RSV-specific airway resident memory CD8+ T cells and differential disease severity after experimental human infection

Abstract

In animal models, resident memory CD8+ T (Trm) cells assist in respiratory virus elimination but their importance in man has not been determined. Here, using experimental human respiratory syncytial virus (RSV) infection, we investigate systemic and local virus-specific CD8+ T-cell responses in adult volunteers. Having defined the immunodominance hierarchy, we analyse phenotype and function longitudinally in blood and by serial bronchoscopy. Despite rapid clinical recovery, we note surprisingly extensive lower airway inflammation with persistent viral antigen and cellular infiltrates. Pulmonary virus-specific CD8+ T cells display a CD69+CD103+ Trm phenotype and accumulate to strikingly high frequencies into convalescence without continued proliferation. While these have a more highly differentiated phenotype, they express fewer cytotoxicity markers than in blood. Nevertheless, their abundance before infection correlates with reduced symptoms and viral load, implying that CD8+ Trm cells in the human lung can confer protection against severe respiratory viral disease when humoral immunity is overcome.

Figures

Figure 1. Inoculation of healthy adults with RSV causes upper respiratory tract infection.

Forty-nine healthy adult volunteers were inoculated with RSV Memphis 37. (a) The study design is shown. (b) Clinical outcomes are shown. (c) Self-reported symptoms scores of 49 subjects were recorded daily. Cumulative data are shown as mean±s.e.m. Infected individuals were defined by having had RSV detected from nasal lavage on at least 2 consecutive days by qPCR of nasal lavage. (d) Symptoms from 49 subjects were categorized as originating from the upper respiratory tract (blue), lower respiratory tract (red) or as systemic (green) according to previously described criteria (see Methods). Data are shown as mean± s.e.m. (e) Viral load of 49 individuals was determined by N gene qPCR from nasal lavage. Individual data points and means are shown.

Figure 2. RSV infection of healthy adults causes widespread lower respiratory tract inflammation.

Twenty-four volunteers underwent serial bronchoscopic procedures 14 days before inoculation with RSV M37, 7 or 10 days and 28 days post inoculation. (a) The macroscopic appearance of the airways was noted by bronchoscopists blinded to the infection status of the subject. One representative subject is shown. (b) Viral load was determined from bronchial brushings (solid line) and bronchoalveolar lavage (dashed line) by qPCR. Cumulative data are shown as mean±s.e.m. (c) RSV antigen (brown) was detected in bronchial biopsies by immunohistochemistry. One representative donor is shown. (d) CD8+ cells (brown) were identified in bronchial biopsies by immunohistochemistry. Scale bar, 20 μm. (e) CD8+ cells were enumerated in bronchial biopsies of infected individuals (with RSV detected by qPCR in bronchial brushings) and PCR− individuals. Data are presented as number of positive cells per square millimetre of subepithelium or per 0.1 mm2 of epithelium. Significant P values for two-tailed Wilcoxon matched pairs tests are shown (*P<0.05, **P<0.01, ***P<0.001) comparing (with baseline) epithelial PCR+ samples at day 7–10 (P=0.0012) and day 28 (P=0.004), epithelial PCR− samples at day 7–10 (P=0.0469), subepithelial PCR+ samples at day 7–10 (P=0.0004) and day 28 (P=0.0419), and subepithelial PCR− samples at day 7–10 (P=0.0156).

Figure 3. RSV infection induces short-lived activation and proliferation of CD8+ T cells in peripheral blood.

Forty-nine adult volunteers were inoculated with RSV Memphis 37, serial blood (n=49) and BAL (n=24) samples were stained with anti-CD3, CD8, Ki-67 and CD38, and analysed by flow cytometry. (a) Plots are gated on CD3+CD8+ lymphocytes. Numbers represent percentage of CD8+ T cells. One representative subject is shown. (b) Mean and s.e.m. of Ki-67+CD38+ CD8+ T-cell frequencies in the blood of infected (PCR+) and challenged but uninfected (PCR−) are shown. The P value for a two-tailed Wilcoxon matched pairs test is shown (***P<0.001). (c) Frequencies of Ki-67+CD38+ cells as a proportion of CD8+ T lymphocytes were determined at day 10 post inoculation in infected and uninfected individuals. Mean±s.e.m. are shown with the P -value for a two-tailed Mann–Whitney test (***P=0.001). (d) Mean and s.e.m. of Ki-67+CD38+ CD8+ T-cell frequencies in the BAL of infected (PCR+) and challenged but uninfected (PCR−) are shown. The significant P value for a two-tailed Wilcoxon matched pairs test is shown (*P=0.0317). (e) Frequencies of Ki-67+CD38+ cells, as a proportion of CD8+ T lymphocytes were determined at day 10 post inoculation in matched blood and BAL samples. The significant P value for the Wilcoxon matched pairs tests is shown (**P=0.0059). Correlations between peak frequencies of Ki-67+CD38+ CD8+ T cells 10 days post infection in blood and (f) cumulative viral load and (g) symptoms (trapezoidal area under the curve (AUC)) are shown using non-linear regression and Spearman's rank correlation.

Figure 4. Immunodominance hierarchies in MHC class-I-restricted RSV epitope-specific CD8+ T-cell responses.

Epitopes were predicted in silico for binding and used to stimulate day 10 PBMCs from HLA-matched infected volunteers with sufficient cells (HLA A*01:01, n=4; A*02:01, n=5; and B*07:02, n=4) in interferon-γ (IFN-γ) ELISpot assays. (a) Peptides were pooled according to their originating RSV protein. Two-tailed Mann–Whitney tests were carried out between peptide pools and negative control wells (A*01:01_NS1/N, P=0.0286; A*02:01_NS1/N, P=0.0079; A*02:01_NS2/P, P=0.0119; and B*07:02_NS1/N, P=0.0286). (b) Individual peptides from positive pools were subsequently tested separately. Dotted lines represent a cutoff of 25 spots per 106 PBMCs. Means±s.e.m. and P values for unpaired Mann–Whitney tests are shown (A*01:01_NS1_2, P=0.0286; A*01:01_NS1_4, P=0.0286; A*01:01_M_2, P=0.0294; A*01:01_M_7, P=0.0294; A*02:02_NS2_3, P=0.0002; B*07:02_N_2, P=0.0079; B*07:02_N_8, P=0.0079; B*07:02_G_2, P=0.0278; and B*07:02_L_3, P=0.0079). (c) The median frequencies of IFN-γ+ cells in response to individual RSV epitopes presented by HLA-A*01:01 (blues), A*02:01 (reds) and B*07:02 (greens) are shown in 17 HLA-matched RSV-infected individuals at day 0, 10 and 28 post infection. (d) The frequencies of median total IFN-γ-producing cells at baseline, day 10 and day 28 post infection in response to HLA-A*01:01 (blue)-, A*02:01 (red)- and B*07:02 (green)-restricted epitopes in 17 HLA-matched infected individuals are shown. P values for Wilcoxon matched pairs tests are shown (A*01:01, P=0.0117; A*02:01, P=0.0039; and B*07:02, P=0.125; NS (not significant)=P>0.05, *P<0.05, **P<0.01, ***P<0.001).

Figure 5. RSV antigen-specific CD8+ T-cell kinetics diverge in BAL compared with blood.

Whole blood/PBMCs and BAL from RSV-infected individuals were co-stained with anti-CD3, CD8 and tetramers, and then analysed by flow cytometry. (a) Numbers represent the percentage of A1-M-YLE+ CD8+ T cells as a proportion of CD3+ lymphocytes at day 0, 7, 10, 14 and 28 post inoculation. Representative plots for one subject gated on CD3+ lymphocytes are shown. (b) The frequencies of A1-M-YLE (n=9), A2-NS2-FLV (n=10) and B7-N-NPK (n=8) tetramer-positive cells in blood as a proportion of CD8+ T cells are shown up to 6 months follow-up. P values for Wilcoxon matched pairs tests compared with day 0 are shown (A1-M-YLE day 10, P=0.0039; day 14, P=0.0039; day 28, P=0.0039; A2-NS2-FLV day 10, P=0.0488; B7-N-NPK day 10, P=0.0078; day 14, P=0.0223; and day 28, P=0.0207). (c) The frequencies of A1-YLE, A2-NS2-FLV and B7-N-NPK tetramer-positive cells in BAL as a proportion of CD8+ T cells are shown. P values for Wilcoxon matched pairs tests are shown comparing day 10 (P=0.0156) and day 28 (P=0.0002) frequencies with baseline. Abbreviated P values are shown: NS (not significant)=P>0.05, *P<0.05, **P<0.01. Mean±s.e.m. of epitope-specific CD8+ T-cell responses are shown in (d) blood (n=20) and (e) BAL (n=13).

Figure 6. RSV-specific CD8+ T cells in BAL display a distinctive resident memory phenotype.

Tetramer+ CD8+ T cells in blood and BAL were co-stained for markers to assess their differentiation status. (a) CD69 and CD103 as canonical markers of resident memory CD8+ T cells are shown in blood (n=9) and BAL (n=5) from infected volunteers. Significant P values for two-tailed Wilcoxon matched pairs tests in blood compared with baseline are shown (day 7, P=0.0313; day 10, P=0.0039; day 14, P=0.0313; and day 28, P=0.0313). (b) Memory markers CD45RA and CCR7 are shown in blood (n=19) and BAL (n=8). Mean±significant P values for two-tailed Wilcoxon matched pairs tests compared with baseline are shown in blood for T-effector/effector memory cells (day 7, P=0.0034; day 10, P=0.0002; day 14, P=0.0002; day 28, P=0.0067) and effector memory T cells re-expressing CD45RA (day 7, P=0.0443; day 10, P=0.0025; day 14, P=0.0003; day 28, P=0.0135). (c) Proliferation and activation markers Ki-67 and CD38 are shown in blood (n=19) and BAL (n=8). Significant P values for two-tailed Wilcoxon matched pairs tests are shown compared with baseline in blood (day 7, P=0.0025; day 10, P=0.0001; and day 14, P=0.0005) and BAL (day 7 versus day 10, P=0.0444; and day 10 versus day 28, P=0.0022 as no Ki-67+CD38+ cells were found in any baseline samples). Throughout, representative plots from a single subject at day 0, 10 and 28 post infection are shown with tetramer+ cells as red dots and total CD8+ T cells in grey contours. Summarized data are shown for all RSV-infected subjects (who could be analysed using tetramers) with median frequencies shown in red. Abbreviated P values are shown: NS (not significant)=P>0.05, *P<0.05, **P<0.01.

Figure 7. Reduced expression of cytotoxicity and co-stimulatory markers by RSV-specific CD8+ T cells in the airway.

Tetramer+ CD8+ T cells in blood and BAL were co-stained for phenotypic markers. (a) Cytotoxicity molecules perforin and granzyme B are shown in blood (n=19) and BAL (n=5). Significant P values for two-tailed Wilcoxon matched pairs tests compared with baseline are shown in blood (day 7, P=0.0144; day 10, P=0.0009; and day 14, P=0.0413). (b) Co-stimulatory molecules CD27 and CD28 are shown in blood (n=19) and BAL (n=10). Mean±s.e.m. and significant P values for two-tailed Wilcoxon matched pairs tests compared with baseline are shown in blood for CD27+CD28+ cells (day 14, P=0.0454; day 28, P=0.0068) and CD27−CD28− cells (day 28, P=0.0144). (c) Homing receptors CCR5 and CD62L are shown in blood (n=19) and BAL (n=5). Significant P values for two-tailed Wilcoxon matched pairs tests compared with baseline are shown in blood (day 10, P=0.0009; day 14, P=0.0245). Throughout, representative plots from a single subject at day 0, 10 and 28 post infection are shown with tetramer+ cells as red dots and total CD8+ T cells in grey contours. Summarized data are shown for all RSV-infected subjects (who could be analysed using tetramers) with median frequencies shown in red. Abbreviated P values are shown: NS (not significant)=P>0.05, *P<0.05, **P<0.01.

Figure 8. RSV-specific CD8+ T cells in blood show limited polyfunctionality.

PBMCs from individuals inoculated with RSV were stimulated with peptide epitopes (YLE, n=3; NPK, n=4; and influenza M1-GIL, n=6) and subsequently intracellularly stained for IFN-γ, tumour necrosis factor (TNF) and interleukin-2 (IL-2) for analysis by flow cytometry. (a) Flow cytometric data gated on CD3+CD8+ lymphocytes from one representative donor are shown. Numbers represent percentage of CD8+ lymphocytes. (b) The mean (±s.e.m.) frequencies of cytokine-producing CD8+ T cells as determined by Boolean gating are shown as percentages of total responding cells. (c) The proportion of single, double and triple cytokine producers is shown.

Figure 9. Pre-existing RSV-specific memory CD8+ T cells in the airway correlate with reduced disease severity.

RSV-specific CD8+ T-cell responses to HLA-A*01:01-, A*02:01- and B*07:02-restricted epitopes were analysed by tetramer staining before inoculation. (a) Baseline frequencies of tetramer+ cells in infected HLA-A*01:01+ (n=9), A*02:01+ (n=6) and B*07:02+ (n=11) in matched blood and BAL samples were compared. The medians and P values for Wilcoxon matched pairs tests (A1-M-YLE, P=0.0234; A2-NS2-FLV, P=0.0313; B7-N-NPK, P=0.0195) are shown (*P<0.05). (b) Baseline frequencies of tetramer+ cells in BAL from infected (n=14) and uninfected (n=12) individuals were compared and tested for significant difference using the Mann–Whitney test (P=0.9256). Medians are shown. (c) Non-linear regression was used to assess the association between baseline RSV-specific CD8+ cell frequencies in BAL and disease severity in infected individuals as measured by (c) cumulative symptom score, (d) nasal viral load (trapezoidal area under the curve (AUC)), (e) cumulative lower respiratory tract symptoms and (f) peak viral load in bronchial brushings. Spearman's rank correlation coefficient (rs) and P values are shown.