A cardinal feature of multiple sclerosis (MS) is the persistent intrathecal synthesis of antibodies. Our previous finding that a large fraction of B cells infiltrating the MS brain are infected with Epstein-Barr virus (EBV) raises the possibility that this virus, because of its ability to establish a latent infection in B cells and interfere with their differentiation, contributes to B-cell dysregulation in MS. The aim of this study was to gain further insight into EBV latency programs and their relationship to B-cell activation in the MS brain. Immunohistochemical analysis of postmortem MS brain samples harboring large EBV deposits revealed that most B cells in white matter lesions, meninges, and ectopic B-cell follicles are CD27+ antigen-experienced cells and coexpress latent membrane protein 1and latent membrane protein 2A, 2 EBV-encoded proteins that provide survival and maturation signals to B cells. By combining laser-capture microdissection with preamplification reverse transcription-polymerase chain reaction techniques, EBV latency transcripts (latent membrane protein 2A, EBV nuclear antigen 1) were detected in all MS brain samples analyzed. We also found that B cell-activating factor of the tumor necrosis factor family is expressed in EBV-infected B cells in acute MS lesions and ectopic B-cell follicles. These findings support a role for EBV infection in B-cell activation in the MS brain and suggest that B cell-activating factor of the tumor necrosis factor family produced by EBV-infected B cells may contribute to this process resulting in viral persistence and, possibly, disruption of B-cell tolerance.
Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system (CNS) resulting from a complex interplay between risk-conferring genes and environmental factors, most likely infectious agents (1-3). Among the latter, Epstein-Barr virus (EBV), a B-lymphotropic DNA herpesvirus that infects approximately 95% of the general population before adulthood, shows a strong association with MS mainly on the basis of epidemiological and serological evidence (4, 5). A role for EBV in the development of MS is supported by EBV seropositivity and seroconversion before disease onset in all adult MS patients (4, 6) and 83% to 99% of children with MS (7-9) and by an increased risk of developing MS in individuals with elevated anti-EBV antibodies (mainly immunoglobulin G [IgG] to EBV nuclear antigen 1 [EBNA-1]) and a previous history of symptomatic EBV infection (infectious mononucleosis) (4). There are increased levels of EBNA-1 IgG and higher EBV-specific T-cell responses in MS patients compared with healthy EBV carriers (4, 10-12), and the presence of oligoclonal IgG binding to EBV antigens in the cerebrospinal fluid of some MS patients (10, 13-15) suggests a persistently perturbed EBV infection. It has also been argued that EBV dysregulation in MS patients might result from defects in EBV-specific T-cell immunity (16). The recent finding of a correlation between elevated EBNA-1 IgG serum titers and gadolinium-enhancing lesions on magnetic resonance imaging in MS patients indicates an association between EBV infection and disease activity (17).
The possible mechanistic links between EBV infection and MS are unclear. Some studies suggest that EBV may break immune tolerance to CNS antigens through molecular mimicry (18, 19). Bystander immune activation, immortalization of B-cell clones producing myelin autoantibodies, and detrimental effects of the antiviral immune response have been proposed as alternative, yet to be investigated, mechanisms (20). The possibility that EBV might establish a persistent infection in the CNS and reactivate periodically bolstering an immunopathologic response has been explored by several groups with conflicting results. Using in situ hybridization for EBV-encoded small RNA (EBER) and immunohistochemistry (IHC) for EBV latent (EBNA-2, latent membrane protein 1 [LMP-1]) and lytic (BFRF1, gp350/220, p160) proteins, we detected B cells infected with EBV in postmortem brain tissue from 21 of 22 MS patients with different courses (i.e. acute, relapsing-remitting, primary progressive, and secondary progressive), disease durations, and degrees of CNS inflammation, but not from other inflammatory neurological diseases (15). Of note, the frequency of EBV-infected cells in different MS brain samples correlated with the extent of immune infiltration and numbers of CD20+ B cells (15). In a subset of patients with secondary progressive MS, meningeal B-cell follicle-like structures were identified as main intracerebral niches for EBV persistence and reactivation. These findings, together with the observation that CD8+ T cells accumulating in the MS brain show signs of cytotoxicity toward EBV-infected plasma cells (15), support the possibility that an immunopathologic response targeting abnormal intracerebral EBV deposits could be a major cause of tissue injury in MS. In contrast with our data, however, several studies have failed to detect EBV infection in all or most of the MS brain samples analyzed using in situ hybridization for EBER (21-24), IHC (23-25), and polymerase chain reaction (PCR) techniques (23-25). Such a disparity of findings calls for further investigations to assess the specificity and sensitivity of the methods used to detect EBV infection in the MS brain. Elucidation of this issue is critical for understanding the elusive link between increased immune reactivity to EBV and the pathogenesis of MS lesions (5).
Because EBV almost exclusively infects B cells and uses the B-cell differentiation program to establish a persistent, usually asymptomatic, infection in humans (26), 1 reason to search for EBV in the MS brain is to determine whether intracerebral B-cell abnormalities in MS might be the direct consequence of a locally dysregulated EBV infection (27). Intrathecal synthesis of Ig, mainly oligoclonal IgG with largely unknown specificity, is found in more than 90% of MS patients (28, 29). Polyspecific antibodies reacting with several pathogens are also synthesized intrathecally in 80% to 100% of MS patients and are thought to represent a bystander reaction of the chronic inflammatory process (30-35). Such a heterogeneous response in MS contrasts with findings in infectious diseases of the CNS in which most antibodies produced intrathecally bind with high affinity to the causative agent (33, 36-38).
The EBV latency is maintained by expression of a set of viral proteins that deliver activation, growth, and survival signals to infected B cells (39, 40). Among latent EBV proteins, LMP-1 and LMP-2A exert a positive effect on B-cell activation and survival. Latent membrane protein 1 mimics B-cell activating signals delivered through activation of CD40 (41, 42), whereas LMP-2A constitutively activates proteins involved in the B-cell receptor signal transduction cascade (43-45). When LMP-2A is expressed in B cells, it bypasses B-cell tolerance checkpoints, leading to the production of high levels of autoantibodies and the development of lupus-like disease in transgenic mouse models (46, 47). In addition, LMP-1 and LMP-2A upregulate the expression of 2 B-cell growth factors, a proliferation-inducing ligand and B cell-activating factor of the tumor necrosis factor family ([BAFF] also known as B-lymphocyte stimulator) in B-cell lines (48). The BAFF is essential for maintenance of the peripheral B-cell repertoire and initiation of T cell-independent B-cell responses (49) and has been implicated in the development of autoimmunity in experimental settings and in several human B cell-related autoimmune diseases (50). Increased BAFF expression was found in the brain of patients with MS, suggesting a role for BAFF in intrathecal B-cell dysregulation (51).
The aim of this study was to provide further evidence of persistent EBV infection in the MS brain and gain insights into EBV latency programs and their relationships to intracerebral B-cell abnormalities. Using IHC and postmortem brain tissue from MS cases with abundant B-cell infiltrates, we analyzed the distribution and localization of LMP-2A, an EBV-encoded protein that was not investigated in previous studies from our (15) and other groups (21-25). We also used laser-capture microdissection in combination with very sensitive real-time reverse transcription PCR (RT-PCR) techniques to analyze the expression of EBV latency genes in intracerebral B-cell reservoirs. We then asked whether CNS-infiltrating B cells produce BAFF, which could act in an autocrine and paracrine manner to favor B-cell activation and EBV persistence in the MS brain.
Materials and Methods
The EBV-positive human Burkitt lymphoma-derived cell lines Raji and P3HR-1 were used as positive controls for LMP-2A immunostaining and RT-PCR for viral transcripts, respectively. The EBV-negative Burkitt lymphoma-derived cell lines Ramos and BJAB were used as negative controls. The cell lines were kindly provided by Dr Pankaj Trivedi, University of Rome “La Sapienza,” Rome, Italy. The EBV-immortalized lymphoblastoid cell line GCi, with 8% of the cell population showing spontaneous lytic replication (52), was kindly provided by Dr Henry-Jacques Delecluse, German Cancer Research Center, Heidelberg, Germany. Cell lines were cultured at 37°C in 5% CO2 in RPMI 1640 supplemented with 10% fetal calf serum.
Patients and Tissues
The study was performed on postmortem brain tissue from 9 cases with secondary progressive MS and 3 control subjects who died of cardiac failure, ovarian cancer, and aspiration pneumonia, respectively, without evidence of neuropathologic alterations. Tissues were provided by the UK Multiple Sclerosis Tissue Bank at Imperial College London (confirmation of MS diagnosis provided by Drs F. Roncaroli and R. Nicholas) after collection via a prospective donor program with fully informed consent. Based on the available clinical documentation, all patients with secondary progressive MS were in the progressive phase of the disease, and no treatment is reported in the 6 months before death. This study was approved by the ethics committee of the Istituto Superiore di Sanità.
A total of 13 brain tissue blocks (4 cm3 each; 10 snap frozen, 3 fixed frozen) from 9 MS cases and 2 blocks (1 snap frozen, 1 fixed frozen) of cerebral tissue from 2 control subjects were used for IHC. Lesion inflammatory activity was assessed as previously described (53). All tissue blocks examined contained at least 1 chronic active white matter (WM) lesion (n = 22). Tissue blocks from 2 cases each contained an active lesion. Ectopic B-cell follicles (n = 12) were detected in the inflamed meninges of 8 of the 9 MS cases examined. Snap-frozen brain tissue blocks from 7 MS cases were examined for EBV transcripts by nested and/or quantitative real-time RT-PCR.
Paraffin-embedded brain sections of an EBV-associated diffuse large B-cell lymphoma, used as positive control for LMP-1 and LMP-2A IHC, and of other inflammatory neurological diseases (3 tuberculous and 1 mycotic meningoencephalitis, 1 luetic meningitis) were obtained from Dr Romana Höftberger, Clinical Institute of Neurology, Vienna, Austria. Paraffin-embedded brain sections of 1 case with herpes simplex virus 1 encephalitis, 1 with mycotic meningitis, and 1 with primary cerebral vasculitis, and 1 snap-frozen control lymph node were obtained from Dr Egidio Stigliano, Policlinico A. Gemelli, Rome, Italy. Clinical and autopsy data of MS and control cases are shown in Table, Supplemental Digital Content 1, .
For MS and control cases, we used both snap-frozen brain tissue blocks and tissue blocks fixed in 4% paraformaldehyde (PFA) for several days, cryoprotected in sucrose, frozen by immersion in isopentane precooled on a bed of dry ice, and stored at −80°C until use. Ten-micrometer-thick cryostat sections were cut from the blocks, mounted on Super Frost plus slides, and stored at −80°C until use. For bright-field IHC, air-dried acetone- or methanol-fixed (only for myelin oligodendrocyte glycoprotein staining) cryostat sections were rehydrated with PBS, subjected to microwave in citrate buffer antigen unmasking procedure (only for PFA-fixed tissue blocks; no antigen retrieval was performed for myelin oligodendrocyte glycoprotein staining), and incubated for 20 minutes with 0.1% H2O2 in PBS to eliminate endogenous peroxidase activity. Sections were then preincubated with 10% normal sera and immunostained with the following monoclonal antibodies (mAbs): anti-myelin oligodendrocyte glycoprotein (AbD Serotec, Oxford, UK), anti-HLA-DR (1:50, clone CR3/43; Dako, Carpinteria, CA), anti-CD20 (clone L26 prediluted; ScyTeK Laboratories, Logan, UT), anti-LMP-2A (1:250, clone 4E11; Ascenion, Munich, Germany) against the N-terminal cytoplasmic tail of the LMP-2A protein (54) (snap-frozen tissues only), anti-LMP-1 (1:100, clone CS.1-4; Dako), and anti-BAFF (1:50, clone B1A3.2; Biogen Idec Inc, Cambridge, MA). All primary antibodies were diluted in PBS plus 2% bovine serum albumin and incubated overnight (ON) at 4°C or 1 hour at room temperature (RT) (anti-LMP-1, anti-LMP-2A). After extensive washing with PBS, sections were incubated with biotinylated secondary antibodies (rabbit anti-mouse or donkey anti-rat; Jackson Immunoresearch Laboratories, Cambridgeshire, UK) and avidin-biotin horseradish peroxidase complex using the ABC Vectastain Elite kit (Vector Laboratories, Burlingame, CA), according to the manufacturer's instructions. Staining reactions were performed with 3,3′-diaminobenzidine (Sigma, St Louis, MO) as chromogen. Negative controls included the use of IgG-isotype controls or omission of primary antibody. Sections were counterstained with hematoxylin and viewed under an Axiophot microscope (Carl Zeiss, Jena) equipped with a digital camera (Axiocam MRC); images were acquired using Axiovision 4 AC software.
For immunostaining on paraffin-embedded and fixed-frozen brain sections, anti-LMP-2A rat mAb (1:200, clone 15F9; AbD Serotec) was used. Antigen retrieval in citrate buffer 0.1 mmol/L in a microwave at medium power for 30 minutes was performed before incubation with anti-LMP-2A mAb. For LMP-1, antigen retrieval was performed in Dako retrieval solution (pH 9) for 20 minutes at 95°C.
Immunofluorescence and Confocal Microscopy
Cryosections of MS brain and control tissues were stained using double-immunofluorescence techniques, and images were acquired and analyzed using a laser scanning confocal microscope (LSM 5 Pascal, Carl Zeiss). Sections were air-dried and postfixed in 4% PFA for 5 minutes at RT. After an initial blockade with 10% normal serum in PBS, snap-frozen sections were incubated ON at 4°C with purified anti-LMP-2A rat mAb (clone 4E11) alone or in combination with anti-CD20 mouse mAb or anti-BAFF mouse mAb. Paraformaldehyde-fixed frozen sections were incubated ON with anti-CD79a rabbit polyclonal antibody (Abcam Inc, Cambridge, MA) and anti-CD27 mouse mAb (Novocastra Laboratories, Newcastle upon Tyne, UK), or after 20 minutes in microwave in citrate buffer, ON at 4°C with anti-LMP-1 or anti-BAFF mouse mAb and then for 90 minutes at RT with anti-LMP-2A rat mAb (clone 15F9). Antibody binding was visualized using a mixture of Alexa Fluor 488-conjugated goat anti-mouse IgG and biotinylated donkey anti-rat IgG followed by tetramethyl rhodamine isothiocyanate (TRITC)-conjugated streptavidin (for LMP-2A) or Cy3-conjugated donkey anti-rabbit (for CD79a) diluted in PBS (all from Jackson Immunoresearch Laboratories). For LMP-1/LMP-2A double immunostaining, a mixture of biotinylated rabbit anti-mouse and TRITC-conjugated goat anti-rat secondary antibodies was used followed by fluorescein isothiocyanate (FITC)-conjugated streptavidin (for LMP-1). Incubation with secondary antibodies was carried out for 1 hour at RT. For CD20/BAFF double immunostaining, 2 different procedures were used for snap-frozen and PFA-fixed sections, respectively. Snap-frozen sections were incubated ON at 4°C with anti-BAFF mAb, followed by biotinylated rabbit anti-mouse IgG for 1 hour at RT and then incubation with TRITC-conjugated streptavidin (Jackson Immunoresearch Laboratories) for 30 minutes. At the end of this procedure, sections were incubated for 2 hours with FITC-conjugated anti-CD20 mAb (1:5, clone BCA-B/20; Aczon, Bologna, Italy). Paraformaldehyde-fixed sections were incubated with a mixture of purified anti-CD20 and anti-BAFF mAbs ON at 4°C and then 1 hour with a mixture of Alexa Fluor 488-conjugated goat anti-mouse IgG1 (Invitrogen, Eugene, OR) for BAFF and Alexa Fluor 568-conjugated goat anti-mouse IgG2a (Invitrogen) for CD20. For double-immunofluorescence staining for BAFF and ionized calcium binding adapter molecule 1 (Iba-1), PFA-fixed sections were incubated with a mixture of anti-BAFF mAb (1:50) and anti-Iba-1 rabbit polyclonal antibody (1:300, Wako Pure Chemical Industries, Osaka, Japan) ON at 4°C and then with Alexa Fluor 488-conjugated goat anti-mouse IgG (Invitrogen) and TRITC-conjugated goat anti-rabbit IgG (Jackson Immunoresearch Laboratories) for 1 hour at RT. For double-immunofluorescence staining for BAFF and glial fibrillary acidic protein, PFA-fixed frozen sections were incubated ON with anti-glial fibrillary acidic protein rabbit polyclonal antibody (1:300; Dako) and anti-BAFF mouse mAb and then with Alexa Fluor 488-conjugated goat anti-mouse IgG and TRITC-conjugated goat anti-rabbit IgG for 1 hour at RT. Sections were then washed in PBS and sealed with Vectashield (Vector Laboratories) or ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen). Purified antibodies were diluted in PBS plus 2% bovine serum albumin. For negative controls, primary antibodies were replaced with preimmune serum or IgG-isotype control.
For immunofluorescence staining of cell lines, methanol-fixed cells were incubated for 1 hour at RT with anti-LMP-2A rat mAb (1:120, clone 4E11) and then with TRITC-conjugated goat anti-rat antibody for 50 minutes at RT. For double immunofluorescence for LMP-2A and CD20, cells were further incubated with FITC-conjugated anti-CD20 mouse mAb (Aczon) for 1 hour. Cells were then washed, sealed with ProLong Gold antifade reagent with DAPI, and analyzed with a digital epifluorescence microscope (Leica Microsystem, Wetzlar, Germany). Negative control stainings were performed using a mixture of rat IgG1 and FITC-conjugated mouse IgG2a followed by incubation with TRITC-conjugated goat anti-rat antibody.
In Situ Hybridization
In situ hybridization for EBER was performed on paraffin-embedded brain sections from 1 EBV-associated CNS lymphoma and 8 cases with other inflammatory neurological diseases using the Epstein-Barr virus (EBER) PNA Probe/Fluorescein and the PNA ISH detection kit (Dako), according to the manufacturer's instructions. Incubation with the probe was carried out for 2 hours or ON at 55°C with similar results. A sense probe was always included as negative control for in situ hybridization and gave no signal. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe was included as positive control. Sections were sealed in aqueous medium (Dako) and viewed and photographed with an Axiophot Zeiss microscope equipped with an Axiocam digital camera using the Axiovision 4 AC software.
Snap-frozen brain tissue blocks from 6 of the 9 MS cases listed in Table, Supplemental Digital Content 1, (MS79, MS92, MS121, MS176, MS180, MS342) and control lymph node were used for laser-capture microdissection and subsequent RNA analysis. For each tissue block, the integrity and quality of total RNA extracted with the SV Total RNA Isolation System (Promega, Madison, WI) were checked on ethidium bromide containing 1% agarose gels in Tris-borate/EDTA buffer.
To identify specific subregions, 10 to 20 serial brain sections for each MS case were mounted on membrane slides for laser-capture microdissection (MMI AG, Glattbrugg, Switzerland), air-dried for 20 seconds at RT, fixed in RNase-free 75%, 95%, and 100% ethanol (30 seconds for each step), dehydrated in xylene for 5 minutes, and air-dried for at least 5 minutes. Sections before and after these series were stained for CD20 to identify B cell-containing regions in the inflamed meninges and WM lesions. Some sections of each series were stained with HistoGene LCM staining kit (Arcturus Engineering, Mountain View, CA) to visualize the areas of interest. The staining procedure was performed between the fixation steps in 75% and 95% ethanol, following the manufacturer's instructions. Ten-micrometer-thick control lymph node sections were microdissected after histological staining to isolate B-cell follicles. Unstained and stained sections were kept at RT in a sealed box containing silica gel until cutting (usually <24 hours).
Using a laser microdissector SL Cut (MMI AG) equipped with a Nikon Eclipse TE2000-S microscope, we separately cut areas containing meningeal immune infiltrates and B-cell follicles, lesioned gray matter (GM), B cell-enriched perivascular cuffs in WM lesions, lesioned WM surrounding inflamed blood vessels, and normal-appearing WM (NAWM). The same brain areas were cut in 3 to 10 serial sections and pooled in a single cap. The mean values ± SD of pooled microdissected areas were 449.800 ± 236.200 μm2 for meningeal immune infiltrates and B-cell follicles, 1.074.800 ± 184.800 μm2 for lesioned GM, 233.760 ± 107.000 μm2 for intralesional perivascular cuffs, 730.188 ± 348.920 μm2 for lesioned WM and NAWM, and 570.000 ± 14.140 μm2 for lymph node follicles.
The isolated tissue fragments of each series were collected in 50 μL of lysis buffer (PicoPure RNA isolation kit, Arcturus Engineering), incubated at 42°C for 30 minutes and centrifuged at 800 × g for 2 minutes. Lysates were stored at −80°C until use.
Quantitative Real-Time RT-PCR
Total RNA was extracted from four 20-μm-thick snap-frozen brain sections for each MS (n = 4) and control (n = 2) case using RNeasy Mini Kit (Qiagen, Valencia, CA), according to the manufacturer's instructions. Total RNA was also extracted from microdissected samples obtained from MS brain and control lymph node sections, as previously described, using PicoPure RNA isolation kit (Arcturus Engineering) according to the manufacturer's instructions. DNA was removed using RNase-Free DNase set (Qiagen). One microgram of RNA isolated from brain sections, measured by Nanodrop (Thermo Fisher Scientific, Wilmington, DE), and all RNA obtained from microdissected areas were reverse transcribed with oligo (dT) and random hexamers using Murine Leukemia Virus Reverse Transcriptase (Invitrogen Life Technologies, Carlsbad, CA). The obtained cDNA was used to perform quantitative real-time PCR for GAPDH, CD19, LMP-2A and EBNA-1. In preliminary experiments, the EBV-positive P3HR-1 cell line and the EBV-negative Ramos cell line were used to test the specificity of primer pairs and set PCR conditions. Primer sequences and PCR conditions are listed in Table, Supplemental Digital Content 2, . To test the sensitivity of the quantitative PCR assay for detection of LMP-2A and EBNA-1, serial dilutions of the EBV-positive lymphoblastoid cell line GCi were prepared in a background of 5 × 10−5 EBV-negative BJAB cells.
To increase the sensitivity of the assay, PreAMP Master Mix Kit (Applied Biosystems, Foster City, CA) was used to enrich for cellular and viral gene transcripts. The cDNAs obtained from whole brain sections and microdissected brain and lymph node samples were preamplified according to the manufacturers' instructions using 90 nmol/L of each primer in a mix containing the same forward and reverse primers for GAPDH, CD19, LMP-2A, and EBNA-1 used for real-time RT-PCR. Quantitative PCR assays were performed in triplicate using 2 μL of cDNA template in a final volume of 20 μL containing the specific primer pairs, Light Cycler Fast Start DNA Master SYBR Green I (Roche Diagnostics, Mannheim, Germany) in the presence of 3 mmol/L MgCl2 on a LightCycler Instrument (Roche Diagnostics). Serial dilutions (4-log range) of PCR products derived from cDNA of EBV-positive P3HR-1 cells and human primary B cells in 10 μg/mL sonicated salmon sperm DNA were included in each run as positive controls for EBV (LMP-2A, EBNA-1) and CD19 gene expression, respectively, together with a negative control (positive control in the absence of RT enzyme). Sample values were normalized by calculating the relative quantity of each mRNA to that of GAPDH using the formula 2−ΔCt, where ΔCt represents the difference in cycle threshold (Ct) between target mRNA and GAPDH mRNA. The Ct values greater than 31 to 35 were not considered within the optimal range for good PCR efficiency, and samples with Ct values greater than 42 were considered negative. To verify the specificity of each amplicon, products of the PCR reactions were purified and sequenced (Eurofins MWG Operon Sequencing Service, Martinsried, Germany).
Total RNA was extracted from microdissected MS brain tissue samples as previously described. Reverse transcription was performed by using avian myeloblastosis virus reverse transcriptase and random hexamers (Promega). Latent membrane protein 2A primer sets and PCR conditions are listed in Table, Supplemental Digital Content 3, . Primers and conditions for PCR and nested PCR were as previously described (55). A second nested PCR reaction was carried out with inner primers for 30 cycles using 1 of 25 μL of the nested PCR product as DNA template; the forward inner-inner primer (56) was used in combination with the reverse inner primer used for the nested PCR (55). P3HR-1 and Ramos cell lines were used as positive and negative controls, respectively. From these cell lines, 3 μg of RNA was retrotranscribed. For the first PCR, the starting material was 1 μL of a 1:1000 diluted RT. The PCR products were visualized by electrophoresis in 3% ReadyAgarose wide-Mini Gel with ethidium bromide (BioRad). Images of the gels were acquired using the Fluor-S Multimager CCD camera. Products of PCR reactions were purified and sequenced to confirm the specificity of the reaction (Eurofins MWG Operon Sequencing Service). As internal loading control, PCR for GAPDH was carried out for 35 cycles using primers and conditions listed in Table, Supplemental Digital Content 3, . Negative controls lacking template RNA or RT were included in each experiment.
To be certain that the samples analyzed using IHC and RT-PCR techniques contained substantial numbers of EBV-infected B cells, we selected brain tissue blocks from 9 cases with secondary progressive MS that had been characterized for robust inflammatory activity in the meninges and the presence of immunologically active lesions in the WM and GM (15, 53).
Immunohistochemical Evidence of LMP-2A Protein Expression in MS Lesions and Meninges
In all MS brain specimens analyzed, LMP-2A immunoreactivity localized to the surface of many lymphocytes within perivascular cuffs of active and chronic active WM lesions, meningeal immune infiltrates, and ectopic B-cell follicles, but was not detected in brain parenchyma or vascular endothelial cells (Fig. 1). In control experiments, the EBV-negative Ramos cell line and sections from 2 nonpathological brains and 1 normal lymph node did not stain for LMP-2A (not shown), whereas LMP-2A immunoreactivity was readily detected in the EBV-positive Raji cell line and in an EBV-positive CNS lymphoma (Figure, Supplemental Digital Content 4, ). Latent membrane protein 2A was not detected in brain sections from 8 cases with other inflammatory neurological diseases (1 primary cerebral vasculitis, 1 viral encephalitis, 3 tuberculous meningoencephalitis, 1 mycotic meningoencephalitis, 1 luetic meningitis, and 1 mycotic meningitis), despite the presence of numerous infiltrating B cells (Figure, Supplemental Digital Content 4, ). Accordingly, in situ hybridization for EBER revealed no EBV-infected cells in 6 of these cases and the presence of rare scattered EBER-positive cells in 1 luetic meningitis case and 1 tuberculous meningoencephalitis case (Figure, Supplemental Digital Content 4, ).
The identity of LMP-2A-positive cells in MS brain inflammatory infiltrates was confirmed by double-immunofluorescence staining with anti-LMP-2A and anti-CD20 mAbs (Fig. 2). Although virtually all LMP-2A-positive cells were CD20+, the percentage of CD20+ B cells immunoreactive for LMP-2A ranged between 40% and 80% depending on the brain sample and area analyzed. This indicates that a high proportion of MS brain-infiltrating B cells is latently infected with EBV and is in line with data obtained with combined in situ hybridization for EBER and CD20 immunostaining (15).
Double-immunofluorescence staining for LMP-2A and LMP-1, another EBV-encoded latent membrane protein, revealed coexpression of these 2 viral proteins in a large number of cells infiltrating WM lesions and meninges (Figs. 3A-C). Control LMP-1 immunostaining in an EBV-positive CNS lymphoma and other inflammatory neurological diseases is shown in Figure, Supplemental Digital Content 4, . Cells expressing LMP-2A, but not LMP-1, were only rarely observed (Fig. 3B). This finding is consistent with the simultaneous expression of LMP-1 and LMP-2A in EBV-infected cells during latency III and II programs (39, 40). Together with the detection of a low frequency of cells immunoreactive for EBNA-2 (15), the first viral protein expressed in newly infected naive B cells during the latency III program but shut off in memory B cells during latency II (57), these data suggest that EBV mainly uses the latency II program to persist in the MS brain. Consistent with this, double immunofluorescence for the B-cell/plasma cell marker CD79a, one of the transducing units of the B-cell receptor (58), and CD27, an antigen expressed in memory B cells (59), showed that the majority (80%-98%) of B cells in WM lesions, inflamed meninges, and ectopic B-cell follicles are CD27+ antigen-experienced cells (Figs. 3D-F). These data are consistent with the finding that most B cells in the cerebrospinal fluid of MS patients are memory B cells and short-lived plasma blasts (60).
Quantification of LMP-2A and EBNA-1 Transcripts in MS Whole-Brain Sections Using Preamplification Real-Time RT-PCR
To assess the expression of the EBV latency genes LMP-2A and EBNA-1 (the latter being expressed during all phases of EBV latent infection) in a first set of experiments, whole-brain sections from 4 MS (MS46, MS79, MS180, MS342) and 2 control (C28, C30) cases were analyzed by quantitative real-time RT-PCR. This technique could detect LMP-2A and EBNA-1 transcripts in a single EBV-positive lymphoblastoid cell (Figure, Supplemental Digital Content 5, ), but in none of the MS and control brain samples analyzed (Fig. 4A). Expression of the B-cell marker CD19 was very low relatively to GAPDH and varied between different MS brain samples; in MS46 and MS180, the CD19 expression level was very close to that found in control brains, indicating negligible B-cell infiltration (Fig. 4A). In line with previous studies (23-25), these findings indicated that real-time RT-PCR is not sufficiently sensitive to detect small amounts of EBV transcripts even if the brain sections analyzed are derived from MS cases with a substantial degree of inflammation. The negative results could be caused by a combination of factors including suboptimal RNA preservation in autopsy tissue samples, differences in B-cell content between brain sections from the same tissue block, and small numbers of B cells relative to the total brain and inflammatory cell population.
To increase the sensitivity of the PCR assay, we next performed selective preamplification of LMP-2A, EBNA-1, and GAPDH cDNAs on the same brain sections previously analyzed, and quantitative real-time PCR was performed on the amplified templates. With cDNA preamplification, very low levels of LMP-2A and EBNA-1 transcripts became detectable in 3 and 2 of the 4 MS brain samples analyzed, respectively, but not in control brain samples (Fig. 4B). Of note, the highest amount of both EBV transcripts was found in the MS case with the highest relative amount of CD19 transcripts (MS342), which is indicative of more B-cell infiltration.
Detection and Quantification of EBV Transcripts in MS Brain Immune Infiltrates by Combining Laser-Capture Microdissection With Real-Time RT-PCR
To precisely match EBV gene expression with the presence of B-cell infiltrates, we next used laser-capture microdissection to select for brain regions enriched in B cells (sparse immune infiltrates and ectopic B-cell follicles in the meninges and perivascular cuffs in the WM) and devoid of B cells (lesioned WM surrounding inflamed blood vessels, subpial GM lesions, and NAWM) in combination with real-time RT-PCR to detect CD19, LMP-2A, and EBNA-1 transcripts in the selected areas. Areas of interest were identified by IHC and microdissected from adjacent sections after histological nuclear staining of samples from 4 MS cases (MS79, MS176, MS180, and MS342) (Figure, Supplemental Digital Content 6, ). In some cases, ectopic B-cell follicles were easily identified and dissected from dried unstained sections (Figure, Supplemental Digital Content 6, ). The B-cell follicles isolated from a nonpathological lymph node were used as positive control for CD19 and as negative control for LMP-2A and EBNA-1.
Because of the very low amounts of RNA recovered from laser-cut brain tissue, all MS brain samples analyzed had a moderate-to-low signal for the housekeeping gene GAPDH (26-33 Ct) (Table). CD19 was detected at very low levels (33-39 Ct) in 8 of 10 microdissected B cell-containing samples. Latent membrane protein 2A and EBNA-1 were detected in all and 7 of 10 B cell-containing samples, respectively (Table). With the exception of 1 meningeal infiltrate (in MS176), the signal for LMP-2A (29-35 Ct) was generally higher than that for EBNA-1 (31 to >40 Ct) and CD19 in the same sample (Table). Notably, in MS180, the only case with an active WM lesion in this series, EBNA-1 and LMP-2A signals were similar and higher than those of GAPDH, respectively (Table). Conversely, neither CD19 nor viral transcripts were detected in any of 9 laser-cut MS brain samples devoid of B cells (Table). Only CD19 but no viral transcripts were detected in B-cell follicles of the control lymph node (Table).
For an accurate comparative analysis of B-cell and EBV transcripts within a suitable Ct range using real-time RT-PCR, we performed selective preamplification of CD19, LMP-2A, EBNA-1, and GAPDH cDNAs on the same laser-cut samples previously analyzed. cDNA preamplification lowered Ct values for GAPDH in all MS brain samples and Ct values for CD19, LMP-2A, and EBNA-1 in all B cell-containing areas, whereas these transcripts remained undetectable in the samples devoid of B cells (Table). This increased sensitivity was consistent and reproducible, indicating that preamplification does not distort relative expression levels; rather, it enhances sensitivity. In B cell-containing samples from MS79, MS180, and MS342, the LMP-2A transcript levels were always higher than those of EBNA-1 (2- to 25-fold; median value, 10) and slightly (2- to 3-fold in MS79 and MS243) or markedly (10- to 67-fold in MS180; median value, 14.3) higher than those of CD19. Conversely, in the sparse meningeal infiltrate of MS176 LMP-2A, the expression was approximately 8-fold lower than that of EBNA-1 and CD-19 (Fig. 5). Relative to CD19, the expression of EBNA-1 was lower, similar or 2- to 4-fold higher in different microdissected brain samples (Fig. 5). In addition to confirming substantial EBV infection within MS brain immune infiltrates, these findings suggest heterogeneity in the level of latent EBV gene transcription in different MS cases.
Expression of LMP-2A in B cell-containing microdissected areas from 5 MS brain samples (MS79, MS92, MS121, MS176, MS180) was confirmed by double-nested RT-PCR with different primer sets (Figure, Supplemental Digital Content 7, ). Latent membrane protein A transcripts were detected in 6 meningeal B-cell follicles and 1 of 2 B cell-containing perivascular cuffs in a chronic active WM lesion, but not in meningeal infiltrates containing sparse B cells, indicating a lower sensitivity of this method as compared with real-time RT-PCR. No LMP-2A transcripts were detected in lesioned GM (n = 4) or WM (n = 1) areas.
BAFF Production by EBV-Infected B Cells
The BAFF plays a critical role in supporting B-cell maturation and survival and T cell-independent antibody production and has been implicated in the induction of self-reactive antibodies in experimental settings (50). In a previous study, BAFF immunoreactivity was detected in reactive astrocytes but not in WM perivascular immune infiltrates of MS brains (51). The BAFF expression was not investigated thoroughly in MS meninges and active WM lesions, however. Because BAFF is produced by EBV-infected B cells and transduction of EBV-negative Ramos cells with LMP-1- and LMP-2A-expressing vectors upregulates BAFF mRNA and protein (48), we re-examined BAFF expression in the MS brain, focusing our IHC analysis on ectopic B-cell follicles and active WM lesions highly enriched in EBV-infected B cells.
Strong BAFF immunoreactivity was detected in the cytoplasm of immune cells infiltrating the meninges of all the MS brain samples analyzed (n = 6) (Figs. 6A-C). In 2 cases, the frequency of BAFF-positive cells was high both in sparse meningeal infiltrates and ectopic B-cell follicles (Figs. 6A, B), whereas in the other 4 cases, BAFF-positive cells had a more scattered distribution (Fig. 6C).
Double-immunofluorescence staining with anti-CD20 and anti-BAFF mAbs revealed that virtually all BAFF-positive cells in ectopic B-cell follicles and less organized meningeal infiltrates were B cells, whereas the proportion of BAFF-positive B cells in the total B-cell population largely varied depending on the brain sample and area analyzed (Figs. 7A-C). Within ectopic follicles, the percentage of B cells expressing BAFF ranged between 10% and 50% (Figs. 7A, B), and there was also variability between 2 different follicles of the same sample. Sparse meningeal B cells were all BAFF positive in some areas (Fig. 7C) and largely BAFF negative in others (not shown). This variability may reflect the detection limits of IHC or differences in the activation state of EBV-infected B cells. By double-immunofluorescence staining with anti-BAFF and anti-LMP-2A mAbs, we obtained direct evidence that BAFF is expressed in EBV-infected cells (Figs. 7D, E). The percentage of intrafollicular LMP-2A-positive cells expressing BAFF ranged between 15% and 90% in the 4 MS samples examined (15% in MS109, 30% in MS 180 [Fig. 7D], 50%-90% in MS121, and approximately 90% in MS92 [Fig. 7E]).
Because BAFF is known to be produced by activated macrophages (61), we also performed double-immunofluorescence staining with mAbs specific for BAFF and for the macrophage marker Iba-1. The immunoreactivity of BAFF was rarely detected in meningeal Iba-1-positive macrophages (Fig. 7F), indicating that B cells are the main source of BAFF in the inflamed meninges. Several BAFF-positive cells with characteristic astrocyte morphology but no Iba-1-positive/BAFF-positive microglia were present in subpial cortical lesions (Fig. 7F) and in the surrounding nondemyelinated GM (data not shown).
The expression of BAFF in reactive astrocytes was confirmed in WM lesions (Figs. 8A-C). Moreover, nearly all B cells were found to express BAFF in the large perivascular cuffs of active MS lesions in 2 MS cases (Figs. 8C-E). Conversely, no BAFF immunoreactivity was detected in perivascular immune infiltrates in chronic active WM lesions (Fig. 8A), despite the fact that in the same brain sections, BAFF-positive B cells were present in the inflamed meninges and/or active WM lesions.
This study confirms and extends previous evidence of perturbed EBV infection in the MS brain by showing that LMP-2A, a latency protein used by EBV to stimulate B-cell activation and survival, is expressed in a large proportion of brain-infiltrating B cells and that transcripts of EBV latency genes (LMP-2A and EBNA-1) can be consistently detected in WM intralesional and meningeal immune infiltrates using very sensitive RT-PCR techniques in combination with laser-capture microdissection. Another major finding of this study is that BAFF is produced by EBV-infected B cells in the MS brain, suggesting that EBV activation signals may be amplified by autocrine and paracrine actions of BAFF. These findings support the hypothesis that a persistently dysregulated EBV infection could play a key role in sustaining chronic B-cell activation, which is itself essential for viral persistence and an immunopathologic response.
Other studies have failed to detect EBV in postmortem brains of all or most MS cases examined (21-25). We suggest that these discrepancies are most probably explained by technical differences rather than by analysis of different MS cohorts. Indeed, brain samples from the same MS cases (UK MS Tissue Bank collection) were EBV positive in our studies and EBV negative when analyzed by other groups (23,24). The use of differently processed brain tissues (frozen versus paraffin-embedded), extensive sampling and selection of brain sections with substantial B-cell infiltration, sensitivity of in situ hybridization and PCR techniques used to detect EBV nucleic acids, and differences in IHC protocols and type of antibodies used to visualize viral proteins are all critical issues that need to be carefully evaluated when searching for EBV in MS brain tissue.
By adding LMP-2A to the list of EBV antigens that we have already analyzed by IHC (i.e. EBNA-2 and LMP-1 for the latent phase, and BFRF1, gp350/220, and p160 for the lytic phase of viral infection) (15), we provide further evidence that this technique can reliably detect EBV and be used to investigate the status of viral infection in the MS brain. The highest density of LMP-2A-positive B cells was found in the perivascular cuffs of active WM lesions and in meningeal B-cell follicles, matching the expression pattern of LMP-1 (15, and this study) and EBER (15), 2 other markers of EBV latent infection.
Using quantitative real-time PCR techniques, several recent studies failed to detect EBV DNA and/or EBER transcripts in brain sections from all (24, 25) and most (23) MS brain samples analyzed. Because RNA preservation in autopsy brain samples might not be optimal and because EBV infection seems confined to a subset of MS brain-infiltrating immune cells, it should be suspected that the detectability of EBV nucleic acids in whole-brain sections might be well below the sensitivity of real-time PCR techniques. Here, we show that selective cDNA preamplification allows detecting small amounts of EBV RNA that could not be detected by conventional real-time PCR on whole-brain sections. Moreover, laser-capture microdissection in conjunction with real-time RT-PCR provided consistent and reproducible detection of EBV latency transcripts in MS brain tissue. Selective cDNA preamplification allowed us to perform an accurate comparative analysis of cellular and viral genes even if the amount of starting RNA was very small. With these optimized techniques, LMP-2A and EBNA-1 gene expressions were demonstrated in all microdissected MS brain regions containing B-cell infiltrates, and quantitation of viral gene transcription was performed. We found that the expression of LMP-2A and EBNA-1 was comparable and in some cases higher than that of the B cell-associated gene CD19, which is compatible with the high proportion of EBV-infected B cells assessed with EBER in situ hybridization and IHC (15, and this study). Without the sensitivity of laser-capture microdissection and/or preamplification real-time PCR techniques, we suspect that EBV detection would be unattainable in MS brain samples.
The EBV uses 4 different gene programs to manipulate the normal B-cell differentiation process and establish a lifelong latent infection in B cells (39, 40). In the initial phase of infection in the tonsil, EBV induces extrafollicular naïve B cells to undergo activation and proliferation through 3 LMPs (LMP-1, -2A, and -2B) and 6 EBV-encoded nuclear proteins (EBNA-1, -2, -3a, -3b, -3c, and -LP). This growth program (or latency III) allows expansion of the viral episome in the B-cell compartment. Subsequently, EBV induces infected lymphoblasts to switch to a default program (or latency II), which entails a more limited set of latency proteins (EBNA-1, LMP-1, and LMP-2A) and allows infected B cells to enter a germinal center reaction and acquire a memory phenotype (62). Latent membrane protein 1 and LMP-2A mimic the signals that are normally delivered to the B cell through CD40 and the B-cell receptor, respectively, allowing infected B cells to survive and mature independently of antigen and T-cell help (41, 42, 63, 64). The virus persists in rare memory B cells that express no viral proteins (latency 0) or only EBNA-1 (latency I) and are present at very low frequency in the blood (39, 40). Downregulation of viral latent proteins allows EBV to escape immune surveillance that is otherwise very efficient in targeting latently and lytically infected B cells (65). By showing that most B cells present in MS lesions, meninges, and ectopic follicles are CD27+ antigen-experienced cells and express LMP-1 and LMP-2A, but not EBNA-2 as previously reported (15), and by detecting EBNA-1 and LMP-2A transcripts in the same infiltrated areas, we demonstrate that EBV mainly uses the latency II or default program to establish a persistent infection in the MS brain.
Intrathecal production of antibodies in MS is thought to reflect an antigen-targeted and T cell-dependent immune response because antibodies synthesized by B cells in MS brain lesions and cerebrospinal fluid are isotype switched (mainly IgG), somatically mutated, and clonally expanded (66-70). Based on our findings that a large proportion of B cells accumulating in the MS brain are infected with EBV and express LMP-1 and LMP-2A, which drive B cells into a germinal center reaction by inducing antiapoptotic and proliferation signals (47, 71-73) as well as expression of activation-induced cytidine deaminase, the enzyme responsible for somatic hypermutation and Ig class-switch recombination in germinal center B cells (48), we propose that dysregulated EBV infection, rather than chronic antigenic stimulation, contributes to intrathecal oligoclonal and polyclonal B-cell activation and IgG synthesis in MS. Abundant expression of the antiapoptotic protein Bcl-2, presence of proliferating B cells and plasma cell accumulation in MS immune infiltrates (15), as well as the formation of meningeal B-cell follicles with germinal center-like features (53, 74), including activation-induced cytidine deaminase expression (15), could all be the manifestations of EBV-induced B-cell stimulation. The fact that meningeal follicles lack the typical structure of lymphoid follicles with a germinal center and a mantle zone and comprise mainly memory B cells is, however, suggestive of a dysfunctional B-cell maturation process. Within these structures, EBV-infected B cells could undergo somatic hypermutation uncoupled from the selection processes that occur physiologically in germinal centers and allow B-cell clones producing high affinity Ig variants to survive, whereas low-affinity or self-reactive clones are eliminated by apoptosis (75). This dissociation could explain the difficulty of identifying the target antigens of oligoclonal IgG, which represent the main part of intrathecally produced antibodies.
By showing that most B cells accumulating perivascularly in active MS lesions are infected with EBV and express BAFF, and that BAFF is expressed in a variable proportion of EBV-infected cells in sparse meningeal infiltrates and ectopic B-cell follicles, our results support the idea that this cytokine, through its ability to regulate proliferation, maturation, and survival of B cells, represents an important link between EBV infection and B-cell dysregulation in autoimmune diseases (76). Our observations in MS brain tissue are consistent with the results of in vitro studies showing that EBV-infected B-cell lines produce BAFF, and that transfection of B-cell lines with LMP-1- and LMP-2A-expressing vectors upregulates BAFF mRNA and protein (48). In a previous study, both BAFF mRNA and protein were shown to be expressed in the normal human brain and to be upregulated in the brain of MS patients (51). Astrocytes were identified as the main source of BAFF in the CNS parenchyma, and BAFF receptor was shown to be expressed in MS immune infiltrates (51). Because of its localization in the perivascular space of inflamed blood vessels and in meningeal infiltrates/follicles, BAFF secreted by or bound to the membrane of EBV-infected B cells could play a major role in promoting B-cell survival and differentiation (50) as well as survival of a pool of long-lived plasma cells (77), the most likely source of intrathecally produced antibodies in MS. Because not all B cells in the MS brain express EBV markers, it is conceivable that BAFF produced by latently EBV-infected B cells might act in an autocrine and paracrine manner, supporting the survival and maturation of both infected and uninfected B cells.
Increased BAFF production is a common finding in several B cell-related autoimmune diseases, and BAFF overexpression in transgenic mice leads to rescue of self-reactive B cells and development of lupus-like autoimmune disease (78-80). Interestingly, studies in transgenic mice indicate that LMP-2A overexpression also facilitates autoreactive B-cell activation through its ability to mimic B-cell receptor signaling and increase the sensitivity of B cells to toll-like receptor activation (46, 47). Although it is still uncertain whether autoantibodies to CNS antigens have a pathogenic role in MS (27), the possibility should not be excluded that EBV, through LMP-2A and induction of BAFF, might contribute to disruption of B-cell tolerance checkpoints, resulting in expansion of autoreactive B-cell clones.
In conclusion, the present findings strengthen the concept that intrathecal B-cell activation in MS might result from and be instrumental for persistent EBV infection in the CNS. Continuous attempts of the host immune system to eliminate intracerebral EBV-infected B cells would result in perpetuation of a destructive immunopathologic process as indicated by a recent study showing intrathecal enrichment in EBV-specific CD8+ cytotoxic T cells in a cohort of early MS patients (81). Although it may prove difficult to identify the early events and cofactors that enable EBV-infected B cells to colonize the brain in MS, acquiring new information on EBV latency patterns, the switch from viral latent infection to lytic replication, and the molecular pathways (e.g. BAFF) induced by EBV in the inflamed brain should allow us to better understand the contribution of EBV to abnormal B-cell activation and chronic inflammation in MS and identify novel molecular targets for therapeutic intervention. This knowledge may also help shed light on the mechanisms underlying the beneficial effects of B-cell depletion with rituximab in relapsing-remitting MS (82) and interpret the outcome of therapies targeting BAFF function (83).
The authors thank Dr Diego Franciotta for helpful comments on the manuscript. All MS brain tissue samples used in this study were supplied by the UK Multiple Sclerosis Tissue Bank (www.ukmstissuebank.imperial.ac.uk), funded by the MS Society of Great Britain and Northern Ireland (registered charity 207495). The authors also thank Dr Romana Höftberger (Clinical Institute of Neurology, Wien, Austria) and Dr Egidio Stigliano (Policlinico A. Gemelli, Rome, IT) for providing brain sections from cases with other inflammatory neurological diseases, Dr Susan Kalled (Biogen Idec, Boston, MA) for providing the anti-human BAFF antibody, and Ms Estella Sansonetti and Mr Luigi Nicoletti for graphical work.