Clonally diverse CD38+HLA-DR+CD8+ T cells persist during fatal H7N9 disease

Abstract

Severe influenza A virus (IAV) infection is associated with immune dysfunction. Here, we show circulating CD8⁺ T-cell profiles from patients hospitalized with avian H7N9, seasonal IAV, and influenza vaccinees. Patient survival reflects an early, transient prevalence of highly activated CD38⁺HLA-DR⁺PD-1⁺ CD8⁺ T cells, whereas the prolonged persistence of this set is found in ultimately fatal cases. Single-cell T cell receptor (TCR)-αβ analyses of activated CD38⁺HLA-DR⁺CD8⁺ T cells show similar TCRαβ diversity but differential clonal expansion kinetics in surviving and fatal H7N9 patients. Delayed clonal expansion associated with an early dichotomy at a transcriptome level (as detected by single-cell RNAseq) is found in CD38⁺HLA-DR⁺CD8⁺ T cells from patients who succumbed to the disease, suggesting a divergent differentiation pathway of CD38⁺HLA-DR⁺CD8⁺ T cells from the outset during fatal disease. Our study proposes that effective expansion of cross-reactive influenza-specific TCRαβ clonotypes with appropriate transcriptome signatures is needed for early protection against severe influenza disease.

Fig. 1

Prolonged expression of CD38⁺HLA-DR⁺ on CD8⁺ T cells in fatal H7N9 cases. The frequency of CD38⁺HLA-DR⁺CD8⁺ T cells and IFNγ/TNF production was assessed in the peripheral blood of patients hospitalized with severe H7N9 infection at different times after infection. a Representative FACS plots are shown for IFNγ/TNF production, cell gating shown in the previous report⁵, and CD38⁺HLA-DR⁺ expression by CD8⁺ T cells for patient a79 (recovered from H7N9) and patient a33 (died from H7N9). The gating strategy is shown in Supplementary Figure 7. b–i Data are graphed for all the patients available for this study: eight “recovery” patients (open symbols) and three patients with fatal disease outcomes (solid symbols) as b frequency of CD38⁺HLA-DR⁺ expression in CD8⁺ T cells; c numbers of CD38⁺HLA-DR⁺ CD8⁺ T cells in 1 ml of blood; d numbers of IFNγ⁺CD8⁺ T cells in 1 ml of blood; e frequencies of activated CD38⁺HLA-DR⁺CD8⁺ T cells, as mean values of all the assayed time-points, are shown for patients who survived versus patients who died; f numbers of CD38⁺HLA-DR⁺CD8⁺ T cells (data from all the assayed time-points) in 1 ml of blood were analyzed according to the disease outcome; g numbers of IFNγ⁺CD8⁺ T cells in 1 ml of blood (data from all the assayed time-points) were analyzed according to the disease outcome; h, i binary comparisons between the survival and fatal groups were performed for the hospital stay time (in days). h Numbers of activated CD38⁺HLA-DR⁺CD8⁺ T cells in 1 ml of blood, as peak values of all the assayed time-points, are shown for patients who survived versus patients who died; and i numbers of activated IFNγ⁺CD8⁺ T cells in 1 ml of blood, as peak values of all the assayed time-points, are shown for patients who survived versus patients who died. Analyses were implemented using MATLAB (version R2014b; the MathWorks, Natick, MA). Error bars represent standard errors of the mean. Standard two-tail Student’s t-test was performed with p value as indicated

Fig. 2

Dynamics of CD38⁺HLA-DR⁺PD-1⁺CD8⁺ cells segregate H7N9 disease outcome. a Representative FACS plots of PD-1 expression within the CD38⁺HLA-DR⁺CD8⁺ T cells (in red) and the total CD8⁺ T cells (in blue) in patient a79 (recovered from H7N9) and patient a33 (died from H7N9). b Correlation between the frequency of CD38⁺PD-1⁺CD8⁺ T cells and a proportion of CD38⁺HLA-DR⁺CD8⁺ T cells, as measured by Spearman test. c Representative histograms of PD-1 expression within CD38⁺HLA-DR⁺CD8⁺ T cells (red) and the remaining non-CD38⁺HLA-DR⁺ CD8⁺ T cells (blue). d High and prolonged expression of PD-1 in patients who died, as triple-positive CD38⁺HLA-DR⁺PD-1⁺ and mean fluorescence intensity (MFI) of PD-1 expression within CD38⁺HLA-DR⁺CD8⁺ T cells and non-CD38⁺HLA-DR⁺CD8⁺ T cells for survivals and fatal cases (a standard two-tail Student’s t-test), with samples from individual patients being acquired on an LSR Fortessa on different days. e CD45RA and CD27 profiles on CD38⁺PD-1⁺CD8⁺ T cells (in red) and on total CD8⁺ T cells (in blue). f Similar to CD38⁺HLA-DR⁺CD8⁺ T cells, the frequency of CD38⁺PD-1⁺CD8⁺ T cells declined with time in the H7N9 survival group, which was maintained or increased in H7N9 patients who died. The frequency of CD38⁺PD-1⁺CD8⁺ T cells was also analyzed in g patients hospitalized with seasonal pH1N1 (5 patients; dotted lines, solid triangles), H3N2 (10 patients in dashed lines, open diamonds), influenza B (6 patients in dotted black lines, open triangles), non-influenza/non-viral hospitalized patients (5 patients; gray lines, open gray circles), and 1 H3N2 patient with fatal outcome (black line, solid black circle); and h healthy individuals during inactivated influenza vaccination in 2015 (18 individuals vaccinated with a trivalent (TIV) vaccine; black lines) and 2016 (27 individuals vaccinated with quadrivalent (QIV) vaccine; gray lines). i Frequency of CD38⁺PD-1⁺CD8⁺ T cells during the first (T₁) and last time-point (T_L) of hospitalization or vaccination in H7N9 survival cases, fatal influenza patients (3 H7N9, 1 H3N2); seasonal influenza survival cases, and seasonal influenza vaccinees. Means are shown as horizontal bars. Standard two-tail Student’s t-test was performed with indicated p values

Fig. 3

TCRαβ repertoire diversity and stability within A2-M1₅₈⁺CD8⁺ T cells in H7N9. a FACS profiles are shown for A2-M1₅₈⁺CD8⁺ T cells after tetramer staining directly ex vivo for a9, a10, and a79 patients. A2-M1₅₈⁺CD8⁺ T cells were single cell-sorted for TCRαβ repertoire analysis using a multiplex RT-PCR and sequencing. Populations were gated on viable Dump⁻tetramer⁺CD3⁺CD8⁺ events. Circos plots of frequencies of Vβ-Jβ (in black) and Vα-Jα (in blue) usage in paired TCRαβ sequences are shown for b memory A2-M1₅₈⁺CD8⁺ T cells in five healthy donors at one time-point or c–e activated A2-M1₅₈⁺CD8⁺ T cells in hospitalized patients (a9, a10, and a79) at different time-points after influenza infection. The width of the band is proportional to the frequency and each band represents a unique clone. Red segments represent private dominant A2-M1₅₈-specific TCR clones, black segments represent public A2-M1₅₈-specific TCR clones. Blue segments represent private dominant A2-M1₅₈-specific TCR clones present across different time-points in the H7N9 patients. Prominent TRBV19 and TRAV27 segments are underlined. Circos plots were generated with the Circos software package

Fig. 4

Diverse usage of TCRα–TCRβ pairings within CD38⁺HLA-DR⁺CD8⁺ T cells. The frequency of TCRαβ clonotypes within particular TRAV–TRBV paired segments is shown for individual patients: a non-HLA-A*02:01 surviving individuals; b HLA-A*02:01 surviving individuals; c patients with a fatal disease outcome. Pulled data from all the time-points is shown, as summarized in Supplementary Table 5. d The number of specific TRBV–TRAV paired segments utilized per donor is shown for non-HLA-A*02:01 surviving patients, HLA-A*02:01 surviving patients and fatal patients for CD38⁺HLA-DR⁺CD8⁺ T cells as well for H7N9 surviving patient and healthy individuals for A2-M1₅₈⁺CD8⁺ T cells

Fig. 5

Distinct clonal expansion patterns within CD38⁺HLA-DR⁺CD8⁺ T cells in H7N9. Longitudinal single-cell TCRαβ repertoire analysis of CD38⁺HLA-DR⁺CD8⁺ T cells reveals delayed early clonal expansions within fatal patients. For comparison, TCRαβ characteristics of M1₅₈⁺CD8⁺ T cells (directed at a single influenza T cell specificity) are shown. a Simpson’s Diversity Index (SDI) shows differential patterns of TCRαβ clonal expansions longitudinally within CD38⁺HLA-DR⁺CD8⁺ T cells in patients who survived (in green) and died (in pink). SDIs for survival M1₅₈⁺CD8⁺ T cells are shown in blue. b Principal component analyses of Vα-chain segments, Jα regions, Vβ-chain segments, and Jβ segments (panels from left to right) were performed to identify increased clustering of TCRαβ elements for survival patients but more distinct TCR signatures for fatal cases, based on TCRdist analysis. The size of specific dot points reflects the size of specific TCR genes used. c V and J gene segment usage and pairing landscape is shown by average-linkage dendrogram of TCRdist clusters for Jα/Vα-chain segments and Jβ/Vβ-chain segments (panels from left to right). Segments are colored based on the frequency with which they occur within the repertoire with a color sequence, beginning with red (most frequent), then green (second most frequent), blue, cyan, magenta, and black. d CDR3 length against prominent Vα-chain segments, Jα regions, Vβ-chain segments, and Jβ regions is shown for CD38⁺HLA-DR⁺CD8⁺ T cells within both the surviving and fatal H7N9 patients, as well as M1₅₈⁺CD8⁺ T cells within the surviving patients. The left right ordering of the segment types is chosen so that VA and JA are on the left, VB and JB are on the right, and the alpha–beta pairing with the largest adjusted mutual information is in the middle

Fig. 6

Global differences at the transcriptome levels within CD38⁺HLA-DR⁺CD8⁺ cells. Single-cell RNAseq (scRNAseq) was performed on CD38⁺HLA-DR⁺CD8⁺ T cells from longitudinal samples from two subjects, surviving a11 (d15, d29) and fatal a33 (d12, d31) patients. a Unsupervised principal-component analysis (PCA) of CD38⁺HLA-DR⁺CD8⁺ T cells reveals a clear segregation across the two patients, especially between survival (a11) and fatal (a33) at the early time-points (a11, d15 and a33, d12). Additional information between expanded versus non-expanded TCRαβ clonotypes (further defined in Supplementary Table 7) is shown using different symbols: triangles for expanded TCRαβ clonotypes and circles for non-expanded TCRαβ clonotypes; b RNAseq data were used to assess the similarity of CD38⁺HLA-DR⁺CD8⁺ T cells within each patient across different time-points. The average Pearson’s correlation coefficient was calculated for each pair of CD38⁺HLA-DR⁺CD8⁺ T cells within the same time-point using FPKM values for all detected genes, revealing cell-to-cell similarity over time for surviving patient a11, and a marked loss of similarity (or increase in transcriptomic heterogeneity) in fatal patient a33 in the late stage of infection. c Unsupervised hierarchical single-cell consensus clustering (SC3) heatmap plot highlights the expanded TCRαβ clonotypes, patient, and time-points with fixed colors and legend. Differentially expressed genes and a priority list of genes associated with exhaustion show four clusters segregating cells according to the patient and the time-point (survival a11, d15, and d29; fatal a33, d12, and d31). FPKM: fragments per kilobase of transcript per million mapped reads