a proapoptotic mitochondrial Bcl-2 family protein) or phosphorylates mTOR that then activates p70S6K to phosphorylate BAD. Other downstream targets of AKT such as IκBα/NF-κB and GSK3 are involved in regulating cell survival as well (Fresno Vara et al., 2004).

Accordingly, the diverse phosphoproteomic changes in early DON-induced RSR associated with regulation of apoptosis discovered here recapitulate previous in vitro observations that this toxin evokes competing apoptotic and survival pathways (Zhou et al., 2005). Before apoptosis onset at 30min after DON exposure at 5mg/kg bw, phosphorylation changes occur on proteins in the proapoptotic pathways, such as PML and BCLAF1 that regulate p53 and Bcl-2 family, which are previously known to be phosphorylated in DON-induced RSR, as well as downstream caspase substrates that have directly lead to physiological and morphological alterations characteristic of apoptosis. Proteins modifying the activity of the antiapoptotic pathway, PI3K and PEA15, were also differentially phosphorylated. These early phosphorylation changes likely fine-tune the balance between apoptotic and survival signaling before the immunosuppressive effects take place with extended DON exposure.

Categorization of DON-Regulated Phosphopeptides into Distinct Temporal Profiles

FCM clustering analysis was employed to trace and categorize the temporal changes of protein phosphorylation observed (Supplementary dataset 2). Based on the pattern of phosphorylation changes and the average probability of the clusters, six distinctive temporal profiles (rapid, intermediate, and prolonged) of the regulated (phosphorylated, dephosphorylated) phosphopeptides were recognized (Figs. 9A–F). DON-induced RSR initiated with rapid (≤ 5min) dephosphorylation of RNA processing proteins (e.g., SFRS1, SFRS9, and DKC1) (Fig. 9A) and phosphorylation of apoptosis regulatory proteins (e.g., STMN1, PDCD4, SLTM, and STEAP3) (Fig. 9B). The stress response proceeded with a cascade of phosphorylation events involving proteins in transcriptional regulation that are phosphorylated (e.g., PML, TCOF1, HDGF, and TRP53BP1) (Fig. 9C) and RNA processing that are dephosphorylated (e.g., RBM39, SRRM2, and SF1) (Fig. 9D) up to approximately 15min. As RSR progressed, prolonged dephosphorylation of cellular protein complex assembly-related proteins (e.g., FLNA and MYH11) occurred to organize intracellular and intercellular signaling (Fig. 9E), whereas prolonged phosphorylation of cell cycle proteins (e.g., NIPBL and EEF1D) was evident, possibly arresting the cell cycle under DON-induced RSR (Fig. 9F).

Fig. 9.

Temporal profiles of DON-regulated phosphosites. Based on the phosphorylation dynamics, FCM clustering analysis generated clusters representing early, intermediate, and late responders with phosphorylation levels up- or downregulated were selected from 20 clusters based on the average probability (A–F). Each trace depicts the natural log value of relative phosphorylation level as a function of time and is color coded according to its membership value (i.e., the probabilities that a profile belongs to different clusters) for the respective cluster. Each cluster is designated by the function of prominent members, with examples of such members given for each cluster.

CONCLUSIONS

The global, quantitative, and kinetic phosphoproteomic analysis presented here expands our understanding of signaling events leading to DON-elicited immunotoxicity from simply MAPK activation to an intricate finely tuned signaling network involving the MAPK and PI3K/AKT signaling cascades. The DON dose of 5mg/kg bw was chosen to evoke robust RSR in the mouse, a species relatively less sensitive than human, which could be tracked by phoshoproteomic analysis. This dose would correspond to consumption of a single bolus of food containing approximately 30 ppm. Although this is much higher than the current FDA guideline of 1 ppm in grain products, DON concentrations ranging from 2 to 93 ppm were detected in food samples associated with a single massive Chinese outbreak in which over 130,000 people were affected (Pestka, 2010b). Phosphoproteome changes were also reflected in proteins associated with downstream regulation of the immunostimulatory and immunosuppressive sequelae. Multiple immune cell populations were targeted as indicated by the identification of altered phosphoproteins associated with immune system activation and development. Competing apoptotic and survival pathways were both fired-in early in DON-induced RSR before the balance shifts toward apoptosis in case of prolonged stress. Extensive protein phosphorylation changes take place in lymphoid tissue very early within 30min, preceding previously reported immunotoxic effects of DON such as proinflammatory gene activation and apoptosis that occur later after 1–3h and 6h, respectively (Fig. 10).

Fig. 10.

Summary of pathways and biological processes potentially leading to the immunotoxicity caused by DON. In response the ribosome damage caused by DON and signaling pathways including MAPK and PI3K/AKT pathways are activated, which mediate key biological processes regulating different aspects of immunotoxicity elicited by DON. Proteins in bold indicate novel mediators identified in this study, proteins in bold and italics indicate previously known mediators confirmed here, and proteins in italics indicate mediators previously determined in other studies but not identified here.

It must be recognized that the spleen is composed of T cells, B cells, macrophages, dendritic cells, mast cells, natural killer cells, red blood cells, and cells associated with splenic vasculature and supporting tissues. Although it would be desirable to dissect the contribution to total phosphoproteome changes by cell type and to understand the different impact of DON on different cell populations in vivo, it is not possible to dissociate and sort the diverse cell types in the spleen without disturbing protein phosphorylation. In addition, DON concentration in the spleen rises and falls during absorption, distribution, metabolism, and excretion, whereas in the cell culture, cells as a single layer are exposed to toxin treatment instantaneously. Finally, analysis of homogenous cell cultures neglects the importance of cell communication across cell types in mounting an immune response. DON affects proteins, such as claudin (Pinton et al., 2010), ZO-1 (Diesing et al., 2011), SRC, ZAK (Pan et al., 2013), AKT as previously reported, myosins (MYH9 and MYH11), CTNNB1, and EPB4.1 as identified in this study, that are involved in tight junctions. They allow intercellular communication and interaction by metabolically and electrically coupling cells together (Denker and Nigam, 1998), which could not be otherwise captured in an in vitro model.

It is known that the detection limit of Western blot is in the low picogram per milliliter range, whereas that of LC-MS is in the nanogram per milliliter range (Mann, 2008). Due to the relatively lower sensitivity of the proteomics approach compared with immunoblotting, some of the signaling proteins known to be phosphorylated in DON-induced RSR, such as p38, ERK1/2, and JNK1/2, were not identified in our proteomic analysis. Although sensitivity is a limitation of the proteomics approach, traditional Western blot analyses are also limited by the largely intuitive nature of candidate selection. Proteomics provides a global view of the key signaling pathways and biological processes affected in DON-induced RSR. Recognition of key phosphorylation events facilitates future targeted exploration of proteins impacted by DON, such as elements of PI3K/AKT pathway or tight junctions in leukocyte populations. It further underlines the importance of melding large-scale proteomics discoveries with previous small-scale studies when modeling the signaling networks mediating DON-induced immunotoxicity.

Compared with metabolic labeling, dimethyl labeling, as employed in this study, involves an addition chemical derivatization step and, therefore, could introduce additional sample handling and concomitant sample loss, which might compromise the coverage of the phosphorylation events. It should be further emphasized that, except for those proteins that are discussed above, the functions of many phosphosites identified here have not yet been functionally characterized, which prevents further functional interpretation of the effect of the phosphorylation changes on immunotoxicity. Functional annotation of the phosphoproteome awaits further studies, and characterization of phosphosites is often confined to protein location without understanding their role in relation to cell functions. Full characterization of phosphosite function will require genetic mutation of these sites to genes encoding other amino acids; however, it exceeds the scope of this study. Nevertheless, our research provides a hypothesis-generating tool and inventories proteins with changed phosphorylation status for future focused studies.

Taken together, this investigation provides a foundation for future exploration of the function of phosphoproteins identified and their associated pathways in DON-induced RSR in the spleen. This study advances the knowledge in the molecular mechanisms of the early development of immunotoxicity elicited by an immunotoxicant frequently encountered in human and animal food. Over the long term, these findings could ultimately contribute to designing strategies for the treatment and prevention of the adverse effects of DON and other ribotoxins.

SUPPLEMENTARY DATA

Supplementary data are available online at http://toxsci.oxfordjournals.org/.

FUNDING

National Institutes of Health (ES003358); USDA NIFA Award 2011-D635 and Wheat and Barley SCAB Initiative Award 59-0206-9-058.