Antigenic Restimulation of Virus-Specific Memory CD8+ T Cells Requires Days of Lytic Protein Accumulation for Maximal Cytotoxic Capacity

Abstract

In various infections or vaccinations of mice or humans, reports of the persistence and the requirements for restimulation of the cytotoxic mediators granzyme B (GrB) and perforin (PRF) in CD8⁺ T cells have yielded disparate results. In this study, we examined the kinetics of PRF and GrB mRNA and protein expression after stimulation and associated changes in cytotoxic capacity in virus-specific memory cells in detail. In patients with controlled HIV or cleared respiratory syncytial virus (RSV) or influenza virus infections, all virus-specific CD8⁺ T cells expressed low PRF levels without restimulation. Following stimulation, they displayed similarly delayed kinetics for lytic protein expression, with significant increases occurring by days 1 to 3 before peaking on days 4 to 6. These increases were strongly correlated with, but were not dependent upon, proliferation. Incremental changes in PRF and GrB percent expression and mean fluorescence intensity (MFI) were highly correlated with increases in HIV-specific cytotoxicity. mRNA levels in HIV-specific CD8⁺ T-cells exhibited delayed kinetics after stimulation as with protein expression, peaking on day 5. In contrast to GrB, PRF mRNA transcripts were little changed over 5 days of stimulation (94-fold versus 2.8-fold, respectively), consistent with posttranscriptional regulation. Changes in expression of some microRNAs, including miR-17, miR-150, and miR-155, suggested that microRNAs might play a significant role in regulation of PRF expression. Therefore, under conditions of extremely low or absent antigen levels, memory virus-specific CD8⁺ T cells require prolonged stimulation over days to achieve maximal lytic protein expression and cytotoxic capacity.IMPORTANCE Antigen-specific CD8⁺ T cells play a major role in controlling most virus infections, primarily by perforin (PRF)- and granzyme B (GrB)-mediated apoptosis. There is considerable controversy regarding whether PRF is constitutively expressed, rapidly increased similarly to a cytokine, or delayed in its expression with more prolonged stimulation in virus-specific memory CD8⁺ T cells. In this study, the degree of cytotoxic capacity of virus-specific memory CD8⁺ T cells was directly proportional to the content of lytic molecules, which required antigenic stimulation over several days for maximal levels. This appeared to be modulated by increases in GrB transcription and microRNA-mediated posttranscriptional regulation of PRF expression. Clarifying the requirements for maximal cytotoxic capacity is critical to understanding how viral clearance might be mediated by memory cells and what functions should be induced by vaccines and immunotherapies.

Keywords: CD8+ T cells; kinetics; perforin; virus-specific immunity.

FIG 1

Low baseline perforin expression, based on staining with the B-D48 and δG9 antibodies, in the HLA-B5701/HIV Gag tetramer⁺ CD8⁺ T cells of an LTNP/EC increased over days upon restimulation. Concatenated flow plots gated on HLA-B5701/HIV Gag tetramer⁺ CD8⁺ T cells from HIV LTNP/EC 1 depict changes in perforin (PRF) protein expression, measured by staining with B-D48 (A) or δG9 (B) monoclonal antibody, from baseline through 6 days of stimulation with overlapping 15-mer Gag peptides.

FIG 2

In contrast to IFN-γ production, CD8⁺ T cells under in vivo conditions of reduced antigen levels required days after antigenic restimulation for maximal lytic protein expression. (A and B) The frequencies of PRF B-D48⁺ (A) or PRF δG9⁺ (B) virus-specific CD8⁺ T cells are shown for HLA-5701/HIV Gag tetramer⁺ cells (red, n = 4) from HIV⁺ LTNP/EC and HLA-A0201/influenza virus MP1 dextramer⁺ (cyan, n = 4) and HLA-B0702/RSV NP pentamer⁺ (black, n = 2) cells from individuals without a recent history of respiratory illness at baseline and following stimulation with peptide pools spanning HIV Gag, influenza virus MP1, or RSV NP, respectively, at various time points. Results are also shown for HLA-5701/HIV Gag tetramer⁺ cells (gray) from chronically infected, viremic progressors (Vir P; n = 2) and ART-suppressed progressors (Rx < 40 P; n = 2). (C) PRF expression by B-D48 and δG9 staining was significantly correlated by the Spearman rank method. (D) GrB expression is shown for 13 of the same antigen-specific CD8⁺ T cells at the same time points. (E) Frequencies of IFN-γ⁺ (solid lines) and PRF⁺ (dotted lines) multimer⁺ CD8⁺ T cells specific for influenza virus (cyan), RSV (black), and HIV in an LTNP/EC (red) and Vir P (dark blue) are shown at baseline and following stimulation. Background IFN-γ percentages from contemporaneous medium controls were subtracted. Only significant P values are shown adjacent to graphs for comparisons of expression averaged over time between controlled/cleared infections versus HIV-specific responses in progressors (A, B, and D) or above graphs for single time point comparisons of GrB between controlled/cleared infections versus HIV-specific responses in progressors (D) and between PRF and IFN-γ expression (F), as follows: *, P ≤ 0.05; **, P ≤ 0.01; and ***, P ≤ 0.001.

FIG 3

Baseline surface phenotype of multimer⁺ CD8⁺ T cells was consistent with prior antigen exposure, did not predict the capacity to upregulate PRF, and tended toward a programmed cell death protein 1⁺ (PD-1⁺) effector memory phenotype with stimulation. (A) PD-1 expression on HLA-5701/HIV Gag tetramer⁺ (red, n = 2 LTNP/EC; dark blue, n = 4 progressors, 2 Rx < 40 P, and 2 Vir P), HLA-A0201/influenza virus MP1 dextramer⁺ (cyan, n = 3), and HLA-B0702/RSV NP pentamer⁺ (black, n = 1) CD8⁺ T cells is shown at baseline and following stimulation with peptide pools spanning HIV Gag, influenza virus MP1, or RSV NP, respectively. (B) PD-1 expression of multimer⁺ CD8⁺ T cells correlated significantly with PRF expression over 6 days (n = 80) by the Spearman rank method. (C) Baseline expression of the CD45 family splice variant RO and C-C chemokine receptor type 7 (CCR7) on the same multimer⁺ CD8⁺ T cells as in panel A enabled grouping into phenotypic subsets as naive (CD45RO⁻ CCR7⁺), central memory (CD45RO⁺ CCR7⁺), effector memory (CD45RO⁺ CCR7⁻), and effector (CD45RO⁻ CCR7⁻). Only significant P values for subset differences between HIV-specific cells from LTNP/EC plus progressors versus influenza virus/RSV-specific cells are shown, as follows: *, P ≤ 0.05, and **, P ≤ 0.01. (D to G) Changes in the frequencies of these subsets over time are shown for the same multimer⁺ CD8⁺ T cells as in panel C following stimulation with peptide pools spanning HIV Gag, influenza virus MP1, or RSV NP.

FIG 4

Lytic protein upregulation correlated with proliferation and occurred more rapidly and to a higher degree in proliferating than in nonproliferating CD8⁺ T cells. (A) The degree of proliferation (based on CFSE dilution) is shown for HLA-5701/HIV Gag tetramer⁺ (red, n = 3 LTNP/EC; dark blue, n = 4 progressors, 2 Rx < 40 P, and 2 Vir P), HLA-A0201/influenza virus MP1 dextramer⁺ (cyan, n = 4), and HLA-B0702/RSV NP pentamer⁺ (black, n = 2) CD8⁺ T cells at baseline and following stimulation. (B) PRF and GrB expression of antigen-specific CD8⁺ T cells significantly correlated with the fraction of CFSE^lo cells by the Spearman rank method. (C and D) Representative flow plots gated on total (C) or HLA-5701/HIV Gag tetramer⁺ (D) CD8⁺ T cells from a B*57⁺ LTNP/EC show fractions of proliferating CFSE^lo versus nonproliferating CFSE^hi CD8⁺ T cells (C) and PRF expression in proliferating (D, left column) versus nonproliferating (D, right column) HLA-5701/HIV Gag tetramer⁺ cells after stimulation for 1 to 6 days with pooled Gag peptides. (E and F) Curves are shown summarizing the changes in PRF (E) and GrB (F) percent expression in proliferating (solid black lines) versus nonproliferating (dotted gray lines) HLA-5701/HIV Gag tetramer⁺ (LTNP/EC, n = 3), HLA-A0201/influenza virus MP1 dextramer⁺ (n = 4), and HLA-B0702/RSV NP pentamer⁺ (n = 2) CD8⁺ T cells following stimulation for 1 to 6 days with pools of overlapping peptides spanning HIV Gag, influenza virus MP1, or RSV NP, respectively. P values are shown for comparisons of expression averaged over time between controlled/cleared infections versus HIV-specific responses in progressors (A) and between PRF (E) or GrB (F) percent expression in CFSE^lo versus CFSE^hi CD8⁺ T cells specific for controlled/cleared virus infections, as follows: ***, P ≤ 0.001.

FIG 5

Increases in HIV-specific CD8⁺ T-cell cytotoxicity occurred over several days after antigenic restimulation with kinetics similar to those of lytic protein upregulation. (A and B) Representative flow plots of cells from a B*57⁺ LTNP/EC depict cytotoxicity, measured by GrB activity in HIV-infected CD4⁺ T-cell targets (A) and their elimination (B), after coincubation for 1 h with CD8⁺ T-cell effectors that were rested overnight (day 0) or had been stimulated for 1 to 6 days. Red numbers on plots represent percentages of live targets with increased GrB substrate fluorescence after subtracting background (A) and calculated infected CD4⁺ T-cell elimination (ICE [B]; see Materials and Methods). The first column (controls) shows background GrB substrate fluorescence when uninfected targets were coincubated with effectors (A) or the initial frequencies of live p24-expressing targets before adding effectors (B). (C) Flow plots showing percentages of CD107a⁺ and/or IFN-γ⁺ CD8⁺ T cells from the same LTNP/EC that had been rested overnight (day 0) or stimulated with infected targets for 1 to 6 days prior to restimulation with uninfected (controls) or infected (other columns) targets for 6 h in the presence of fluorescently labeled CD107a monoclonal antibody, brefeldin A, and monensin. Red numbers on plots represent responses after subtracting background response to uninfected targets. (D) CD8⁺ T-cell cytotoxic responses (blue circles, GrB activity; red squares, ICE) and the frequencies of CD107a⁺ and/or IFN-γ⁺ CD8⁺ T cells (black diamonds) from 3 LTNP/ECs are plotted over time. (E) Cytotoxic responses measured by GrB activity and ICE were significantly correlated by the Spearman rank method. (F to I) PRF and GrB percent expression (F and G) and PRF and GrB MFI (H and I) in HIV-specific CD8⁺ T cells from 3 LTNP/ECs over days 0 to 6 were significantly correlated by the Spearman rank method with cytotoxicity, measured by GrB activity (GrB Act, blue circles) and ICE (red squares).

FIG 6

Fold increases of GrB mRNA levels were greater than for PRF in HIV-specific CD8⁺ T cells after 5 days of stimulation. (A) Heat map depicts Euclidean distance row clustering of log₂ fold change in the expression of 96 mRNA transcripts in sorted HIV tetramer⁺ CD8⁺ T cells from 7 LTNP/ECs at 0.5, 1, 3, or 5 days of stimulation with pooled Gag peptides relative to unstimulated conditions. (B to E) Relative to unstimulated conditions, mRNA levels were grouped according to whether transcripts exhibited significant upregulation (>4-fold [B and C]) with an early (B, ≤24 h) or late (C, ≥72 h) peak or significant downregulation (<0.25-fold [D and E]) with an early (D) or late (E) nadir. The P value boundary was <0.05. (B and C) Significantly upregulated genes included those for interleukin 2 receptor α (IL-2RA), tumor necrosis factor receptor superfamily member 18 (TNFRSF18), IFN-γ-induced protein 10 or C-X-C motif chemokine 10 (CXCL10), macrophage inflammatory protein 1-alpha or chemokine (C-C motif) ligand 3 (MIP-1α, CCL3), IL-10, tumor necrosis factor (TNF), CCL4, B-cell lymphoma 2 (BCL2), lymphotoxin alpha (LTA) or TNF-β, CXCL11, IL-13, CD86, granzyme B (GZMB), CD38, IFN-γ, cytotoxic T-lymphocyte-associated protein 4 (CTLA4), FASL (Fas ligand), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), C-C chemokine receptor type 5 (CCR5), inducible T-cell costimulator (ICOS), and CD40 ligand (CD40L). Perforin 1 (PRF1) levels increased significantly relative to baseline, but only 2.8-fold. (D and E) Significantly downregulated genes included those for granzyme K (GZMK), platelet factor 4 (PF4), IL-12A, eomesodermin (EOMES), granzyme A (GZMA), CCL5, granzyme M (GZMM), IL-2, cytochrome P450 family 7 subfamily a member 1 (CYP7A1), spleen-associated tyrosine protein kinase (SYK), CD4, CD34, IL-1β, E-selectin (SELE), TNFRSF5 (CD40), and IL-8. (F) Granzyme B (GZMB, blue squares) and PRF1 (red circles) mRNA levels were also quantified relative to endogenous controls at time zero and then compared with values obtained under unstimulated (dotted lines) or stimulated (solid lines) conditions at subsequent time points.

FIG 7

Expression of miR-17, miR-155, and miR-150 in HIV-specific CD8⁺ T cells changed significantly during 72 h of stimulation. (A to C) For time course analysis of miRNA data, maSigPro v1.54.0 was used to test linearized mean DCq values for significant differences in variation through time between the stimulated and unstimulated conditions and also to cluster loci based on patterns of variation through time. Based on an FDR threshold of <0.1, miR-17 (A) and miR-155 (B) log₂ transformed expression was significantly upregulated and miR-150 (C) was significantly downregulated in the sorted HIV tetramer⁺ CD8⁺ T cells of 4 LTNP/ECs during 72 h of pooled peptide stimulation relative to unstimulated conditions. Trend lines represent mean levels in control (orange) versus stimulated (green) conditions at 12, 24, 48, and 72 h.