Identification of Epstein-Barr virus proteins as putative targets of the immune response in multiple sclerosis (original) (raw)
Dissecting the antibody repertoire in MS patients by a human cDNA protein-expression array. To investigate the antibody specificity of IgG antibodies from the CSF of MS patients, we applied a novel protein array. The array was generated from a human brain cDNA expression library comprising 37,000 expression clones. CSF samples from 12 MS patients and 5 controls were adjusted to 1 mg IgG/l and each applied to a separate protein array. Immunoreactivity was visualized by HRP-conjugated anti-IgG antibodies. From 0 to 10 expression clones that specifically stained above background were identified in each patient (Figure 1A). After comparing the staining pattern between MS patients and controls, we selected expression clones that showed strong reactivity in MS patients but not in controls.
Analysis of CSF IgG immunoreactivity in MS patients by protein expression arrays. (A) Incubation of the protein expression array with CSF from a representative MS patient (left) and a control donor (right). A 3 cm × 3 cm section of the 24 cm × 24 cm array is shown. IgG immunoreacitivity of the MS CSF to the expression clone B3 (spotted in duplicate) is marked by a circle. IgG concentration was adjusted to 1 mg/l IgG in MS and control CSF. (B) Western blot with purified protein B3. Immunoractivity was observed in the CSF of a representative MS patient (left) but not the NIND (middle) or OIND (right) control donors. All CSF samples were adjusted to 10 mg/l IgG. IgG binding was developed with ECL. M, molecular weight marker. (C) Analysis of immunoreactivity to B3 protein (left) and control protein GAPDH (right) with CSF (1:5 diluted) of 132 MS patients and 125 NIND patients by ELISA. Antibody titers were significantly higher in MS patients. Dot points represent the OD of a single CSF sample. Cut-off point, OD > 0.3 (mean ± 6 SEM). (D) CSF from a patient with high immunoreacitivty to patterns I and II proteins was separated by IEF and blotted against proteins B3, C6, C5, F4, and G4 as well as the control protein GAPDH. Binding of proteins from both patterns was similar with little overlap between the patterns. Similar results were obtained with CSF from another patient (data not shown). (E) IEF immunoblot with CSF (C) and serum (S) from 3 MS patients (adjusted to 10 mg IgG/l) was performed to compare IgG binding patterns to proteins B3, C5, and G4. A stronger and more focused immune response to the proteins was observed in the CSF of 10 patients analyzed in total. *P < 0.001, Fisher’s exact test.
Identification and characterization of primary immunoreactivity in MS. Among the arrayed expression clones, a total of 54 clones were selected. cDNA inserts were sequenced and the corresponding AA sequences determined in all of them. Forty-two unique sequences were defined; these comprised proteins expressed in the correct reading frame but also sequences which were expressed out of frame generating artificial protein sequences. The 21 expression clones that showed the strongest reactivity in 2 or more MS patients were selected for further analyses (Table 1). All of these proteins, irrespective of whether they were expressed in the correct frame (10 clones) or not (11 clones), and the control protein GAPDH were purified for further analyses. The size of all expressed proteins was verified by SDS-PAGE. Specific binding of CSF IgG from MS patients was confirmed by Western blot analysis (Figure 1B). ELISAs were established for all proteins. In initial experiments, a group of 46 MS patients and 28 controls were analyzed for CSF reactivity to the proteins. If a higher immunoreactivity was observed in the MS group by ELISA, additional samples from MS patients and controls were evaluated. For certain proteins, a significant difference with regard to antibody reactivity was observed between MS patients and controls. While more than 13% of the 132 MS patients had CSF antibodies against the expression product of clone B3, no such antibody responses were detected in any of the controls. In contrast, MS patients and controls showed similar immunoreactivity to the control protein GAPDH (Figure 1C). A summary of immunoreactivities to the 21 proteins and GAPDH in patients and controls as determined in the initial ELISA experiments is provided in Table 1.
Summary of immunreactivities to expression clones identified by protein array
Next, the immunoreactivity to the selected 21 proteins in individual MS patients was compared. Interestingly, immunoreactivities to certain sets of distinct proteins occurred consistently in different patients. Patients with IgG reactivity against protein B3 usually also had antibodies against C5, C6, D6, and F3 (pattern I; Table 1), suggesting the presence of a similar epitope that was targeted by the IgG response of the patients. All 5 proteins of this pattern represent artificial products that were expressed out of frame. Three other expression clones (G4, F4, and H5), which contained different cDNA fragments of Myc-associated zinc finger protein (MAZ), also showed high and largely overlapping immunoreactivities (pattern II; Table 1).
To demonstrate that 1 single epitope was responsible for the immunoreactivity observed within the 2 groups, we used isoelectric focusing (IEF) to separate CSF IgG antibodies to OCBs from patients with immunoreactivity to proteins of the 2 patterns. After separation, IgGs were blotted onto membranes coated with the different proteins of patterns I and II as well as the control protein GAPDH (Figure 1D). We found that proteins in each group always bound the same OCBs. In contrast, little overlap was observed between the immune responses to pattern I and II proteins. As expected, no binding to GAPDH was observed.
Finally, we compared the extent of antibody reactivity against selected proteins of each pattern by IEF and immunoblot using the serum and CSF of MS patients, adjusted to the same IgG concentration. We found a qualitatively or quantitatively enhanced CSF antibody reactivity against proteins from both patterns in the majority of patients (Figure 1E), which confirmed the existence of an intrathecal response to these antigens in MS.
Identification of 2 EBV proteins as targets of the CSF IgG response in MS. Epitopes responsible for the immunoreactivity to patterns I and II were mapped by peptide-scan analysis. Short 13-mer peptides with 11-mer overlappings covering the entire sequence of the smallest proteins from each pattern were synthesized onto membranes. CSF from patients with immunoreactivity to these proteins was used to screen for binding to the peptides. We identified IgG binding to the same 2 linear 8-AA minimal epitopes in 3 patients with CSF antibodies against pattern I and pattern II (Figure 2A). Then we performed a fine mapping of both identified peptide sequences by substitution analysis in which 20 AAs were substituted for each position of the 8-mer minimal epitope sequence. Again, CSF from patients with reactivity to the identified proteins was applied to define the best binding AA at each position (shown for pattern I in Figure 2B). In all patients studied, we noted highly similar binding motifs of CSF IgG antibodies. The motifs were identified in the proteins in patterns I and II. However, a complete match was not observed. Therefore, the Swiss-Prot database (http://au.expasy.org/sprot/) was searched for full matches using the identified motifs. We found 10 matches with the first and 13 with the second motif. Among the matches were some irrelevant proteins, as they were from nonpathogenic organisms, but also epitopes derived from human, bacterial, or viral proteins (Figure 2, C and D). We determined binding of CSF IgG to 13-mer peptides derived from proteins that were found by database search. CSF antibody staining to all peptides was observed but was strongest to the 2 peptides, Epstein-Barr nuclear antigen-1 (EBNA-1) and BRRF2, derived from EBV, as determined in 3 patients (Figure 2D). Interestingly, the genes of both EBV proteins are located adjacently in the EBV genome. One, EBNA-1, is known to be expressed in the latent phase of EBV infection. The function of the other, BRRF2, has not been characterized. Recently, this protein was shown to be present in the EBV virion as a putative part of the tegument (26). To ensure that the gene is expressed, we studied transcript expression in latently EBV-infected and EBV-transformed cell lines (Figure 3A). We confirmed transcript expression of both partial and full-length BRRF2 RNA, demonstrating that the genes are transcribed in latently EBV-infected and EBV-transformed cell lines (Figure 3B).
Identification of the CSF IgG-binding epitope. (A) Peptide scan analysis with 13-mer peptides that overlapped 11 AAs, covering the entire sequence of protein B3 of pattern I (upper membrane) and protein H5 of pattern II (lower membrane), was used to define the epitopes. Membranes were incubated with CSF (in 1:100 dilution) from MS patients immunoreactive to B3 or H5. Binding of IgG was visualized by anti-human IgG-HRP and TMB substrate. The minimal peptide epitopes were EPARSRSR for motif 1 and EAGAGGGA for motif 2. Similar results were obtained with CSF from 2 additional patients. (B) Substitution analysis was performed in order to define the optimal binding motif for the 8 AA epitopes defined in A. Binding of IgG was visualized by anti-human IgG-HRP and TMB substrate. A representative example for pattern I is shown. 1–8, the substituted AA-positions of the minimal epitope; 1–20, the 20 naturally occurring acids A–Y; *original peptide sequence. Similar results were obtained with 2 additional CSF samples from MS patients. (C) Definitions of 2 consensus motifs were based on the epitope mapping in 3 MS patients. These motifs were used to search the Swiss-Prot database. Database searching revealed 10 proteins matching with motif 1 and 13 proteins with motif 2. Two identified EBV proteins and the genomic locations according to http://www.ncbi.nlm.nih.gov are displayed. (D) Qualitative comparison of CSF IgG binding to peptides matching motif 1 (left) and motif 2 (right). Antibody binding was quantified by gel densitometry (highest signal and integrated density), which revealed the strongest binding to the 2 EBV epitopes BRRF2 and EBNA-1. The analysis was performed with similar results in 2 additional patients.
Expression and immunoreactivity to EBNA-1 and BRRF2. (A) BRRF2 and EBNA-1 RNA transcripts were detected in the B95 cell line (B95) and the EBV-transformed B cell line (BC) by BRRF2- (61 bp) and EBNA-1–specific (107 bp) RT-PCR. To exclude contamination with residual DNA, a control sample without reverse transcription (no RT) was included in the experiment. (B). Expression of full-length and partial BRRF2 RNA in the B95 cell line verified by RT-PCR. (C) BRRF2 expression in E. coli. The BRRF2 protein was cloned as partial (16 kDa) and full-length (58 kDa) protein with a 30-kDa GST tag resulting in 46-kDa partial and 88-kDa full-length band on SDS-PAGE (SDS) after Coomassie staining (left). Western blot (WB) and immunostaining with CSF of a control donor (Ctr; middle) and of a representative MS patient (right) confirmed the specific binding of CSF IgG to the BRRF2 proteins. (D) Immunoreactivity to recombinant BRRF2 (upper panels) and EBNA-1 (lower panels) was investigated by ELISA in CSF (left, 1:5 dilution) and serum (right, 1:100 dilution) of 130 MS patients compared with 115 NIND and 85 OIND patients. The immunoreactivities for each sample are given as OD values. Mean ODs and P values comparing the extent of immunoreactivity by Student’s t test are displayed above each group. Fisher’s exact test was applied to compare the frequencies of reactive patients: *Significant compared with NIND patients; #significant compared with OIND patients. RT, reverse transcription.
Increased antibody responses to EBV, EBNA-1, and BRRF2 in MS patients. EBV peptides comprising the dominant epitope were synthesized and the CSF IgG response determined in a large group of MS patients and controls. We found a significant difference in the response to EBNA-1 and BRRF2 peptides between MS patients and controls (data not shown). Therefore, protein fragments of EBNA-1 (AAs 302–641) and BRRF2 containing the peptide sequence of interest were used to investigate the antibody response. For BRRF2, we established a recombinant expression as full (AAs 1–537) and partial (AAs 385–537) protein. Western blotting confirmed specific binding of CSF IgG to the recombinant BRRF2 proteins (Figure 3C). ELISAs with the recombinant proteins confirmed that MS patients had higher levels of IgG reactivity to EBNA-1 in CSF and serum compared with control donors who had other inflammatory neurological diseases (OINDs) or noninflammatory neurological diseases (NINDs). A significant difference was also observed for BRRF2 in serum. In CSF, the difference was only statistically significant between patients with MS and those with NINDs but not between patients with MS and OIND patients (Figure 3D). To determine whether antibodies to BRRF2 are intrathecally produced, CSF and sera of MS and OIND patients were adjusted to the same IgG concentration, and BRRF2 antibody titers were determined. In most MS patients, higher titers of antibodies against BRRF2 were detected in CSF than in serum, indicating an intrathecal antibody response to the protein (Figure 4A). In contrast, only 1 OIND patient had an intrathecal IgG response to BRRF2. The number of patients with intrathecal synthesis was significantly higher in MS patients than in controls (P = 0.0054).
Specific intrathecal IgG response to EBV proteins. (A) Intrathecal IgG response to BRRF2 proteins in MS patients and control donors detected by ELISA. CSF and serum were adjusted to 10 mg/l IgG and the ratio of OD CSF/OD serum determined. Ratios above 1.2 indicate intrathecal synthesis. MS patients showed intrathecal BRRF2-specific IgG synthesis more frequently than controls. (B) IEF-immunoblot demonstrating specific binding of CSF oligoclonal IgG bands from 2 MS patients to the EBV proteins. The membranes were coated with BRRF2, EBNA-1, or a solution containing 10% milk alone as indicated. Affinity-blotted IgG was detected with anti-human IgG-HRP and visualized by TMB substrate. (C) IEF-immunoblot for BRRF2-specific OCBs (right) and total OCB pattern (left). Detection of bound IgG was performed as described in B. BRRF2-specific OCBs correspond to some of the major bands in the OCB pattern of the MS patient. (D) Loss of OCBs by preabsorption with EBNA-1 but not GAPDH. 2D-electrophoresis and IgG immunoblot of CSF from a patient with EBNA-1 immunoreactivity. No, no preabsorption of CSF-IgG was performed with EBNA-1 or GAPDH. (E) Solution phase assays demonstrate high affinity and specificity of CSF antibodies to EBNA-1 (left) and BRRF2 (right). Soluble EBV proteins at different dilutions were incubated with CSF and, subsequently, the remaining immunoreactivity measured by ELISA. Competition by soluble antigens is displayed. No competition was measured with GAPDH protein. For each assay, 4 MS patients were analyzed. *Fisher's exact probability test.
Finally, we investigated how antibody titers relate to other clinical and laboratory parameters. Besides a weak correlation of EBNA-1 serum antibody titers with age in MS patients (P = 0.0261, r = –0.1959; Pearson test without Bonfferoni adjustment, n = 129), no other correlations between BRRF2 and EBNA-1 serum antibody levels and clinical or other laboratory parameters were observed. To rule out the possibility of nonspecific EBV reactivation following CNS damage and inflammation, we determined the EBNA-1 antibody levels in 9 stroke patients. No change in EBNA-1 antibody titers was observed 4 weeks (n = 9) and 8 weeks (n = 3) after onset of stroke (data not shown).
EBNA-1 and BRRF2 bind oligoclonal IgG from the CSF of MS patients. To further investigate the role of the humoral immune response against EBV in MS patients, we needed to clarify whether OCBs in CSF bind specifically to these proteins. For this purpose, IgG antibodies in CSF were separated by IEF and blotted onto membranes precoated with the 2 EBV proteins. We found oligoclonal IgG patterns binding to both proteins in MS patients who were antibody positive in ELISA experiments (Figure 4B). The OCB reactivity of BRRF2 and EBNA-1 corresponded to the binding pattern of the initial antigens found on the protein array (compare left panels of Figures 1D and 4B). Furthermore, we demonstrated that oligoclonal IgG antibodies binding BRRF2 and EBNA-1 correspond to the OCBs that are observed by regular IEF immunoblot of CSF IgG (shown for 1 patient with BRRF2 in Figure 4C). In 3 patients with high EBNA-1 antibody titers in CSF, part of the oligoclonal bands were absorbed by preincubation of CSF with EBNA-1 but not GAPDH (Figure 4D). Finally, we confirmed the specificity and high affinity of EBV-specific CSF antibodies by solution phase assays, in which EBNA-1 and BRRF2 reactivity could be competed with the EBV proteins in soluble phase (Figure 4E).
Increased T cell responses to latent EBV protein in MS patients. To determine the extent of T cell responses to latent EBV proteins, we transformed B cells from MS and control donors with EBV. The transformed cell lines expressed latent EBV proteins, including EBNA-1 and BRRF2 (Figure 3A), in the context of the autologous HLA molecules. These cell lines, which process the endogenous proteins into the HLA class I pathway, were used to determine the CD8+ T cell response to EBV proteins expressed in infected and transformed human B cells. Peripheral blood mononuclear cells of each patient were incubated with autologous EBV-transformed B cell lines, and the number of EBV-specific T cells in the CD4, CD8, and CD28 T cell subsets was determined by intracellular cytokine staining (Figure 5). We found a higher number of EBV-specific CD8+ T cells in MS patients than in controls. In particular, a significant difference (P = 0.005) was found for the CD8+CD28+ T cells. In contrast, no difference was observed for CD8+CD28– and CD4+ T cells.
Frequency and phenotype of EBV-specific T cells in MS patients and healthy donors. (A) Strategy for detection of EBV-specific T cells in PBMC samples by intracellular IFN-γ staining and flow cytometry. PBMCs and autologous EBV-transformed B cell lines were either cultured separately (nonspecific activation, left graph) or in short-term coculture (EBV-specific activation, right graph) before staining. Shown are analyses of CD8+ T cells in the CD28+ and CD28– compartments of an MS patient. One percent of CD8+CD28+ T cells were EBV-specific in this patient. (B) Frequency of EBV-specific CD4+ and CD8+ T cells in 11 MS patients and 14 healthy donors (HD). No significant differences were observed in the CD4+ T cell compartments, whereas a higher frequency of EBV-specific CD8+ T cells was observed in MS patients. Further characterization revealed a significantly higher frequency of EBV-specific CD8+CD28+ in MS patients compared with healthy controls. Student’s t test was applied.