Background

B cell depleting therapies are highly effective treatments for multiple sclerosis (MS). B cells are more numerous in active than inactive lesions, and their intrathecal clonal expansion and oligoclonal band production are hallmarks of MS. B cells also present antigens to T cells and secrete inflammatory cytokines. The antigenic specificity of individual B cells in cerebrospinal fluid (CSF) obtained from patients with early MS may help further clarify the role of B cells in MS biology.

Objectives

To determine the viral and autoantigen repertoire of CSF-derived, class-switched B cells from untreated, early MS patients.

Methods

We performed single cell immunoglobulin sequencing on CSF plasma cells, plasmablasts, and class switched memory B cells from 9 untreated patients: five with relapsing remitting MS (RRMS) and four with clinically isolated syndrome (CIS). The interval between the first attack and lumbar puncture ranged from 1 - 222 days (median 67 days). Brain and spinal cord MRIs performed concurrently with lumbar punctures revealed 5/9 patients with gadolinium enhancing lesions.

Using paired heavy and light chain immunoglobulin sequences, we generated 75 monoclonal antibodies (mAbs) and screened them on a suite of unbiased antigen discovery platforms: 1) mouse brain tissue staining, 2) whole human proteome programmable phage display, 3) pan-viral programmable phage display, 4) mouse and human brain immunoprecipitation mass spectrometry.

Results

The mAbs showed diverse antigen specificities. Candidate antigens were primarily ubiquitously expressed, intracellular proteins; however, a minority were macromolecules associated with the plasma membrane and/or enriched in brain tissue. Shared antigenic targets were occasionally identified within subjects but were rarely identified across subjects, with the latter including cytoskeletal proteins. For two mAbs, high-confidence antigens with prima facie relevance to MS were identified: 1) a white matter-restricted lipid species, and 2) an Epstein-Barr virus-interacting host protein.

Conclusions

Using our panel of 75 mAbs derived from plasma cells, plasmablasts, and class-switched memory B cells found in the CSF of early, untreated RRMS/CIS patients, we identified a diverse repertoire of antigenic targets, with a majority comprised of intracellular host proteins.

Background

The International Multiple Sclerosis Genetics Consortium (IMSGC) uncovered the contribution of inherited variants to multiple sclerosis (MS) in 115,801 individuals. Polygenic risk profiling intends to summarize and represent the genetic architecture of complex disorders and identify groups of individuals who can benefit from the knowledge of their increased susceptibility. In this context, it is important to explore the relationships between polygenic risk scores (PRS) in MS with disease status and severity in population-level and familial cohorts, in order to leverage potential clinical utilities.

Objectives

To develop and assess quantifiable measures of MS susceptibility and examine their association with phenotypic variability.

Methods

We employed both the Bayesian LDPred algorithm and Pruning and Thresholding to develop multiple MS-PRS from a multi-cohort GWAS comprising 41,505 participants. Models were validated in the UK Biobank phase 1 dataset and tested in both the UK Biobank phase 2 and the Kaiser Permanente Northern California (KPNC) MS datasets. PRS of families was tested in a cohort of 34 families with one affected parent and at least one affected child. Clinical phenotype data was used in the UCSF EPIC cohort including 742 MS patients. Standard quality control of the base (IMSGC) and target datasets was performed prior to final analyses.

Results

We observed a statistically significant difference between PRS distributions of cases and controls in both the UK Biobank and KPNC cohorts (P < 1e-70), and identified individuals at greater risk versus the rest of the population (OR > 3). We confirmed that an increased PRS in siblings of disease discordant parents is associated with a higher risk of MS and showed an enhanced power for disease prediction among siblings in a small cohort of 152 individuals. These results suggest that PRS metric shows promise for prediction of MS within sibships, but needs to be further tested in larger familial cohorts. The predictive prognostic value of PRS for selected MRI metrics and disability scores suggests that PRS modestly explain phenotypic variations.

Conclusions

Polygenic risk scores are currently the best estimate of the complex genetic architecture of MS and, when clinically implemented, could facilitate recognition and management of MS in early stages of the disease. These results provide a direction for translation of MS-GWAS studies into relevant biology and clinically meaningful outcomes.