ABSTRACT
Arenaviruses are important emerging pathogens and include a number of hemorrhagic fever viruses classified as NIAID category A priority pathogens and CDC potential biothreat agents. Infection of guinea pigs with the New World arenavirus Pichindé virus (PICV) has been used as a biosafety level 2 model for the Lassa virus. Despite continuing research, little is known about the molecular basis of pathogenesis, and this has hindered the design of novel antiviral therapeutics. Modulation of the host response is a potential strategy for the treatment of infectious diseases. We have previously investigated the global host response to attenuated and lethal arenavirus infections by using high-throughput immunoblotting and kinomics approaches. In this report, we describe the differential nuclear proteomes of a murine cell line induced by mock infection and infection with attenuated and lethal variants of PICV, investigated by using two-dimensional gel electrophoresis. Spot identification using tandem mass spectrometry revealed the involvement of a number of proteins that regulate inflammation via potential modulation of NF-κB activity and of several heterogeneous nuclear ribonuclear proteins. Pathway analysis revealed a potential role for transcription factor XBP-1, a transcription factor involved in major histocompatibility complex II (MHC-II) expression; differential DNA-binding activity was revealed by electrophoretic mobility shift assay, and differences in surface MHC-II expression were seen following PICV infection. These data are consistent with the results of several previous studies and highlight potential differences between transcriptional and translational regulation. This study provides a number of differentially expressed targets for further research and suggests that key events in pathogenesis may be established early in infection.
The arenaviruses are small, enveloped, bipartite RNA viruses with a unique ambisense coding strategy. A number of arenaviruses are important human pathogens, including the type species, Lymphocytic choriomeningitis virus (LCMV), which has recently been implicated in a number of posttransplant fatalities (3, 13, 26), and several hemorrhagic-fever-causing viruses, such as Lassa virus, Junin virus, and Machupo virus. The hemorrhagic arenaviruses are NIAID category A priority pathogens and CDC potential biothreat agents. Lassa virus is endemic in West Africa, where it causes significant morbidity and mortality (58). Currently, the only treatments available for these infections are supportive care and the broad-spectrum antiviral ribavirin, which must be administered within the first week of infection to be most efficacious (47). Little is known about the molecular basis of the pathology, and this has hindered the design of novel therapeutics.
Pichindé virus (PICV) is a New World arenavirus that infects guinea pigs, resulting in hemorrhagic fever. Two passage variants of this virus have been developed that cause either a mild, self-limiting infection (the P2 variant) or a severe hemorrhagic fever (the P18 variant) (33). This matched virus pair allows the differential signaling events and responses following infection to be determined. Understanding these events may allow us to develop novel therapeutic strategies which act to modulate the response seen following infection with the virulent virus in order to simulate the cellular responses seen following infection with the attenuated virus that lead to viral clearance and recovery.
Previous research suggests that the severity of arenavirus pathogenesis may be due to dysregulation of innate immune signaling and the cytokine response (6, 8, 9, 16, 17, 24, 59); however, we do not yet fully understand how arenaviruses induce this dysregulation or how this leads to disease. Modulating the host immune response to a pathogen may well prove to be an efficacious form of treatment against a number of infections, and current strategies are focused on a broad-spectrum “boosting” of the immune response to a pathogen (2). However, some infections, in which the disease is immune response mediated, may be exacerbated by this approach, and so a more-complete understanding of specific host-pathogen interactions is required. Also, as the type of broad-spectrum immune modulation required may change at different stages of infection, an understanding of how host cell signaling changes over time is also needed. Many viruses modulate cell signaling pathways to induce a cellular state that can facilitate productive infection (4, 7) or to evade the immune response (29, 42, 62). A more-complete understanding of the interactions between virus and host may allow the development of novel broad-spectrum antiviral therapeutics that act at the level of the host.
We have previously shown a number of differential cell-signaling events in attenuated and virulent PICV infections by using a murine macrophage-like cell line, as a guinea pig macrophage cell line is not available. While mice do not succumb to PICV infection, this experimental system has revealed significant differences between P2 and P18 infections. Our initial global study was at the level of the proteome, using high-throughput immunoblotting (8); this was followed by a kinomics analysis in which the activities of cellular kinases following infection were investigated (9). While both of these approaches have allowed us to determine many of the cellular signaling events which may be involved in pathogenesis, the assays have an intrinsic bias in the selection of the proteins and substrates whose level or activity was measured. A recently published study used genomics arrays to assay gene expression profiles following the infection of macaques with a virulent or attenuated variant of LCMV (22). This study has further complemented our understanding of the cellular response to arenavirus infection. We sought to expand our knowledge of these virus-host interactions in an unbiased fashion by using two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) to assay the differential proteomes induced by these infections and over time and to determine the identity of the spots that showed the greatest differences by using tandem mass spectrometry (MS-MS) peptide fragment sequencing. An understanding of the phosphoproteome, proteome, and transcriptome will allow us to understand both the transcriptional and translational regulation of the host response to arenavirus infection.
For this report, we used 2D-PAGE to investigate the host cell response to infection with attenuated and virulent isolates of PICV in vitro over a time course of infection. Protein identification using MS-MS revealed a number of proteins implicated in the regulation of the central mediator of inflammation NF-κB. We also show that heterogeneous nuclear ribonuclear protein (hnRNP) family proteins show significant differential expression in attenuated and lethal infections and provide promising avenues for further investigation.
MATERIALS AND METHODS
Cell culture and viruses.P388D1 (murine macrophage-like) cells were maintained in RPMI medium supplemented with 5% fetal bovine serum and 2 mM glutamine. Cells were infected with purified P2 or P18 PICV at a multiplicity of infection of 1 or mock infected with an equivalent fraction of virus purification medium. Cells were harvested at various times postinfection, and cytoplasmic and nuclear extracts prepared. Time points are relative to the time of initial virus addition to the culture medium.
Viruses were from serially spleen-passaged stocks from inbred strain 13 guinea pigs infected with PICV Munchique strain CoAn 4763 (72). Virus was quantified in a standard plaque assay on Vero cells as described previously (5). Viruses were purified by using polyethylene glycol gradients to remove potential contamination from cytokines and other soluble factors.
Cell extract preparation.Cytoplasmic and nuclear extracts were prepared from P388D1 cells as previously described (23), with the addition of a nuclear purification step using Optiprep (Axis-Shield) gradients. Briefly, lysates were underlayered with 10 ml 30% Optiprep and 5 ml 35% Optiprep and centrifuged at 4°C for 30 min at 4,300 × g. The interface was removed and placed in a fresh tube which was filled with sucrose buffer I as previously described (27), plus 1.5 mM CaCl2. Following centrifugation at 4°C for 15 min at 1,900 × g, the pellet was resuspended in sucrose buffer I and the centrifugation repeated. Nuclear lysis was completed by following the referenced protocol. Protein concentrations were determined by using a bicinchoninic acid assay kit (Pierce Biotechnology, Rockford, IL).
Electrophoresis and immunoblotting.Protein samples were denatured in sodium dodecyl sulfate (SDS)-PAGE loading buffer (100 mM Tris, pH 6.8, 4% SDS, 5% 2-mercaptoethanol [vol/vol], 30% glycerol) and boiled for 5 min. A 5-μg sample per lane was loaded onto Invitrogen Novex 10% gels (Invitrogen, Carlsbad, CA); the gels were electrophoresed, and then transferred to polyvinylidene difluoride membranes (Millipore, Billerica, MA) and immunodetected following standard protocols. Antibodies to hnRNP A2/B1 and A1 were purchased from Abcam (Kendall, MA).
Separations and differential analysis.Nuclear extracts were used for proteomics analysis. Samples with a low protein concentration were concentrated on Microcon 10-kDa molecular-weight-cutoff (Millipore, Billerica MA) spin filters, according to the manufacturer's protocol, prior to 2D-gel electrophoresis. All samples were desalted on spin desalting columns (Pierce Biotechnology, Rockford, IL) prior to isoelectric focusing (IEF). A 200-μg amount of protein was used for each sample replicate. DeStreak rehydration buffer and 0.5% immobilized pH gradient (IPG) buffer, pH 3 to 10, (GE Healthcare) were added to each replicate for rehydration of 11-cm, pH 3 to 10 IPG strips (GE Healthcare). IEF was accomplished in a stepwise fashion by first applying 50 V for 11 h to assist in rehydration of the IPG strips, followed successively by 250 V for 1 h, 500 V for 1 h, 1,000 V for 1 h, 8,000 V for 2 h, and, finally, 8,000 for 6 h.
Following IEF, the strips were frozen at −80°C. Strips were then thawed in 4 ml 6 M urea, 50 mM Tris, 2% SDS, 20% glycerol, and 5 mM Tris (2-carboxyethyl) phosmine for 15 min. The strips were then incubated in the same buffer and 25 mg/ml iodoacetamide for 15 min. The strips were rinsed in running buffer and placed in IPG wells of 8-to-16% Tris Criterion polyacrylamide gels (Bio-Rad, Hercules, CA). Agarose 0.5% was overlaid, and the second dimension was run at 150 V for 2.25 h at 4°C. The running buffer was 25 mM Tris, 192 mM glycine, 0.1% SDS, pH 8.3. The gels were fixed overnight in 50% methanol, 10% acetic acid. The gels were stained with Pro Q Diamond stain, followed by Sypro ruby gel stain (Molecular Probes, Eugene, OR), according to the manufacturer's protocol. The gels were imaged either on a Fuji FLA-5100 laser scanner or a Perkin Elmer ProXpress proteomics imaging system.
Forty-five nuclear extracts from mock-, P2-, and P18-infected cells at 2 h, 4 h, 8 h, 12 h, and 16 h in triplicate were prepared for 2D electrophoresis. However, five samples had insufficient protein for 2D analysis. This led to the exclusion of the 12-h time point from further analysis. The 40 2D gels were analyzed by using Progenesis Discovery software, version 2005 (Nonlinear Dynamics Ltd., Newcastle upon Tyne, United Kingdom). The software automatically detects spots on each gel based upon a proprietary algorithm. Spot filtering and editing is then manually performed to remove those spots whose intensity values are insufficient for identification via mass spectrometry. Average gels were created from the four gels of each type of treatment, and the maximum number of gels where the spots may be absent was one in four. The automatic matching of spots between the gels was manually reviewed and adjusted as necessary. Due to the large number of gels in this experiment, the analysis for each time point was performed separately for all the conditions to ensure that the maximum number of spots were accurately matched between gels and conditions. Normalization of spot volumes was then performed on the gels based on total spot volume for each gel. The ratio of normalized spot volumes was then calculated across all relevant average gels.
Spots for subsequent identification were selected based upon a relative abundance of more than twofold. The protein gel spots were excised and prepared for matrix-assisted laser desorption ionization (MALDI) MS-MS analysis by using ProPic and ProPrep robotic instruments (Genomic Solutions, Ann Arbor, MI), following the manufacturer's protocols. Briefly, dried gel pieces were incubated with 20 μg/ml trypsin (Promega, Madison WI) in 25 mM ammonium bicarbonate, pH 8.0, at 37°C for 4 h. The peptide solution is drawn through a ZipTip by the ProPrep robot, and the peptides eluted with 50% acetonitrile-water containing α-cyano-cinnapinic acid and spotted onto an MALDI plate for MS analysis.
MS.MALDI-tandem time-of-flight (TOF-TOF) MS on an Applied Biosystems 4800 MALDI TOF/TOF proteomics analyzer (Foster City, CA) was used to analyze the samples and determine protein identification. The Applied Biosystems software package included 4000 series Explorer (version 3.6 RC1) with Oracle database schema version (version 3.19.0) and data version (3.80.0) to acquire both MS and MS-MS spectral data. The instrument was operated in positive ion reflectron mode, the mass range was 850 to 3,000 Da, and the focus mass was at 1,700 Da. For MS data, 1,000 to 2,000 laser shots were acquired and averaged from each sample spot. Automatic external calibration was performed by using a peptide mixture with reference masses of 904.468, 1,296.685, 1,570.677, and 2,465.199 Da.
Following MALDI MS analysis, MALDI MS-MS was performed on several (4-9) abundant ions from each sample spot. A 1-kV positive-ion MS-MS method was used to acquire data under post-source decay conditions. The instrument precursor selection window was ±3 Da. For MS-MS data, 2,000 laser shots were acquired and averaged from each sample spot. Automatic external calibration was performed by using reference fragment masses of 175.120, 480.257, 684.347, 1,056.475, and 1,441.635 Da (from a precursor mass of 1,570.700 Da).
Applied Biosystems GPS Explorer (version 3.6) software was used in conjunction with MASCOT to search the respective protein databases using both MS and MS-MS spectral data for protein identification. Protein match probabilities were determined by using expectation values and/or MASCOT protein scores. MS peak filtering included the following parameters: mass range, 800 Da to 4,000 Da; minimum signal/noise filter = 10; mass exclusion list tolerance = 0.5 Da; and the mass exclusion list (for some trypsin- and keratin-containing compounds) included masses of 842.51, 870.45, 1,045.56, 1,179.60, 1,277.71, 1,475.79, and 2,211.1. For MS-MS peak filtering, the minimum signal/noise filter was 10.
For protein identification, the mouse taxonomy was searched in either the mouse NCBI or SwissProt database. Other selection parameters included the following: enzyme as trypsin; maximum missed cleavages = 1; fixed modifications included carbamidomethyl (C) for 2D-gel analyses only; variable modifications included oxidation (M); precursor tolerance was set at 0.2 Da; MS-MS fragment tolerance was set at 0.3 Da; mass = monoisotopic; and peptide charges were only considered as +1. The significance of a protein match, based on both the peptide mass fingerprint in the first MS and the MS-MS data from several precursor ions, is based on expectation values; each protein match is accompanied by an expectation value. The expectation value is the number of matches with equal or better scores that are expected to occur by chance alone. The default significance threshold is a P value of <0.05, so an expectation value of 0.05 is considered to be on this threshold. We used a more-stringent threshold of 10−3 for protein identification; the lower the expectation value, the more significant the score.
Student's t test significance testing was not performed on these protein spots for multiple reasons. First, our sample size was small; we had either three or four replicates for each protein spot, so the statistics of such a small sample size was underpowered for this study. Second, the outline shape of the same protein spot as defined by the Progenesis software varied from one 2D gel to the next, further complicating statistical comparison.
Ingenuity pathway analysis.The Ingenuity pathway analysis software (IPA) tool has been described in detail previously (11). Briefly, IPA consists of a large, manually curated knowledge base of cellular interactions and regulatory events mined from the peer-reviewed literature. An experimental data set can be used to query this knowledge base, and IPA will construct functional signaling networks based on known interactions within the knowledge base. The software uses Fisher's exact test to assign a statistical score of the probability that these in silico networks could be generated by chance. For our studies, a score of 5 (P = 10−5) was used as the cutoff for significance. Networks were constructed using the Ingenuity knowledge base as the reference set.
Electrophoretic mobility shift assay.The following single-stranded oligonucleotides were synthesized by BioSynthesis, Inc. (Lewisville, TX): DRAX, CCTAGCAACAGATGCGTCATCTCAAAA, and DPBX, CCTAGTGAGCAATGACTGATACAAAGC. Duplex oligonucleotides were annealed at 1.75 μM in 10 mM Tris-HCl (pH 7.6), 2 mM MgCl2, 50 mM NaCl, 1 mM EDTA by heating to 95°C and cooled slowly. Probes were labeled with γ-32P by using T4 polynucleotide kinase (Promega, Madison, WI) under standard reaction conditions. A 5-μg amount of nuclear extract was incubated with 0.1 pmol of labeled oligonucleotide probe in a 15-μl volume for 15 min under standard reaction conditions [20 mM HEPES (pH 7.5), 50 mM KCl, 2.5 mM MgCl2, 20 mM dithiothreitol, 10% glycerol with 0.5 μg poly(dI-C) per ml]. Samples were then resolved by electrophoresis on a standard 6% acrylamide gel in 0.25× Tris-borate-EDTA. The gels were then dried and imaged on a Packard Instant Imager and by exposure to Autorad film (ISC Bioexpress, Kaysville, UT).
Flow cytometry.P388D1 cells were cultured in six-well plates and mock infected or infected with P2 or P18 PICV for 24 h. Cells were then harvested and washed in fluorescence-activated cell-sorting buffer (phosphate-buffered saline plus 0.05% sodium azide, 0.5% bovine serum albumin, and 2% fetal bovine serum). Cells were stained with anti-major histocompatibility complex II (MHC-II) conjugated to fluorescein isothiocyanate (eBioscience, San Diego, CA), washed three times in fluorescence-activated cell-sorting buffer, and analyzed by using a BD Canto flow cytometer (BD Biosciences, San Jose, CA). Data were analyzed by using FCS Express (De Novo Software, Los Angeles, CA).
RESULTS
Analysis of 2D-gel electrophoresis.Cell extracts from mock-infected or PICV P2 or P18 variant-infected P388D1 cells from triplicate cultures were prepared at 2, 4, 8, 12, and 16 h postinfection. While cell extracts from infected guinea pigs would have been preferable, the large amount of protein required for this study (milligram quantities) meant that in vitro infections were required. Additionally, at the time of this study the guinea pig genome had not been sequenced and so proteins could not be definitively identified by MS-MS peptide sequencing. Although mice infected with PICV do not exhibit symptoms, our previous studies have shown that murine cells infected in vitro exhibit similar responses in the signaling pathways that we have compared with those of cells from primary guinea pig macrophages, support viral replication, and show significant differences in their responses to P2 and P18 infections (8, 9, 24). Samples were resolved by 2D electrophoresis; triplicate gels were run per sample. There was insufficient protein in the 12-h samples to run triplicate gels, and so this time point was excluded from further analyses. Spots were considered to be up- or downregulated if the difference was seen on duplicate gels and was greater than twofold. A summary of the changes identified is shown in Fig. 1. As can be seen, infection with the P18 variant induced fewer total differences from mock infection than infection with the attenuated P2 variant at all time points, although P18 did induce more downregulation at 4 and 8 h postinfection. This finding is consistent with the results of our previous kinomics study, which showed significantly more-complex signaling networks induced during attenuated P2 infection in vivo (9). Of particular interest is the finding that the disparity between P2- and P18-induced changes is greatest at the earliest time point (2 h). This is consistent with our hypothesis that critical events early in infection are likely to be of key importance in determining the host's fate. Figure 1 also shows that there is a greater number of differences between P2 and P18 infections than between infection with either virus variant and mock infection at 2, 4, and 8 h postinfection. This result shows that, despite having very few genetic differences (72), the P2 and P18 variants of PICV produce strikingly different host responses.
Summary of protein level changes over time following infection with P2 or P18 PICV. All up- and downregulated spots with a twofold or greater difference which were identified in duplicate or triplicate gels were tallied for each time point. Spots which were present or absent in one gel were recorded as upregulated or downregulated, respectively. Comparisons refer to the observation for the virus-infected sample or, in the case of P2 versus P18, for the P2-infected sample; M refers to mock infection. The y axis shows the number of >twofold differences. Individual protein changes can be found in Table S1 in the supplemental material. v, versus; hpi, hours postinfection.
Another point of interest is the difference in the number of proteins which show expression or modification changes following P2 or P18 infection. For example, at 2 h postinfection, the P2 variant induces an approximately equivalent number of upregulated and downregulated proteins; in contrast, P18 infection induces mainly upregulation of proteins. At 4 h postinfection, the cellular response to both P2 and P18 infection appears to predominantly involve upregulation of host proteins. By 8 h postinfection, the pattern is very different, with P2 infection resulting in a predominant upregulation of cellular proteins, similar to P18 at 2 h postinfection, and with P18 exhibiting a majority of downregulated events. By 16 h postinfection, both viruses produce a broadly similar response, with the vast majority of protein level changes being upregulation; accordingly, at this time point there are few significant protein level differences between P2 and P18 infections.
Spot identification by MS-MS.We next attempted to identify the proteins which showed differential expression between infections. We selected 190 spots for sequencing using MALDI-TOF-TOF MS-MS. This number included all of the spots that showed differential expression in P2 and P18 infections at all time points (118 spots) and the spots which showed the greatest increases or were present/absent between mock, P2, and P18 infections. A summary of all significant spot identifications (expectation value of <10−3 and/or protein score of >53) for all treatments and time points is shown in Table S1 in the supplemental material. As can be seen there, the majority of proteins identified with high confidence are members of the hnRNP family. Additionally, other key cellular proteins likely to be important in the cellular response to infection, including proteins involved in energy production and several heat shock proteins, were also identified. Figure 2 shows representative gels showing identified proteins from mock- and P2-infected cells. Unfortunately, a significant number of proteins sequenced were not identified with significant confidence scores or expectation values. We hypothesize that this was in part due to protein degradation due to the extensive time taken to perform the appropriate image warping and spot-matching analysis on the large number of gels used in this study.
Representative gels showing proteins identified by MS-MS following 2D-PAGE. 2D electrophoresis was performed in triplicate on nuclear extracts from mock-, P2-, and P18-infected cells harvested at various times postinfection. Expression changes of proteins which showed >twofold differences between treatments in two or three of the gels were considered significant. One hundred ninety spots were selected for protein identification by MS-MS. The figure shows representative gels from mock- and P2-infected samples. The spots which were identified as having an expectation threshold of <10−6 are annotated. The proteins shown here correspond to the first two sections in Table S1 in the supplemental material. Boxed numbers are the spot identification numbers allocated by the software.
Construction of functional signaling networks.Due to the small number of proteins identified with high confidence in each comparison at each time point, the proteins identified for each comparison (mock- versus P2-infected cells, mock- versus P18-infected cells, and P2- versus P18-infected cells) at all time points were combined into a single data set using the SwissPROT accession number (http://www.expasy.org/sprot/ ) and severalfold-change value. Protein pathways that met the significance threshold were merged to produce the networks shown in Fig. 3.
Pathway analysis reveals protein networks and key “hubs” of interaction. Significant proteins identified at all time points for each comparison were uploaded to the Ingenuity pathway analysis software using SwissPROT identifiers. Networks which met the cutoff threshold (score of >5) were merged to form the interaction networks shown. Proteins shown backed by gray symbols are those which were present in the data set, and proteins backed by white symbols were included in the networks on the basis of known interactions. Shaded regions highlight the subnetworks of interaction within each global network. Proteins in shaded regions are those directly identified from differentially expressed spots, and proteins in unshaded regions are those brought into the networks on the basis of known interactions mined from the literature. The networks are described in Results. (a) Mock versus P2. (b) Mock versus P18. (c) P2 versus P18.
Analysis of the mock versus P2 data set produced one significant (score of 39) network with protein functions related to lipid metabolism, molecular transport, and small-molecule biochemistry (Fig. 3a). Interestingly, the in silico-created network incorporated the cystic fibrosis transporter protein, which we have previously shown by kinome analysis to be involved in PICV-induced networks (9). The network also included the Fos proto-oncogene, included in our earlier networks constructed by high-throughput immunoblotting (8), the heat-shock protein 70 complex, and the transcriptional regulator CCAAT-enhancer binding protein alpha. The mock versus P18 data set produced a significant network (score of 30), shown in Fig. 3b, with protein roles in cellular assembly and organization, carbohydrate metabolism, cell signaling, and energy metabolism. As can be seen, there is a discrete, highly interconnected subnetwork involved in energy production and another, broader network, which shows a lower degree of interconnectivity, including the proto-oncogene c-Myc, previously identified in our kinome assay, and an hnRNP family member which can function as a transcription factor, hnRNP K.
As characterizing the pathogenesis-associated differences between P2 and P18 infections was the main aim of this study, all P2 versus P18 differences were sequenced. Thus, the data set constructed from the P2 versus P18 differences was considerably larger than that produced from data from infection with either virus compared to mock infection. Analysis of these data produced three significant networks (scores of 25, 25, and 13) which were merged to produce an integrated signaling network (Fig. 3c). As with the network produced for mock versus P18 infection, discrete subnetworks were observed, with roles in energy metabolism, carbohydrate metabolism, and molecular transport. As with previous analyses, comparison of the P2- and P18-induced proteomes produced signaling networks which included central nodes, such as c-Myc and c-Fos. The transcription factor SP1, previously shown to be differentially activated in P2 and P18 infections (9), is also a central, highly interconnected node in this analysis. The hnRNP K protein shows a high level of interconnectivity in this network. Interestingly, the nucleolar phosphoprotein nucleophosmin is incorporated into this network. Nucleophosmin has been shown to be an NF-κB coactivator (21, 40), suggesting a further mechanism for NF-κB regulation, consistent with the results of our earlier studies in which we identified differential NF-κB activity in P2 and P18 infections (24). Several other proteins identified in these networks, such as the insulin receptor (INSR) and annexin A2 (ANXA2), were found in our earlier network analyses.
We next looked at the functional roles of the differentially expressed/modified proteins in order to better understand the potential roles of these proteins in disease. Figure 4 shows significance scores of how these proteins map to known functional processes and diseases. The histogram represents a significance score, indicating the number of proteins which correspond to the particular process rather than an indication of activity. Both P2 and P18 infection led to significant changes in proteins involved in carbohydrate metabolism, energy production, cell signaling, and RNA trafficking compared to the results for mock infection. Of interest is the increased number of proteins induced by P2 that are involved in amino acid metabolism and protein degradation, folding, and synthesis. Consistent with the disease's pathology, a significant number of proteins with roles in hepatic-system disease, immune development, response and disease, and inflammatory disease are also differentially expressed in P2 and P18 infections. These functions centered on HSPD1/Hsp60 (downregulated in P2), HSPA5/78-kDa glucose-regulated protein (upregulated in P2), and hnRNP A2/B1 (all spots downregulated in P2).
Functional significance of identified proteins. The Ingenuity pathway analysis software was used to perform a comparison analysis on the proteins that were differentially expressed/modified in P2- or P18-infected cells compared to their levels in mock-infected cells. These proteins are assigned roles in functional and disease processes on the basis of published reports. The bar represents the statistical involvement of proteins in these processes rather than an increase or decrease in process activity. The dashed line indicates the threshold of statistical significance.
Cytoplasmic shuttling of hnRNP proteins.As the majority of the spots we identified were hnRNP family members (representative gels are shown in Fig. 5) and these spots showed differential expression between infections, we further investigated the subcellular location of these proteins during time courses of infections by nuclear/cytoplasmic fractionation of infected cells and detection of hnRNP A2/B1 by immunoblotting (Fig. 6). Of particular interest is the observation that at 2 h postinfection, hnRNP A2/B1 is nuclear in mock- and P18-infected cells, whereas the protein localizes to the cytoplasm following infection with the attenuated P2 variant. At later stages of infection, P2 and P18 induce similar patterns of hnRNP A2/B1 expression and localization. Similar nuclear localization and expression was observed for hnRNP A1 (data not shown). These results are consistent with many of our previous data showing that P18-infected cells resemble mock-infected cells at early times postinfection, suggesting a failure to activate and/or active suppression of host cell responses.
Example of differential hnRNP expression in P2- and mock-infected cells. Representative gels showing differences in hnRNP expression. The spots found to be hnRNP family members in the 2-h-time-point gels were identified with the Progenesis software. The spots shown in the figure correspond to the hnRNP proteins shown in the 2-h section of Table S1 in the supplemental material.
hnRNP proteins show differential subcellular locations in P2- and P18-infected cells. P388D1 cells were mock (M), P2, or P18 infected in triplicate, and nuclear and cytoplasmic extracts prepared at various times postinfection. Samples were denatured in SDS-containing buffer and resolved by PAGE. Samples were immunoblotted with antibodies against hnRNP A1 and hnRNP A2/B1.
Differential activity of XBP-1.Network analysis of proteins which showed differential expression in P2 and P18 infections (Fig. 3c) implicated the transcription factor X-box binding protein 1 (XBP-1), a transcription factor involved in transactivating the MHC-II promoter. XBP-1 is a leucine zipper protein and is structurally similar to the AP-1 members c-Fos and c-Jun (41). To further investigate the role of this protein in PICV infection, nuclear extracts were prepared from P2- and P18-infected cells at 4 and 16 h postinfection and used to assay differences in binding to two probes previously characterized as XBP-1 binding sites (53). Differential binding was assayed by mobility shift assay (Fig. 7a). For both probes, P2 infection induced increased binding of the upper band (marked) at 4 h postinfection compared to the levels of binding induced by mock and P18 infections. By 16 h postinfection, both P2 and P18 led to increased levels of binding of the upper band.
P2 and P18 infection induced differential XBP-1 binding to DNA and MHC-II surface expression. (a) Nuclear extracts from mock-, P2-, and P18-infected cells were harvested at 4 and 16 h postinfection and assayed by gel shift assay for their ability to bind two MHC-II promoter elements. Probes DRAX and DPBX are described in Materials and Methods. M, mock infection. (b) Surface expression of MHC-II on mock-, P2-, or P18-infected macrophages was assayed at 12 and 48 h postinfection by flow cytometry.
To determine whether these changes in MHC-II promoter binding led to a change in the surface expression of MHC-II, cells were mock infected or infected with P2 or P18 PICV for 12 or 48 h, and the levels of surface expression of MHC-II were determined by flow cytometry. Figure 7b shows an increase in MHC-II type I-Ek induced by P2 infection by 12 h compared to the levels induced by mock and P18 infections (median fluorescence intensity of 2,876, compared to 1,920 and 1,737 for mock and P18 infections, respectively). By 48 h, infection with P18 induces a striking increase in the surface expression of MHC-II, with P2 also inducing increased expression, with a more-heterogeneous population of MHC-II upregulation as shown by the wider distribution following staining with both MHC-II antibodies. These results correlate with the early change in XBP-1 binding induced by P2 infection, consistent with the early increase in MHC-II expression induced by P2 infection.
DISCUSSION
In this report, we have investigated the host response to attenuated and virulent arenavirus infection using 2D electrophoresis of the nuclear proteome and MS-MS. We have identified with high confidence proteins which show differential expression or modification in three classes of pairwise comparison and over a time course of infection. By using pathway analysis, we have illustrated how these identified proteins may be involved in functional signaling networks. While our protein identification revealed a number of proteins which are canonically cytoplasmic, many of the proteins, such as ATP synthase (ATP5A) or EIF5A, can also be found in the nucleus or its membrane. Additionally, we cannot rule out contamination of the nuclear fraction with mitochondrial proteins or newly synthesized proteins on the rough endoplasmic reticulum. However, the comparative nature of this study and the fact that all samples were treated identically may mean that any changes in contaminating proteins are indicative of the changes in their levels in infected cells. Pursuant to this is the large number of proteins involved in energy metabolism which showed significant increases following infection; this finding is consistent with cells increasing energy production in response to challenge, a result also seen following inflammatory responses induced by lipopolysaccharide treatment (11).
A finding of particular interest and a result which shows an intrinsic advantage of nonbiased, global approaches in understanding virus-host interactions is the identification of several proteins, which are not often studied in this field, which may prove to be of importance in furthering our understanding of viral pathogenesis. In this study, we have identified several proteins involved in regulating NF-κB activity and the inflammatory response, a central component in hemorrhagic-fever pathogenesis.
One group of proteins which may play a central role in regulating the inflammatory response in arenavirus infections are the 70-kDa heat shock proteins (HSP70). We have previously identified the HSP70 HSPA5 protein as downregulated following infection with PICV P2 (8); our kinomics analysis showed that the peptide corresponded to the phosphorylation sequence of the HSP70 homologue in Escherichia coli (9). This finding demonstrates that the kinase(s) upstream of HSP70 in guinea pig macrophages are active and capable of phosphorylating their target site. A recent study by Djavani et al. has shown by using microarray analysis that a number of HSP family mRNAs are differentially transcribed during LCMV infection (22). Interestingly, infection with both the virulent WE and attenuated Armstrong LCMV strains led to 2- to 3.5-fold reductions in HSPA8 transcript levels. While this correlates with the decrease in protein levels seen following P2 infection in our previous report, it does not correlate with the lack of HSP70 protein reduction following P18 infection in this assay. There are a number of explanations for this observation. First, this could be due to the differences between a cell culture system and an animal model or a difference between the PICV and LCMV virus effects on the host cell. Another possible reason could be differences between transcriptional and translational regulation or a change in protein half-life and turnover.
The increased levels of HSPA8 following P2 infection presented in this report may seem to contradict our earlier report and that of Djavani et al. (22) which show decreases in both transcript and protein levels. However, as this study was performed on nuclear extracts, the increase in nuclear HSPA8 is indicative of activation (20, 36, 38), which is regulated by phosphorylation, resulting in nuclear translocation of the protein (14). Nuclear translocation of HSP70 has also been shown to be induced early following cytomegalovirus infection (50). HSP70 activation has been shown to have an anti-inflammatory effect. In cells in which HSP70 was activated prior to tumor necrosis factor alpha (TNF-α) treatment, NF-κB nuclear translocation and DNA binding were blocked, concomitant with a stabilization of IκB (71). The proteins described above are all involved in regulating NF-κB activation. As we have previously shown, NF-κB is differentially active in P2 and P18 infections (24). This observation, coupled with the fact that NF-κB is a central regulator of the immune response, makes this pathway a key target for therapeutic modulation. As this pathway has been extensively studied due to its role in a number of disease pathologies which are mediated by inflammation, a number of pharmacological agents have already been identified and may prove to be important in developing immunomodulatory treatments for infectious diseases. Figure 8 summarizes the results discussed above and illustrates how the proteins described may act to regulate NF-κB activity.
A regulatory network for the inflammatory response. Several proteins identified in this study (described in Discussion) with a known function in regulating the NF-κB response were used to construct a potential regulatory network for NF-κB. Grids next to the proteins indicate nuclear protein and transcriptome expression differences characterized in this study and that of Djavani et al. (22). The upper row of each box indicates changes in levels of nuclear proteins in this study, and the lower row transcriptional expression changes in LCMV-infected macaques in the study of Djavani et al.; red indicates upregulation, and green indicates downregulation. The columns represent, from left to right, mock versus attenuated infection, mock versus severe infection, and attenuated versus severe infection with PICV P2 and P18 in our study and LCMV Armstrong and WE in the transcriptome study. CREB3L1, cyclic-AMP-responsive element binding protein 3-like 1; VCP, valosin-containing protein; HSPA5, 78-kDa glucose-regulated protein precursor (heat shock 70-kDa protein 5); PPARG, peroxisome proliferative activated receptor gamma; NR3C1, nuclear receptor subfamily 3 group C member 1; CREBBP, CREB binding protein; Hsp70, heat shock protein 70 (group); NFKBIA, nuclear factor κB inhibitor α (IκB-α); RELA, NF-κB p65 subunit. Solid lines indicate direct interactions between proteins, broken lines indicate an indirect interaction, an ellipse represents a transcriptional regulator, a rectangle represents a ligand-dependent nuclear receptor, a diamond represents an enzyme, concentric circles represent a group or complex, and a single circle represents other functions.
Several other proteins identified may play additional roles in regulating inflammation. Valosin-containing protein (VCP) has been identified as a target of Akt (68), identified as a key signaling node in our earlier global analyses. A small interfering RNA knockdown of VCP was shown to inhibit NF-κB activation following growth factor stimulation (68); additionally, VCP has been shown to be a multiubiquitin chain-targeting factor and targets the ubiquitinated form of IκB, the inhibitor of NF-κB, to the proteasome for degradation (18, 19). VCP may also play a role in the apoptotic pathway induced by endoplasmic reticular stress (57). The transcriptional regulator CREB-binding protein 3-like 1 (CREB3L1), identified in the P2 versus P18 network, may also be involved in NF-κB regulation via its regulation of HSPA5. HSPA5 can inhibit the formation of the HSP90/glucocorticoid receptor complex, required for steroid binding (32) which leads to suppression of the inflammatory response via inhibition of NF-κB-mediated transactivation (10, 60). The glucocorticoid receptor (NR3C1) is also implicated in the mock infection versus P18 network, further suggesting a role for this protein in pathogenesis. Other CREB family proteins have also been shown to play a role in NF-κB signaling and the inflammatory response. The pleiotropic cytokine interleukin-6 (IL-6), important in the production of acute-phase proteins, is transactivated by NF-κB, which is sufficient for transactivation in response to TNF-α stimulation; however, maximal transactivation requires association with the CREB family protein CREBBP (CREBBP/CBP or CBP/p300) (67). Interestingly, CREBBP/p300 has been shown to play an important role in modulating the glucocorticoid receptor and in the NF-κB antagonism (48, 49) described above. The peroxisome proliferator-activated receptor (PPAR) family proteins are involved in regulating the NF-κB pathway. PPARγ is shown in the P2 versus P18 network. Interestingly, our previous study identified a differential significance for PPAR signaling between P2 and P18 infections in vivo (9). PPARγ has been shown to be involved in the inhibition of NF-κB activity by butyrate (61) and in reducing the nuclear translocation of p65 in response to IL-1 and gamma interferon treatment by inhibiting the degradation of IκB-α (35). The role of the PPAR pathway in infection requires further investigation since a more-complete understanding of the regulation of NF-κB in infection and its role in pathogenesis may be of critical importance in terms of developing therapeutics which act at the level of host response modulation.
An interesting finding of this study was the inclusion of the transcription factor XBP-1 in our signaling networks. This transcription factor is involved in the transactivation of the MHC-II promoter. Downregulation of surface MHC-II on dendritic cells has previously been reported, with a virulent variant of LCMV leading to a significant reduction in MHC-II expression compared to the levels of expression of dendritic cells infected with an attenuated variant or mock infected (63, 64). Our data show an increase in surface expression of MHC-II following PICV infection of murine macrophages and also show differential DNA-binding activity of a transcription factor involved in MHC-II expression. The difference between these findings may be due to differences between macrophage and dendritic-cell responses, the difference in cell type used for the study, differential effects of PICV and LCMV infections on host immune signaling, or differences due to high multiplicities of infection in in vitro infections and cells taken from infected animals in which only a proportion of cells are infected. The upregulation of MHC-II late in infection may, at least in part, explain why infected mice do not exhibit disease.
Consistent with the hnRNP shuttling data presented here, the P2/P18 differences appear to be most striking early in infection, suggesting that the initial effects of infection, such as binding and entry, may be important in determining appropriate immune response development and disease outcome. These data serve to link the early molecular-signaling events following infection with the known impairment in the adaptive immune response, such as is seen in LCMV infection and also at the level of antigen presentation (1). Now that the receptors for both Old World and New World hemorrhagic arenaviruses are known (12, 56), we can begin to investigate the downstream signaling pathways of these receptors for differential activation and which responses they induce following virus binding and during entry.
This study identified a large number of differentially expressed or modified hnRNP family members. While the fact that these proteins are abundant in the cell, particularly compared to proteins such as transcription factors, may mean that we are more likely to identify these proteins than others due to sampling effects, the fact that the differences between P2 and P18 infections are so striking means that these proteins may have a biologically significant role in determining pathogenesis. This conclusion is reinforced by the discovery of hnRNP proteins in our previous assay and network analysis (8). Validation of hnRNP expression by immunoblotting confirmed significant differences between P2 and P18 infections in hnRNP shuttling. However, the nuclear cytoplasmic shuttling data assayed by immunoblotting did not correlate with our observations from the 2D-PAGE analysis, in which we observed increased hnRNP A2/B1 in P2-infected cells at 2 h postinfection; there could be a number of reasons for this, such as the 2D results relating to highly specific posttranslationally modified forms of the protein or different splice variants. However, both assays show clear differences between the infections in hnRNP shuttling. Other viral systems have also been shown to induce relocalization of hnRNP proteins. Infection of cells with poliovirus has been shown to cause redistribution of the hnRNPs A1 and K from the nucleus to the cytoplasm (30). Additionally, this study showed that this relocalization was a direct effect of virus infection, rather than simply an indirect effect of transcriptional-repression-induced hnRNP cytoplasmic accumulation (55) mediated by poliovirus 3C protease cleavage of the cellular TATA-binding protein (15). Herpes simplex virus infection has also been shown to induce changes in hnRNP distribution within the nucleus, with the hnRNP moving from a diffuse nuclear distribution to a punctate staining pattern following infection (44).
hnRNPs can also bind and/or directly regulate viral RNAs. For example, hnRNP A1 has been shown to bind to the transcription-regulatory region of mouse hepatitis virus RNA (39) and regulate RNA synthesis (65). Additionally, hnRNP A1 is able to bind directly to the mouse hepatitis virus nucleocapsid protein (70). More recently, it was shown that multiple A/B hnRNPs can replace hnRNP A1 in mouse hepatitis virus RNA synthesis (66). Human immunodeficiency virus type 1 RNA trafficking is mediated by hnRNP A2, and it has also been shown that the immediate early gene 2 protein of human cytomegalovirus interacts with hnRNP A1. These studies, in addition to the identified roles for hnRNPs as transcription factors, suggest a number of ways in which hnRNPs could be involved in arenavirus infection. Further experiments will be required to define the role hnRNP proteins are playing in arenavirus infection and whether the functional effects of hnRNPs may be implicated in the differential pathogenesis of these viruses.
We have previously shown that the thioaptamer XBY-S2 is able to protect guinea pigs from lethal PICV infection (25). Interestingly, when XBY-S2 was used in a novel sandwich assay (69), it was also able to bind the hnRNP proteins A1, A2/B1, A3, A/B, and D0. The hnRNP A1 protein has been shown to be involved in regulating the NF-κB pathway by binding to IκB-α (31), suggesting a further mechanism by which hnRNPs may be important in mediating the host response to infection. While this may be an artifact due to the relative abundance of hnRNP proteins, we cannot exclude the possibility that this thioaptamer may act, if only in part, by binding to and modulating the function of hnRNP proteins.
This study suggests that hnRNPs are highly likely to be involved in partially determining the pathogenesis of PICV infection. Due to the involvement of hnRNPs in a number of viral infections, continued investigation of the role of hnRNPs in viral replication and pathogenesis may prove to be of key importance in the development of novel therapeutic approaches, as well as further our understanding of the complex, multifunctional role these proteins play in the molecular biology of the cell. In addition, this report has identified a number of proteins, differentially expressed in attenuated and virulent infections, which are associated with the regulation of the inflammatory response. These proteins, and the pathways in which they act, require further investigation to clearly determine their role in infection and to define targets for therapeutic modulation.
Lassa fever is often characterized by a delay in the production of antibody (34, 54), suggestive of an inhibition of the early innate responses that are required to develop an effective adaptive response. Consistent with this is the observation that the infection of macrophages with Lassa virus does not induce TNF-α expression and inhibits TNF-α production following lipopolysaccharide stimulation, whereas infection with the apathogenic Mopeia virus does not (43). This could be explained by an inhibition of NF-κB activation by virulent arenaviruses, as we have previously shown for PICV (24). Inhibition of IRF-3 has been reported for LCMV (45), providing another mechanism for inhibition of an innate immune signaling pathway. In this study, we have identified differential expression of nucleophosmin. This protein has been shown to inhibit PKR activity (28), CXCR4-mediated chemotaxis (73), and DNA binding of IRF-1 (37). IRF-1 is implicated in the control of numerous cellular genes with important roles in establishing early immune responses and mediating apoptosis, such as type I interferons, the TAP transporter, Fas ligand, caspases, IL-1, IL-8, IL-12, STAT 1 and 2, and 2′-5′-oligoadenylate synthetase. Modulation of the expression of these genes would have a profound effect on the development of the immune response and could, in combination with the other reported inhibitory effects, be a critical mechanism for establishing a pathogenic infection. Experiments are currently in progress to further characterize and define the role of nucleophosmin in arenavirus infection.
Interestingly, nucleophosmin, XBP-1, NF-κB p65 (previously shown to be perturbed by virulent arenavirus infection) (24), and IRF-1 have been implicated in a hormone response network (75). Transcriptional analysis of peripheral blood mononuclear cells from macaques infected with LCMV also identified changes in hormone response genes (22). Dysregulation of endocrine function has been reported for persistent LCMV infection following viral infection of the pituitary and inhibition of growth hormone (GH) production (51, 52). GH and gamma-aminobutyric acid have been shown to act in an autocrine fashion to regulate GH expression (27, 46, 74). Modulation of the responses to these hormones could be a mechanism for the reduction of GH seen in persistent LCMV infection. Given that hormone response networks have now been implicated at the genomic and proteomic level and nucleophosmin additionally identified at the kinomics level (9), further investigation of hormone response dysfunction and of the role of nucleophosmin in acute arenavirus disease is required.
We are now beginning to define arenavirus interactions with the host cell at a number of different levels, from the phosphoproteomic to the genomic. Further analysis of these datasets in combination will reveal how these pathways are regulated and how arenavirus infection modulates host responses to infection. This will allow future experiments to clearly define not only potential diagnostic and prognostic biomarker signatures but also key determinants of viral pathogenesis. As we begin to construct functional signaling networks and map virus-induced perturbations, we can begin to link these molecular changes to disease pathology at the tissue level and the level of the whole organism. This may identify novel targets for therapeutic intervention, particularly novel immunotherapies which aim to restore the functions inhibited by viral proteins.
ACKNOWLEDGMENTS
We acknowledge Barry Elsom for technical assistance and Susan Stafford, Zheng Wu, and Robert English of the Biomolecular Resource Facility at UTMB for assistance with the 2D gels and protein identification.
This work was supported by grants from the NIH (UO1 AI05487 and R01 A127744) and by a grant to N.K.H. through the Western Regional Center for Excellence for Biodefense and Emerging Infectious Disease Research (NIH grant number U54 AI057156).
D.G.G. and B.A.L. declare financial interests in AM Biotechnologies and AptaMed, UTMB spin-off companies involved in thioaptamer technologies, including those using thioaptamer XBY-S2.
FOOTNOTES
- Received 19 June 2008.
- Accepted 24 October 2008.
- Copyright © 2009 American Society for Microbiology