Protoarray Analysis Essay

Citation: Kim SH, Kim JY, Lee HJ, Gi M, Kim BG, Choi JY (2014) Autoimmunity as a Candidate for the Etiopathogenesis of Meniere's Disease: Detection of Autoimmune Reactions and Diagnostic Biomarker Candidate. PLoS ONE 9(10): e111039. https://doi.org/10.1371/journal.pone.0111039

Editor: Bernd Sokolowski, University of South Florida, United States of America

Received: February 18, 2014; Accepted: September 23, 2014; Published: October 17, 2014

Copyright: © 2014 Kim et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This study was supported by a grant of the Korea Healthcare technology R&D Project, Ministry for Health & Welfare Affairs, Republic of Korea (HI08C2149) to JYC and by the Basic Science Research Program through the National Research Foundation of Korea (NRF) grant funded by the Ministry of Science, ICT & Future planning (2012R1A1A1042980) to SHK. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

In 1861, Prosper Meniere first described Meniere's disease as an inner ear disorder that manifests as fluctuating vertigo, sensorineural hearing loss, tinnitus, and aural fullness. The prevalence of Meniere's disease is 3.5–513 per 100,000, which is higher than the prevalence of systemic lupus erythematosus (SLE) and multiple sclerosis [1]. The unpredictable nature of Meniere's disease has a serious effect on patients' daily life. During active episodes, the quality of life score of patients with Meniere's disease is thought to be lower than that of AIDS patients treated with AZT, that of patients with severe chronic obstructive pulmonary disease, and that of non-institutionalized patients with Alzheimer's disease [2]. The main pathologic site is thought to be the inner ear, which consists of the cochlea, vestibule, and endolymphatic sac. A characteristic finding of Meniere's disease is the dilatation of the endolymphatic compartment of the inner ear caused by an increase in endolymph (endolymphatic hydrops, Fig. 1) [3]. The proposed etiologies of endolymphatic hydrops are autoimmune, allergic, genetic, traumatic, and infectious (viral) [4]–[9]. These finally result in endolymphatic hydrops by deteriorating ion homeostasis and fluid volume regulation in the inner ear [3]. However, the exact pathologic mechanism underlying endolymphatic hydrops is still unknown.

Figure 1. Schematic drawing of the inner ear and endolymphatic hydrops as a mechanism for Meniere's disease.

The inner ear consists of the cochlea, vestibule, and endolymphatic sac (ES). The utricle (U), saccule (S), and semicircular canals (SCCs) form the vestibule. A. Normal inner ear structure. B. Endolymphatic hydrops in patients with Meniere's disease.

https://doi.org/10.1371/journal.pone.0111039.g001

Certain findings have provided evidence that autoimmunity may underlie the pathology of Meniere's disease. The prevalence of systemic autoimmune diseases such as rheumatoid arthritis, ankylosing spondylitis, and SLE in patients with Meniere's disease is 3- to 8-fold higher than in the general population [10]. In addition, autoantibodies such as the anti-heat-shock protein 70, anti-68 kD inner ear protein antibody, anti-myelin peroxidase zero antibody, and anti-thyroid peroxidase antibody have been detected in the serum of patients with Meniere's disease [11]–[14]. However, these autoantibodies were not found in all of the patients.

Previous studies tended to investigate only a select few target proteins instead of conducting mass screening; in addition, many of these studies used western blot analyses to detect antigen-antibody reactions between patient serum and animal inner ear tissues, which can demonstrate the existence of an antigen-antibody reaction but provides no information on the identity of the autoantibody. Few studies demonstrated increased proteins in the serum of Meniere's disease patients that were reported to be related with inflammatory reaction or inner ear disorders by proteomics technique [15]. But, there was no evidence if these materials existed in the inner ear fluid of Meniere's disease patients. Studies using human inner ear tissue are rare, and no studies have investigated autoimmunity using human inner ear fluid. To overcome the limitations of previous studies and to understand the autoimmune pathologic mechanisms underlying Meniere's disease, mass screening-based studies of autoimmune reactions using human inner ear fluid and sera of patients should be conducted.

In this study, we tried to provide evidence for the involvement of autoimmunity in Meniere's disease and identify the candidate antigens that react with autoantibodies, which can suggest diagnostic biomarker candidates for Meniere's disease. Several studies were performed. First, the protein composition of inner ear fluid from control patients and patients with Meniere's disease was compared using proteomic analysis. Second, candidate autoantigens that reacted with circulating autoantibodies in patient serum were investigated using protein array (Protoarray, Invitrogen, Life Technologies, Grand Island, NY). Third, western blots using patient serum and mouse inner ear tissues were performed to investigate whether the circulating autoantibodies reacted with inner ear tissue. The results of this study can provide the basic information for the development of diagnostic biomarkers as well as the understanding of pathologic mechanisms of Meniere's disease.

Methods

Selection of patients and controls

Thirteen patients diagnosed with Meniere's disease according to the criteria of the American Academy of Otolaryngology Head and Neck Surgery (1995) [16] were enrolled in the patient group. Samples of inner ear fluid (endolymphatic sac luminal fluid) were taken from 3 patients undergoing endolymphatic sac surgery to treat their intractable disease, and peripheral blood was sampled from the other 10 patients. Three patients with acoustic tumors and 10 patients with simple tympanic membrane perforation who planned to undergo myringoplasty were enrolled as controls. Endolymphatic sac luminal fluid was sampled from the 3 controls during the acoustic tumor surgery via the translabyrinthine approach, and peripheral blood was sampled in the other 10 controls. The controls with acoustic tumors had severe sensorineural hearing loss but no history of sudden vertigo. The controls had no history of sensorineural hearing loss or vertigo, and their audiograms showed mild conductive hearing loss. None of the patients or controls had a history of systemic disease, and all of their laboratory parameters, including the electrocardiogram, chest radiography, blood cell counts (red blood cells, white blood cells, and platelets), liver and kidney function tests, and urinalysis, were normal. The gender distribution and mean age of the patients and controls were not significantly different (p>0.05 for the chi-square test and t-test, Table 1).

Sampling of inner ear fluid and sera

Three patients with Meniere's disease underwent endolymphatic sac surgery for their intractable disease and the 3 controls with acoustic tumors underwent tumor removal via the translabyrinthine approach. To obtain inner ear fluid, endolymphatic sac luminal fluid was sampled during surgery. In each surgical procedure, the endolymphatic sac should be fully exposed. Because the amount of luminal fluid was very small (<4 µl), we infused 200 µl of normal saline into the endolymphatic sac and aspirated the diluted luminal fluid from the endolymphatic sac, as previously described [17]. The fluid samples were immediately stored at −80°C until analysis.

Blood was sampled from the other 10 participants in each group for serum protein analysis. The blood was stored in a tube containing ethylenediaminetetraacetic acid. The plasma was immediately separated and stored at −80°C until analysis.

One-dimensional electrophoresis (1-DE) of endolymphatic sac luminal fluid

1-DE was performed to compare the protein composition of the fluid of patients with Meniere's disease and that of controls. Thirty micrograms of protein from the diluted luminal fluid was used for 1-DE for each sample. The samples were lyophilized and dissolved in 15 µl of distilled water. Samples were subjected to sodium dodecyl sulfate gel electrophoresis on an 8–16% Tris/Glycine gel and stained with Coomassie Brilliant Blue.

Identification of endolymphatic sac luminal fluid proteins by liquid column mass spectrometry (LC-MS/MS)

The entire 1-DE gel lane was cut into 8 pieces according to molecular weight for digestion. After reduction with dithiothretol and alkylation with iodoacetamide, each piece of gel was treated with trypsin for in situ digestion. It was then washed with 10 mM ammonium bicarbonate and 50% acetonitrile, swollen in digestion buffer containing 50 mM ammonium bicarbonate, 5 mM CaCl2, and 1 µg of trypsin. Next, it was incubated at 37°C for 12 h. Peptides were recovered over the course of 2 extraction cycles with 50 mM ammonium bicarbonate and 100% acetonitrile. The resulting peptide extracts were pooled, lyophilized, and stored at −20°C.

Nano LC-MS/MS analysis was performed on an Agilent 1100 Series nano-LC and linear trap quadrupole (LTQ)-mass spectrometer (Thermo Electron, San Jose, CA). The capillary column used for LC-MS/MS analysis (150 mm ×0.075 mm) was obtained from Proxecon (Odense M, Denmark) and slurry-packed in-house with 5 µg, 100 Å pore size Magic C18 stationary phase (Michrom Bioresources, Auburn, CA). The mobile phase A for LC separation was 0.1% formic acid in deionized water and the mobile phase B was 0.1% formic acid in acetonitrile. Chromatography was performed using a linear gradient from 5% B to 35% B over 100 min, from 40% B to 60% B over 10 min, and from 60% B to 80% B over 20 min. The flow rate was maintained at 300 nl/min after splitting. Mass spectra were acquired using data-dependent acquisition with full mass scan (400–1800 m/z) followed by MS/MS scans. Each acquired MS/MS scan represented the average of one microscan on the LTQ. The temperature of the ion transfer tube was controlled at 200°C and the spray was 1.5–2.0 kV. The normalized collision energy was set at 35% for MS/MS.

The MASCOT and SEQUEST (BioWorks software version 3.2, Thermo Electron) search engines were used to search the UniProt human protein databases (release 14.8; 82728 sequences) for the tandem mass spectra. Mass tolerances of 1.2 Da and 0.6 Da were used for precursor and fragment ions, respectively. The search included variable modification of oxidation on methionine and carbamidomethyl of cysteine. PeptideProphet and ProteinProphet were used to estimate the false discovery rate (FDR) for any minimum probability used as a cut-off for MASCOT and SEQUEST search results.

Protoarray analysis of serum samples (Immune Response Biomarker Profiling)

To investigate the presence of autoantibodies and their target antigens, Protoarray (Human Protein Microarray v5.0 containing 9400 human proteins, Invitrogen) analysis was performed with sera from patients and controls according to the manufacturer's protocol. Briefly, after blocking array slides with blocking buffer for 1 h, we washed the slides with washing buffer for 5 min, and 5 ml of diluted serum (1∶500) was placed on the slides. The slides were incubated for 90 min and washed 4 times for 5 min with washing buffer. After washing, Alexa Fluor 647 (final concentration of 1 µg/ml) was added on the slide, and the slides were incubated for 90 min. The antibody was aspirated, and the slides were washed for 5 min 4 times. These steps were performed at 4°C. The slides were dried immediately by centrifugation at 200×g for 1 min and stored in a slide box to protect them from light until scanning was performed. The dried arrays were scanned using a GenePix 4000B microarray scanner (Molecular Devices, Sunnyvale, CA). Genepix Pro microarray data acquisition software was used to align the scanned image with the template and to acquire the pixel intensity data for each spot on the array. The reported pixel intensity was calculated as the average of duplicate signals obtained after subtracting the background signal. Protoarray Prospector software (Life Technologies) was used to analyze the data, perform background subtraction, and normalize the signals. The normalized signal intensities obtained for controls and patients were compared with a t-test, and differences were considered significant at p<0.05.

Western blotting

To investigate whether antibodies in the patient serum reacted with inner ear tissue, western blots using patient serum and mouse inner ear tissue were performed. We used mouse inner ear tissue for ethical reasons and because it would have been technically difficult to harvest the entire human cochlea and vestibule. Eight-week-old male C57BL/6 mice were used. The entire inner ear was separated from the temporal bone of the mouse. The cochlea and vestibule were separated, and the membranous labyrinth of the cochlea and vestibule were carefully dissected. Each membranous labyrinth was lysed with 2× sample buffer (250 mM Tris-HCl [pH 6.5], 2% sodium dodecyl sulfate (SDS), 1% DTT, 0.02% bromophenol blue, and 10% glycerol). Protein levels were quantified by comparing the absorbance of the lysate with that of serially diluted bovine serum albumin (0, 0.2, 0.4, 0.6, 0.8, and 1 mg/ml) in the VersaMax ELISA plate reader (Molecular Devices). Samples were heated for 5 min at 95°C. Equal total amounts of protein were prepared for each gel lane. A colored marker mixture was used to estimate the molecular weights of the bands. Proteins were separated using 10% SDS-polyacrylamide gel at 125 V for 4 h with a running buffer (25 mM Tris-Base, 192 mM glycine and 0.1% SDS) and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA) using a semi-dry transfer cell (Bio-Rad, Hercules, CA) for 2 h at 200 mA and a transfer buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.05% Tween 20). The membranes were blocked with 5% skim milk in Tris-buffered saline (TBS; 50 mM Tris-HCl [pH 7.5] and 150 mM NaCl) for 2 h at room temperature. The blot was incubated overnight with patient serum diluted at 1∶200 in 0.5% Tween-20 in TBS. The blot was washed with TTBS, incubated with a secondary anti-human antibody (Cell Signaling Technology, Danvers, MA) in TTBS for 45 min at room temperature, and visualized by enhanced chemiluminescence (Amersham Biosciences, Champaign, IL).

Ethics statement

This study was approved by the institutional review board of Severance Hospital (approval number 4-2011-0871 and 4-2013-0483), and written informed consent was obtained from all of the participants. Institutional Animal Care and Use Committee of Yonsei University College of Medicine approved this study and all mice were treated in accordance with the guidelines for the Care and Use of Laboratory Animals of Yonsei University College of Medicine (approval number 2011-0084).

Results

Difference in the protein composition of the ES luminal fluid of controls and patients

The density and distribution of protein bands in the 1-DE of ES luminal fluid from controls and patients with Meniere's disease varied (Fig. 2); the density and molecular weight of bands from the patients and controls were identical in some cases and different in others. For example, the bands located between 63 and 75 kDa associated with control 1 and 2 (C1 and C2) and patient 1 (P1) appeared similar, but the bands associated with control 3 (C3) and patient 2 and 3 (P2 and P3) appeared denser. The bands located between 17 and 35 kDa and between 75 and 180 kDa also varied. Consequently, the distribution and density of protein bands differed for each individual, and it was difficult to identify disease-specific protein bands.

To identify differences in the protein composition of the ES luminal fluid of controls and patients with Meniere's disease and to identify disease-specific proteins, LC-MS/MS was performed with proteins extracted from 1-DE gels from each control and patient. A total of 6784 (C1, FDR of 1.86%), 6108 (C2, FDR of 0.95%), 3114 (C3, FDR of 1.68%), 6961 (P1, FDR of 1.35%), 7740 (P2, FDR of 3.11%), and 763 (P3, FDR of 2.00%) proteins were detected in the LC-MS/MS analysis. Immunoglobulin and its variants were the most commonly identified proteins (49% and 58% of all proteins in the luminal fluid of controls and patients, respectively).

We analyzed the proteins that were found only in the luminal fluid of patients with Meniere's disease. First, common proteins found in the luminal fluid of all of the patients with Meniere's disease were analyzed. A total of 180 proteins were identified in this analysis, 76% of which were immunoglobulin and its variants; albumin, keratin, globin, transferrin, protease inhibitor, and complement were also detected (Fig. 3). Then, proteins that were also detected in controls were excluded from the 180 proteins. As a result, nine proteins were identified: 8 consisted of immunoglobulin and its variants, and 1 was interferon regulatory factor 7 (Table 2 and Table S1). With the exception of the immunoglobulin kappa light chain variable region (gi|4323924), which had a molecular weight of 17–26 kDa, all of the proteins had a molecular weight of 55–63 kDa. Consequently, the proteins detected only in the luminal fluid of Meniere's disease patients were those involved in the inflammatory or immune reactions.

Reaction of antigens with patient sera in the Protoarray analysis

We investigated the presence of circulating autoantibodies in the peripheral blood of the 10 patients with Meniere's disease using Protoarray Immune Response Biomarker Profiling. Eighteen proteins had more than 2-fold greater signal intensity in the patients with Meniere's disease than in the controls (p<0.05); the signal intensity of 8 proteins was more than 10-fold higher in the patients than in the controls (Table 3, Fig. 4). Among them, the signal intensity of the immunoglobulin heavy constant gamma 1 (IGHG1) was approximately 205-fold higher in the patients than in the controls. A number of antigens had a sensitivity and specificity of >60% and >80%, respectively: IGHG1, the regulator of G-protein signaling 10 (RGS10), transcript variant 2, chromosome 2 open reading frame 34 (C2orf34), and SH3-domain GRB2-like endophilin B1 (SH3GLB1) had a sensitivity and specificity of 80%; the cDNA clone IMAGE:4155919, complete cds, calcium/calmodulin-dependent protein kinase IV (CAMK4), GSG1-like (GSG1L), transcript variant 2, mRNA, and NIMA (never in mitosis gene a)-related kinase 7 (NEK7) had a sensitivity of 70% and a specificity of 80%; and neural cell adhesion molecule 2 (NCAM2) had a sensitivity of 60% and a specificity of 90%. The 8 proteins with more than 10-fold higher signal intensity in the patients had a sensitivity and specificity of >70% and 80%, respectively, for Meniere's disease except Aminoacylase 1 (ACY1) which had a sensitivity and specificity of 100% and 50%, respectively (Fig. 4). More information on antigens with signal intensity more than 2-fold higher in the patients is provided in Table S2. These results provide evidence for the existence of circulating autoantibody and enhanced autoimmune/immune reactions in patients with Meniere's disease.

Figure 4. Difference in signal intensity of controls and patients in the Protoarray experiment.

A. Raw signals of Protoarray chips of control and Meniere's disease patient. B. Normalized signal intensities of the antigens with a signal intensity more than 10-fold higher in the patients with Meniere's disease than in the controls. IGHG1, immunoglobulin heavy constant gamma 1; RGS10, regulator of G-protein signaling 10, transcript variant 2; C2orf34, chromosome 2 open reading frame 34; SH3GLB1, SH3-domain GRB2-like endophilin B1; ACY1, aminoacylase 1; CAMK4, calcium/calmodulin-dependent protein kinase IV; GSG1L, GSG1-like (GSG1L), transcript variant 2, mRNA. Red bars and error bars represent the mean normalized signal intensity and the SE, respectively.

https://doi.org/10.1371/journal.pone.0111039.g004

Reaction of inner ear tissue antigens with patient sera

Western blotting (Fig. 5) was performed to investigate whether the sera from patients with Meniere's disease could produce an antigen-antibody reaction with inner ear tissue. There were no single disease-specific bands for an antigen-antibody reaction in the patients; instead bands corresponding to a molecular weight of 63–75 kDa and 25–48 kDa were more frequently found in the patients with Meniere's disease (Fig. 5).

Figure 5. Western blot of the reaction of serum from controls and patients with mouse inner ear.

Red, blue, and yellow arrows represent the detected inner ear antigens with molecular weights of 63–75 kDa, 35–48 kDa, and 25–25 kDa, respectively. C, control; P, patient. Co, mouse cochlear tissue protein, V, mouse vestibular tissue protein.

https://doi.org/10.1371/journal.pone.0111039.g005

Bands corresponding to a molecular weight of 63–75 kDa were detected in 6 of the patients and in 2 of the controls (red arrows in Fig. 5). Evidence of an antigen-antibody reaction both in the cochlear and vestibular tissues was detected in 3 of the 6 samples from patients with Meniere's disease; evidence of an antigen-antibody reaction only in the vestibular tissues was observed in 2 of the 6 samples, whereas evidence for this reaction only in the cochlear tissues was observed in 1 of the 6 samples. Evidence for an antigen-antibody reaction both in the cochlear and vestibular tissues was observed only in one of the controls (C7) and a weak reaction was observed only in the cochlear tissue in another one of the controls (C9) (Fig. 5).

Bands distributed between 25–48 kDa were detected in 7 of the patients with Meniere's disease and in 2 of the controls (blue and yellow arrows in Fig. 5). The reaction was detected only in the vestibular tissues in 5 of the 7 patient samples and both in the cochlear and vestibular tissues in 2 of the patient samples. The reaction was detected only in the cochlea in 1 of the controls and both in the cochlea and vestibule in the other control.

We evaluated whether the antigens that reacted with patient serum in the Protoarray experiment had molecular weights of 25–48 kDa and 63–75 kDa, as shown in our western blot data. The following proteins had molecular weights in the ranges of interest: C12orf48 (65.1 kDa), PCLO (46 kDa), ACY1 (45.9 kDa), NPY2R (42.7 kDa), B3GALT4 (41.5 kDa), SH3GLB1 (40.8 kDa), HNRPH3, transcript variant 2H9A (36.9 kDa), GSG1L, transcript variant 2, mRNA (36.8 kDa), C2orf34 (36.1 kDa), IGHG1 (36.1 kDa), and NEK7 (34.6 kDa). These antigens could be involved in inner ear autoimmune reactions. Among the proteins, C12orf48 was the only protein which produced a significantly lower signal intensity in the patients without an antigen-antibody reaction in western blots (P1, P2 in Fig. 5) than in the patients with a detectable antigen-antibody reaction in western blots (P3–P10). However, without peptide sequencing, we cannot determine whether the protein was a specific antigen involved in an immune reaction.

Our results imply that multiple autoantibodies or antigens rather than a specific antibody or antigen can cause autoimmune reactions in the inner ear that result in Meniere's disease.

Discussion

The pathophysiology of Meniere's disease is still unknown, and various etiologies have been proposed. One of the proposed etiologies of Meniere's disease is autoimmunity; this putative etiology is supported by the fact that this disease often occurs bilaterally (in 25–40% of patients), responds to glucocorticoids and anti-inflammatory treatments, and is characterized by elevated levels of autoantibodies or circulating immune complexes and antigen-antibody reactions between patient serum and animal inner ear tissues [18]. However, Meniere's disease does not always occur bilaterally, and experimental studies have only been performed on small numbers of patients or have only targeted a restricted number of autoantibodies or inflammatory markers.

Although the number of patients enrolled in this study was small, we found reliable evidence for the inner ear immune/inflammatory reaction in patients with Meniere's disease by investigating the protein composition of the inner ear fluid in diseased patients and controls. We also mass screened for autoantibody-antigen reactions using the Protoarray system and detected antigen-antibody reactions using patient serum and mouse inner ear tissues. Most of these methods have not been used in previous experiments. The results of this study will contribute to the development of more cost-effective and efficient methods for screening and detecting autoimmune reactions in large numbers of patients with Meniere's disease.

Analysis of ES luminal fluid and evidences of immune reaction

Autoimmune reactions in the inner ear may cause damage to the epithelial layers surrounding the endolymphatic space. In such an instance, fluid in the endolymphatic space should contain evidence of immunologic reactions, including autoantibodies such as those found in the synovial fluid of patients with rheumatoid arthritis. Thus, analyses of the endolymph can provide evidence for the involvement of autoimmunity in the pathogenesis of Meniere's disease.

We analyzed the protein composition of the inner ear fluid of diseased and control groups to find evidence suggesting that increased immune or inflammatory reaction is involved in the pathogenesis of Meniere's disease. Proteomic techniques (LC-MS/MS) were used to analyze the protein constituents of the inner ear fluid. We used ES luminal fluid because the protein concentrations in this fluid are very high [17] and because the ES is the site where most immunologic reactions occur in the inner ear [19]–[21]. In addition, sampling the ES luminal fluid does not usually affect inner ear function; indeed, the 3 patients enrolled in this experiment who underwent ES surgery had preserved inner ear functions after sampling.

The use of cochlear and vestibular endolymph in addition to ES luminal fluid would have improved our study; however, sampling the cochlear and vestibular endolymph in patients with Meniere's disease would have been impossible because the inner ear function of the patients would have been destroyed and the protein concentrations in these compartments would have been too low for the analysis.

The most commonly encountered proteins in the ES luminal fluid were immunoglobulins and their variants. This is expected, as the ES is a known site of immunologic responses in the inner ear. However, the only proteins that were detected exclusively in the patients with Meniere's disease were immunoglobulins, their variants, and interferon gamma regulatory factor, suggesting that increased inflammatory reactions in the inner ear may contribute to the pathology of Meniere's disease. The increased inflammatory reaction could be caused by various etiologies such as allergy, viral infection, genetic cause, or autoimmunity. Although direct evidence of autoimmunity was not found in the luminal fluid analysis, our results about the presence of circulating autoantibodies and increased immune reaction between patients' sera and mouse inner ear tissue could support the possibility of autoimmunity as a cause of increased immune/inflammatory reaction in the inner ear. If autoimmunity was the cause of the increased inflammatory reactions the autoantibodies responsible for these reactions have a molecular weight between 17 and 26 kDa and between 55 and 63 kDa as revealed by LC-MS/MS in our results.

Evidences for autoimmune reactions in the inner ear of patients with Meniere's disease have been reported. One report demonstrated the presence of focal inflammation with intraepithelial invasion by mononuclear cells in the ES of patients with Meniere's disease that altered the normal structures in the endolymphatic sac [22]. The authors of this study suggested that autoimmune reactions may have triggered the inflammatory changes in the ES.

Another report showed that an antigen-antibody reaction between sera from patients with Meniere's disease and human endolymphatic sac tissue was detected by immunohistochemistry in 10% of patients [23]. However, no study has comprehensively analyzed the protein components of the inner ear fluid of patients with Meniere's disease. This is the first study to investigate the difference in protein constituents of the ES luminal fluid of controls and patients with Meniere's disease and to provide supportive evidence for the involvement of autoimmunity in the pathogenesis of Meniere's disease.

However, there were limitations in our study associated with the use of LC-MS/MS to analyze the ES luminal fluid. First, we could not compare the amounts of certain proteins or peptides, as this technique could only identify the types of proteins contained in the fluid. A comparative analysis with the iTRAQ technique could have quantified the amounts of protein in the controls and patients. However, the amount of ES luminal fluid sampled from each person was too small (less than 200 µl after dilution) to be quantified by iTRAQ and exact quantification after dilution is impossible. In addition, the development of less-invasive therapies such as the use of intratympanic steroids or gentamicin injections has limited the need for ES surgery and therefore the opportunities to sample ES luminal fluid. Additionally, although many immunoglobulins and their variants were detected in the LC-MS/MS analysis, it remains unclear whether these proteins acted as autoantibodies. Immunoprecipitation using human inner ear fluid and the inner ear or using human inner ear fluid and animal inner ear samples would need to be performed, and the identity of the autoantibody would need to be confirmed with mass spectrometry. However, these experiments would be difficult to perform for ethical reasons and because of the limited amounts of human inner ear fluid. Microanalysis techniques with very small amounts of sample should be performed to confirm the occurrence of autoimmunity in the future.

Evidence of circulating autoantibodies in patient sera

We sought to determine whether circulating autoantibodies were present in the sera of the patients, as inner ear fluid is difficult to sample for practical use; this limitation precludes the study of the pathologic mechanisms underlying Meniere's disease in large numbers of patients. Detectable autoantibodies or evidence of autoimmunity in the serum could be used as diagnostic biomarkers in conjunction with mass screening. There were several reports investigating the difference of serum protein profiles between Meniere's disease patients and controls using proteomics as a screening method. Proteins related to immune reaction or inner ear disorder such as complement factor H, β2 glycoprotein 1, vitamin D binding protein, and β actin were previously reported to be increased in the serum of Meniere's disease by proteomic analysis [15]. Additionally, fibrinogen α and γ chains, β2 glycoprotein, and complement factor B and H were revealed to be differently expressed according to hearing threshold of patients [24]. However, this study eliminate serum abundant proteins such as albumin and immunoglobulins for 2-DE analysis and performed LC-MS/MS in the several spots that were different between Meniere's disease and control after image analysis. Therefore, most circulating antibodies could be excluded in the analysis and only targeted spots were analyzed by LC-MS/MS; the study did not focused on the Ag-Ab reaction between circulating antibodies and corresponding antigens, but analyzed proteins exclusively increased in the serum of Meniere's disease patients. In our study, we tried to investigate Ag-Ab reaction between circulating antibodies and target antigens using Protoarray and consequently detected circulating antibodies to 18 candidate proteins which could be involved in autoimmune reactions in the patients; the signal intensity of these proteins was more than 2-fold higher in the patients with Meniere's disease than in the controls. With the exception of IGHG1, all of these proteins were located in the membrane or subcellular areas (information on the localization of these proteins can be found at http://www.uniprot.org). These proteins are known to be involved in cell signaling, mitochondrial structure, receptor gating, cell mitotic activity, ganglioside biosynthesis, fasciculation, neuropeptides, DNA splicing and repair, cytokinesis, and maintaining neurotransmitter release sites (Table 3). However, the functions of some of these proteins, including cDNA clone IMAGE:4155919, complete cds, and Mitochondrial coiled-coil domain protein 1 (MCCD1), are unknown. Among those proteins, only mRNA of CAMK4, NEK 7, and PCLO were reported to exist in the inner ear sensory epithelium and ribbon synapse [25], but the existence of the other proteins in the human inner ear has not been demonstrated. PCLO is thought to be involved in the synaptic neurotransmission in the inner ear, although its functional relevance is still unclear [26]. If autoantibodies for PCLO in Meniere's disease deteriorate the inner ear function by reacting with PCLO in the ribbon synapse, it could be one of the candidate autoantibodies for the disease. However, a truncated splice variant of PCLO, piccolino, which could maintain the integrity of synaptic transmission in the absence of PCLO in the retina ribbon synapse, was also found in the sensory ribbon synapse of the cochlea [27]. Therefore, the synaptic transmission can remain intact, even with the Ag-Ab reaction between autoantibodies and PCLO in the inner ear. The function of the other proteins which were reported to exist in the inner ear still remains unknown.

The enhanced immune reaction associated with IGHG1 could be explained in several ways. First, an immune complex autoimmune disorder could explain this occurrence. Circulating autoantibodies to immunoglobulins could form immune complexes that may cause inner ear damage via a type III hypersensitivity reaction. Indeed, several studies have reported elevated circulating immune complexes in 54–94% of patients with Meniere's disease [28], [29]. However, other studies have shown evidence of circulating immune complexes in only 4–7.4% of patients [30], [31]. Several methods can be used to detect circulating immune complexes; however, none of these methods can detect all types of circulating immune complexes. Therefore, the prevalence of patients with circulating immune complexes may vary from study to study. Differences in the race and number of patients enrolled in a study can also influence the prevalence of circulating immune complexes. Evidence of damage to the inner ear associated with circulating immune complexes is important in the pathogenesis of Meniere's disease. However, evidence for inner ear pathology is insufficient, even though histopathological studies of the human temporal bone have demonstrated that patients with Meniere's disease have C3 and C1q deposits in their inner ear [32], [33].

Second, increased amounts of anti-IGHG1 antibodies in patients with Meniere's disease may be a result of an excessive autoimmune or inflammatory reaction in the inner ear. It is also possible that increased concentrations of anti-IGHG1 antibody are involved in regulating the immune system and suppressing excessive immune reactions by reducing B cell activity. Other antibodies may cause autoimmune reactions at a cellular level and affect the function of the epithelial cells of the inner ear that regulate inner ear homeostasis via the disruption of cell signaling or cellular structures. Such autoimmune reactions could result in secondary increases in the concentrations of anti-immunoglobulin antibodies.

Evidence of an immune reaction between patient serum and the inner ear

Demonstrating the presence of immune reaction between circulating autoantibodies in the serum and the inner ear tissue is important. We used western blots with proteins from mouse inner ear tissue and patient serum to demonstrate the existence of this type of immune reaction. The western blots showed that more antigen-antibody reactions occurred in the patients with Meniere's disease than in the controls. Animal inner ear antigens that have been reported to react with sera from patients with Meniere's disease have molecular weights in the 32–35 kDa, 42–46 kDa, 52–59 kDa, and 79–80 kDa ranges [18]. Microsequencing showed that the 28 KDa and 42 KDa antigens corresponded to Raf-1 and beta actin, respectively [18]. These antigens have been detected in a variety of species, including guinea pigs, cows, and humans, suggesting that the antigens might be organ-specific rather than species-specific. In our experiment, inner ear antigens with molecular weights in the 25–35 kDa, 35–48 kDa, and 57–63 kDa ranges were detected. These antigens are likely to be similar to those detected in previous studies. Although the identity of these antigens can be conjectured from the antigens with similar molecular weight detected in the Protoarray experiment as suggested in the result, it should be further studied using immunoprecipitation of patient serum and inner ear tissues followed by mass spectrometry of the corresponding protein bands.

We divided the mouse inner ear tissue into cochlear and vestibular tissues and investigated whether an antigen-antibody reaction between these tissues and patient serum occurred. In contrast, previous studies tended to use whole inner ear tissue. We found that each antigen reacted with the serum differently; samples of patient serum could react with the cochlear tissue, with the vestibular tissue, or with both. Clinically, cochlear and vestibular symptoms in Meniere's disease are different for each patient. In general, vestibular symptoms tend to coincide with cochlear symptoms. However, the progression of each cochlear and vestibular symptom and function varies from patient to patient. The varying antigen-antibody reactions observed in each tissue may be associated with the varying clinical features of the disease.

Because a variety of inner ear antigens could react with patient serum, it appears that multiple target antigens and autoantibodies (rather than a single antigen-antibody combination) may be responsible for the autoimmune reaction associated with Meniere's disease. The 1-DE findings examining the protein composition of the ES luminal fluid of patients with Meniere's disease also support this hypothesis: the distribution of bands was different in the 3 patients, suggesting that the protein composition of the ES luminal fluid of each patient was different and that different antibodies or inflammatory materials are present in each patient.

Clinical and future implications

The diagnosis of Meniere's disease is based on clinical symptoms of vertigo and fluctuations in hearing; the diagnosis is confirmed by showing evidence of sensorineural hearing loss using the pure tone audiogram, as recommended by the AAO-HNS [1995] [16]. Other clinical tests, including vestibular function tests, electrocochleography, and the glycerol test, are not as useful for diagnosing Meniere's disease. Because the diagnostic criteria are primarily symptom-based, differentiating Meniere's disease from other vestibular disorders such as vestibular migraine, vestibular paroxysmia, and sudden sensorineural hearing loss with vertigo can be difficult. Diagnostic markers that can more accurately diagnose Meniere's disease are therefore needed. Biologic markers can potentially decrease the cost of diagnosing Meniere's disease by avoiding unnecessary laboratory and imaging work-ups and promoting proper treatment after accurate diagnosis. If autoimmunity is one of the causes of Meniere's disease, detecting autoantibodies or inflammatory materials can be useful.

Highly sensitive experimental chips using candidate antigens (such as the 18 antigens detected in our study) can be manufactured to avoid the need to use the expensive conventional screening chips used in our study. These diagnostic chips should contain multiple candidate antigens, as multiple candidate antigens were detected in our study and in previous reports. A prospective study using experimental chips should be undertaken in a large population, and highly sensitive and specific markers should be chosen. This type of study will enable subtypes and the prognosis of patients with Meniere's disease to be classified. Understanding the pathophysiology underlying Meniere's disease can also contribute to the development of new treatment methods. None of the current treatments for Meniere's disease can prevent the progression of the disease. Prospective, randomized controlled studies using anti-inflammatory agents such as TNF-α or steroids would be performed in a large population in the patients with autoimmune pathology if it can be confirmed by the autoimmunity screening.

Many studies have attempted to describe the pathophysiology of Meniere's disease; however, our understanding of the pathophysiology of this disease remains limited. In fact, Meniere's disease is a syndrome which may be caused by multiple factors. Autoimmunity is one of the candidate etiologies and thought to represent less than 20%. Recently, familial aggregation for Meniere's disease was reported to be as high as 10–20% [6], [8]. This represents a significant role for genetics in Meniere's disease. Among the various candidate genes associated with Meniere's disease, several genes related to immune response were reported to determine an increased susceptibility of Meniere's disease. Genes that were revealed to be associated with bilateral Meniere's disease, chronic balance/hearing loss were allelic variants of DRB1, PTPN22, TLR 10, MICA genes [34]–[36]. Genetic factors can be one of the causes of autoimmunity or increased immune reaction in Meniere's disease. If various candidate genes associated with Meniere's disease are revealed in the future, it can also be used as evidence for developing new treatment method as well as diagnostic and prognostic markers. However, so far, at least 60–70% of etiologies of Meniere's disease remains unknown, and considerable efforts should be taken to investigate the etiopathogenesis of Meniere's disease using new molecular technologies.

Conclusion

The findings of this study suggest that autoimmunity could be one of the pathologic mechanisms behind Meniere's disease. Multiple autoantibodies and antigens may be involved in the autoimmune reaction. Specific antigens that caused immune reactions with patient's serum in Protoarray analysis can be candidates for the diagnostic biomarkers of Meniere's disease. Further studies with mass screening using candidate antigen-antibody reactions are needed to identify future treatment modalities and to determine the true prevalence of autoimmune pathologic mechanisms underlying Meniere's disease.

Acknowledgments

We wish to acknowledge technical support from Yonsei Proteomie Research Center (www.proteomix.org) and professor Won-Sang Lee who provided human samples of inner ear fluid. We also thank to professor Eui-Cheol Shin, director of Laboratory of Immunology & Infectious Disease, Graduate School of Medical Science and Engineering, KAIST, for his helpful comments about the experiment.

Author Contributions

None. Conceived and designed the experiments: JYC SHK. Performed the experiments: SHK JYK HJL MG. Analyzed the data: JYC SHK JYK HJL MG BGK. Contributed reagents/materials/analysis tools: JYC SHK. Wrote the paper: SHK.

References

  1. 1. Alexander TH, Harris JP (2010) Current Epidemiology of Meniere's Syndrome. Otolaryngologic Clinics of North America 43: 965–970.
  2. 2. Anderson JP, Harris JP (2001) Impact of Meniere's disease on quality of life. Otol Neurotol 22: 888–894.
  3. 3. Sajjadi H, Paparella MM (2008) Meniere's disease. Lancet 372: 406–414.
  4. 4. Banks C, McGinness S, Harvey R, Sacks R (2012) Is allergy related to Meniere's disease? Curr Allergy Asthma Rep 12: 255–260.
  5. 5. Fife TD, Giza C (2013) Posttraumatic vertigo and dizziness. Semin Neurol 33: 238–243.
  6. 6. Hietikko E, Kotimaki J, Sorri M, Mannikko M (2013) High incidence of Meniere-like symptoms in relatives of Meniere patients in the areas of Oulu University Hospital and Kainuu Central Hospital in Finland. Eur J Med Genet 56: 279–285.
  7. 7. Minor LB, Schessel DA, Carey JP (2004) Meniere's disease. Curr Opin Neurol 17: 9–16.
  8. 8. Requena T, Espinosa-Sanchez JM, Cabrera S, Trinidad G, Soto-Varela A, et al. (2014) Familial clustering and genetic heterogeneity in Meniere's disease. Clin Genet 85: 245–252.

The host cell plays a major role during the replication of small plus-stranded RNA viruses with limited coding capacity. A yet-unknown number of host proteins, termed host factors, are likely exploited by viruses to facilitate their replication and spread. Indeed, viruses likely coopt or alter many cellular processes, such as translation and RNA and protein degradation, via their coded proteins (1, 27, 29). The host cells also activate antiviral defense mechanisms, such as gene silencing, to destroy viral RNAs (20). Recent systematic genome-wide screens of a single-gene knockout library of yeast (Saccharomyces cerevisiae) conducted with Brome mosaic virus (BMV) (19) and Tomato bushy stunt virus (TBSV) (17, 35) revealed that their replication is affected by ∼100 mostly unique host genes. Additional genome-wide screens revealed that 32 host factors affected tombusvirus evolution via stimulating or inhibiting RNA recombination (50, 51). However, the above-described genome-wide screens could have missed the identification of those host genes, which are part of multimember gene families with overlapping functions. Therefore, additional screens might lead to the identification of new host genes affecting virus replication.

Genomic RNA of TBSV and the closely related Cucumber necrosis virus (CNV) codes for two replication proteins, termed p33 and p92pol, and three additional proteins involved in encapsidation, cell-to-cell movement, and suppression of gene silencing (63). p92pol is the RNA-dependent RNA polymerase (28, 31, 34, 37), whereas p33 is an essential replication cofactor involved in RNA template selection/recruitment (26, 33, 41) and in the assembly of the viral replicase (38), and it is an integral part of the tombusvirus replicase complex (45, 49). Interestingly, yeast cells expressing tombusvirus p33 and p92pol replication proteins can efficiently replicate a TBSV-derived replicon RNA (repRNA), which is a defective interfering (DI) RNA identified in TBSV-infected plants (34, 39). The tombusvirus repRNA serves not only as a template for replication but also as an assembly platform for the viral replicase complex (38, 39). Altogether, yeast serves as a useful model host to study replication and recombination of tombusviruses, allowing the utilization of powerful genomics and proteomics tools developed for yeast.

The p33 replication protein likely interacts with many host proteins that could affect various steps during tombusvirus infections. Host proteins are likely involved in intracellular trafficking of p33 and in posttranslational modification of p33, such as phosphorylation, which has been shown to affect the ability of p33 to bind to the viral RNA (55, 58). To identify those host proteins that interact with the tombusvirus p33 replication cofactor, we took a global approach based on the yeast proteome microarray (66, 67). Previous studies using the yeast proteome microarray have identified numerous yeast proteins involved in protein-protein interactions, lipid binding, DNA binding, and small-substrate binding, thus demonstrating the usefulness of the global analysis approach (14, 15, 56, 66, 67). In addition, a yeast protoarray approach was used successfully to identify many host proteins interacting with a 3′ fragment of the BMV RNA (68). Two of those host proteins were confirmed to play a role during BMV infection in plants.

This work has led to the identification of 58 host proteins that bound to the purified recombinant p33, whereas 11 bound to the unique C-terminal portion of p92pol. The identified proteins include helicases, ubiquitin (Ub) ligases and proteases, translation factors, RNA-modifying enzymes, and proteins with unknown functions. Additional data from in vitro binding experiments and the split-Ub two-hybrid assay confirmed many of the host protein-p33 interactions. Further detailed work with Cdc34p Ub-conjugating enzyme demonstrated that it is present within the viral replicase complex. Downregulation of Cdc34p decreased the accumulation of TBSV repRNA in yeast and reduced the in vitro activity of the tombusvirus replicase. Ubiquitination of p33 has been performed in vitro with purified Cdc34p, which likely plays a regulatory role in TBSV replication. In addition, we have shown that p33 becomes mono- and biubiquitinated in yeast cells. Overall, the presence of Cdc34p within the tombusvirus replicase validates the usefulness of the protoarray approach.

Previous SectionNext Section

MATERIALS AND METHODS

Yeast and Escherichia coli plasmids.To study the effect of overexpression of selected yeast proteins on viral RNA replication and recombination, we used the yeast open reading frame (ORF) collection from Open Biosystems. In this yeast ORF collection, each ORF is expressed from 2μ plasmid BG1805 under the control of the GAL1 promoter and fused to a tandem affinity tag that includes a hemagglutinin tag and the zz domain of protein A at the C terminus. For the replication assay, the parental strain (BY4741) was cotransformed with three separate plasmids: (i) pHisGBK-His33/DI72-CUP1 [coexpressing CNV p33 from the ADH1 promoter and DI72(+) RNA from the CUP1 promoter], (ii) pGAD-His92-CUP1 (containing the CNV p92pol gene behind the CUP1 promoter), and (iii) one of the individual yeast ORF clones (Open Biosystems) or 2μ plasmid pYES-NT-C (Invitrogen) as a control. For the recombination assay, pHisGBK-His33-CUP1/DI-AU-FP-GAL1 (coexpressing CNV p33 from the CUP1 promoter and DI-AU-FP from the GAL1 promoter) (7) was used together with pGAD-His92-CUP1 and one of the yeast ORF plasmids. For FLAG purification of virus replicase from yeast, the yeast plasmids pGBK-33HF and pGAD-92HF, expressing His6- and FLAG-tagged p33 and p92, and pGBK-His33 and pGAD-His92, expressing only His6-tagged p33 and p92, were described previously (49).

The full-length CDC34 and the regions of CDC34 corresponding to the N-terminal 170-amino-acid (aa) and C-terminal 125-aa sequences were amplified by PCR (the sequences of the primers are available upon request). The PCR products were treated by BamHI and XhoI and ligated into pYES-NT-C or pYC2/CT (a centromeric, low-copy-number plasmid) (Invitrogen) digested with the same enzymes. The PCR products were also cloned into pGEX-2T (Novagen) to construct protein expression plasmids in E. coli. The CDC34 C95S mutation was constructed by use of a QuikChange XL site-directed mutagenesis kit (Stratagene), changing the Cys95 TGT codon to TCT (serine). The mutated DNA was cloned into pYES-NT-C, pYC2/CT, and pGEX-2T as described above.

To construct the expression vector for the glutathione S-transferase (GST)/Ub fusion, we amplified the yeast Ub sequence with PCR using primers 2227 (GGCGGGATCCATGCAGATTTTCGTCAAGACTTTG) and 2228 (GGCCCTCGAGTTAACCACCTCTTAGCCTTAGCACAAG). The PCR product was digested with BamHI and XhoI and cloned into pGEX-2T at the BamHI/XhoI sites.

Yeast strains and culturing.Saccharomyces cerevisiae strains BY4741 (MATahis3Δ1 leu2Δ0 met15Δ0 ura3Δ0) and InvSc1 were obtained from Invitrogen. The CDC34 strain (pCDC34::Kanr Tet07 TATA URA3::CMV-tTA MATahis3-1 leu2-0 met15-0) with the titratable yeast Tet promoters from the Hughes collection (yTHC) was obtained from Open Biosystems. Yeast cultures were grown at 29°C in YPD (1% yeast extract, 2% peptone, and 2% glucose) or defined synthetic medium (SC) supplemented with appropriate amino acids and 2% glucose or 2% galactose as a carbon source. Yeast transformation was done by using the standard lithium acetate-single-stranded DNA-polyethylene glycol method, and transformants were selected by complementation of auxotrophic markers (34, 36). The URA3 gene in the CDC34 yTHC strain was mutated by transforming yeast with an SdaI-truncated ura3 fragment and then growing yeast on SC medium with 2% glucose in the presence of 1 g/liter 5-fluororotic acid. The 5-fluororotic acid-resistant colonies were selected and the colonies were further checked by their inability to grow on SC medium lacking uracil (SC-U).

Expression and purification of recombinant tombusvirus replication proteins.The sequence of a v5 epitope tag was added at the C terminus of TBSV p33 in plasmid pMAL-p33 (44) by use of PCR with p33-specific primers 788 (GAGGGATCCGAGACCATCAAGAGAATG) and 1621 (CGCGTCTAGATTTGACACCCAGGGACTCCTGTGA) and v5-specific primers 1619 (CGCGTCTAGAGGGCCCTTCGAAGGTAAGCCT) and 1620 (CGGGCTGCAGTCAATGGTGATGGTGATGATGACCGGT). The obtained PCR products were digested with BamHI-XbaI and with XbaI-PstI, respectively. The two PCR products were cloned simultaneously into BamHI-PstI-digested pMAL-c2 vector (NEB). The obtained plasmid was used to express the maltose binding protein (MBP)-p33-v5 fusion protein in E. coli (44, 46), followed by the addition of 100 μg of RNase A to the bacterial pellet prior to resuspension in 50 mM HEPES, pH 7.4, and 100 mM NaCl. Affinity purification was performed on maltose binding resin in 50 mM HEPES and 100 mM NaCl according to the manufacturer's instructions (NEB). The MBP tag of the MBP-p33-v5 fusion protein was cleaved off with 1 μl factor Xa at 4°C in 1× protoarray buffer plus 0.5 M NaCl and 0.5% Triton X-100 (46). The final concentration of p33-v5 was 0.7 μg/μl. The v5 epitope tag was added to the C terminus of pMAL-p33N82 and pMAL-p92C (Fig. 1A) (43) by using a strategy as described above for pMAL-p33.

FIG. 1.

Yeast protoarray approach to identify host proteins interacting with TBSV replication proteins. (A) Schematic representation of the overlapping TBSV p33 and p92pol replication proteins and their derivatives expressed in E. coli. TMD, predicted transmembrane domains; RPR, arginine-proline-rich RNA-binding region; RDRP, RNA-dependent RNA polymerase; S1 and S2, p33:p33 interaction subdomains. Note that the sequence in the N-terminal region in p92pol is identical with the p33 sequence due to the gene expression strategy of tombusviruses. (B) Identification of host proteins binding to p33 and a truncated p33 (termed p33C) (panel A) replication protein based on the protoarray. The whole chip contains 4,088 purified yeast proteins in duplicate. For the binding assay, we used E. coli-expressed v5-tagged p33 and biotinylated p33C cleaved from the MBP tag. Two representative subarrays are shown to illustrate the binding of host proteins to p33 but not to p33C (top) or to both p33 and p33C (bottom). Note that the protoarray contains various amounts of yeast proteins (the actual values are supplied by Invitrogen), and this information was used to calculate the relative binding of each host protein to p33/p33C (Table 1).

Biotinylation of p33C for protoarray analysis.Twenty microliters of the affinity-purified MBP-p33C (2.5 μg/μl) (44) in 50 mM HEPES, pH 7.4, and 100 mM NaCl was biotinylated by the addition of 1.6 μl of 5 nmol/μl biotin-sulfosuccinimidyl ester sodium salt and 2 μl of 1 M NaHCO3, pH 8.4. The biotinylation reaction was performed at room temperature for 60 min, followed by cleaning through a gel filtration minicolumn (exclusion limit, 6,000 Da). The final protein concentration was 0.5 μg/μl. The quality of the protein biotinylation was assessed by Western blotting using streptavidin-alkaline phosphatase conjugate and a fluorescent substrate, CDP-Star, according to the manufacturer's instructions (Invitrogen).

Yeast protoarray analysis.The protein array slide (Invitrogen) was blocked for 1 h at 4°C in the blocking buffer (1× phosphate-buffered saline, 1.0% bovine serum albumin, and 0.1% Tween 20). Then, 120 μl of the p33-v5 probe (5 ng/μl) or the biotinylated p33C probe (50 ng/μl) in the probing buffer (1× phosphate-buffered saline, 5 mM MgCl2, 0.5 mM dithiothreitol [DTT], 0.05% Triton X-100, 5% glycerol, and 1.0% bovine serum albumin) was pipetted on each protoarray slide. The slide was then covered and incubated at 4°C for 1.5 h. After three 1-min washes with the cold probing buffer, the protein array was incubated in the dark on ice with 0.1 μg/ml streptavidin-Alexa Fluor 647 for the detection of the biotinylated protein probe or with anti-v5-Alexa Fluor 647 conjugate for the v5-tagged p33 probe. The unbound Alexa Fluor 647 conjugates were washed off three times with 1-min washes in the probing buffer. The array was centrifuged for 4 min at 800 × g and left in the slide holder to air dry for 30 min. Scanning and quantification were performed with a ScanArray 4000 (Packard Bioscience, Billerica, MA) scanner and QuantArray V3.0 software. We scored those host proteins that had a relative value of higher than 150, as calculated on the basis of the actual amounts of yeast proteins spotted on the protoarray and the measured signal in repeated experiments, as shown in Table 1.

TABLE 1.

The names and functions of yeast proteins bound to TBSV p33

Pulldown assay.The recombinant TBSV p33C was expressed as a fusion with MBP in E. coli Epicurian BL21-CodonPlus (DE3)-RIL (Stratagene) as described previously (43). E. coli cells were sonicated in MBP column buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 10 mM β-mercaptoethanol), and the fusion protein was immobilized on amylose resin (100 μl). The unbound material was drained and washed with MBP column buffer twice. Yeast strains expressing individual GST-His6-tagged proteins were selected from a GST-His6 ORF library (a generous gift from Brenda Andrews) (57). Yeast cells were first pregrown in SC-U with 2% glucose at 29°C and then grown in SC-U with 2% galactose until reaching mid-logarithmic phase (optical density at 600 nm, 0.8 to 1.0). Yeast cells were suspended in binding buffer (20 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, 10% glycerol, 0.1% Nonidet P-40 [NP-40], 10 mM β-mercaptoethanol, and 1% [vol/vol] yeast protease inhibitor mix) and were broken by glass beads in a Genogrinder. The total yeast lysate was cleared by centrifugation at 21,000 × g for 10 min at 4°C (performed twice) and loaded on the affinity column with immobilized MBP-p33C or MBP. The binding reactions were performed at 4°C for 1 h with continuous rotation. After the binding, the beads were washed five times with 1 bed volume of binding buffer containing 200 mM NaCl. Proteins bound to the beads were eluted by incubation in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer at 85°C for 10 min. Samples (10 μl) were loaded onto SDS-PAGE gels. Standard Western blotting was performed with anti-His/anti-mouse immunoglobulin G antibodies (39).

Host protein overexpression studies.For inducible launching of TBSV replication, we used the copper-regulatable CUP1 promoter. The CUP1 promoter was amplified form the yeast genomic DNA with primers 1779 (CGCGGAATTCGACATTTGGGCGCTATACGTGCATATGT) and 1780 (CGCGCTCGAGTACAGTTTGTTTTTCTTAATATCTATTTCGA), followed by digestion of the PCR product with XhoI/EcoRI. The sequence of DI72 in combination with the 3′-located ribozyme was amplified with primers 471 (CCCGCTCGAGGGAAATTCTCCAGGATTTCTC) and 1069 (CCGGTCGAGCTCTACCAGGTAATATACCACAACGTGTGT) and digested with XhoI/SacI. Both fragments were cloned simultaneously into pHisGBK-His33-DI72 (17) digested with EcoRI/SacI to yield pHisGBK-His33-CUP-DI72.

Yeast strain BY4741 was transformed with the selected ORFs under the GAL1 promoter (Open Biosystems) along with pGAD-His92 and pHisGBK-His33-CUP-DI72. Selection was done in synthetic complete medium lacking Ura, Leu, and His (SC-ULH medium), whereas the culturing of yeast transformants was done in 1.5 ml of SC-ULH plus 2% galactose for 20 h at 30°C to produce the particular host protein. Then, TBSV repRNA replication was launched by adding 50 μM CuSO4 to the media, followed by additional culturing for 24 h at 30°C. The final optical density was ∼0.7 to 1.0. For the detection of the expressed host and viral proteins, we performed Western blotting as described previously (39).

Split-Ub yeast two-hybrid assay.The split-Ub assay was based on the Dualmembrane kit 3 (Dualsystems Biotech). The His6-tagged CNV p33 was amplified from pGBK-His33 (39) with primers 2261 (GTCGCTGCAGTACTAGTAGGCCTGGAGGTTCTCATCATCATC) and 2262 (GTCGCCATGGAGGCCTCTATTTCACACCAAGGGAC) and cloned into PstI/NcoI-digested pBT3-N (Dualsystems). From the resulting plasmid, the expression cassette comprising CYC1pro-LexA-VP16-Cub-6xHis-p33 was amplified by PCR using primers 2236 (CGGCCTGCAGGCTCATTTGGCGAGCGTTGG) and 2262 and cloned into SdaI/NcoI-digested pGAD-H (40), creating pGAD-BT2-N-His33. pPR3-N and pPR3-C (Dualsystems) were modified by the addition of a synthetic polylinker. To do that, CNV p33 was amplified from pGBK-His33 by use of primers 2181 (GTGGGATCCGAATTCAGATCTGGGCCCGGGATGGATACCATCAAGAGGATG) and 2182 (GTCGTCGACTTAATCGATGCTAGCCCATGGCCCGGGTTTCACACCAAGGGACTC). The resulting PCR product was digested with BamHI and SalI and ligated into similarly digested pPR3-N (Dualsystems) to generate pPR3-N-p33. Alternatively, the PCR product was digested with BamHI and ClaI and ligated into similarly digested pPR3-C (Dualsystems) to generate pPR3-C-p33. Plasmids pPR3-N-p33 and pPR3-C-p33 were digested with XmaI and gel purified to excise the portion containing p33 and religated to create pPR-N-RE and pPR-C-RE. Host genes were amplified by PCR using specific primers and cloned into pPR-N-RE or pPR-C-RE in frame with the hemagglutinin-NubG cassette. Yeast strain NMY51 [MATahis3Δ200 trp1-901 leu2-3,112 ade2 LYS2::(lexAop)4-HIS3 ura3::(lexAop)8-lacZ ade2::(lexAop)8-ADE2 GAL4; Dualsystems] was transformed with pGAD-BT2-N-His33 and pPR-N-RE (NubG) or one of the pPR host gene constructs and plated onto Trp/Leu minimal medium plates. Transformed colonies were picked with a loop, resuspended in water, and streaked onto Trp/Leu/His/Ade plates to test for p33-host protein interactions.

Replicase purification.Yeast SC1 strain transformed with expression plasmids for His6- and FLAG-tagged p33 and p92 and DI72 were grown at 29°C until reaching an optical density at 600 nm of 0.8 to 1.0. Replicase purification was performed as described earlier with modification (49). Briefly, 200 mg of yeast cells was broken in TG buffer (50 mM Tris-HCl [pH 7.5], 10% glycerol, 15 mM MgCl2, and 10 mM KCl) supplemented with 0.5 M NaCl, 0.1% NP-40, and 1% (vol/vol) yeast protease inhibitor cocktail. The enriched membrane fraction was solubilized in 1 ml TG buffer with 0.5 M NaCl, 1% NP-40, 5% caprylyl sulfobetaine (SB3-10; Sigma) via gentle rotation for 2 h at 4°C. Then, the tube was incubated at 37°C for 5 min and centrifuged at 21,000 × g at 4°C for 15 min. Supernatant was incubated with 50 μl anti-FLAG M2-agarose affinity gel (Sigma) for 2 h at 4°C. The unbound materials were removed by centrifugation at 500 × g, and the resin was washed three times with 1 ml TG buffer with 0.5 M NaCl, 1% NP-40, and 1% SB3-10 and twice with 1 ml TG buffer with 150 mM NaCl and 0.1% NP-40. Proteins bound to affinity beads were eluted by incubating in SDS-PAGE loading buffer at 85°C for 10 min and then subjected to SDS-PAGE on a 10% gel by use of anti-His (Amersham) and anti-cdc34 (generous gift from Mark Goebl) antibodies.

In vitro ubiquitination.The yeast strain expressing UBA1 was from the GST-His6 ORF library (57). The purification of GST-His6 UBA1 was performed from yeast, while the GST-CDC34 and GST/Ub was obtained from E. coli by use of glutathione Sepharose beads (Novagen). The obtained GST fusion proteins were dialyzed in 40 mM Tris-HCl (pH 7.5), 10% glycerol, 50 mM NaCl, 2 mM MgCl2, and 0.5 mM DTT overnight at 4°C (two changes of 0.5 liter buffer each). The purified His6-tagged human Ub was obtained from Boston Biochem. In vitro ubiquitination was carried out in 20 μl ubiquitination buffer (40 mM Tris-HCl [pH 7.5], 50 mM NaCl, 5 mM MgCl2, 2 mM ATP, 1 mM DTT, 10 mM creatine phosphate [Roche], and 1 U creatine kinase [Roche]) containing 50 nM GST-His6-UBA1, 500 nM GST-CDC34, 100 μM His6-Ub, or 30 μM GST/Ub and purified recombinant 0.5 μg MBP-p33 or MBP as substrates. The ubiquitination assays were performed at 30°C for 60 min and then terminated by boiling for 5 min with SDS sample buffer containing 0.1 M DTT, followed by Western blot analysis with anti-MBP antibody.

Analysis of p33 ubiquitination in yeast.To construct pGBK-FLAG-p33, we used primers 2450 (GGCAAGCTTACCATGGGTCGGGATTACAAGGAC) and 992B (GAGCTGCAGCTATTTCACACCAAGGGA) and pGBK-33HFH as a template. The PCR product was digested with NcoI and PstI and ligated into similarly digested pGBK-33HFH. S. cerevisiae strain Sc1 was transformed with pYES2/NT-C and YEp105 (8), which expresses a copper-inducible c-Myc-tagged Ub and pGBK-33HFH or pGBK-FLAG-p33. Transformed yeast cells were pregrown in medium lacking Ura, Trp, and His (UTH minimal medium) supplemented with 2% glucose for 24 h at 29°C and then pelleted and grown in UTH minimal medium plus 2% galactose and 50 mM CuSO4 for an additional 24 h. Cultures were centrifuged and washed in 20 mM Tris-HCl, pH 8.0. Four hundred microliters of yeast pellet was resuspended in 600 μl of ice-cold TG buffer supplemented with 0.5 M NaCl, 0.25% yeast protease inhibitor cocktail, and 10 mM N-ethylmaleimide. Cells were broken with glass beads in a homogenizer for 6 min at 1,500 strokes/min. Two volumes of cold extraction buffer was added and the extracts were transferred to Eppendorf tubes and centrifuged at 21,000 × g for 15 min at 4°C. The supernatant was discarded and the pellet was resuspended in 1 ml ice-cold TG buffer plus 1% NP-40 and 5% SB3-10. This mixture was incubated for 1 h at 4°C with rotation and 5 min at 37°C to solubilize membranes, followed by centrifugation at 21,000 × g for 15 min at 4°C. The supernatant containing the solubilized membrane proteins was loaded onto columns containing 25 μl of anti-FLAG M2 agarose from mouse (Sigma) (equilibrated in TG buffer plus 0.5 M NaCl and 0.1% NP-40) and incubated for 2 h at 4°C with rotation. Columns were drained and filled with 1 ml cold washing buffer I (TG buffer plus 0.5 M NaCl and 1% NP-40) and rotated for 20 min at 4°C. Columns were drained and washed three additional times with washing buffer I and two times with washing buffer II (TG buffer plus 0.05 M NaCl and 0.1% NP-40). After the last wash, columns were centrifuged at 1,000 × g for 1 min to remove excess liquid and proteins were eluted with 50 μl of SDS-PAGE loading buffer without β-mercaptoethanol at 85°C. Columns were centrifuged for a short period to recover the proteins, and then 2.5 μl of β-mercaptoethanol was added to each sample and samples were boiled for 5 min and subjected to SDS-PAGE and Western blotting using monoclonal anti-FLAG M2 antibody from mouse (1:5,000 dilution; Sigma) and alkaline phosphatase-conjugated anti-mouse (1:5,000) followed by Nitro Blue Tetrazolium-BCIP (5-bromo-4-chloro-3-indolylphosphate) detection. c-Myc-tagged Ub was detected using anti-c-Myc antibody from rabbit (1:10,000 dilution; Bethyl) and alkaline phosphatase-conjugated anti-rabbit (1:10,000) followed by Nitro Blue Tetrazolium-BCIP detection.

Previous SectionNext Section

RESULTS

Yeast protoarray-based identification of host proteins that bind to the tombusvirus p33 replication protein.The viral replication proteins likely interact with many host proteins to facilitate viral replication. Also, some host proteins could interact with and inhibit the function of the viral replication proteins as an antiviral mechanism. To determine host proteins interacting with the viral replication proteins on a proteome-wide scale, we have utilized a protoarray-based approach with purified host proteins. Since TBSV repRNA replicates efficiently in yeast (34, 39), we took advantage of the yeast proteome microarray carrying 4,088 purified yeast proteins (∼70% coverage of all the yeast genes). To probe the yeast proteome microarray for the identification of host proteins binding to the tombusvirus p33 replication cofactor protein, we purified recombinant p33 fused to MBP from E. coli by use of affinity chromatography as described previously (44). After purification, the N-terminal MBP fusion was cleaved off from p33 by use of factor Xa protease. Importantly, p33 also carried a C-terminal v5 tag to facilitate its detection. The highly purified p33 preparation was applied (in the presence of an excess amount of bovine serum albumin as a nonspecific competitor) onto the yeast protoarray to promote protein-protein interactions, followed by thorough washing. The bound p33 was detected via v5-specific antibody conjugated with Alexa Fluor dye and visualized with a microchip scanner (Fig. 1B). It is worth noting that the v5 tag in p33 could interfere with binding to host proteins and/or the binding of a host protein to this region of p33 could block the detection of the v5 tag on the array.

We found that 58 host proteins bound to the recombinant p33 efficiently and reliably under the in vitro conditions employed (see Materials and Methods). Among these host proteins, there are 3 protein chaperones (Gim3p, Jjj1p, and Jjj3p), 5 proteins involved in protein ubiquitination (Cdc34p, Rsp5p, Uba1p, Ubp10p, and Ubp15p), 6 translation factors involved in mRNA translation (Bfr1p, Efb1p, Hbs1p, Rpl8Ap, Tif1p, and Tif11p), and 10 proteins involved in RNA processing and metabolism (Ala1p, Bud21p, Erb1, Rib2p, Sas10p, Stm1p, Trm1p, Trz1p, Tsr2p, and Urn1p) (Table 1). Additional proteins are known to be involved in various cellular processes and the functions of nine proteins are not yet defined (Table 1).

To obtain further insight into p33-host protein interactions, we also performed similar binding experiments with purified recombinant p33N82 (v5 tagged), which contains only the N-terminal 82 aa of p33, as well as with p33C (Fig. 1A) with defined RNA-binding and protein interaction domains (43-45) by use of the yeast protoarray. However, for p33C, we used biotin labeling followed by detection via streptavidin conjugated with Alexa Fluor dye and visualization with a microchip scanner. Also, because p33C is more soluble than full-length p33, which carries long hydrophobic stretches (Fig. 1A), we could use p33C at a concentration ∼10 times higher than that used for p33 to increase the sensitivity in the protoarray experiments (see Materials and Methods). Interestingly, 29 host proteins (50% of the 58 bound to the full-length p33 on the chip) bound to p33C, and 12 (21%) bound to p33N82, whereas 8 (14%) bound to both p33C and p33N82 (Table 1). The remaining nine (15%) host proteins did not bind to p33N82 and p33C, suggesting that this group of proteins might bind to the hydrophobic central domain in p33, which is missing from both p33N82 and p33C (Fig. 1A).

Yeast protoarray-based identification of host proteins that bind to the unique C-terminal fragment of replication protein p92pol.To identify host proteins binding specifically to the unique, nonoverlapping portion of p92pol, we also performed protoarray-based binding experiments with purified recombinant p92C (v5 tagged), which contains the RNA-dependent RNA polymerase motifs in the unique C-terminal segment of p92pol (Fig. 1A). We found that 21 host proteins bound both to p33 and to p92C (Table 1), whereas 11 host proteins bound only to p92C (Table 2) and not to p33. The list includes an RNA helicase (Dpb3p), a methylase (Dot1p), an aminopeptidase (Map1p), an RNA-binding protein (Npl3p), and a translation factor (Tef2p) (Table 2). Thus, the protoarray approach allowed the identification of host proteins that selectively bound to p33 or the nonoverlapping portion of p92pol in vitro.

TABLE 2.

The names and functions of yeast proteins bound to TBSV p92C

Interaction of recombinant p33 with selected yeast proteins in vitro.Since the protoarray approach might be prone to identify false positives (i.e., proteins that interact with the viral p33 only on the chip under the conditions applied), we used additional approaches to confirm that the recombinant p33 could bind to the identified host proteins as found in the above-described protoarray experiments. To this end, we performed protein pulldown experiments with purified recombinant p33C and 16 host proteins chosen based on their intriguing functions in yeast. For the protein pulldown experiments, yeast lysates containing soluble proteins prepared from yeast overexpressing one of the 16 GST/His6-tagged host proteins were applied to columns containing immobilized MBP-p33C fusion protein. After elution of the bound proteins from the column, we analyzed whether the particular host protein was present in the eluted fraction by using Western blotting (Fig. 2). Similarly prepared recombinant MBP bound to beads was the negative control to exclude nonspecific binders. These experiments are thus complementary to the protoarray experiments, which had the yeast proteins fixed to a solid surface, while the p33C was the probe protein present in solution.

FIG. 2.

Binding of selected host proteins to p33 replication protein in vitro. MBP-tagged p33C and MBP (1 μg each) were separately immobilized on amylose beads, followed by incubation with cytosolic extracts prepared from yeast individually expressing the given GST-His6-tagged host proteins. The bound host proteins were eluted from the beads and were analyzed by 10% SDS-PAGE and detected via anti-His6 antibody. We tested a total of 17 host proteins out of the 58 host proteins identified via the protoarray for their abilities to bind to p33C. Protein bands of the expected sized are depicted with arrows. Molecular mass markers are shown on the left. Note that Hbs1p was expressed as a His6-tagged protein from plasmid pYES.

As expected, GST control bound to neither MBP nor p33C (Fig. 2, lanes 15 and 16). In contrast, 14 host proteins, including Arp8p, Iwr1p, Tif1p, Tif11p, Gsy2p (not shown), Cdc34p, Hbs1p, Rpl8Ap, Nap1p, Spt16p, Jjj1p, Sas10p, Elf1p, and Stm1p bound to p33C-MBP much more efficiently than to MBP (Fig. 2), confirming that these host proteins can interact with the C-terminal domain of p33 in vitro. The binding of Trm1p to p33C-MBP was only about twice as efficient as that to MBP (Fig. 2, lanes 33 and 34). Smk1p mid-sporulation-specific mitogen-activated protein kinase, which bound weakly to p33 on the protoarray (not shown), did not interact with p33C in vitro (Fig. 2, lanes 27 and 28). Overall, data on 16 yeast proteins from the protein pulldown experiments are mostly in agreement with the protoarray experiments, demonstrating efficient interactions between p33 and the selected yeast proteins under in vitro conditions. The major difference in the results from the protoarray and protein pulldown experiments is that Tif1p did not bind to p33C (only to the full-length p33) in the protoarray (Table 1), while it did bind efficiently to p33C based on protein pulldown experiments (Fig. 2, lanes 5 and 6).

In vivo interaction between identified host proteins and the tombusvirus p33 based on the split-Ub two-hybrid assay.To confirm that the identified p33-host protein interactions can also take place on subcellular membrane surfaces within yeast cells, where p33 is localized (22, 33), we used the split-ubiquitin yeast two-hybrid assay. This assay is based on the ability of the N-terminal (NubG) and C-terminal (Cub) halves of Ub to reconstitute a functional protein (9, 10). When NubG and Cub, both fused separately to interacting proteins, are brought into close proximity and reconstitute a functional Ub protein, cleavage by endogenous Ub-specific proteases (UBPs) leads to the release of an artificial transcription factor, LexA-VP16, fused to Cub. This allows the activation of LexA-driven HIS3 and ADE2 expression in the nucleus. In summary, the split-Ub system, unlike the original yeast two-hybrid system, does not require interacting proteins to be localized to the nucleus, allowing an analysis of protein interactions on the cytosolic surfaces of membranes, which are the natural subcellular locations of the membrane-bound p33 protein (22, 33).

The split-Ub assay with 19 of the identified host proteins revealed that Rpl8Ap, Arp8p, Ubp15p, Elf1p, Nap1p, Tif11p, Ubp10p, Gsy2p, Tif1p, and Stm1p interacted with the membrane-bound p33 when used as N-terminal fusion with NubG (Fig. 3A). Additional host proteins, such as Sas10p, Hbs1p, and Iwr1p, interacted with the membrane-bound p33 when used as C-terminal fusion with NubG (Fig. 3B). The interaction of Efb1p, Trm1p, Spt16p, and Jjj1p with p33 was detectable, but weak, in this assay. Only Uba1p, which is a Ub-activating E1 protein, and Cdc34p (not shown), which is an E2 Ub-conjugating enzyme, have not been found to interact with p33 in the split-Ub assay (Fig. 3B). Since Uba1p and Cdc34p are involved in ubiquitination, they might interfere with Ub cleavage in this assay (putative false negatives). Altogether, the split-ubiquitination assay confirmed that 18 of the identified host proteins interact with membrane-bound p33 in yeast.

FIG. 3.

Confirmation of host protein-p33 interactions via the split-Ub two-hybrid assay. (A) The full-length sequences of 19 host proteins were fused to NubG as N-terminal (N-term) fusions. Heat shock protein 70 (SSA1) was used as a positive control because it is known to interact with p33 (49). The number of colonies formed reflects the strength of protein-protein interaction. Those host factors that had positive interactions with p33 are boxed in black, whereas those with weak interactions are shown in gray boxes. Negative interactors are in boldface on a plain background. Note that the split-Ub two-hybrid assay is based on the ability of N-terminal (Nub) and C-terminal (Cub) halves of Ub to reconstitute a functional protein. A single-amino-acid mutation in Nub (NubG) reduces its affinity for Cub. However, when NubG and Cub are fused to interacting proteins they are brought to close proximity and reconstitute a functional Ub protein, which is cleaved by endogenous UBPs. This cleavage releases an artificial transcription factor, LexA-VP16, which is fused to Cub, allowing the activation of LexA-driven HIS3 and ADE2 genes in the nucleus. Unlike the yeast two-hybrid system, the split-Ub system does not require interacting proteins to be localized to the nucleus, allowing the analysis of protein interactions involving membrane-bound proteins in their natural cell location. (B) Sequences of those host proteins from panel A that interacted weakly or did not interact with p33 were fused to the C terminus (C-term) of NubG and tested for their interactions with p33 as described in the legend to panel A.

Effect of overexpression of selected host proteins on TBSV repRNA replication in yeast.To test if the host proteins that interacted with p33 could affect tombusvirus RNA replication, we took advantage of the previously developed efficient tombusvirus replication system in yeast (34, 35) with some modifications. First, we separately overexpressed 44 of the identified host proteins from the galactose-inducible GAL1 promoter (12) in yeast cells for 20 h. Subsequently, we launched TBSV repRNA replication from the CUP1 promoter in the same cells. Comparable amounts of yeast cells were harvested 24 h later, followed by Northern blotting to measure the level of TBSV repRNA produced. We used rRNA as a loading control for the normalization of data on repRNA accumulation in yeast. The accumulation level of repRNA in yeast carrying pYES plasmid, which expresses only a short peptide, was taken as 100% (Fig. 4). As an additional control, we overexpressed a pseudogene (APT2) which has no enzymatic activity when expressed (2) and failed to interact with p33 based on the protoarray experiments (not shown). Overexpression of Apt2p led to 79% ± 9% repRNA accumulation compared with that seen for yeast carrying the pYES control (Fig. 4). This suggests that protein overexpression in general might reduce the ability of yeast cells to support repRNA accumulation under the protein overexpression condition. Based on the pYES and APT2 controls, we considered the overexpression of a protein to be inhibitory if it significantly reduced repRNA accumulation below 70% and stimulatory if it increased repRNA accumulation significantly, i.e., above 130% (Fig. 4).

FIG. 4.

Effect of overexpression of selected host proteins interacting with p33 replication proteins on TBSV repRNA accumulation in yeast. A total of 44 zz domain-tagged yeast proteins (Open Biosystems) were expressed separately from the galactose-inducible GAL1 promoter 20 h prior to launching TBSV repRNA replication. The accumulation of repRNA was measured with Northern blotting using total RNA extracts from yeast 24 h after the addition of 50 μM copper sulfate, which induced TBSV repRNA replication. The accumulation levels of repRNA in yeast strains overexpressing the APT2 pseudogene and a short peptide from pYES were used as controls to calculate what values were not significantly different (shaded areas). Host factors whose overexpression significantly enhanced repRNA accumulation are boxed in black, whereas one host factor that decreased repRNA accumulation is boxed in gray. Each value represents data from 6 to 12 independent samples. Note that Cdc34p was expressed without the zz tag from a pYES vector.

These protein overexpression experiments revealed that two of the host proteins tested affected repRNA accumulation dramatically (Fig. 4). These were Hbs1p translation factor and Cdc34p E2 Ub-conjugating enzyme, which increased repRNA levels by ∼3.5- and ∼2.2-fold, respectively. An additional five host proteins, namely, the Jjj1p cochaperone, the Pol30p proliferating cell nuclear antigen transcription factor, the Arp8p actin-related protein, the Erb1p rRNA maturation protein, and the Trz1p tRNA-processing protein, increased repRNA accumulation by 40 to 60% compared to what was seen for yeast carrying the pYES expression vector (Fig. 4). In contrast, overexpression of Ddr48p, a DNA damage-responsive protein, led to a 63% ± 5% reduction in repRNA accumulation in yeast (Fig. 4), which is only slightly (but significantly) less than the inhibitory effect of the Apt2p control. Overexpression of the remaining 36 host proteins affected repRNA accumulation by less than 40% (Fig. 4). Overall, the above-described experiments demonstrated that ∼18% of the 44 host proteins tested could affect tombusvirus repRNA accumulation when overexpressed in yeast. We could not exclude the possibility that several of the overexpressed tagged proteins are not fully functional under the expression conditions.

Effect of overexpression of selected host proteins on TBSV RNA recombination in yeast.The effect on tombusvirus RNA recombination was tested for 44 of the identified host proteins that interacted with p33 by taking advantage of the previously developed tombusvirus recombination system in yeast (51) with some modifications. The recombination assay was similar to the above-described replication assay, except that it was based on a highly recombinogenic repRNA termed DI-AU-FP, which contains an AU-rich recombination hot spot (51, 54). In this assay, recombination takes place between two molecules of DI-AU-FP RNAs (50, 51). The accumulation of the recombinant RNAs (recRNAs) was estimated using Northern blotting (Fig. 5). Importantly, we calculated the ratio of recRNA to repRNA (DI-AU-FP), which is more informative about recombination than recRNA levels alone (50, 51). Among the 44 host proteins tested, 4 affected TBSV RNA recombination dramatically by enhancing the ratio of recRNA by ∼8- to 12-fold (Fig. 5). Overexpression of other host proteins tested had lesser effects on TBSV recRNA accumulation (not shown).

FIG. 5.

Effect of overexpression of selected host proteins interacting with p33 replication proteins on recombination by a TBSV repRNA in yeast. (A) Northern blotting of total RNA obtained from yeast overexpressing zz-tagged yeast proteins as shown. The host proteins were overexpressed separately from the galactose-inducible GAL1 promoter 20 h prior to launching the replication of DI-AU TBSV repRNA, a highly recombinogenic repRNA (7). Then, the yeast culture was grown in the presence of 50 μM copper sulfate at 29°C for 24 h, followed by a 1:10 dilution into the same medium. The recombinant repRNAs were measured in samples obtained from yeast cultures 48 h after the dilution step. Note that the ratio of recRNA to repRNA was calculated, since that reflects the frequency of recombination more accurately than does the amount of recRNA, which is dependent on both the frequency of recombination and the amount of the template (original parental repRNA). The ratio of recRNA to repRNA in the control yeast expressing a short peptide was set to 100%. Each experiment was performed six to eight times. (B) Western blotting showing the overexpressed zz-tagged host proteins (marked with arrows) and the tombusvirus p33 and p92pol (marked with asterisks) replication proteins.

Among the identified proteins, Gsy2p glycogen synthase enhanced the recRNA ratio by 12-fold, the most among the 44 host proteins tested (Fig. 5A). Overexpression of the Arp8p actin-related protein, the Tif11p translation initiation factor eIF1A, and Iwr1p (unknown function) increased the ratio of recRNA to repRNA by 8- to 9.5-fold (Fig. 5A). Because a high amount of p92pol has been shown to increase TBSV recombination (16), we also tested p33 and p92pol levels in yeast overexpressing host proteins. The Western analysis showed no significant increase in p92pol levels in comparison with p33 in the Gsy2p, Arp8p, Tif11p, and Iwr1p overexpression strains (Fig. 5B), suggesting that these host proteins do not affect RNA recombination through changing the p33: p92pol ratio but via a different, yet-uncharacterized mechanism(s).

Cdc34p is present within the purified tombusvirus replicase complex.To further test the functional relevance of the identified host proteins interacting with p33, we selected Cdc34p based on its intriguing function as an E2 Ub-conjugating enzyme. Moreover, Cdc34p might be present within the tombusvirus replicase complex, since Cdc34p matches well with an unidentified yeast protein (termed X factor) with a pI value of ∼4 and a molecular mass of ∼35 kDa that has been detected in the highly purified tombusvirus replicase complex via two-dimensional gel electrophoresis (49). On the other hand, the other intriguing host protein, Hbs1p, is far less characterized and its function will be studied in the future.

To test if Cdc34p is present within the highly purified tombusvirus replicase complex, we performed Western blotting with Cdc34p-specific antibody by use of a purified preparation of the detergent-solubilized tombusvirus replicase obtained from yeast grown at 23°C (optimal for TBSV replication) or 29°C (optimal for protein expression) (Fig. 6). The Flag affinity-purified tombusvirus replicase containing Flag-tagged p33 and p92pol replication proteins (Fig. 6A, lanes 1 and 3) contained the native Cdc34p expressed from its natural chromosomal position (Fig. 6B, lanes 1 and 3). Copurification of Cdc34p with the Flag-tagged replication proteins indicated that Cdc34p is a component of the membrane-bound replicase complex. The preparation was RNase treated prior to loading to the affinity column, suggesting that Cdc34p was likely retained on the Flag column via its binding to the replication proteins and not via the viral repRNA bound to the replication proteins. The control Flag-purified preparations from yeast expressing His6-tagged p33 and p92pol (Fig. 6A, lanes 2 and 4) lacked these replication proteins as well as Cdc34p (Fig. 6B, lanes 2 and 4), excluding the possibility that Cdc34p is a contaminating protein bound nonspecifically to the beads. Therefore, we propose that the previously detected X factor in the tombusvirus replicase complex (49) is Cdc34p, based on its copurification with the viral replicase (Fig. 6) and its physical properties (pI of 3.96 and molecular mass of 34 kDa).

FIG. 6.

Copurification of Cdc34p with the tombusvirus replicase complex. (A) FLAG- and His6-tagged p33 (p33HF) and p92pol (p92HF) replication proteins were purified after detergent-based solubilization of a membrane-enriched fraction of yeast on a FLAG affinity column. Western blotting with anti-His6 antibody detected the presence of p33 and p92pol in the purified replicase complex active in an in vitro replication assay (not shown), as demonstrated by Serva and Nagy (49). The control samples were from yeast expressing His6-tagged p33 (p33H) and p92pol (p92H) prepared as described above for p33HF and p92HF. Asterisks show p33 dimers that are partly resistant to denaturing conditions. (B) Western blotting of the same samples as in panel A with anti-Cdc34p antibody. Note that native Cdc34p was expressed from its original promoter and its original location on the chromosome. Occasionally, the Cdc34p formed a dimer when the yeast was grown at 29°C (marked with an asterisk in lane 3). (C) Western blotting analysis with anti-Cdc34p antibody showing similar amounts of Cdc34p present in the membrane-enriched fractions prior to affinity purification.

Downregulation of the Cdc34p level decreases TBSV repRNA accumulation in yeast.To test if downregulation of the Cdc34p level affects TBSV repRNA accumulation, we used doxycycline-regulatable expression of Cdc34p from its original chromosomal location (25). Because our preliminary experiments suggested that Cdc34p has a long half-life (not shown), we downregulated Cdc34p expression 6 h prior to launching repRNA replication. Indeed, the level of Cdc34p expression was ∼10 times lower in yeast grown in the presence of doxycycline than it was in its absence (Fig. 7B, bottom, lanes 3 and 4 versus 1 and 2). The accumulation of repRNA decreased ∼3-fold in yeast grown in the presence of doxycycline (Fig. 7B, top, lanes 4 to 6), whereas the amount of p33 replication protein did not change (Fig. 7B, bottom, lanes 1 to 4), demonstrating that Cdc34p is important for tombusvirus replication.

FIG. 7.

The effect of downregulation and overexpression of WT and mutant Cdc34p on TBSV repRNA accumulation in yeast. (A) Schematic representation of the known domains in Cdc34p. The active-site mutation that abolishes the E2 Ub conjugation activity of Cdc34p is shown with a black box. (B) Northern blotting of repRNA accumulation in yeast with downregulated expression of Cdc34p. The native Cdc34p was expressed from its original location but via the doxycycline (Dox)-regulatable Tet expression promoter. The yeast also expressed a short peptide from pYC2 (as a control), the WT Cdc34p, and the C95S active-site mutant of Cdc34p from low-copy-number plasmid pYC2. Yeast was grown in the absence or presence of 10 μg/ml doxycycline as indicated. Doxycycline was added 6 h prior to launching TBSV repRNA replication. The accumulation level of repRNA was measured 22 h after induction using ImageQuant software. rRNA was used as a loading control (see bottom panel). Western blotting results show the levels of Cdc34p and p33 for the above-described yeast samples. (C) Overexpression of Cdc34p increases repRNA accumulation in yeast. (Top) Northern blotting shows the accumulation of TBSV repRNA in yeast overexpressing of a short peptide from pYC2 or WT Cdc34p in BY4741; (bottom) Western blotting results show the levels of Cdc34p and p33 in the above-described yeast samples. Further details are as described for panel B. (D) Effect of overexpression of WT or mutated His6-tagged Cdc34p from high-copy-number plasmid pYES on repRNA accumulation in BY4741 yeast. The shown Cdc34p mutants are the E2 site mutant (C95S) and the C-terminally (N170; deletion of aa 171 to 295) and the N-terminally (C125, deletion of aa 1 to 170) truncated versions (Fig. 7A). (Bottom) Western blotting to show the overexpressed His6-tagged Cdc34p mutants. Further details are as described for panel B.

Downregulation of Cdc34p level decreases the activity of the tombusvirus replicase.To test if a decreased amount of Cdc34p could directly affect the activity of the tombusvirus replicase, we performed in vitro replicase assays with a membrane-enriched fraction derived from yeast. The membrane-enriched fraction contains the tombusvirus replicase and is capable of performing TBSV repRNA synthesis in vitro using the copurified repRNA as a template. To compare similar amounts of replicase complexes, we adjusted the p33 content in the membrane-enriched fractions obtained from yeast cultured with or without doxycycline (17, 38, 39). These in vitro experiments have shown that the tombusvirus replicase obtained from yeast with downregulated Cdc34p 22 h after launching TBSV replication was only 17% as active as the control preparation from yeast grown in the absence of doxycycline (Fig. 8C, lanes 3 and 4 versus 1 and 2). Similarly prepared replicase preparations at earlier time points after launching TBSV replication, such as 6 and 12 h, showed only 73% and 54% decreases (Fig. 8A and B), suggesting that Cdc34p was still more readily available during the formation of the replicase complex at the early time points than at the late time point. Altogether, the in vitro assay demonstrated that Cdc34p is critical for efficient tombusvirus replicase activity.

FIG. 8.

Decreased replicase activity in the presence of a low Cdc34p level. A replicase activity assay with membrane-enriched preparations obtained from yeast expressing high or low levels of Cdc34p, based on the addition of 10 μg/ml doxycycline (Dox) to the growth medium. Doxycycline was added 6 h prior to launching TBSV replication. The yeast samples were taken 6 h (A), 12 h (B), or 22 h (C) after the induction of TBSV replication. The membrane-enriched fraction contained the endogenous repRNA template that was used during the in vitro replicase assay in the presence of 32P-UTP and the other unlabeled ribonucleotide triphosphates. Panel C also shows the replicase assay with WT and C95S Cdc34p proteins. Note that the in vitro activities of the tombusvirus replicase were normalized based on p33 levels (to adjust for some differences in the p33 levels, as shown in the bottom panel).

The Ub conjugation function of Cdc34p is important for TBSV repRNA replication.Cdc34p has two functional domains: the 170-aa N-terminal UBC Ub conjugation domain conserved in E2s and the unique C-terminal acidic domain involved in protein/substrate binding (18). To test which function/domain of Cdc34p is important for tombusvirus replication, we used Cdc34p mutants in a complementation assay. We found that overexpression of Cdc34-C95S with inactive Ub-conjugating function from a plasmid could not complement TBSV repRNA accumulation in yeast with downregulated Cdc34p expression from the chromosome, whereas the full-length wild-type (WT) Cdc34p could complement repRNA accumulation (Fig. 7B, lanes 16 to 18 versus 10 to 12). Moreover, the N-terminal and C-terminal portions of Cdc34p rendered Cdc34p nonfunctional in a complementation assay (Fig. 7D, lanes 13 to 20), suggesting that both the UBC and the acidic domains of Cdc34p are important during TBSV replication.

In addition, we found a normal level of replicase activity obtained from yeast expressing the WT Cdc34p from a centromeric, low-copy-number plasmid, while Cdc34p was downregulated from the chromosomal location (Fig. 8C, lanes 7 and 8). Thus, the plasmid-borne WT Cdc34p can function in the viral replicase. On the other hand, the mutated Cdc34-C95S protein with inactive Ub-conjugating function expressed from a plasmid could not complement the downregulated Cdc34p from the chromosomal location in a replicase assay (Fig. 8C, lanes 11 and 12). These experiments suggest that the Ub conjugation function of Cdc34p is important for the function of the tombusvirus replicase.

Ubiquitination of p33 by Cdc34p in vitro.Since Cdc34p is part of the tombusvirus replicase complex, it also binds to p33 directly and the ubiquitination function of Cdc34p is important for the replicase function (Fig. 8); therefore, we wanted to test if Cdc34p could directly ubiquitinate p33 replication protein in the absence of Skp1p/cullin/F-box (SCF) E3 ligase complex, which is involved in substrate selection during normal cellular functions (48). To this end, we performed an in vitro ubiquitination assay with purified recombinant proteins. The in vitro assay demonstrated that Cdc34p could ubiquitinate a fraction of p33 replication protein, causing ∼9-and ∼18-kDa shifts in the mobility of p33 (Fig. 9A, lane 1). This shift in mobility is expected if a single Ub or two Ubs are added to p33 (mono- and biubiquitination). The higher-molecular-mass products detected on the gels could represent p33 proteins with multi- and/or polyubiquitination (Fig. 9A, lane 1). The ubiquitination of p33 by Cdc34p required Uba1p E1 protein, Ub, and ATP, and it did not take place if one of these components was missing in the assay (Fig. 9A, lanes 2 to 5). A similar assay with the purified MBP did not yield ubiquitination of MBP at a detectable level (Fig. 9B, lanes 6 and 7), suggesting that p33 is specifically ubiquitinated by Cdc34p. Moreover, the increase in the molecular mass of p33 was ∼35 kDa when we used GST-tagged Ub during a standard in vitro ubiquitination assay (Fig. 9C), confirming that Ub was added to p33 in the in vitro ubiquitination assay. A mutated Cdc34-C95S protein with inactive Ub-conjugating function (Fig. 9C), as well as the N- or C-terminally truncated Cdc34p variants (not shown), could not ubiquitinate p33, supporting the model that the ubiquitination of p33 was a specific feature of Cdc34p.

FIG. 9.

Ubiquitination of p33 by Cdc34p in vitro is independent of SCF E3 ligase complex. The in vitro ubiquitination was carried out in reaction mixtures containing 50 nM GST-Uba1p E1-activating enzyme (purified from yeast), 500 nM purified recombinant GST-Cdc34p (purified from E. coli), 100 μM His-Ub, 2 mM ATP, and purified recombinant MBP-p33 (A) or MBP as the negative control (B). The ubiquitination was analyzed by Western blotting using an anti-MBP antibody. The mono- and biubiquitinated p33s are marked with arrows. Note that monoubiquitination introduces an increase of ∼9 kDa to the molecular mass of p33. (C) The in vitro ubiquitination of p33 by Cdc34p was carried out as described above except that GST-tagged Ub (GST/Ub) was used, leading to a 35-kDa increase in the size of MBP-p33 (shown as p33) as well as higher-molecular-mass products indicating mono-, bi-, and multiubiquitination of p33. Note that we also tested an ubiquitination-deficient mutant of Cdc34p (C95S).

Ubiquitination of the p33 replication protein in yeast.To demonstrate the possible ubiquitination of p33, we FLAG affinity purified His6/FLAG-tagged p33HF from yeast coexpressing Ub tagged with c-Myc from a plasmid. In these experiments, p33HF was solubilized from the membrane fraction, followed by purification and Western blotting to detect the addition of c-Myc-Ub to p33HF via anti-c-Myc antibody. Interestingly, we found mono- and biubiquitinated p33HF based on detection by anti-c-Myc antibody as well as a shift in the molecular mass of p33HF (mono- and biubiquitination cause ∼8- and 16-kDa increases, respectively) (Fig. 10, lane 2). The identified bands likely represent ubiquitinated p33HF because similar experiments with p33 tagged with FLAG only (termed p33F, which is about 2 kDa smaller than p33HF) resulted in slightly faster-migrating mono- and biubiquitinated p33F (Fig. 10, lane 1). The change in the migration pattern of the purified p33HF versus p33F proteins supports the idea that these bands represent various ubiquitinated p33 and excludes the possibility that they represent ubiquitinated contaminating host proteins in our FLAG affinity-purified samples. Moreover, the control sample containing His6

One thought on “Protoarray Analysis Essay

Leave a Reply

Your email address will not be published. Required fields are marked *