B cell transcriptomics and immunoglobulin genetics in rheumatoid arthritis

Abstract

Rheumatoid Arthritis (RA) is a complex autoimmune disease characterized by inflammation of the peripheral joints. The exact disease pathology remains to be elucidated, but it has been hypothesized that autoreactive B cells and autoantibodies play a major role in the etiology. In this thesis, I studied synovial B cells at the time of RA diagnosis using single cell RNA sequencing, spatial transcriptomics and isolation of monoclonal antibodies. In a separate track, I optimized Next Generation Sequencing library preparation methods to study antibody repertoires, infer germline antibody alleles and I applied these methods to a local disease study group.

Using single cell RNA sequencing and spatial transcriptomics from patient paired tissue pieces, we found mostly memory B cells and naïve B cells in lymphoid rich biopsies. CD27++ plasma cells were found to have a strong identity signal as shown by both methods. Similarily, both methods showed a strong signature for T-B cell crosstalk that we discuss to drive differentiation to memory and plasma cells. Furthermore, we found the CXCL12-CXCR4 axis to be an active component of a plasma cell niche in synovial tissue. Studies of the B cell receptor from the single cell RNA sequencing data imply a clonal recall response from memory cells differentiating into plasma cells due to recurrent antigen exposure. We also identified an anticitrullinated protein reactivity towards modified vimentin in the tissue biopsies.

Moreover, we compared 5’ rapid amplification of cDNA ends (5’RACE) antibody library preparation methods to a 5’ multiplex (5’MTPX) approach. We detected a 5’RACE amplicon length limitation that occurs when the sequences of the antibody heavy chain repertoire have long 5’ untranslated regions (UTRs) or long complementary determining regions 3 (CDR3s). Furthermore, we tested a single lambda constant chain primer versus a mix of single lambda constant chain specific primers and found that there was good detection of independent variable-joining segment (V-J) recombination that allows identification of lambda chain germline alleles.

Finally, we used the 5’MTPX approach to study the antibody heavy chain alleles of thirty individuals from the local Epidemiological Investigation of Rheumatoid Arthritis (EIRA) study. We found common structural variations as well as few novel alleles. In haplotype analysis, we observed that the IGHV4-34 gene was seemingly hemizygous. This could be linked to a polymorphism in the recombination signal sequence of one chromosome leading to the absence of the allele in the expressed repertoire. Notably, this variation was also found in thirty control subjects from the EIRA study group and in the SNP data from the Finnish population group in the 1000 Genomes Project. Hence, we stress the importance of population stratification when performing disease association studies.

In summary, I discuss RA therapy with respect to T-B cell interaction and a plasma cell survival niche, suitable antibody library generation methods for deep repertoire studies and germline gene inference studies, and the relevance of this for larger disease association studies.