Specificity, frequency, and phenotype of citrullinated-specific T cells vary with disease activity in rheumatoid arthritis
Cliff Rims, Hannah A. DeBerg, Sylvia E. Posso, Virginia S. Muir, Hannes Uchtenhagen, Anne M. Hocking, Heather Bukiri, Jeffrey Carlin, Bernard Ng, Peter S. Linsley, Eddie A. James, Jane H. Buckner
Cliff Rims, Hannah A. DeBerg, Sylvia E. Posso, Virginia S. Muir, Hannes Uchtenhagen, Anne M. Hocking, Heather Bukiri, Jeffrey Carlin, Bernard Ng, Peter S. Linsley, Eddie A. James, Jane H. Buckner
View: Text | PDF
Research Article Immunology

Specificity, frequency, and phenotype of citrullinated-specific T cells vary with disease activity in rheumatoid arthritis

Abstract

In rheumatoid arthritis (RA), CD4+ T cells specific for citrullinated antigens (cit-antigens) are key drivers of disease, but knowledge about epitopes and phenotypes remains limited. We characterized the frequency and phenotype of cit-specific CD4+ T cells in peripheral blood using HLA class II tetramers combined with computational analysis of phenotypic clusters to simultaneously detect peptides derived from 5 cit-antigens (aggrecan, vimentin, fibrinogen, cartilage intermediate layer protein, and α-enolase) previously implicated in RA pathogenesis. In a cross-sectional cohort, cit-aggrecan–, cit-vimentin–, and cit-fibrinogen–specific T cells were more frequent in participants with RA than healthy volunteers, associated with active disease, and had Th1-like and stem-like lineages in RA. In a longitudinal cohort investigating response to therapy, the frequency of cit-aggrecan–, cit-vimentin–, and cit-fibrinogen–specific CD4+ T cells was significantly higher at baseline and further elevated in responders. Furthermore, the frequency of cit-specific Th1-like cells in responders decreased over time. In contrast, the frequency of Th1-like cells in non-responders increased over time. Collectively, these findings demonstrate that cit-specific CD4+ T cells are expanded in RA and target a broad number of antigens across a breadth of phenotypes. Furthermore, the predominant antigen specificities associate with disease activity and exhibit dynamic changes in phenotype that reflect response to therapy.

Authors

Cliff Rims, Hannah A. DeBerg, Sylvia E. Posso, Virginia S. Muir, Hannes Uchtenhagen, Anne M. Hocking, Heather Bukiri, Jeffrey Carlin, Bernard Ng, Peter S. Linsley, Eddie A. James, Jane H. Buckner

×

Figure 4

The frequency of cit-specific Th1-like cells is decreased individuals who respond to therapy.

Options: View larger image (or click on image) Download as PowerPoint
The frequency of cit-specific Th1-like cells is decreased individuals wh...
The frequencies of cit-aggrecan– and cit-VimFib–specific T cells were combined from RA participants who had blood draws before and after starting a new therapy. At first draw, all individuals had active disease based on a weighted wRAPID3 score of 2.3 or higher. These individuals were subsequently defined as Responder or Non-Responders based on their weighted RAPID3 at second draw (Responder wRAPID3 < 2.0, Non-Responder wRAPID3 ≥ 2.3; see Supplemental Figure 11). (A) Frequency of cit-Agg– and cit-VimFib–specific CD4+ T cells in Responders (n = 9) and Non-Responders (n = 11); 2-tailed non-parametric Wilcoxon test. (B) The frequency of cit-Agg– and cit-VimFib–specific CD4+ T cells in Responders (n = 20) and Non-Responders (n = 33) was normalized to initial draw and then plotted against the number of months thereafter; simple linear regression model, 95% confidence intervals, and 2-tailed Spearman’s correlation where non-responders P = 0.0020 and slope compared to responders calculated as P = 0.0098. (C) The baseline frequency of cit-antigen–reactive Th1-like (EM CXCR3+) cells in responders (N = 9) were compared to non-responders (N = 11); 2-tailed non-parametric Wilcoxon test. (D) The frequency of cit-antigen–reactive Th1-like (EM CXCR3+) cells in Responders (n = 20) and Non-Responders (n = 33) was normalized to initial draw and then correlated with months after baseline draw prior to new therapy; non-parametric Spearman correlation was used to determine correlation to time (responder P = 0.044, non-responder P = 0.018) and simple linear regression was used to determine slope differences between the 2 groups (P = 0.0027). For C and D, data points are those samples where parent event count is at least 8. Graphs A–D are log2(x +1) transformed. *P ≤ 0.05, ***P ≤ 0.001.