Background/Purpose: Vimentin is a dominant target of B-cells selected in lupus tubulointerstial inflammation (TII), a predictor of renal failure. The origins of anti-vimentin antibodies (AVA) are unknown. Technologies quantifying specific antigenic targeting do not exist. Objectives:₁ Characterize reactivities of germline BCR precursors of TII AVAs; ₂ Develop software to quantify mAb reactivity with subcellular regions of cultured cells and tissue; ₃ Quantify selectivity of mAbs for vimentin vs other cellular antigens; ₄ Determine somatic hypermutations (SHMs) influential in Vimentin selectivity.

Methods: 10 human IgG1 AVA mAbs (from 6 TII patients) in selected (sel) and germ-line reverted (rev) forms, and 21 mAbs from naïve healthy B-cells were generated. Recombinant human vimentin (native and citrullinated) reactivity was assayed by ELISA. HEp-2 cells were co-stained with TII AVAs (sel and rev) and subcellular markers (nuclear [Hoechst], vimentin [mAb V9] and cytoplasm [mAb eno-1]), and imaged by multicolor immunofluorescence laser scanning confocal microscopy. Single amino acid variants of TII mAbs “PB4” (n=16) and “PB5” (n=11), representing single SHM reversions were also assayed. Lupus TII tissue was co-stained with novel FLAG-tagged variants of PB4 and PB5 (sel and rev). Mean pixel intensity (MPI) was a measure of magnitude of human mAb reactivity. Vimentin selectivity was quantified by co-variance of binding (Pearson Correlation coefficient of MPI with anti-vimentin channel), or ratio of MPI of vimentin positive : other respective subcellular region. A machine learned software (“CytoSkaler”) was developed to identify individual cultured cells, and subcellular areas, allowing measurement of human mAb reactivities (MPIs) with subcellular regions. CytoSkaler was also used for quantifying reactivity with subcellular regions of TII tissue.

Results: 9/10 reverted TII AVA mAbs had reduced vimentin reactivities by ELISA, comparable to those of naïve healthy B-cell derived mAbs. All TII mAbs had lower reactivities with citrullinated antigen. CytoSkaler yielded highly accurate cell segmentation estimations (accuracy =0.97, intersection of union =0.96), allowing high throughput quantification of human mAb reactivities with subcellular regions. Typically, CytoSkaler demonstrated progression from low magnitude, broad cellular reactivity at germline, to high preference, and magnitude of reactivity, for vimentin positive regions following affinity maturation (i.e “PB4”). MAb PB5 was unique, maturing from a germline precursor with a speckled nuclear pattern and purified histone reactivity. Staining patterns observed with TII tissue using Flag-PB4 and -PB5 matched those generated with HEp-2 cells. Reactivity of PB4 with vimentin was due to a combination of multiple SHMs. In PB5, S36D reversion was responsible for nuclear (Histone) reactivity and G55D lost vimentin reactivity.

Conclusion: From low reactivities comparable to naïve healthy B-cells, or Histone reactivity, TII AVAs gain both affinity and selectivity for vimentin. We have developed image analysis software for quantifying mAb selectivity for candidate antigens, and a flag human IgG1 mAb construct enabling analysis of human tissue rich in IgG.

« Back to 2019 ACR/ARP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-novel-image-analysis-program-cytoskaler-demonstrates-that-anti-vimentin-antibody-affinity-maturation-in-lupus-tubulointerstitial-nephritis-also-results-in-more-selective-antigen/