Radioactive in situ hybridization for Epstein–Barr virus–encoded small RNA supports presence of Epstein–Barr virus in the multiple sclerosis brain

Sir, despite the association between Epstein–Barr virus infection and multiple sclerosis being supported by an increasing number of epidemiological, serological and immunological studies (Ascherio and Munger, 2010; Maghzi et al., 2011; Pender, 2011), the mechanisms linking increased immune reactivity to this ubiquitous herpes virus to CNS pathology in patients with multiple sclerosis remain to be demonstrated (Lünemann and Münz, 2009; Giovannoni, 2011). A point of contention is whether an active and persistent Epstein–Barr virus infection in the CNS of patients with multiple sclerosis may sustain a detrimental immunopathological response (Pender, 2009; Ascherio and Bar-Or, 2010). In support of this scenario, we have shown high frequencies of Epstein–Barr virus latently infected B cells and more sporadic signs of viral reactivation in post-mortem brain tissue from patients with multiple sclerosis but not patients with other inflammatory neurological diseases (Serafini et al., 2007, 2010; Aloisi et al., 2010). Presence of Epstein–Barr virus latently infected cells in active multiple sclerosis brain lesions has been confirmed recently (Tzartos et al., 2012). In contrast, other groups did not find Epstein–Barr virus in all or the majority of post-mortem multiple sclerosis brain samples analysed (Hilton et al., 1994; Opsahl and Kennedy, 2007; Willis et al., 2009; Peferoen et al., 2010; Sargsyan et al., 2010; Torkildsen et al., 2010). As recently discussed in Brain (Lassmann et al., 2011), no consensus has been reached on technical and interpretation issues relating to the detection of Epstein–Barr virus in the multiple sclerosis brain. Although a variety of techniques [in situ hybridization for Epstein–Barr virus-encoded small RNA (EBER), immunohistochemistry for Epstein–Barr virus latent and lytic proteins and PCR assays for Epstein–Barr virus nucleic acids] have been used to search for Epstein–Barr virus in multiple sclerosis brain, comparing and interpreting the contrasting results from different studies has been difficult owing to the lack of common procedures and tools. EBER in situ hybridization is considered the most powerful standard method to study the anatomical localization of Epstein–Barr virus–infected cells (Gulley and Tang, 2008). EBERs (EBER1 and EBER2) are small untranslated Epstein–Barr virus transcripts that are abundantly expressed during all phases of Epstein–Barr virus latent infection and generally localize in the cell nucleus. Despite high target specificity and sensitivity of EBER probes, false negative results due to RNA degradation or presence of non-specific staining can invalidate the conclusions of in situ hybridization studies (Gulley and Tang, 2008). Regarding Epstein–Barr virus presence in the multiple sclerosis brain, criticism has been raised toward the high number of EBER-positive cells we found in multiple sclerosis white matter lesions and meninges (particularly inside B cell follicles) (Serafini et al., 2007) and the presence of both nuclear and cytoplasmic EBER signals observed using a peptide nucleic acid (PNA) probe and detection kit from DakoCytomation (Lassmann et al., 2011). The reported absence of EBER signals in brain tissues from cases with other inflammatory neurological diseases or in non-pathological brain and lymphoid tissues (Serafini et al., 2007, 2010) was not deemed sufficient to interpret the EBER signals observed in the multiple sclerosis brain as specific (Lassmann et al., 2011). We reasoned that a sound strategy to confirm or disprove the findings obtained with the commercial EBER in situ hybridization kit (Serafini et al., 2007) was to take advantage of the high sensitivity of a radioactive in situ hybridization protocol routinely used by one of the authors (L.M.) at S. Raffaele Hospital in Milan (Muzio et al., 2002, 2005a, b, 2010; Centonze et al., 2009) and the availability of non-commercial EBER probes validated in previous studies (Niedobitek et al., 1991) to analyse EBER transcripts in the same multiple sclerosis brain tissue samples that yielded contrasting results (Serafini et al., 2007; Peferoen et al., 2010; Sargsyan et al., 2010; Willis et al., 2009). The signals obtained with radioactive EBER in situ hybridization were then confronted with those obtained in adjacent brain sections with the commercial, non-isotopic EBER in situ hybridization used in the laboratory at Istituto Superiore di Sanità in Rome. Multiple sclerosis brain tissue samples characterized by a high degree of inflammation and presence of numerous B cells and plasma cells—the main Epstein–Barr virus reservoirs—were selected starting from a case series provided by the UK MS Tissue Bank. A total of 108 sections were cut from 13 paraformaldehyde-fixed frozen brain blocks from six multiple sclerosis cases (Cases MS79, MS92, MS121, MS154, MS180 and MS234), all of which were reported to be Epstein–Barr virus negative in other studies (Peferoen et al., 2010; Sargsyan et al., 2010; Willis et al., 2009). The selected brain samples contained well-preserved RNA, as assessed by in situ hybridization for glyceraldehyde 3-phosphate dehydrogenase (Fig. 1A). Fifty-two brain sections were processed for CD20 and/or CD138 immunohistochemistry (to detect B cells and plasma cells, respectively) and for EBER in situ hybridization using the PNA probe (which hybridizes with both EBER1 and EBER2) and detection kit from DakoCytomation (Serafini et al., 2007). Fifty-six brain sections were coded and processed for radioactive in situ hybridization at S. Raffaele Hospital. Sections were exposed to αP33-labelled probes synthesized from plasmids containing complementary DNAs for EBER1 and EBER2 (kindly provided by Prof. G. Niedobitek, Sana Klinikum Lichtenberg/Unfallkrankenhaus Berlin, Germany) (Niedobitek et al., 1991) and then to the autoradiography emulsion for 5, 9 or 18 days (strong signals were detected at all times). Sections were analysed with a darkfield-equipped microscope after counterstaining with 4′,6-diamidino-2-phenylindole (DAPI) or in brightfield microscopy after counterstaining with haematoxylin. The protocol for radioactive in situ hybridization is described in detail in the Supplementary material. Sections stained with the radioactive and non-radioactive in situ hybridization methods were assessed independently. Pictures of relevant areas were acquired and the distribution of EBER signals was compared and related to that of B cells and/or plasma cells in adjacent sections. Detection of EBER+ cells in inflamed multiple sclerosis meninges: comparison of radioactive and non-radioactive in situ hybridization. In situ hybridization with a PNA glyceraldehyde 3-phosphate dehydrogenase (GAPDH) probe reveals labelling of most cells in a brain section from Case MS180 (A, moderately inflamed meninges and adjacent grey matter are shown). In situ hybridization with pooled αP33-labelled EBER1 and EBER2 antisense (as) probes reveals presence of scattered EBER+ cells in the meninges and in the grey matter (arrows) (B, darkfield image with DAPI counterstaining). In darkfield, silver grains accumulating over EBER+ cells appear as red dots, or black spots in higher density areas, whereas EBER− cells are only positive for DAPI nuclear staining. No signal is present when using radiolabelled EBER1/EBER2 sense (s) probes (C). The distribution of EBER+ cells in the meninges and grey matter is similar in an adjacent section that has been incubated with PNA EBER1/EBER2 probe (D, the arrows indicate EBER+ cells in the grey matter), whereas no signal is detected with a PNA irrelevant probe (E). (A to E) Serial brain sections. Numerous EBER+ cells are detected in a dense perivascular immune infiltrate located in the subarachnoid space at the tip of a cerebral sulcus in Case MS234 when using pooled αP33-labelled EBER1 and EBER2 antisense probes (darkfield and brightfield images in F and G, respectively; the insets in G show EBER+ cells at higher magnification). Individual αP33-labelled EBER1 (H) and EBER2 (I) antisense probes reveal a similar expression pattern although the signal intensity and the number of positive cells are reduced in the EBER2 probed section. No signal is present when using radiolabelled sense probes (J). The distribution and density of intrameningeal EBER+ cells observed in an adjacent section exposed to PNA EBER1/EBER2 antisense probes (K) match the images shown in F–I. The area marked with the frame in K is shown at high power magnification in L, highlighting the nuclear localization of EBER. Immunostaining for CD20 reveals presence of numerous B cells in the same meningeal immune infiltrate (Linset). No signal is detected with an irrelevant PNA probe (M). Pictures in panels F to M are from serial sections. Sections shown in B, C, F, H, I and J are counterstained with DAPI. Sections shown in A, D, E, G, K, L and M are counterstained with haematoxylin. Scale bars: A–K and M = 100 μm; L = 50 μm; Linset = 20 μm, Ginset = 10 μm. The specificity of the signal obtained with radioactive EBER in situ hybridization was first checked in Epstein–Barr virus positive and negative control tissues (Supplementary Fig. 1). Using pooled antisense EBER1 and EBER2 probes, numerous EBER+ cells were detected in a tonsil from a patient with infectious mononucleosis (paraffin-embedded tissue). Sections of a post-mortem, non-pathological abdominal lymph node (fixed-frozen tissue) contained a few scattered EBER+ cells in a restricted perifollicular region, whereas no EBER signals were detected in a non-neuropathological brain (fixed-frozen tissue). After incubation with sense EBER probes, no signals were detected in any of the above tissues (Supplementary Fig. 1). Non-radioactive EBER in situ hybridization yielded similar findings (Supplementary Fig. 1). Then, radioactive EBER in situ hybridization was performed on multiple sclerosis brain sections adjacent to those selected for presence of B cell/plasma cell–containing immune infiltrates in chronic active white matter lesions and in the meninges. Initially, sections were incubated with pooled EBER1/EBER2 antisense probes and in a second round of experiments with the individual probes to check for preferential expression of EBER1 or EBER2. A problem that was frequently encountered when processing sections cut from post-mortem fixed-frozen multiple sclerosis brain tissue was detachment of dense immune infiltrates (i.e. medium/large intraparenchymal perivascular cuffs and meningeal B cell follicles) from histology slides after high stringency washing and during autoradiography emulsion processing (an example is given in Fig. 2I). Nevertheless, less prominent immune infiltrates in the meninges (Fig. 1) and white matter lesions (Fig. 2) could be evaluated for the presence of EBER-positive cells. The highest frequency of EBER+ cells was detected in diffuse meningeal infiltrates (Fig. 1B), meningeal B cell aggregates (Fig. 1F and G) and perivascular immune infiltrates of white matter lesions (Fig. 2A, B and G). Isolated EBER+ cells were rarely seen in the grey matter (Fig. 1B) and more frequently in white matter lesions (Fig. 2E). When adjacent sections were exposed to individual radiolabelled EBER1 and EBER2 probes, both transcripts were detected although signal intensity and number of positive cells were higher in sections incubated with EBER1 riboprobes (Fig. 1H and I). In general, the localization of EBER+ cells mirrored that of CD20+ B cells (inset in Fig. 1L and Fig. 2C) or CD138+ plasma cells (inset in Fig. 2F). High amounts of silver grains accumulated over EBER+ cells, impairing the identification of single positive cells when these were more tightly packed (Figs 1F, G and 2B, G). Moreover, owing to the dotted distribution of radioactive EBER signals in positive cells, it was difficult to discriminate between nuclear or cytoplasmic localization of the signal. Multiple sclerosis brain sections incubated with radiolabelled EBER1 and EBER2 sense probes and processed with the same protocol did not show any discernible signal (Fig. 1C and J). Detection of EBER+ cells in chronic active multiple sclerosis lesions: comparison of radioactive and non-radioactive in situ hybridization. A small cluster of perivascular EBER+ cells in a chronic active white matter lesion (Case MS79) is visualized with pooled αP33-labelled EBER1 and EBER2 antisense probes (A). (B) The same immune infiltrate at higher magnification. Immunostaining for CD20 (C) and in situ hybridization with PNA EBER1/EBER2 antisense probe (D) performed in adjacent sections show perivascular accumulation of several B cells and EBER+ cells, respectively. Single scattered EBER+ cells are detected in a periventricular white matter lesion (different tissue block of Case MS79) with radiolabelled (E) and PNA (F) EBER1/EBER2 probes. The inset in F shows presence of scattered CD138+ plasma cells in the same area. Perivascular EBER+ cells in a white matter lesion from case MS121 are visualized with radiolabelled (G) and PNA (H) EBER1/EBER2 probes. Panel I shows that the dense, intralesional perivascular cuff enriched in CD20+ B cells (inset) is lost after in situ hybridization with radiolabelled EBER probes; only one EBER+ cell is visible in the adjacent parenchyma. Sections in A, B, E, G and I are counterstained with DAPI; sections in C, D, F and H are counterstained with haematoxylin. Scale bars: E, F, I and Iinset = 100 μm; A, D, G and H = 50 μm; B and C = 20 μm. When the radioactive EBER signals were compared with the signals obtained in adjacent multiple sclerosis brain sections with the PNA EBER probe (complementary to EBER1 and EBER2) (Figs 1D, K, L and 2D, F, H), there was a good correspondence in the distribution and frequency of EBER+ cells between the two sets of sections. Provided that tissue preservation was comparable after processing with the two in situ hybridization methods, the same pattern of aggregated or isolated EBER+ cells was observed in the inflamed meninges (Fig. 1) and in chronic active white matter lesions (Fig. 2). No signal was observed with an irrelevant PNA probe (Fig. 1E and M). Several conclusions can be drawn from this study. First, absence of Epstein–Barr virus RNA in the multiple sclerosis brain should be interpreted with caution. The ability of a radioactive in situ hybridization protocol to consistently detect EBER signals in the same multiple sclerosis brain samples that other groups reported to be devoid of EBER+ cells (Peferoen et al., 2010; Willis et al., 2009) points to lack of sufficient sensitivity of the non-radioactive in situ hybridization protocols used in those studies and/or degradation of target RNA in the tissue blocks or sections analysed. Most protocols developed for the diagnosis or study of Epstein–Barr virus-associated malignancies and primary acute Epstein–Barr virus infection in bioptic material may not be optimal to detect Epstein–Barr virus in autopsy tissue samples from patients with non-tumoral, chronic pathologies that could be characterized by a relatively lower level of Epstein–Barr virus gene expression in the infected cells. Better RNA preservation in fixed frozen brain samples than in archival paraffin-embedded material could also account for inability to visualize EBER signals in the latter (Peferoen et al., 2010; Willis et al., 2009; Lassmann et al., 2011). However, Tzartos et al. (2012) have demonstrated presence of EBER+ cells in sections from paraffin multiple sclerosis brain tissue. A combination of highly sensitive EBER in situ hybridization and optimally processed multiple sclerosis brain samples is more likely to provide the best sensitivity and reproducibility of the results. Second, the similar distribution of EBER+ cells detected in serial multiple sclerosis brain sections using radiolabelled EBER probes and the PNA EBER probe/in situ hybridization kit excludes that the high frequency of Epstein–Barr virus infected B cells revealed with the latter method is attributable to the use of an oversensitive chromogen-based detection system, as discussed in Lassmann et al. (2011). Although atypical, the cytoplasmic localization of EBER observed in a fraction of B cells/plasma cells accumulating in the multiple sclerosis brain should not be dismissed as non-specific (Lassmann et al., 2011). This finding should rather be evaluated in light of recent findings showing that EBERs can be released from the nucleus and interact with cytoplasmic proteins of the infected cells as well as with receptors, such as Toll-like receptor 3, expressed on neighbouring innate immune cells and mediating their activation (Iwakiri et al., 2009). Whether EBER leakage from the nucleus in the multiple sclerosis brain is related to a viral survival strategy, exposure to a chronically inflamed environment, abortive viral reactivation or just loss of cellular integrity in autopsy tissue, remains to be determined. Third, by confirming that an active Epstein–Barr virus infection can spread across the CNS through infection of B cells, this study strengthens the idea that the more robust humoral and cellular immune responses to Epstein–Barr virus found in patients with multiple sclerosis (Sumaya et al., 1985; Höllsberg et al., 2003; Lünemann et al., 2006, 2010; Jilek et al., 2008; Farrell et al., 2009; Sundqvist et al., 2012) are directly linked to inflammatory brain pathology. The findings that primary acute Epstein–Barr virus infection and raised anti-Epstein–Barr virus antibody serum titres are risk factors for multiple sclerosis strongly suggest that immune dysfunction caused by prior Epstein–Barr virus infection plays a role in multiple sclerosis development (Ascherio and Münger, 2010; Handel et al., 2010). Although it has to be considered (and investigated) that altered immune reactivity to Epstein–Barr virus might trigger or result from a still poorly defined autoimmune dysfunction in multiple sclerosis, the available evidence converges to indicate that patients with multiple sclerosis have a problem with the control of Epstein–Barr virus infection (Pender, 2011). An analogy can be drawn between multiple sclerosis and human T-lymphotropic virus type 1 (HTLV-1) associated myelopathy/tropical spastic paraparesis, a chronic inflammatory and demyelinating disease that develops in a minority of individuals infected with the HTLV-1 retrovirus as a consequence of the migration of HTLV-1 infected T cells in the CNS and the ensuing immunopathological response (Jacobson, 2002). In individuals who are predisposed to develop multiple sclerosis because of a susceptible genetic background and/or exposure to other environmental risk factors (Giovannoni and Ebers, 2007), circulating infected B cells could act as Trojan horses for Epstein–Barr virus entry into the CNS during primary Epstein–Barr virus infection or later as a consequence of some other insult to the CNS (Warner and Carp, 1981; Pender, 2003; Franciotta et al., 2008; Maghzi et al., 2011). Epstein–Barr virus appears well suited to support intracerebral survival, expansion and maturation of B cells because of its ability to activate latently infected B cells and undergoes reactivation when these differentiate into plasma cells (Serafini et al., 2007). CNS colonization by Epstein–Barr virus–infected cells would promote a chronic highly detrimental immunopathological response, a scenario supported by accumulation of Epstein–Barr virus–specific CD8+ T cells in the CSF of patients with a recent diagnosis of multiple sclerosis (Jaquiéry et al., 2010) and evidence of a cytotoxic attack on Epstein–Barr virus lytically infected cells in multiple sclerosis brain lesions and meninges (Serafini et al., 2007; and unpublished results). However, as the search for Epstein–Barr virus has been performed mostly in autopsy brain material obtained from cases with multiple sclerosis dying during the secondary progressive phase of multiple sclerosis, it remains to be shown that intracerebral Epstein–Barr virus infection is an early event in multiple sclerosis. An answer to this question can be provided by searching for Epstein–Barr virus nucleic acids and protein antigens in biopsies from patients undergoing brain surgery for differential diagnosis purposes or in the CSF from patients with clinically isolated syndrome or early definite multiple sclerosis. The authors’ research was funded by grants from European Union (FP6 Integrated project NeuroproMISe—LSHM-CT-2005-01863) and Italian Ministry of Health (Project 107-Ricerca Finalizzata 2007) to F.A. Supplementary material is available at Brain online. All post-mortem samples were supplied by the UK MS Tissue Bank (www.ukmstissuebank.imperial.ac.uk). The authors would like to thank members of the UK MS Tissue Bank (R. Reynolds, G. Dveric, S. Fordham, J. Steele) for the assistance in the collection of the material used in this study.