|Home | About | Journals | Submit | Contact Us | Français|
Where there is life, there are viruses. The impact of viruses on evolution, global nutrient cycling, and disease has driven research on their cellular and molecular biology. Knowledge exists for a wide range of viruses, however, a major exception are viruses with archaeal hosts. Archaeal virus-host systems are of great interest because they have similarities to both eukaryotic and bacterial systems and often live in extreme environments. Here we report the first proteomics-based experiments on archaeal host response to viral infection. Sulfolobus Turreted Icosahedral Virus (STIV) infection of Sulfolobus solfataricus P2 was studied using 1D and 2D differential gel electrophoresis (DIGE) to measure abundance and redox changes. Cysteine reactivity was measured using novel fluorescent zwitterionic chemical probes that, together with abundance changes, suggest that virus and host are both vying for control of redox status in the cells. Proteins from nearly 50% of the predicted viral open reading frames were found along with a new STIV protein with a homolog in STIV2. This study provides insight to features of viral replication novel to the archaea, makes strong connections to well described mechanisms used by eukaryotic viruses such as ESCRT-III mediated transport, and emphasizes the complementary nature of different omics approaches.
The relatively recent demarcation of archaea as a third domain of life and the exotic viruses associated with these organisms are currently very active research topics.1–5 Our understanding of archaeal viruses and how they interact with their hosts lags well behind viruses associated with bacterial and eukaryotic hosts and this is especially true of viruses that infect members of the Crenarchaea. The basic viral replication cycle is just beginning to be understood, even for the best described viruses that infect Sulfolobus spp., such as Sulfolobus spindle-shaped virus (SSV)6 and Sulfolobus islandicus rod-shaped virus (SIRV).7 What is clear at this point is that studies of archaea are bringing insight to evolution of the domains of life and exciting new biology such as virus-associated pyramids (VAPs) on infected cell surfaces and viruses that change morphology after release.8–10
STIV was originally isolated from enrichment cultures of a high temperature (~80°C) acidic (~pH 3) hot spring in Yellowstone National Park (YNP).11 It was the first icosahedral virus described from the archaeal domain of life. It can infect S. solfataricus (P2), originally isolated in Italy, as well as Sulfolobus species found in YNP. Structural models based on cryo-electron microscopy and image reconstruction revealed a capsid with pseudo T=31 symmetry, turret structures at each of the five-fold axes, and an internal lipid layer. 12 Surprisingly, STIV has a clear common ancestry at the structural level with both prokaryotic and eukaryotic viruses.13 Subsequent analysis determined that the capsid is composed of nine viral proteins and an internal layer of cyclic tetraether lipids.14 The 17.6 kb double stranded DNA genome has 37 open reading frames that code for proteins that, for the most part, lack homologs at the sequence level. Structural models based on X-ray diffraction are now available for 4 proteins including the major capsid protein.13, 15–17
A majority of the viruses with archaeal hosts are believed to be non-lytic.18–21 However, recent evidence that STIV is lytic suggests that this topic may need to be revisited.22 The basics of the STIV replication cycle and a transcriptome analysis of the host response to infection have been reported22, providing the first glimpse of an archaeal response to viral infection. The prior transcriptome analysis and the data presented herein are from the same sample set of a near synchronous infection of > 95% of S. solfataricus strain (SsP2-2-12). Transcription of STIV genes was evident 8 hours post infection (hpi) and peaked at 24 hpi, with little temporal variation of viral gene transcription. The microarray analysis of infection detected changes in expression for 177 host genes (~ 5%), with 124 up-regulated and 53 down-regulated. The up-regulated genes were primarily involved with DNA replication and repair or of unknown function, while the down-regulated genes were associated with energy production and metabolism. A surprising discovery was made from a time-course study of cells after STIV infection when dramatic pyramids appeared on the cell surface beginning at 32 hpi.8 Soon after, a second report of viral associated pyramids (VAPs) on the closely related S. islandicus infected by SIRV1 was published.9 An investigation of how infection is manifest at the protein level has yet to be reported on any of these systems.
The central objective of this study was to extend our understanding of the STIV-Sulfolobus viral-host interaction beyond the level of transcriptional regulation, using proteomics and activity-based protein profiling. To accomplish this, a time course analysis of the proteins that change in STIV infected S. solfataricus was performed using 1D and 2D-DIGE, followed by protein identification using mass spectrometry. Viral proteins were detected at 24 hpi and peaked at 32 hpi. Gel image analysis of protein abundance revealed nine host proteins that increased and one that decreased over the course of infection. This data was augmented with the first use of activity-based protein profiling in archaea. Viral proteins were found in both cytoplasmic and membrane fractions. No co-regulated changes at the protein and mRNA levels were found, a result that has been reported for other systems23, however, the identified cellular networks were the same. Additionally, whereas nearly half of the regulated genes were of unknown function, changes in the proteome highlight known proteins with strong connections to eukaryotic processes such as cell division and vesicular transport.
STIV production and purification was carried out by growing the S. solfataricus strain P2-2-12 in DSMZ Medium 182 (http://www.dsmz.de/microorganisms/medium/pdf/DSMZ_Medium182.pdf) at pH 2.5 as previously described.22
Batch cultures of three independent pairs of S. solfataricus strain P2-2-12 cultures were infected with STIV at a multiplicity of infection (MOI) of ~1.5 to 2 and sampled over time as previously described 22. Samples were removed from the cultures and processed in parallel for mRNA and protein assays. Cells used for proteomics were pelleted, resuspended in 0.5 ml of sterile PBS, flash-frozen in liquid nitrogen, and stored at −80°C until analysis. Microarray hybridizations and analysis were conducted as described in details by Ortman et al.22
Sample extracts were prepared from cell suspensions washed with PBS. Cells were lysed with a combination of freeze-thaw and lysis buffer (30 mM Tris-HCl pH 8.5 (4°C), 7 M urea, 2 M thiourea, 4% CHAPS, 50 mM DTT, 0.5% IPG carrier ampholytes and a cocktail of protease inhibitors) for 60 min. and the supernatant was clarified by centrifugation. Proteins were purified and concentrated by precipitation with a 5-fold volume of cold acetone, and re-suspended for 1hr in lysis buffer without DTT. Protein concentration was measured with the RC/DC Protein Assay Kit (Bio-Rad). Samples were kept frozen until use. Protein samples were labeled with CyDyes using the minimal labeling method according to the manufacturer’s protocol (GE Healthcare, http://www.gelifesciences.com/aptrix/upp00919.nsf/Content/EF8087141F244296C1257628001D0763/$file/18_1164_84_AB.pdf). Approximately 1 out of 100 lysine side chains are labeled. Briefly, 50 μg of each protein extract was labeled separately at 0°C in the dark for 30 min with 400 pmoles of the N-hydroxysuccinimide esters of cyanine dyes (Cy3 and Cy5 CyDyes;) dissolved in 99.8% DMF (Sigma). The internal standard, an equimolecular mixture of all the protein extracts, was labeled with Cy2. Labeling reactions were quenched by the addition of 1 μL of a 10-mM L-lysine solution (Sigma) and left on ice for 10 min. Cy2, Cy3 and Cy5 labeled samples were combined appropriately and mixed with rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS) containing 50 mM DTT and 0.5 % IPG.
2-DE was performed as described elsewhere24, using precasted IPG strips (pH 3–11 NL, non-linear, 24 cm length; GE Healthcare) in the first dimension (IEF). Labeled samples were combined with a maximum of 450 μl rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 0.5 % IPG buffer pH 3–11 NL, 40 mM DTT, and a trace of bromophenol blue) and loaded onto IPG strips. Typically, 150 μg proteins were loaded on each IPG strip and IEF was carried out with the IPGPhor II (GE Healthcare). Focusing was carried out at 20°C, with a maximum of 50 μA/strip. Active rehydration was achieved by applying 50 V for 12 h. This was followed by a stepwise progression of 500 V for 500 Vhr, gradient ramp from 500 to 1000 V for 1 h, gradient ramp from 1000 to 3000 V for 1 h, gradient ramp from 3000 to 5000 V for 1 h, gradient ramp from 5000 to 8000 V for 1 h, gradient ramp to 8000 V over 1 h, then 8000 V constant for a total of 44,000 Vhr. After IEF separation, the strips were equilibrated twice for 15 min with 50 mM Tris-HCl, pH 8.8, 6 M Urea, 30% glycerol, 2% SDS and a trace of bromophenol blue. The first equilibration solution contained 65 mM DTT, and 153 mM iodoacetamide was added in the second equilibration step instead of DTT. The strips were sealed on the top of the gels using a sealing solution (0.75% agarose in SDS-Tris HCl buffer). The second-dimension SDS-PAGE was performed in a Dalt II (GE Healthcare), using 1 mm-thick, 24-cm, 13% polyacrylamide gels, and electrophoresis was carried out at a constant current (45 min at 2 W/gel, then at 1 W/gel for ~16h at 25°C). When relevant, gels were stained with SYPRO® Ruby or Coomassie® Brilliant Blue stains for spot picking.
Gels were scanned using a Typhoon Trio Imager according to the manufacturer’s protocol (GE Healthcare) at 100 μm resolution. Images were subjected to automated difference in gel analysis using Progenesis SameSpots software version 3.0.2 (Nonlinear Dynamics Ltd.). The Cy3 gel images were scanned at an excitation wavelength of 532 nm with an long pass emission wavelength of 570 nm, Cy5 gel images were scanned at an excitation wavelength of 633 with a long pass emission wavelength of 670 nm, while the Cy2 gel images were scanned at an excitation wavelength of 488 nm with 520 nm emission filter. Gel spots were co-detected as DIGE image pairs, which were linked to the corresponding in-gel Cy2 standard. Between gel comparisons were performed utilizing the Cy2 in-gel standard from each image pair, using Progenesis SameSpots. The gels are then stored in 1% acetic acid at 4°C until spot excision.
To check for protein specific bias in Cydye labeling, a separate analysis in which uninfected SsP2 lysate was separately labeled with all three Cy-dyes was performed. Labeled samples were mixed and 2D-DIGE was performed. Five S. solfataricus proteins with a bias toward Cy-dye 3 which could result in a false positive using a 1.5 fold-cutoff were found. None of these proteins were among the differentially regulated protein spots at the higher fold cut-off used in this study; therefore, dye swapping in the DIGE analysis was not used.
The membrane proteins from S. solfataricus 2-2-12 were prepared using a cell fractionation and ultra-centrifugation protocol modified from Lower et al.25 S. solfataricus was grown to OD650 ~0.3 and cells were harvested at 3,944 × g for 10 minutes. The cell pellet was washed with PBS pH (7.4) and resuspended in 5 ml of 25 mM phosphate buffer (PB) (pH 7.4) containing 25 ul of a 7× Complete protease inhibitor cocktail (Roche), and 50 ul of DNase I/RNase. The cells were lysed by freezing and thawing in liquid N2 followed by probe sonication at 30% duty cycle for 1 min. (repeated 3 times) on ice. The lysate was centrifuged at 1,000 × g for 10 min at 4°C to remove debris. Supernatant liquid was centrifuged at 100,000 × g (30,000 RPM) for 75 min at 4°C to separate the membrane fraction from soluble proteins. The membrane containing pellet was then washed by resuspension in 5 ml of 20 mM sodium acetate (pH 5.0) containing 0.5 M NaCl and the 100,000 × g centrifugation step was repeated. The pellet was resuspended in 5 ml of 25 mM PBS containing 0.4% Triton X-100 and 125 mM NaCl. Membrane material was then pelleted with a third centrifugation leading to a final pellet which was resuspended in 25 mM PBS and and an equal volume of 4× SDS-PAGE gel-loading buffer. Samples were boiled prior to separation on 4–20% SDS-PAGE.
S. solfataricus proteins were labeled with a custom made fluorescent Zdye (ZB-M LC-01-56) coupled to a maleimide (Figure 2) [Dratz, E. A. and P. A. Grieco. Novel Zwitterionic Fluorescent Dyes for Labeling in Proteomic and Other Biological Analyses. USA patent Number 7,582,260, 9/1/2009, Montana State University, Dratz, E. A. and P. A. Grieco. Novel Zwitterionic Fluorescent Dyes for Labeling in Proteomic and Other Biological Analyses. USA patent Number 7,833,799, Montana State University, 11/16/2010.]. A publication detailing synthesis and properties of the dye is forth coming. The labeling reactions were performed on 3 biological replicates using dye at a final concentration of 5 μM in PBS, pH 7.4, for 20 min at room temperature. Reactions were quenched with 2D gel-loading buffer containing 40 mM DTT, loaded onto 24 cm, 3-11NL isoelectric focusing gel strips. 2D gels were scanned on the Typhoon fluorescence imager using a standard 488/510 nm filter set. Gels were then stained with SYPRO®Ruby (Molecular Probes™) to confirm equal protein loading before DIGE analysis. Finally, the gels were stained with Coomassie GelCode™ Blue Safe (Thermo Scientific) Protein Stain for visible spot picking.
C92 (Δ1–27) was cloned in-frame with N-terminal 6His-tag into the pDEST14 expression vector (Gateway system, Invitrogen), according to the manufacturer’s protocol. Escherichia coli BL21 (DE3) PlysS strain (Invitrogen) was used for the protein expression. Protein was purified using Ni-NTA affinity gravity flow column (Novagen). Determination of C92 molecular mass in solution was carried out by size exclusion chromatography using an ÄKTApurifier™ FPLC (GE Healthcare) with a BioSep_SEC_S2000 (PhenomenexR) column, equilibrated with PBS (pH 7.4) that had been calibrated using the molecular mass standards Vitamin B12 (1,350 Da), Myoglobulin (17,000 Da), Ovalbumin (44,000 Da), γ-Globulin monomer (158,000 Da), γ-Globulin dimer (316,000 Da) and Tyroglobulin (670,000 Da).
Protein spots of interest were excised from the gels, washed, in-gel reduced and S-alkylated, followed by digestion with porcine trypsin (Promega) overnight at 37°C14, 24. The solution containing peptides released during in-gel digestion were transferred to sample analysis tubes for mass analysis. LC/MS/MS used an integrated Agilent 1100 liquid chromatography–mass-selective detection (LC-MSD) trap (XCT-Ultra 6330) controlled with ChemStation LC 3D (Rev A.10.02). The Agilent XCT-Ultra ion trap mass spectrometer is fitted with an Agilent 1100 CapLC and Chip Cube under the control of MSD trap control version 5.2 Build no. 63.8 (Bruker Daltonic GmbH). Injected samples were first trapped and desalted on the Zorbax 300SB-C18 Agilent HPLC-Chip enrichment column (40 nl volume) for 3 min with 0.1% formic acid delivered by the auxiliary pump at 4 μl/min. The peptides were reverse eluted and loaded onto the analytical capillary column (43 mm × 75 μm ID, also packed with 5 μm Zorbax 300SB-C18 particles) and connected in-line to the mass spectrometer using the ChipLC ESI spray needle, with a flow of 600 nl/min. Peptides were eluted with a 5 to 90% acetonitrile gradient over 16 min. Data-dependent acquisition of collision induced dissociation tandem mass spectrometry (MS/MS) was utilized. Parent ion scans covered the m/z range 400 to 2,200 at 24,300 m/z-s. MGF compound list files were used to query an in-house SsP2 database using MS and MS/MS ion mass tolerances of 1.2 and 0.5 amu respectively. Positive identification required two significant peptides based on MASCOT (Matrix science, London, UK) MOWSE scores >32 (p<0.05) and Phenyx GeneBio z-Score of >8. Protein fold recognition using 1D and 3D sequence profiles, coupled with secondary structure and solvation potential were performed using DUF26 and PHYRE (Protein Homology/analogY Recognition Engine; http://www.sbg.bio.ic.ac.uk/phyre/index.cgi)27. The STIV genome map was constructed using Vector NTI Advance 11.0 (Invitrogen). Multiple sequences alignments were performed using CLUSTAL 2.1.
To add depth to our understanding of STIV replication and the biology of Archaea, a parallel transcriptomic and proteomic analysis of S. solfataricus strain SP2-2-12 during STIV infection was conducted. Global transcriptome analysis showed that viral mRNA expression rose rapidly after 8 and peaked at 24 hpi, see Figure S1-A. Viral gene expression during infection of SP2 ATCC stock strain and the clonal isolate SP2-2-12 are very similar, except for a delay in the timing of gene expression and viral release in the SP2 ATCC strain, as shown in Figure S1. Two other crenarchaeal viruses, STIV2 and SIRV1, were recently reported to also have similar narrow host selectivity9, 28. Together with our results, this suggests that previous reports that crenarchaeal viruses are predominantly non-lytic should be reinvestigated.
Based on the microarray expression results for Sulfolobus SP2-2-12, parallel samples at 24 and 32 hpi were selected for proteomic analysis. Total soluble protein from control and infected cell lysates was labeled with DIGE dyes, combined and analyzed using standard 2D-DIGE methodology.29 Greater than 900 spots were detected on each gel (Figure 1) using Progenesis SameSpots analysis.30 After filtering to remove spot irregularities, 807 protein spots were used in the analysis across all 9 gels, which is consistent with our previous 2D DIGE experiments.29, 31 Using a 1.8 fold change cutoff, there were 11 regulated protein spots at 24 hpi and 29 at 32 hpi. Protein spots were identified using in-gel proteolysis and LCMS/MS analysis24 followed by searching of a database containing all ORFs in S. solfataricus and STIV greater than 50 amino acids using Phenyx and MASCOT search algorithms. Previous experience with STIV indicated that non-annotated reading frames are used14, therefore, this larger database was adopted to minimize false negatives in the preliminary IDs. A minimum of two peptides with significant scores were the basis for protein identification. Analysis of 20 regulated protein spots led to the identification of 10 host and 9 STIV encoded proteins (Tables 1 & 3). For the host proteins, 9 increased in abundance while 1 decreased (Table 1).
The limited number of host proteins showing statistically significant changes in abundance was initially surprising, considering the synchronous timing of infection and the wide scale changes observed in mRNA22. Because differential expression is based on spots that show statistically significant changes (p=0.05 q=0.1) for a given fold change, highly variable spot intensity between biological replicates could reduce the number of spots that showed a significant difference. However, the calculated standard deviation as a percentage of spot intensity between biological replicates was, on average, less than 20% of spot volume, indicating that variability in spot intensity between biological replicates was not a factor in the differential analysis of protein expression (Figure S2).
The host protein showing the greatest change in abundance was SSO0209, a putative N-acetyltransferase. N-acetyltransferases constitute a superfamily of functionally diverse enzymes that catalyze the transfer of an acetyl group from acetyl-Coenzyme A to the primary amine of a wide range of acceptor substrates. The use of histone acetylation to regulate gene transcription in eukaryotes is well known32, 33 and a number of viruses, including HIV and HPV, regulate N-acetyltransferases.34–38 Global changes in protein acetylation have been associated with metabolic regulation39, 40 which is consistent with the changes in the transcriptomics data.22
Manipulation of host cell machinery involved with cell division is a hallmark of nearly all viruses. The increase in abundance of two proteins that are members of the recently described cell division family of proteins, CdvA and CdvB, SSO0911 and SSO0881 provides the first evidence at the protein level for involvement of Cdv proteins in archaeal viral replication. A three-gene operon composed of CdvA, B, and C has been found in most crenarchaea41–43 and comprises the minimal machinery needed for cell division. CdvA and CdvB form co-localized oligomers segregating chromosomes and are associated with the leading edge of constriction during cytokinesis.42 CdvC encodes an AAA+ ATPase believed to be involved with disassembly of the CdvA and CdvB complexes. CdvB and CdvC are similar to type E endosomal sorting proteins of eukaryotes that form the ESCRT-III sorting complex. CdvB is a positively charged coiled-coil protein and all crenarchaea that have Cdv genes have multiple copies of CdvB-like proteins.42 The third member of the operon, CdvA, appears to be unique to archaea. Structural prediction suggests that CdvA is a coiled-coil protein possibly similar to myosin, tropomyosin, or cingulin-like proteins. Such a structure would be consistent with a role for CdvA proteins in ESCRT-like daughter chromosome sorting processes or possibly as part of the yet to be delineated archaeal cytoskeleton.
The S. solfataricus cdv operon of genes SSO0911, 0910, and 0909 (CdvA, B, and C respectively) were all upregulated at the mRNA level during STIV infection.22 At the protein level, CdvA (SSO0911) was the only member of this Cdv operon that increased in abundance. This could be because SSO0910 and SSO0909 were below the level of detection in the proteomics experiment. Beyond the obvious biological significance of understanding the role of cell division proteins in viral infection, the ESCRT-III family is of particular interest because they are involved in eukaryotic protein sorting, endosomal vesicle formation, and budding of enveloped viruses44, 45 S. solfataricus ESCRT proteins have been found in secreted vesicles 46 and in purified viral samples.14 SSO0881 is one of four CdvB homologs in S. solfataricus (the other genes are SSO0451, SSO0619, and SSO0910).
The CdvB ESCRT-III like proteins of eukaryotes have microtubule interacting and transport (MIT) domains at the C-terminus. This domain is missing from the archaeal homologs, and instead has been reported to be replaced by a helix-turn-helix motif that may be used for DNA binding.42 Secondary structural prediction of SSO0881 was consistent with this idea (Figure S3) and the ability to bind DNA is consistent with a role in cell division and segregation of polynucleotides. The specific cellular role of SSO0881 is currently unknown, but, it was one of only two host proteins that co-purify with STIV virus particles.14 Recent work also puts this protein in the Snf7 protein family that is part of the ESCRT complex which in eukaryotes manages protein sorting and transport from the endosome to the vacuole/lysosome.
Viruses known to use the ESCRT pathway include many retroviruses, such as human immunodeficiency virus (HIV)47, 48 and murine leukemia virus (MLV)47, as well as viruses from other families, such as vesicular stomatitis virus49 and herpes simplex virus (HSV)50. Disruption of ESCRT function or mutation of viral late domains generally prevents the separation of virions from the cell membrane. Structural prediction of the hypothetical protein SSO2632, which was also regulated, using Phyre 3D revealed a strong similarity to myosin/tropomyosine like proteins, which also play crucial roles in eukaryotic cell division and transport, further strengthening this idea. The finding of ESCRT III and myosin-like proteins among the regulated proteins and mRNA, suggests that STIV modulates the ESCRT system for transport or release. The role of ESCRT proteins in viral transport and how this could relate to STIV and other archaeal viruses has been covered in a recent review.51
Fibrillarin-like pre-rRNA processing protein, SSO0940, was another regulated protein with strong ties to viral replication in eukaryotes. Fibrillarin is a small nucleolar RNA-processing protein that catalyzes methyl transfer reactions52. HIV Tat protein specifically co-localizes with fibrillarin in the nucleoli of Drosophila oocyte nurse cells. Tat expression is accompanied by a significant decrease of cytoplasmic ribosomes, which is apparently related to an impairment of ribosomal rRNA precursor processing.53 Other viral proteins such as porcine arterivirus nucleocapsid protein54 and non-structural protein 3b, a protein specifically encoded by the severe acute respiratory syndrome coronavirus (SARS-CoV)55 associate with fibrillarin. A suggested outcome of these interactions is the modulation of host cell function through rRNA precursor processing and control of ribosome biogenesis. Other viruses such as, herpes simplex virus 1 56 and groundnut rosette virus, use fibrillarin to target viral proteins to the nucleolus for assembly of viral ribonucleoprotein particles57, 58. Because Sulfolobus lacks a nucleus, we speculate that SSO0940 is involved in viral replication via rRNA processing/ribosome assembly through specific interactions with one of the STIV proteins.
Disulfide Oxidoreductase SsPDO (SSO0192) is up regulated in the STIV infected cells. SSO0192 was reported to be involved in the reduction of both the Bcp1 and Bcp4 proteins59 and is a substrate of thioredoxin reductase SSO2416 in S. solfataricus60. SSO0192 belongs to the Thioredoxin reductase (TRX)-Glutaredoxin reductase (GRX)-like family of proteins that, among other functions, have a central role in the formation and maintenance of cytoplasmic disulfide bonds.61 Interestingly, Thioredoxin reductase-1 (TR1) negatively regulates the activity of the HIV-1 encoded transcriptional activator, Tat, in human macrophages.62 It was found that Tat-dependent transcription and HIV-1 replication were significantly increased in human macrophages when TR1 activity was reduced and the effect was independent of the redox-sensitive transcription factor, NF-κB. This suggests that upregulation of disulfide oxidoreductase is an antiviral mechanism employed by archaea as well as eukaryotic cells.
Peroxiredoxin (SSO2613), also known as Bcp4, is a general cellular stress response protein and was the only identified host protein that decreased in abundance during STIV infection. Bcp-4 is a substrate of the disulfide oxidoreductase discussed above, and plays an important role in the peroxide-scavenging system in S. solfataricus60, 63, 64. Peroxiredoxin homologs are prevalent in thermophiles and S. solfataricus has four orthologs: Bcp1 (SSO2071), Bcp2 (SSO2121), Bcp3 (SSO225) and Bcp4 (SSO2613).65 It has been proposed that the Bcps are part of an antioxidant system using Bcp1 and Bcp4 to prevent endogenous peroxide accumulation, while Bcp2 and Bcp3 are induced in response to external peroxides.60 Recently our group showed that Bcp2 was significantly up regulated in response to H2O2 oxidative stress.29 STIV apparently counteracts to some extent, host upregulation of SSO0192 by down regulating SSO2613, which is a downstream target of SSO0192. This would be analogous to the increase in HIV replication with TR1 down regulation.62 In addition, proteomic analysis of swine fever virus66 and Respiratory Syncytial Virus (RSV) infection67, confirm changes in peroxiredoxin (Prdx-1, −3, and −4) oligomer status and total abundance during infection. From a cellular standpoint, Prdx-1 and −4 are essential for preventing RSV-induced oxidative damage to a subset of nuclear intermediate filament and actin binding proteins in epithelial cells. Retroviruses may counter such effects, as it was recently shown that peroxiredoxin-1 expression was decreased in H4IIE cells in the presence of a retroviral vector.68 Therefore, this archaeal virus system is behaving very similarly to well studied human viruses with regard to infection induced changes in oxidative stress.
If the host and virus are vying for control over redox sensitive systems in the cell, we reasoned that it might be possible to find specific proteins that are affected. Reactive cysteine residues in proteins are critical components in redox signaling, protein folding, and can be used as reporters of cellular redox state. Using a thiol reactive maleimide probe, we monitored protein redox state across the entire proteome. Changes in the redox state of proteins were visualized by coupling a reactive group to a novel zwitterionic fluorescent tag. The ZB-maleimide (ZBM) probe (Figure 2) was designed to overcome limitations of other cysteine labeling reagents on 2D gels.69
Total protein from control and infected cells at 24 and 32 hrs was reacted with the probe, and then analyzed using 2D DIGE. More than 300 fluorescent spots were visible across replicate gels (Figure 2). For reference, there are 1863 proteins with at least one cysteine in SsP2 (see Figure 2D and Supplemental Figure S4). Ten protein spots exhibited significant differential labeling in response to viral infection, with 4 increasing and 6 decreasing in label. Proteins where identified from six of the spots (Table 2). Two of the identified proteins had increased labeling during STIV infection (SO2044, Glutamate dehydrogenase; SSO0421, Transitional endoplasmic reticulum ATPase) and 4 were less reactive (SSO0564, ATP synthase subunit B; SSO0282, Thermosome beta subunit; SSO1151, Conserved hypothetical putative tldD protein; SSO1134, Heterodisulfide reductase subunit C). Protein identification from the remaining four spots was unsuccessful due to insufficient quantities.
SSO0421 was present in spots 13 and 18 suggesting that the protein was posttranslationally modified. Post-translational modifications (PTMs) are a common and highly rapid way in which protein activity can be regulated. Based on gene annotation and functional characterization, the presence of enzymes involved with the addition and removal of PTMs such as methylation, acetylation, phosphorylation, glycosylation, and N-terminal processing are present in S. solfataricus (P2)29, 65, 70, 71. Finding the targets and frequency of PTM is a challenging task and currently, little is known about the use of PTM in archaea. The ability to detect PTMs at the level of the proteome, without limiting the scope of the experiment, is a strength of the 2D-DIGE approach. Changes in protein pI and/or MW are caused by most PTMs leading to altered gel mobility and changes in spot intensities. Proteins with less than complete modification at a specific site will be found in more than one spot and the ratios can be used to follow modification kinetics.
Glutamate dehydrogenease labeling increased significantly upon infection, indicating greater accessibility of reduced thiols or reduction of previously oxidized cysteine side chains. Importantly, based on DIGE experiments this protein did not change in abundance, so the change detected reports on regulation at the level of activity, independent of abundance. Glutamate dehydrogenase (SSO2044) has only one cysteine residue. Surprisingly, this cysteine is not only conserved in the 4 annotated glutamate dehydrogenases in SP2 (gdhA-1, gdhA-2, gdhA-3, gdhA-4) but also in the human homologues (hGDH1 and 2). This enzyme plays a curial role in amino acid metabolism and transport and is involved with Human Cytomegalovirus (HCMV) infection by increasing the production of glutamine.72 An increase in glutamine in other systems has been revealed by metabolic flux studies73, 74 and cells starved of glutamine failed to produce infectious virions. Also, glutamate dehydrogenase activation can help maintain the cellular energy balance for viral specific lipid production.72 This link is particularly interesting because, as we have shown, the internal membrane of STIV is composed of a subset of host lipids.14
Heterodisulfide reductase subunit C (hdrC-2, SSO1134) had a much lower level of labeling in the virally infected cells (Figure 2 and Table 2). It contains 8 cysteine residues which are grouped into 2 sequence motifs characteristic of Fe4S4 clusters as in ferredoxin iron-sulfur binding proteins. This active-site cluster is directly involved in mediating heterodisulfide reduction 75 and catalyzes the formation of coenzyme M (CoM-SH) and coenzyme B (CoB-SH) by the reversible reduction of the heterodisulfide, CoM-S-S-CoB.76 Decreased labeling of SSO1134 could indicate that more of the protein is in the holo form providing the cells with increased control over disulfide formation. The high levels of disulfide bonds in intracellular proteins in hyperthermophiles and their viruses16, 77, make this a particularly interesting result.
In addition to changes in the host proteome after STIV infection, 13 STIV proteins were identified. Nine viral encoded proteins were identified from 2D gel spots that increased in abundance. These were assigned to the open reading frames A223, B109, B116, B129, B130, B264, B345, C381, and C92. Four of these (B345, B109, B130, and C92) were found in more than one horizontally displaced spot (Figure 1, Table 3), indicating the presence of post-translational modifications of these proteins that create protein isoforms. A previous analysis of purified STIV particles had a similar pattern of spots for B345, the major capsid protein.14 The majority of the 37 STIV ORFs have no close sequence homolog in other known genomes and of those identified by 2D-DIGE, A510, A55, A78, B109, B116, and B130 currently have no assigned function. It is known that A55, A78, B109, B130, A223, B345 and C381 are present in STIV particles14, explaining the high levels of expression that we found. Previously it was found that C381 and A223 have sequence similarity to PRD1 P5 vertex proteins and together fit nicely into the electron density observed for the STIV virus turret.14 Prior sequence analysis and structural prediction indicated that B164 is similar to the Poxvirus ATPase family.14 Structure predication using PHYRE showed that B264 is similar to topoisomerase V (Table 3). Lastly, a small protein with a distinct transmembrane domain, C92 (9.8 kDa, pI 5.19) was found in two spots that migrated only slightly slower than the dye front in the molecular weight dimension (Figure 1). In comparison to these proteomics results, the microarrays showed that all STIV ORFs were actively transcribed (Figure S1). Likely reasons for this discrepancy are low protein abundances for some of the viral transcripts, regulatory functions for RNAs that did not lead to significant translation, and possible incompatibility of some proteins with 2D gel analysis.
An exciting development in archaeal virology was the recent description of novel VAPs that appear on the surface of sulfolobales prior to viral egress. Two recent publications, one looking at STIV in Sulfolobus solfataricus and another at the rod-shaped virus SIRV-2 that infects Sulfolobus islandicus, show that VAPs arise from the Sulfolobus cell surface and apparently lead to rupture of the tough protein-rich S-layer prior to cell lysis.8, 9 The VAPs from SIRV were reported to be composed strictly of protein from SIRV ORF98.78 The mechanism of formation for the VAPs is not known, but the dramatic rearrangements that take place in the S-layer and membrane suggest that proteins possessing membrane interaction domains will be important. A number of the highly expressed STIV proteins (e.g., A223, B130, B264, B345, C381 and C92) are predicted to contain transmembrane sections. Because membrane proteins are typically at low levels relative to cytoplasmic proteins and membrane proteins may behave poorly on standard 2D gels, a second approach was needed to look for changes in membrane protein abundance.
To enhance detection of membrane associated proteins and changes in protein composition, 1D SDS-PAGE analysis of enriched membrane fractions was carried out. A comparison of membrane protein fractions from mock and STIV infected cells showed that the overall 1D gel banding pattern was quite similar (Figure 3). However, there were several visually distinct differences, indicated by asterisks, such as the major band just below the 10-kD marker in the infected cell samples. Each lane was cut into 22 horizontal slices and subjected to in-gel proteolysis and protein identification. From the 22 pairs of samples, 74 host proteins were identified; 57 were common to both, 13 unique to infected, and 5 unique to control (for the complete list see Supplemental Table 1). Protein sequence analysis indicated that approximately 50% of these proteins were predicted to have transmembrane domains. Nine viral proteins were identified from the membrane fraction; A55, A61, A223, A510, B130, B164, B345, C92, and C381. Four of which (A55, A61, A510, and B164) were not found on the 2D gels.
C92 is of particular interest because it is believed to be responsible for VAP formation implicated in cell lysis and viral release.79 It was highly expressed in the membrane fraction (Figure 3). Sequence analysis indicated that C92 has a single transmembrane helix and a soluble domain that is likely to be helical as well (Figure 4). The only proteins in the NCBI nonredundant library with sequence similarity are other crenarchaeal viral proteins; P98 from Sulfolobus islandicus rod-shaped virus 2 (SIRV2); gp49 from Sulfolobus islandicus rod-shaped virus 2 (SIRV2_gp49) and gp42 from Sulfolobus islandicus rod-shaped virus 1 (SIRV1_gp42) (Figure 4B). Recently it has been shown that SIRV2 virus infection of S. islandicus also produces VAPs akin to those from S. solfataricus28 and that viral protein P98 is responsible for formation of the VAPs in the native archeal host as well as on E. coli bacteria after heterologous expression.78, 80
The large quantity of C92 present in membrane fractions after STIV infection, suggested that the protein was present in quantities sufficient for VAP assembly. To investigate properties of the protein, C92 was expressed in E. coli as a full length or as a truncated construct. Full length C92 expressed to moderate levels and, as expected, partitioned with the membrane fraction of the cell lysate. Removal of the transmembrane spanning region, C92Δ1-27, produced a highly soluble protein and both the N and C terminal His-tagged constructs were readily purified.
It seems likely that the VAPs are partially or entirely formed by protein polymerization.78, 79 To test the oligomeric state of C92 and its potential to initiate polymerization, the purified recombinant protein lacking the TM domain (1–27) was analyzed by size exclusion chromatography (SEC). Based on the elution volume, the truncated C92 was present as a distribution of oligomeric species (Figure S5A). The lack of a strong propensity for the soluble portion of C92 to aggregate into megadalton size complexes suggests that either oligomerization is driven by the N-terminal membrane spanning region (residues 1–27) which was deleted, or that C92 acts in concert with other proteins and/or lipids in the cell.
To search for interaction partners of C92, the C-terminal domain was chemically linked to Sepharose resin and incubated with cell lysates from mock and infected S. solfataricus cells. After trying a number of binding and wash conditions, no proteins with a high affinity for C92 were identified by 1D SDS-PAGE or LCMS of a tryptic digest of potentially C92-bound proteins (data not shown). As a final check for oligomerization activity, E. coli cells expressing full length C92 were visualized by negative stain electron microscopy. Many of the cells expressing C92 had a very strange morphology, with large “horn-like” protrusions, not found on any of the control cells, suggesting that the C92 protein has an inherent ability to alter membranes (Figure S5B). A similar experiment using SIRV2 P98 lead to the observation of VAP-like structures in E. coli78. Given that VAP formation would require high levels of C92 expression we compared relative viral mRNA levels during infection of the ATCC stock strain SP2 and strain SP2-2-12 used for this study. Interestingly, ATCC stock cells showed a lower level of C92 transcription relative to other viral transcripts (Figure S1) and are largely resistant to infection (data not shown).
Further investigation into the other virally encoded proteins that were detected revealed some noteworthy findings. BLAST analysis of A61 showed that it has significant similarity to proteins from the Crenarchaeal stygiolobus rod-shaped virus and CopG/DNA-binding domain-containing protein from Metallosphaera sedula DSM 5348. A61 was also one of only two proteins in common between STIV, STIV2, and SIRV2 28. Sequence analysis and protein fold recognition performed for A510 using DUF26 and PHYRE27, suggested a role as a topoisomerase/recombinase. The band in which A223 and C381 were found had an apparent molecular mass of 200 kDa on 1D SDS-PAGE reducing gels (Figure 3, lane I, band 2); however, their predicted masses are 24 kDa and 42 kDa, respectively. These proteins have recently been shown to be part of the turrets in STIV81, confirming our earlier predictions based on a proteomic analysis of purified STIV14. The anomalously high molecular masses of A223 and C381 in 1D gel analysis suggest that they may reside in a complex that resists disruption in reducing SDS. This behavior has been described for similar structural proteins from mesophilic viruses, such as PBCV-1 82. Under the conditions used for 2D gel analysis, the proteins were detected close to their expected masses (Figure. 1). This is consistent with the behavior of these proteins from highly purified STIV14 and rules out the possibility of read through translation which can occur in S. solfataricus (Cobucci-Ponzano 2010).
Recoding of genetic information is well documented in archaea, including the use of nonstandard amino acids, frame shifts, read-through of amber codons, and alternative sites for the initiation of transcription83, 84. To assure that virally expressed proteins were not overlooked, we formatted our protein sequence database such that STIV ORFs began immediately after stop codons and allowed for read-through products of the amber codon, which is a common site of recoding in archaea84. This lead to the identification of a new viral protein from 2D spot 20 (Figure 1). This ORF has an amber stop codon after the first 9 residues, so it was not listed in the original genome annotation12. This ORF has now been annotated B129 (see Figure 5 and Figure S11). A blast search using the amino acid sequence from B129 against STIV2, produced a significant hit to STIV2 A105 which is itself similar to STIV C118 (Supplemental Figure S12). This raises the question of why STIV would have two such genes when only one appears to be present in STIV2. Further sequence analysis failed to produce significant findings and the function of these proteins remains to be discovered.
The combined data from the 1 and 2D gels led to the detection of 13 virally encoded proteins (Table 3). When the ORFs of the detected proteins are viewed on a circular map of the viral genome, they cluster together from 9 to 6, as on a clock (Figure 5). The sequential order of these proteins is indicative of a shared transcriptional regulatory mechanism, which is common for viruses in general and those that replicate in S. solfataricus.22, 85 The temporal variation of RNA expression observed in the microarray experiment on STIV was limited, all the virual proteins detected come from the region of the genome that first appeared after 16 hours, which was on the later side. Because most of the proteins from this region of the viral genome are part of the particle, they should be produced in large quantities, increasing the probability of detection in proteomics experiments. These results are consistent with predictions based on sequence analysis and structural data that the ORFs from 6–9 o’clock are regulatory proteins. This is also consistant with our previous data in that transcripts from this region were the first to be detected by microarray1, 15, 16. This expression scheme then appears to explain the location of C92 and A197, within the late genes, even though they are not structural proteins. A197 was predicted to be a glycosyltransferase, based on the structural model from crystallography15 with the major capsid protein, B345, as a primary target for glycosylation14, 15. Lastly, the level and timing of C92 protein production are consistent with it having a structural role in VAP formation, as discussed earlier.
This is the first study to investigate viral infection in archaeal cells using an untargeted proteomics-based approach despite widespread interest in archaeal organisms and more recently the viruses that infect them. As such, this work brings clarity and significant insights to STIV and archaeal host-virus interactions. A surprisingly consistent connection was found between S. solfataricus proteins that were altered during viral infection and eukaryotic proteins known to be involved with viral infection and egress. For example, general and specific stress response proteins that mediate redox status were found to be involved. The regulation of ESCRT-like proteins by viral infection in both archeal and eukaryotic systems implies a deep evolutionary root for viral trafficking processes. It was recently shown that genome packaging leads to redistribution of STIV particles within cells8, 86. Our data point to the ESCRT-like CdvB homolog, SSO0881, as the likely candidate for delivery of double stranded viral DNA for packaging and a second Cdv operon, SSO0909-9011 for particle transport. The regulation of myosin-like proteins, such as SSO2632, could also be involved with transport or serve as a structural protein for cytoplasmic reorganization or formation of the electron dense spheres associated with VAPs8. The STIV C92 protein is clearly involved with VAP formation, though the details on what triggers polymerization and lysis are still unclear. Future studies now have specific targets for genetic manipulation, in both Sulfolobus and STIV, to elucidate the mechanism of cell division, the nature of the cellular cytoskeleton in the crenarchaea, and how the VAPs are built.
The evolutionary relationship between STIV and viruses with bacterial and eukaryotic hosts has been primarily built on morphological similarities8, 13. The untargeted proteomics methods used here further strengthen the mechanistic connections across domains and suggests targets for future studies. For example, the functional analysis of C92 and the discovery of B129 add two important new pieces to the puzzles of archaeal viruses. The involvement of trafficking, oxidation, and RNA modification systems are akin to those used by eukaryotic viruses, indicating that the investigation of archaeal virus-host interactions will provide evolutionary insights to viruses and biology across all domains of life.
In the bigger picture, this study emphasizes the complimentary rather than competitive nature of different omics approaches23, 29, 87. By integrating data from the transciptome, proteome, and activity assays significant biological meaning is emerging. The challenge now is in how best to properly weight data from different sources and which types of data provide the most added value.
On the surface, the archaeal hyperthermo acidiphile Sulfolobus solfataricus and the virus STIV, appear to be very different from eukaryotic virus-host pairs. After conducting a thorough proteomics analysis, we conclude that the difference is only skin-deep. Below the unique viral associated pyramids, the activated replication, transport, and redox systems are similarly involved in eukaryotic viral infection.
BB is funded by MCB 0646499 and MCB 1022481 from the National Science Foundation, the Center for Bio-Inspired Nanomaterials (Office of Naval Research grant #N00014-06-01-1016), and the Thermal Biology Institute (NASA grant NNG04GR46G) and NSF DEB-0936178 (M.J.Y) and EF-080220 (M.J.Y) We would like to thank the Murdock Charitable Trust and NIH Cobre 5P20RR02437 for support of the Mass Spectrometry Facility at MSU.