Cutting Edge: Insufficient Perforin Expression in CD8⁺ T Cells in Response to Hemagglutinin from Avian Influenza (H5N1) Virus¹

Perforin plays a critical role in cell-mediated cytotoxicity against viral infection and is a key component of the lytic granules in cytotoxic CD8⁺ T lymphocytes (1, 2). Viral infections of perforin-deficient mice or persons with perforin gene defect, such as familial hemophagocytic lymphohistiocytosis, may result in some characteristic presentations of dysregulated immune responses, including the hyperproduction of proinflammatory cytokines (including IFN-γ and macrophage-derived cytokines) and hemophagocytic syndrome (3, 4, 5, 6). These presentations are similar to the clinical and laboratory findings from the severe human cases of H5N1 viral infection, especially those with acute respiratory distress syndrome and multiorgan failure (7, 8, 9, 10). Thus, we hypothesized that the manifestations of severe infection by H5N1 virus may be associated with the insufficient perforin expression.

Ten healthy subjects between 25 and 35 years old were enrolled. They were vaccinated with inactivated influenza vaccines (including H1N1, H3N2, and influenza B Ags) for at least 2 years. The study was approved by the Institutional Review Board, and written informed consent was obtained from all subjects. Not one human case of avian influenza was reported thus far in Taiwan.

DCs are essential for eliciting T cell immunity in vivo (11). To mimic the in vivo condition of Ag presentation and recognition, we prepared monocyte-derived immature DCs as APCs in the 10 healthy subjects. PBMC were prepared from heparinized blood and isolated by differential centrifugation over endotoxin-free Ficoll-Paque (Pharmacia Biotech). Monocytes were negatively isolated from PBMC using immunomagnetic beads coated with mAbs against human CD2, CD7, CD16, CD19, and CD56 (Monocyte Negative Isolation kit; Dynal Biotech) according to the manufacturer’s instructions. Negatively isolated monocytes were suspended in complete RPMI 1640 (Invitrogen Life Technologies) supplemented with penicillin/streptomycin, 10% FCS (Invitrogen Life Technologies), recombinant human IL-4 (1000 U/ml; BD Pharmingen), and GM-CSF (50 ng/ml; BD Pharmingen) in 24-well plates (1 ml/well) with a cell concentration of 8 × 10⁵/ml in triplicate. These monocytes showed the phenotype of immature DCs after 6 days of culture demonstrated by positive staining with mAbs to CD11c (PE-conjugated B-ly6, mouse IgG1; BD Pharmingen), CD80 (FITC-conjugated L307.4, mouse IgG1; BD Pharmingen), CD86 (FITC-conjugated IT2.2, mouse IgG2b; BD Pharmingen), and negative staining by mAb to CD14 (PE-conjugated M5E2, mouse IgG2a; BD Pharmingen) by using flow cytometry with a FACSCalibur cytometer and CellQuest software (BD Biosciences).

We pulsed these immature DCs (10⁵/ml RPMI 1640 in 24-well plates) with recombinant hemagglutinins (300 ng/ml) from three influenza A viruses (A/New Caledonia/20/99 (H1N1), A/Wyoming/3/2003 (H3N2), and A/Vietnam/1203/04 (H5N1), abbreviated as H1, H3, and H5, respectively; Protein Sciences) for another 24 h, and then we coincubated these Ag-presenting DCs with isolated autologous CD8⁺ T cells (as described below) with a ratio of CD8⁺ T cells: DCs = 10:1. According to the certifications from manufacturer, the preparations of H1, H3 and H5 have been tested for endotoxin and the pyrogenicity <20 endotoxin units/ml. No bacterial or fungal contamination was found. The three hemagglutinins have similar molecular mass (∼72,000 kDa) and are purified to >90% purity under the conditions preserving their biological activity and tertiary structure. The hemagglutinin concentrations <30 ng/ml could not induce detectable lymphoproliferation activity of CD8⁺ T cells in our methods. The optimal correlation between lymphoproliferation activity of CD8⁺ T cells and the hemagglutinin concentrations could be achieved with the range of hemagglutinin concentrations between 100 and 1000 ng/ml.

The viability of hemagglutinin-pulsed DCs, before and after incubation with CD8⁺ T cells, were estimated by counting chamber with characteristic morphology of DCs and trypan blue dye exclusion.

CD8⁺ T cells were positively isolated from PBMC and detached from beads by an immunomagnetic method (CD8 Positive Isolation Kit; Dynabeads plus DETACHaBEAD; Dynal Biotech) according to the manufacturer’s instructions. The resulting purity was >99%, and viability was >95%.

The CD8⁺ T cell responses to hemagglutinins of influenza viruses were assessed by incubating CD8⁺ T cells with autologous DCs, which remained at an immature state or already pulsed with H1, H3, or H5, in a final volume of 1 ml of RPMI 1640 (10⁶ CD8⁺ T cells with 10⁵ CD11c⁺ DCs per well of a 24-well plate in triplicate). The CD8⁺ T cells were then harvested for T cell proliferation assay and intracellular perforin expression analysis, and the supernatants were harvested for determination of IFN-γ concentrations on varying duration of incubation.

The proliferation activity of CD8⁺ T cells was assessed by determining the frequencies of CD8⁺ T cells with BrdU incorporation using BrdU Flow Kit (BD Pharmingen) in triplicate. BrdU (final concentration 10 μM) was added 24 h before harvest. Cells with BrdU incorporation were detected by flow cytometry after being fixed with paraformaldehyde, permeabilized with saponin, and stained with anti-BrdU FITC according to the BrdU Flow Kit (BD Pharmingen) protocol from the manufacturer. The hemagglutinin-specific proliferation responses of CD8⁺ T cells were defined as follows: [the CD8⁺ T cell responses elicited by hemagglutinin-pulsed DCs] − [the CD8⁺ T cell responses elicited by immature DCs without being pulsed with hemagglutinin].

Supernatants of the cell cultures were harvested on varying days of incubation and frozen at −70°C until used. Levels of IFN-γ were determined using a commercial ELISA kit (Quantikine; R&D Systems) according to the manufacturer’s instructions. The hemagglutinin-specific IFN-γ production was defined as [the levels of IFN-γ in supernatants from cell cultures containing CD8⁺ T cells and hemagglutinin-pulsed DCs] − [the levels of IFN-γ in supernatants from cell cultures containing CD8⁺ T cells and immature DCs without being pulsed with hemagglutinin].

CD8⁺ T cells were harvested on the varying days of incubation with hemagglutinin-primed DCs for detection of intracellular expression of perforin. CD8⁺ T cells were washed and resuspended in cold Dulbecco’s PBS and then fixed and permeabilized by Cytofix/Cytoperm solution (15 min, 4°C, in the dark; BD Pharmingen) according to the manufacturer’s protocol. These fixed and permeabilized cells were stained with mAb specific for human perforin (FITC-conjugated δG9, mouse IgG2bκ; BD Pharmingen), or isotype control (20 μl/10⁶ cells), and CD45RA (allophycocyanin-conjugated HI100) and CD45RO (PE-conjugated UCHL1) at room temperature for 30 min in the dark, and then analyzed by flow cytometry. The mean fluorescein intensity (MFI) of hemagglutinin-specific intracellular expression of perforin was defined as [the MFI of intracellular expression of perforin in CD8⁺ T cells incubated with hemagglutinin-pulsed DCs] − [the MFI of intracellular expression of perforin in CD8⁺ T cells incubated with immature DCs without being pulsed with hemagglutinin].

Isolated CD8⁺ T cells (1 × 10⁶/well) were prepared as effector cells by incubation with autologous hemagglutinin-primed DCs (as APCs, in a ratio of 10:1) in a final volume of 1 ml of complete RPMI 1640 in 24-well plates for variable duration. In our observation, the APC:T cell ratio of 1:10 was needed to achieve enough stimulation of CD8⁺ T cells. If the APC:T cell ratio is <1:10 (1:20, 1:50, 1:100), the induced cytotoxic activity may be variable or undetectable. Autologous monocytes, isolated from PBMCs, were prepared as target cells (10⁶/ml) by pulsing with hemagglutinins (1 μg/ml) in 24-well plates (1 ml/well) for 24-h incubation. In this study, a nonradioactive lactate dehydrogenase (LDH)-releasing cytotoxicity assay kit (CytoTox 96 Non-Radioactive Cytotoxicity Assay; Promega) was used according to the manufacturer’s instructions with an E:T ratio of 10:1. The E:T ratio of 10:1 has the most interpretable data from the LDH-releasing cytotoxicity assay. E:T ratios of 30:1 or 50:1 may lead to confused LDH-releasing data due to the smaller number of target cells. Control wells for spontaneous LDH release from target cells or effector cells, culture medium background, volume correction, and maximal release from target cells were prepared according to the manufacturer’s instructions. Released LDH in supernatant was measured with a 30-min coupled enzymatic assay (provided by CytoTox 96 kit) that resulted in the conversion of a tetrazolium salt (INT) into a red formazan product. The percentage of cytolysis was calculated as ([(experimental release − background) − (spontaneous target release − background) − (spontaneous effector release − background)[/(maximal target release − spontaneous target release)) ×100%. The percentage of hemagglutinin-specific cytolytic activity was calculated as [percentage of cytolytic activity of CD8⁺ T cells induced by hemagglutinin-pulsed DCs] − [percentage of cytolytic activity of CD8⁺ T cells induced by immature DCs without being pulsed with hemagglutinin].

Statistical significance was determined using a nonparametric test (Wilcoxon signed rank test) if two groups were compared; or one-way ANOVA was used if ≥3 groups were compared. Linear correlation was evaluated by Pearson’s correlation coefficient. All tests were two-tailed. p < 0.05 was considered statistically significant. All data are shown as mean ± SD.

First, we defined the kinetics of hemagglutinin-specific responses of CD8⁺ T cells elicited by autologous monocyte-derived DCs primed with H1, H3, or H5 in 7 days of coincubation of CD8⁺ T cells and DCs from 10 healthy subjects. We found that the kinetics of DC-elicited CD8⁺ T cell responses to H1 or H3 were similar. The remarkable lymphoproliferation (Fig. 1,a) and IFN-γ production (Fig. 1,b) of CD8⁺ T cells were noted after 72 h of coincubation with H1- or H3-pulsed DCs. The expression of intracellular perforin in these CD8⁺ T cells seemed to achieve a peak level on day 3 of coincubation (Fig. 1,c), and was associated with a marked hemagglutinin-specific cytotoxic activity of CD8⁺ T cells since day 3 of coincubation (Fig. 1 d).

However, CD8⁺ T cells that were coincubated with H5-pulsed DCs had a much lower level of intracellular perforin expression (Fig. 1,c) and hemagglutinin-specific cytotoxic activity (Fig. 1,d) when compared with these incubated with H1-pulsed DCs (analysis on day 3 of coincubation; p = 0.002 and 0.002, respectively); however, the responses of lymphoproliferation and IFN-γ production of the CD8⁺ T cells increased markedly since day 5 of coincubation and were significantly higher than those of the CD8⁺ T cells incubated with H1-pulsed DCs on day 7 of coincubation (Fig. 1, a and b; analysis on day 7 of coincubation; p = 0.002 and 0.002, respectively). Because the hemagglutinin-presenting DCs themselves were also the targets of cytotoxicity of CD8⁺ T cells, we assessed the viability of hemagglutinin-pulsed DCs in the coincubation, to evaluate the number of surviving APCs under cytotoxicity of CD8⁺ T cells (i.e., the number of APCs still available to stimulate CD8⁺ T cells) (Fig. 1,e). We found the viability of H1- or H3-pulsed DCs declined rapidly since day 3 of coincubation with CD8⁺ T cells, in correlation with the remarkable H1- or H3-specific cytotoxic activity of CD8⁺ T cells. However, the viability of H5-pulsed DCs was significantly higher than that of H1-pulsed DCs (Fig. 1 e; analysis on day 3 of coincubation; p = 0.002). The data suggested that the higher viability of H5-pulsed DCs may reflect the lower levels of H5-specific cytotoxic activity of CD8⁺ T cells and may lead to the persistence of H5-presenting DCs and sustained stimulation for CD8⁺ T cells.

Among the perforin-expressing CD8⁺ T cells that were incubated with H5-presenting DCs, most of them were naive phenotype (CD45RA⁺; 75.3 ± 8.1%; mean ± SD from 10 healthy subjects; detailed data not shown) rather than memory phenotype (CD45RO⁺). However, most of the perforin-expressing CD8⁺ T cells that were incubated with H1- or H3-presenting DCs were memory phenotype (CD45RO⁺; 68.3 ± 7.2% and 70.3 ± 8.5%, respectively; mean ± SD; detailed data not shown). To differentiate the insufficient expression of perforin in response to H5 was due to the slower response of naive T cells or the direct suppressive effects from H5, we evaluated the impact of recombinant H5 protein on the perforin expression in CD8⁺ T cells. We pulsed immature DCs with varying doses of H5 in a range from 30 to 1000 ng/ml, and assessed the intracellular perforin expression in autologous CD8⁺ T cells that were coincubated with these H5-pulsed DCs for 3 days. We found that the frequencies of intracellular perforin expression in CD8⁺ T cells had a linear correlation with a negative slope along with the H5 doses (Fig. 2,a). To know whether the addition of H5 could reduce the perforin expression in activated CD8⁺ T cells, we assessed the frequencies of intracellular perforin expression of the CD8⁺ T cells that were incubated with autologous DCs pulsed with H1 (300 ng/ml) and varying doses of H5 (Fig. 2 b). The results showed that the higher doses of added H5 could result in reduced levels of perforin expression in CD8⁺ T cells stimulated by DCs, which were pulsed with fixed doses of H1. The data suggest that the insufficient perforin expression in CD8⁺ T cells in response to H5 may result not only from the characters of naive response of T cells, but also from the directly suppressive effect of H5 on perforin expression on CD8⁺ T cells, perhaps through undefined epitopes or H5-related DC modulation.

Infection-associated hemophagocytic syndrome has been frequently associated with EBV. The EBV-associated hypercytokinemia and hemophagocytosis has been suggested to be associated with excessive activation of EBV-infected T cells with the manifestation of hyperproduction of IFN-γ and increased perforin expression in CD8⁺ T cells (3, 12). However, through a different mechanism, the hypercytokinemia and hemophagocytosis in severe human cases of H5N1 virus infection may be attributed to the insufficient expression of perforin in CD8⁺ T cells, according to our data. Thus, we propose that the failure to express enough perforin in CD8⁺ T cells, when encountering H5-presenting DCs, leads to the impairment of cytotoxic activity to clear H5N1 virus- or H5 protein-bearing cells. The persistence of APCs may provide sustained antigenic stimulation to CD8⁺ T cells. The persistent activation of CD8⁺ T cells leads to hyperproduction of IFN-γ, which may result in the overactivation of macrophages and subsequent hyperproduction of proinflammatory cytokines such as TNF-α, a feature similar to the proposed model in perforin-deficient mice or human with familial hemophagocytic lymphohistiocytosis (3, 5). However, this study is limited by the lack of data from animal model experiments or from human cases of avian influenza. This proposed pathophysiology needs to be supported or confirmed by these in vivo data.

Based on the assumption that immune overactivity contributes to the life-threatening complications of human H5N1 infection, corticosteroids frequently have been used clinically to nonspecifically suppress the unfavorable immune responses (9). However, according to the limited data, the use of corticosteroid is not associated with a favorable outcome (9, 13). The knowledge of the mechanisms about how H5 impairs the perforin expression may help us to establish new therapeutics to specifically block the process of hypercytokinemia.

In conclusion, we propose that the hypercytokinemia and hemophagocytosis in severe cases of H5N1 virus infection may be due to H5-related insufficient perforin expression and ineffective cytotoxicity in CD8⁺ T lymphocytes against H5-bearing cells. Detailed mechanism should be clarified to find a way to specifically block this process in severe human cases of avian influenza to reduce the case fatality.

The authors have no financial conflict of interest.