Fig. 4 |. Severe COVID-19 correlates with SLE-like activation of the EF pathway.

a, Heat map of secondary population frequency z-scores by outpatient (purple; n = 7), ICU (green; n = 6) or deceased (red; n = 4) patients who tested positive for COVID-19. Multivariate clustering of patients by Ward’s method is represented by dendrograms. Clusters were designated as CoV-A and CoV-B for downstream analysis. Black boxes highlight transitional B cells or populations previously (DN2, aN) or currently (DN3) implicated in EF responses. The red box and dagger indicate patients analyzed by single-cell V(D)J analysis in Fig. 5. preMZ, precursors of marginal zone B cells. b, Patient sample collection times following symptom onset in CoV-A (n = 9) and CoV-B (n = 8) clusters. c, Representative plots of DN population composition in HD, CoV-A, CoV-B and SLE patient groups. d, DN composition analysis in HD (n = 17), CoV-A (n = 9), CoV-B (n = 8) and SLE (n = 7) patient groups. e, Outer ring represents the mean DN population composition of patient groups. Inner ring represents the mean DN2:DN1 ratios of patient groups. f, DN2:DN1 ratios in HD, CoV-A, CoV-B and SLE patient groups. g, usM frequency of CD19⁺ B cells in HD (n = 17), CoV-A (n = 9) and CoV-B (n = 8) groups. h, Homing receptor surface expression in follicular (rN and DN1) versus EF (aN and DN2) populations observed in CoV-A (n = 9) patients. b, Differences between groups were analyzed by a two-tailed Student’s t-test. d–h, Statistical significance was determined using ANOVA with Tukey’s multiple-comparisons testing between groups. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001; NS, not significant. Summary statistics (b–g): mean ± s.d.