Abstract
Background Epstein Barr Virus (EBV) infection of B cells is now understood to be one of the triggering events for the development of Multiple Sclerosis (MS), a progressive immune-mediated disease of the central nervous system. EBV infection is also linked to expression of Human endogenous retroviruses (HERVs) of the HERV-W group, a further risk factor for the development of MS (Ocrelizumab is a high-potency disease-modifying treatment (DMT) for MS, which depletes B cells by targeting CD20.
Objectives We studied the effects of ocrelizumab on gene expression in peripheral blood mononuclear cells (PBMC) from paired samples from 20 patients taken prior to and 6 months after beginning Ocrelizumab therapy. We hypothesised that EBV and HERV-W loads would be lower in post-treatment samples.
Methods Samples were collected in Paxgene tubes, subject to RNA extraction and Illumina paired end short read mRNA sequencing with mapping of sequence reads to the human genome using Salmon and Differential Gene expression compared with DeSeq2. Mapping was also performed separately to the HERV-D database of HERV sequences and the EBV reference sequence.
Results Patient samples were more strongly clustered by individual rather than disease type (relapsing/remitting or primary progressive), treatment (pre and post), age, or sex. Fourteen genes, all clearly linked to B cell function were significantly down regulated in the post treatment samples. Interestingly only one pre-treatment sample had detectable EBV RNA and there were no significant differences in HERV expression (of any group) between pre- and post-treatment samples.
Conclusions While EBV and HERV expression are clearly linked to triggering MS pathogenesis, it does not appear that high level expression of these viruses is a part of the ongoing disease process or that changes in virus load are associated with Ocrelizumab treatment.
Introduction
Multiple sclerosis (MS) is a disabling immune-mediated, inflammatory disease of the central nervous system (CNS) 1. Pathological damage is directed against CNS myelin and axons, with clinical manifestations that are characterised by relapses and / or progression of neurological deficits. Most patients present with a relapsing-remitting clinical course (RRMS) characterised by partial recovery between bouts of inflammation, followed by secondary progressive disease (SPMS) in which gradual neurological decline is independent of relapses. A smaller number of patients have progressive disease from the start (primary progressive course, PPMS). In spite of variations in clinical presentation, MS is considered a single disease. It is likely that different pathological processes underlie relapses and progression. Numerous immunotherapies have been approved for the treatment of relapsing-remitting disease and only two for progressive disease. There is general consensus that early treatment of MS is advantageous 2,3.
While not solely responsible for MS pathogenesis, viral infections are a known risk factor for the development of the disease and indeed the first established effective therapy for MS was interferon beta, a cytokine with a central role in antiviral immune responses 4. In particular, post pubertal infection with Epstein Barr Virus (EBV), normally a common childhood infection, is strongly linked to disease risk and exposure to the virus appears to be required for disease development 5. Recent large clinical studies including one of repeated long term blood samples from over 10 million US military personnel have demonstrated this epidemiological link beyond doubt 6,7. Over-expression of human endogenous retroviruses (HERVs) of the HERV-W family is also associated with MS disease risk 8. These HERVs are copies of ancestral viral infections that have become integrated into the host’s genome and now perform essential host functions (one HERV-W protein is an essential component of the human placenta) 8.
There is a rapidly increasing body of evidence linking EBV and HERV (particularly the HERV-W family) proteins with MS pathogenesis. EBV replicates primarily in B cells,9 where it establishes latency, associated with the viral protein Epstein Barr nuclear antigen 2 (EBNA2). EBNA2 binds to genetic loci associated with MS risk competing for transcription binding sites with Vitamin D, high levels of which are protective against MS risk and inhibit B cell proliferation 9. HERV-W env is expressed specifically in monocytes, T and B lymphocytes and NK cells and is particularly associated with activation of the non-classical monocyte class (CD14lowCD16+) that are upregulated in MS 10-12.
EBV replication in B cells triggers HERV-W and HERV-K expression 13-16 initiating a cascade of stimulation of inflammation 17-19 and cross reactivity with myelin oligodendrocyte protein (MOG) 20. HERV-W and EBV expression levels are also associated and EBV and HERV-W loads are correlated in MS patients undergoing therapy 21. An additional line of evidence linking HERV-W proteins to MS pathogenesis has shown HERV-W expression in microglia (brain-resident myeloid cells) associates with axons inducing a degenerative phenotype resulting in damage to myelinated axons 22. HERV-W has also been shown to inhibit oligodendrocyte precursor cell formation and remyelination, an effect that can be blocked by the anti-HERV monoclonal antibody GNbAC1 23,24.
Both Epstein Barr nuclear antigen 1 (EBNA1) and HERV-W env demonstrate binding to the HLADR2 allele that is the strongest genetic predisposition to MS (DRB1(*) 15:01) 25-29 and exhibit cross reactivity with myelin components 27,30. EBV specific HLA1 responses are more likely in MS patients and these patients also have EBV-specific memory T cells in their cerebrospinal fluid 31. BCR sequencing of the B cell complement combined with screening of the sequenced antibodies against EBV and CNS proteins of MS patient and controls has demonstrated clonal amplification of EBNA1 and GlialCAM (a protein and chloride channel regulator in glial cells important in CNS repair mechanisms) cross reactive B cells in the PMBC and CNS of MS patients 32. A similar study demonstrated antibodies cross reactive to both EBV and alpha-crystallin B (CRYAB), a molecular chaperone protein involved in glial responses to injury) enriched in MS patients compared with healthy controls 33. Genome wide association studies (GWAS) and transcriptome studies of MS patients have repeatedly indicated antiviral proteins as risk factors in disease occurrence and progression 5,34,35. Clinical trials of T cell therapy specifically targeting EBV have even begun 36 with promising early results for both clinical improvement and decrease in EBV antibody titre.
It is unclear whether these viruses initiate a triggering event creating an aberrant immune response or B cell type that perpetuates itself in the absence of the viral trigger or whether chronic or high viral loads are part of the underlying pathology. There are a range of studies demonstrating that antibody and T cell responses to EBV are consistently higher in MS patients than controls and that these are elevated during relapsing phases of RRMS 25,30,37-46. However, EBV nucleic acid in the blood or shed in saliva is usually not associated with MS 39,42,47-51 though it can be detected in CNS/Brain samples 52.
EBV establishes life-long latency in a subpopulation of memory B cells and there are strong indications that aberrant latency programming in EBV infected cells, indicated by the presence of the EBNA2 protein may be an important factor in the development of MS 53. B cell depletion therapies that broadly target B cell such as cladribine, anti CD-52 antibodies (alemtuzumab) and anti CD-20 antibodies (ocrelizumab, rituxumab) have proven effective in control of clinical disease in MS 54-57. In some cases these therapies also result in decreased EBV antibody titre and cellular immune responses 45,58-61. Those therapies that specifically target naïve and plasma B cells (atacicept) or boost memory T cells such as infliximab (anti TNF-alpha antibody) or lenercept on the other hand enhance disease 54,56,62.
Generation of spontaneous lymphoblastoid cell lines (transformed EBV infected B cells) is more common in MS patients (and in other autoimmune diseases) than in healthy controls 63,64 and genetic variation in EBV latency associated proteins 65 in MS patients has been demonstrated. Expression of the latency-associated protein, EBNA1, is enhanced in B cells from younger patients 16 while “age-associated” B cells (which are expanded in older patients) are also expanded in MS patients and altered based on herpesvirus status 66. This B cell subset are T-bet/CXCR3 + memory B cells that skew immune responses to a Th1 (viral and intracellular pathogen) cellular immune response. They are neuroinvasive and are associated with EBV reactivation 67. This subset of cells can be induced by an atypical latency programme in EBV infected B cells 68 and are currently a key suspect in the cellular triggers of MS.
This study sought to address whether targeting the primary site of viral antigen production for EBV and HERV-W proteins (peripheral B cells) by depletion with the monoclonal antibody Ocrelizumab, which specifically targets the B cell surface protein CD-20 2 reduces viral load and is thereby associated with a reduction in MS pathology and clinical disease.
Materials and Methods
Ethics approval was granted by the University of Nottingham, Faculty of Medicine and Health Sciences Research and Ethical Committee number: MREC 08/H0408/167.
Twenty patients with RRMS or PPMS were recruited prior to beginning Ocrelizumab therapy (Table 1 and Supplementary Information). Patients were recruited from those undergoing routine therapy for MS through the Nottingham University Hospitals NHS trust, including routine clinical assessment of clinical activity (and usually one MRI brain scan a year). Recruitment criteria aimed to be as even as possible while remaining representative of typical MS patients treated with Ocrelizumab within the timeframe of the study. Patients were between 18 and 65 years old, 75% female and 25% male, 75% with RRMS and 25% with PPMS. Patients had no history of other disease modifying therapies (DMTs) (19 patients) or no DMT during the previous 3 months and no treatment with high-dose steroids for MS relapse within the last 30 days (1 patient). This patient had had one dose of Copaxone (glatiramer acetate) several years previously. Seventy five percent of the patients were recently diagnosed (<1 year). A minimum sample size of 13 (pre and post treatment) was estimated for demonstrating significant differences in transcriptomic studies with an FDR of 0.5 and an expected fold change of 4 69.
Five ml of blood was collected into PaxGene tubes (Qiagen). A follow up blood sample was taken 5 months later, when patients had received their first 2 infusions and were reviewed by the MS Team in preparation for the third infusion at 6 months. Blood samples were stored at -20 °C until RNA extraction. RNA extraction was performed with a Paxgene blood RNA kit (Qiagen) as per manufacturer’s protocol. Illumina NovaSeq RNA sequencing with polyA library preparation (150 base pair, paired end reads) was performed by Novogene UK.
The resulting data files were trimmed (quality score 30, min length 150 and adapters removed) with FASTP 70. Mapped to the unmasked ensembl version of the human genome (GrCh38) with Salmon 71. Additional mapping was performed to the EBV reference genome (NC 007605) and a custom Human endogenous retrovirus databases (HERVd) 72. Differential gene expression analysis was performed with the DESeq2 73 pipeline implemented in iDEP 74 with the parameters: false discovery rate 0.1, min fold change 2, model: treatment and patient. Hierarchical clustering and heat maps were also generated in iDEP.
Results
Fourteen genes, of which 13 were clearly linked to B cell function, were downregulated in the post treatment samples; no genes were upregulated. Downregulated genes were: IgG chains (IGHM, 2 variants of IGHD, IGHG2), CD79A and CD79B (part of the B cell receptor complex), CD75 (part of the MHC class II antigen presentation complex), BLK (B lymphocyte tyrosine kinase), MS4A1 (B lymphocyte surface molecule involved in B cell differentiation), VPREB3 (pre assembly of B cell receptor), TCL1A (T cell receptor activator, FCER2 (immunoglobulin E receptor – B cell growth factor) and an unknown transcript (ENSG00000288133).
Hierarchical clustering of sequencing data demonstrated very clear clustering by patient ID (Figure 1). This was a much stronger effect than any other factor in the study (including pre or post treatment, age, sex, or type of disease).
Heat map and Dendogram of hierarchical clustering of patient samples, samples are coloured by patient number (dendrogram) and named by Ocr_patient number_sample number (first or second visit) unique sample ID (two samples were run twice)_unique run ID. Green=lower expression, red=higher expression.
Only one patient had detectable EBV reads (in a pre-treatment sample). Reads for many different HERV groups were detected but there was no differential expression of HERVs in any grouping of patient samples.
Discussion
MS is a complex disease with the triggering events possibly occurring many years before clinical disease onset and no single antigen target identified 75. Between-patient variation is large, necessitating very large epidemiological studies to pin down causal associations like the role of EBV in disease development 6. This is reflected in this study where the strongest factor apparent in hierarchical clustering was the individual patient with pre and post treatment samples from the same patient clustering closely (Figure 1). Other factors such as age, sex, type of disease, time since diagnosis or onset of disease had no statistically evident effect on gene expression in this cohort.
With individual patients controlled for in the differential expression model applied to the transcriptomics data the effects of Ocrelizumab B cell depletion were remarkably targeted to a small number of genes (14) with clear B cell associated functions. There have been a small number of other studies using comparable methods and patient cohorts. Fong et al 2023 76 looked at pre and post treatment gene expression in PBMC from 15 Ocrelizumab treated MS patients at 2 weeks and 6 months post therapy compared with 10 healthy controls, 10 untreated MS patients and 9 MS patients treated with Interferon beta, using microarrays. Perhaps unsurprisingly with a more complicated and less controlled cohort and statistical analysis a much larger number of differentially expressed genes (413 decreased and 184 increased) were identified in their study. Similar to our study however the pathways identified were primarily B cell related and 6 of the same genes were identified, namely CD79A, CD22, CD79B, MS4A1 and two IGHD variants. CD22, CD79A and CD79B can be downregulated by EBNA2 and EBNA3 proteins and this affects BCR signalling 77. This was suggested to be an additional way through which EBV maintains viral latency and controls the survival of infected B cells 77.
Measuring differential expression of HERVs in RNAseq data is not straightforward due to the repetitive nature of transposons, making them not readily distinguishable in some mapping algorithms 78. Mapping success is also heavily dependent on the database chosen as the reference sequence. Schwarz et al 78 compared the existing algorithms used for this kind of work, TEtranscripts 79, SalmonTE 80, Telescope 81, SQuiRE 82and TEtools 83 against a known test dataset and found that SalmonTE and Telescope performed reliably. As the original SalmonTE programme is not currently curated, we recreated its functionality using the Salmon mapping algorithm 71 and the HERVd 72 database of human ERVs, currently the most comprehensive curated database of HERVs. The same approach was taken to EBV mapping. Nali et al 84 looked at differential expression of HERVs in 7 MS patients with secondary progressive MS and 3 healthy control PBMCs with Illumina RNAseq and a very similar bio-informatics pipeline but had quite different results to us. HERV-W and 18 additional HERV families were upregulated in MS patients. These differences however probably reflect the different disease stage, smaller number of samples and choice of control in their study (pre-treatment from the same patient in ours vs healthy control in Nali et al84).
Similar results to ours indicating a lack of differential expression of HERV transcripts (by RNAseq) in the brains of MS patients and controls were reported by Ekljaer et al 2021 85. As EBV detection in blood is low in normal adults and MS patients 48 it is perhaps not surprising that we only detected this in one pre-treatment patient. A lack of differential expression of any HERV family was however unexpected given our original hypothesis that Ocrelizumab depletion of B cells would decrease the opportunity for viral expression in MS patients. This study does however offer support to the body of evidence that suggests that an aberrant cell type or immune response triggered by these viruses, rather than ongoing high viral expression, drives continued pathology in MS. Current evidence points strongly to an aberrant EBV latency programme and resulting in an expanded T-bet/CXCR3 + memory B cell population that is critical in MS pathology 63,68.
Data Availability
RNAseq reads are available at Genbank Bioproject PRJNA1001267 Accession numbers SRR25490470-SRR25490511
Role of Authors
RTar and BG conceptualised the study, BG and RTan recruited patients, RTar performed the RNA extraction and bio-informatics analysis and drafted the manuscript, All authors reviewed and edited the manuscript.
Declaration of Interest and Role of Funders
Funding for this project was provided by Roche UK who played no role in study design, execution, analysis or publication. RTan received support from the UK MRC (CARP MR/T024402/1). The authors declare no other interests.
Footnotes
Radu.tanasescu{at}nottingham.ac.uk
Bruno.gran{at}nottingham.ac.uk
C.ShannonLowe{at}bham.ac.uk
References
