Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Immunol. Author manuscript; available in PMC 2010 May 1.
Published in final edited form as:
PMCID: PMC2714586

Rapid upregulation and granule-independent transport of perforin to the immunological synapse define a novel mechanism of antigen-specific CD8+ T cell cytotoxic activity


Cytotoxic T lymphocytes (CTL) are endowed with the ability to eliminate pathogens through perforin-mediated cytotoxic activity. The mechanism for perforin-mediated antigen-specific killing has been solely attributed to cytotoxic granule exocytosis from activated CD8+ T cells. Here we redefine this mechanism, demonstrating that virus-specific CD8+ T cells rapidly upregulate perforin in response to stimulation temporally with IFN-γ and CD107a expression. Following antigen-specific activation, newly synthesized perforin rapidly appears at the immunological synapse, both in association with and independent of cytotoxic granules, where it functions to promote cytotoxicity. Our work redefines the current mechanism of CTL cytotoxicity and identifies a novel correlate of CD8+ T cell mediated immunity.

Keywords: Human, Cytotoxic T cells, Perforin


Cytotoxic T lymphocytes (CTL) are endowed with the ability to eliminate tumor cells as well as host cells that harbor intracellular pathogens via two principal mechanisms: 1) the exocytosis of pre-formed secretory granules that contain cytotoxic proteins(13), and 2) the engagement of receptors on the target cell that induce cellular apoptosis, such as those specific for Fas ligand(4, 5), tumor necrosis factor (TNF)(6), and TNF-related apoptosis inducing ligand (TRAIL)(6). Cytotoxic granules contain various granzymes(7), a family of highly specific serine proteases that initiate apoptosis(811), and perforin(7), a protein capable of binding phospholipid membranes in a calcium-dependent manner(1214). During activation, perforin monomers polymerize on lipid bilayers to form a pore(12, 14, 15) that permits entry of granzymes into the target cell to induce apoptosis. How perforin enables granzyme entry is not known, however, as the localization of pore formation within the target cell remains controversial. Nonetheless, in both humans and mice, abrogation of perforin activity results in profoundly impaired cellular cytotoxicity and immunodeficiency(16, 17).

CTL are capable of serial killing, the sequential elimination of several target cells(18, 19), but the mechanism that permits this ability is unclear. CTL may 1) ration their granules, 2) rapidly upregulate their cytotoxic proteins to sustain/replenish their killing ability, or 3) need to proliferate in order to recover their cytotoxicity. Cytotoxic granule-mediated killing by CTL occurs within hours of target cell recognition(19), but reconstitution of intracellular perforin has only been detected after several days of proliferation(2022). On this premise, the current paradigm is that CTL must proliferate before perforin recovery, implying that the cytotoxic granule content of a resting CTL dictates the immediate killing potential of that cell. Biologically this seems inefficient, as an infecting virus that could outlast the initial brunt of the effector CD8+ T cell population would continue to replicate in an unfettered manner until the CD8+ T cell population had proliferated. A potential solution to this conundrum is the possibility of rapid recharging of the CTL cytotoxic machinery following activation. Such a mechanism would allow the CTL to continue killing following granule release.

While rapid perforin upregulation was shown to occur in NK cell lines (23), this ability in primary CD8+ T cells has until recently been unclear. Contrary to previous reports, we found that perforin upregulation by antigen-experienced human CD8+ T cells occurs within hours of activation, without a requirement for exogenous cytokines or cellular proliferation(24). However, the role of new perforin remains unknown; is it simply a means to replenish depleted granules, or is it also actively used to mediate cytotoxicity? Here, we provide evidence that newly produced perforin traffics directly to the immunological synapse, where it promotes cytotoxic activity. These results define perforin upregulation ability as a novel correlate of cytotoxic potential in human CD8+ T cells.



Peripheral blood mononuclear cells (PBMC) were obtained from a normal human subject who exhibited an exceptionally strong response to the HLA-B7-restricted hCMV pp65 peptide TM10: TPRVTGGGAM epitope (4235 spot forming cells, SFC, per 106 PBMC). Donor PBMC were collected by the University of Pennsylvania’s Center for AIDS Research Human Immunology Core, in compliance with the guidelines set by the institutions’ internal review board, and cryopreserved in fetal bovine serum (FBS; ICS Hyclone, Logan, Utah) containing 10% dimethyl sulfoxide (DMSO; Fisher Scientific, Pittsburgh, Pennsylvania).


Antibodies for surface staining included anti-CD4 PE Cy5-5 (Invitrogen; Carlsbad, California), anti-CD107a FITC (BD Biosciences; San Jose, California), anti-CD8 Qdot 655 (custom), anti-CD14 Pac Blue (BD Biosciences; San Jose, California), anti-CD16 Pac Blue (BD Biosciences; San Jose, California), and anti-CD19 Pac Blue (Invitrogen; Carlsbad, California). Antibodies for intracellular staining included anti-CD3 Qdot 585 (custom), anti-Granzyme B Texas Red PE (BD Pharmingen; San Diego, California), anti-IFN-γ Alexa 700 (BD Pharmingen; San Diego, California), and anti-TNF-α PE Cy7 (BD Biosciences; San Jose, California). Custom conjugations to Quantum (Q) dot nanocrystals were performed in our laboratory as previously described(25), with reagents purchased from Invitrogen (Carlsbad, California). Anti-human perforin antibodies were purchased from Tepnel (clone D48, Besancon, France) and BD Biosciences (clone δG9, San Jose, California). Additional antibodies used for microscopy experiments are listed in the microscopy section.

FACS Staining assay

Cryopreserved PBMC were thawed, and then rested overnight at 37°C, 5% CO2 in complete medium [RPMI (Mediatech Inc; Manassas, Virginia) supplemented with 10% FBS, 1% L-glutamine (Mediatech Inc; Manassas, Virginia), and 1% penicillin-streptomycin (Lonza; Walkersville, Maryland), sterile filtered] at a concentration of 2×106 cells per ml medium in 12-well plates. The next day, the cells were washed with complete medium and resuspended at a concentration of 1×106 cells/ml with costimulatory antibodies (anti-CD28 and anti-CD48d; 1 µg/ml final concentration; BD Biosciences; San Jose, California), in the presence of monensin (0.7 µg/ml final concentration; BD Biosciences; San Jose, California) and brefeldin A (1 µg/ml final concentration; Sigma-Aldrich; St. Louis, Missouri). Anti-CD107a was always added at the start of all stimulation periods, as described previously(26). As a negative control, 5µl of DMSO was added to the cells, an equivalent concentration compared to the peptide stimulus. SEB served as the positive control (1 µg/ml final concentration; Sigma-Aldrich; St. Louis, Missouri). Peptide stimulations were performed at a final concentration of 2 µM. Stimulation tubes were incubated at 37°C, 5% CO2 for six hours. During the time course experiment, stimulation tubes prepared in parallel were incubated for 1, 2, 4, 6, 8, and 12 hours before being stained with antibodies.

At the end of the stimulation periods, cells were washed once with PBS before being stained for viability with Violet or Aqua amine-reactive viability dye (Invitrogen; Carlsbad, California) for ten minutes in the dark at room temperature. A cocktail of antibodies was then added to the cells to stain for surface markers for an additional twenty minutes. The cells were washed with PBS containing 1% bovine serum albumin (BSA, Fisher Scientific; Pittsburgh, Pennsylvania) and 0.1% sodium azide (Fisher Scientific; Pittsburgh, Pennsylvania), and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences; San Jose, California) according to the manufacturer's instructions. A cocktail of antibodies against intracellular markers was then added to the cells and allowed to incubate for one hour in the dark at room temperature. The cells were then washed once with Perm Wash buffer (BD Biosciences; San Jose, California) and fixed in PBS containing 1% paraformaldehyde (Sigma-Aldrich; St. Louis, Missouri). Fixed cells were stored in the dark at 4°C until the time of collection.

HLA-B7/TM10 tetramers were produced according to standard procedures(27). MHC class I tetramer staining was performed in calcium-free PBS for 15 minutes on ice and in the dark. The cells were then stained for surface and intracellular markers, as described above.

Flow cytometric analysis

For each specimen, between 500,000 and 1,000,000 total events were acquired on a modified flow cytometer (LSRII; BD Immunocytometry Systems; San Jose, California) equipped for the detection of 18 fluorescent parameters. Antibody capture beads (BD Biosciences; San Jose, California) were used to prepare individual compensation tubes for each antibody used in the experiment. Data analysis was performed using FlowJo version 8.7 (TreeStar, Ashland, Oregon). Reported data have been corrected for background.

Metabolic Labeling and Immunoprecipitation

Purified CD8+ T cells were incubated at 37°C in methionine-free medium (Mediatech; Herndon, Virginia) for one hour and then pulsed for 30 minutes with 35S-methionine (Environmental Health and Radiation Safety, University of Pennsylvania). After washing twice with methionine-free medium, the cells were divided and incubated with control media, or stimulated with phorbol 12-myristate 13-acetate (PMA, 50 ng/ml; Sigma-Aldrich; St. Louis, Missouri) and ionomycin (1 µM; Sigma-Aldrich; St. Louis, Missouri) for 4 hours. Subsequent immunoprecipitation was performed as described previously(28).

Western analysis of isolated NK cell lytic granules

1 × 108 YTS cells, either resting or activated via conjugation to KT86 target cells at a 2:1 effector to target cell ratio, were lysed and then subjected to centrifugation at 1000g to eliminate the nuclei. Following removal of the nuclei, the remaining lysate was subjected to centrifugation at 18,000g to pellet the lytic granules. The pelleted lytic granules and the cytoplasmic fraction were isolated and resuspended in PBS containing protease inhibitors and phosphatase inhibitors, and then separated using electrophoresis on a 4–12% Bis-Tris density gradient gel (Invitrogen), then transferred onto a PVDF membrane. After blocking with 3% BSA and 140 mM NaCl in Tris-buffered saline (TBS), the membrane was incubated with the D48 anti-perforin monoclonal antibody (Besancon, France). Bound antibody was detected with peroxidase-conjugated light chain-specific goat anti–mouse IgG (Jackson Immuno-Research Laboratories; West Grove, Pennsylvania) and an enhanced chemiluminescence detection system (GE Healthcare). The membrane was stripped in 0.2 M glycine (pH 2.5), 0.05% Tween 20, and 140 mM NaCl in TBS at 50 ° C for 30 min, blocked with 3% BSA, and re-probed with rabbit beta-actin polyclonal antibody clone 20–33 (Sigma-Aldrich; St. Louis, Missouri), followed by peroxidase-conjugated light chain–specific goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories; West Grove, Pennsylvania).

Confocal microscopy and image analysis

CD8+ T cells were isolated from PBMCs by negative selection (Miltenyi Biotech; Auburn, California), and then stained with Lysotracker Red (Invitrogen; Carlsbad, California) at a concentration of 166 nM in RPMI 1640 with 10% FCS for 40 minutes at 37°C, in the dark, washed twice in RPMI 1640 with 10% FCS, as indicated, and allowed to make conjugates with TM10 peptide or irrelevant peptide (HLA-A1-restricted influenza peptide NP 44–52, CTELKLSDY) pulsed C1R-B7 antigen-presenting cells (APC) at a 2:1 effector to target cell ratio in suspension for either 30 or 240 minutes. The cell conjugates were then adhered to the poly-l-lysine coated glass slides and stained as described previously(28). For experiments involving Golgi staining, purified CD8+ T cells were stimulated with either control media or stimulated with phorbol 12-myristate 13-acetate (PMA, 50 ng/ml; Sigma-Aldrich; St. Louis, Missouri) and ionomycin (1 µM; Sigma-Aldrich; St. Louis, Missouri) for 4 hours before being stained with anti-Golgi marker antibody (Abcam; Cambridge, Massachusetts). For experiments involving HLA staining, the conjugates were stained as described earlier (29) with certain modifications. In brief, conjugates were allowed to adhere on a glass slide as described previously (28) and then blocked with 5% heat-inactivated goat serum in PBS for 20 minutes at room temperature. Anti-HLA- A/B/C FITC antibody (BD Pharmingen; San Diego, California) was diluted in ice cold PBS containing 2% FCS and 0.1% sodium azide and conjugates were stained in the dark at 4°C for 45 minutes. Then the slides were rinsed in PBS containing 2% FCS and the cell conjugates were fixed and permeabilized for staining with anti-perforin antibodies.

Individual CD8+ T cells or conjugated cells were visualized by using a spinning disc confocal microscope (IX81 DSU, Olympus; Center Valley, Pennsylvania). Detection settings were adjusted so that a control stained sample was uniformly negative and fluorescence of experimentally stained samples was not saturating or bleeding through to other channels. The effector and target cells were distinguished by perforin staining. The area of accumulation of fluorescence in the effector cell was measured using the Improvision Volocity software package (Perkin Elmer; Waltham, Massachusetts). A threshold of > 40% of the mean fluorescence intensity of that fluorophore in the cell was used to determine the area occupied. Co-localization between two fluorophores was determined by measuring the area occupied by two different fluorophores. The percent co-localization was then calculated by dividing this co-localized area by the area that is occupied by either of the two fluorophores.

To define the immunological synapse, we measured the percent HLA accumulation at the CTL- target cell interface. The total amount of HLA on the target cell was measured by multiplying the area stained by the HLA-A/B/C antibody (detectable over background) with the mean fluorescence intensity (MFI) of the identified region. The effector and target cell regions forming the CTL-target cell interface were selected and measured separately in addition to the total measurement for each conjugate. The percent HLA accumulation (area × intensity) within the interface of each effector-target cell conjugate was then calculated using the formula: % accumulation = [(synaptic HLA area × synaptic HLA MFI) / (total conjugate HLA area × total conjugate HLA MFI)] × 100. Where specified, the shortest distance of the centroid of the total region of perforin fluorescence and HLA-A/B/C fluorescence at the synapse was measured. All measurements were defined and performed using Improvision Volocity (Perkin Elmer; Waltham, Massachusetts).

Chromium-release assay

Targets were prepared from the EBV-transformed cell line C1R, genetically engineered to express HLA-B7 (a kind gift from Dr. Jeff Frelinger, University of North Carolina, Chapel Hill, North Carolina). C1R cells expressing HLA-A2 (a kind gift from Dr. David Price, Cardiff University, Cardiff, England) were used as a negative control. The cells were incubated with either the B7-restricted peptide TM10 or an irrelevant peptide, [influenza A, NP (44–52), CTELKLSDY] for 1 hour, washed with complete medium, and then labeled with Na51CrO4 for one hour. Cells were washed three times with RPMI (Mediatech Inc; Manassas, Virginia) before being aliquoted into a 96 well plate (10,000 cells/well). Unlabelled target cells were used as cold targets at the start of the experiment. Effector cells were isolated from donor PBMC through negative selection of CD8+ T cells (Miltenyi Biotech; Auburn, California), and plated according to the various effector to target ratios. Where specified, cyclohexamide (10 µg/ml) was added to the effectors at the time of plating. After four hours, 51Cr-labeled target cells were added to all of the wells (10,000 cells/well). All test conditions were performed in triplicate, and the experiment was repeated 4 times. Harvested supernatants were evaluated for the presence of 51Cr using a Top Count scintillation counter (Beckman Coulter, Inc; Fullerton, California). The specific lysis was calculated as follows: % Specific lysis = [(mean of the test wells)-(mean of the spontaneous release wells)/(mean of maximal release wells)-(mean of the spontaneous release wells)] × 100%.


Canvas software, version 10.4.9 (ACD Systems; Miami, Florida) was used to assemble most of the figures. Labels and boxes were added to raw data images in Canvas.

Statistical analyses


Significant differences in areas of co-localization, denoted by an asterisk, were calculated using a two-tailed t-test (95% confidence interval). For fixed cell microscopy, the minimum number of cells evaluated in a given experiment was determined using a sample size calculation with alpha and beta error levels of 1%. Assumptions were based upon data evaluating co-localization of lytic granule contents published by others(30) and our own preliminary data. For all statistical analyses, differences between cell types or conditions were determined using a two-tailed Student’s t-test, and were considered significant if p<0.05.

Chromium-release assays

The data set does not violate the normality assumption. Four independent experiments were conducted to assess the effect of cyclohexamide on specific lysis. Since variability in CD8 T cell isolation created unique E:T ratios for each experiment, the relative change in observed %specific lysis as a result of cyclohexamide addition was calculated for every experiment, averaged, and evaluated using a t-test with t=32.80, 3 degrees of freedom. We used repeated measures ANOVA to take into account the “matched” structure of the experiments to be able to look at all the pair-wise differences between the conditions. Since we only focused on the 2 pair-wise comparisons with and without inhibitor, there was no need to adjust the values for multiple comparisons (Tukey-Kramer).


Sustained cytotoxicity requires new protein synthesis

Rapid perforin upregulation conceivably provides a mechanism by which an antigen-specific CD8+ T cell may retain its cytotoxic capabilities following the exocytosis of pre-formed secretory granules. To this end, we identified a normal human subject whose CD8+ T cells exhibit a high level of cytotoxicity against CMV TM10 peptide-loaded target cells directly ex vivo in a standard chromium release assay (Figure 1A). To address whether newly produced proteins played a role in this killing response, we performed a modified chromium-release assay; purified CD8+ T cells were added to peptide-loaded unlabeled target cells for 4 hours, in the presence or absence of the protein synthesis inhibitor cyclohexamide (CHX). Following this initial incubation intended to expend pre-formed granules, chromium-labeled peptide-loaded C1R-B7 cells were added to the effectors and incubated for an additional 4 hours. As shown in Figure 1B, specific lysis of 40.8% was observed at an initial input E:T ratio of 12.5:1 (effectively a ratio of 2.5:1 when factoring that ~20% of total donor CD8+ T cells are TM10 specific; see Figure 2A) in the absence of cyclohexamide, whereas no appreciable lysis was observed against HLA-mismatched targets or irrelevant peptide. Importantly, in the presence of cyclohexamide, a marked reduction in the killing ability of TM10-specific CD8+ was observed at all E:T ratios [mean reduction = 0.47 (47%); 95% CI = 0.42–0.51, p<0.001; t-test with df=3]. Four independent experiments yielded similar results [TM10 vs. TM10 + CHX p=0.013; paired t-test, t=3.79, df=5]. Thus, de novo protein production generated by antigen-specific CD8+ T cells significantly contributes to continual cytotoxic ability.

Figure 1
New protein synthesis is required for sustained cytotoxicity
Figure 2
Activated CD8+ T cells rapidly upregulate perforin

Antigen-specific CD8+ T cells produce new perforin protein concomitant with cytokine production and degranulation

To verify that perforin upregulation may be a mechanism to explain the sustained cytotoxicity we observed in the CTL assay, we monitored the kinetics of perforin upregulation by flow cytometry relative to degranulation (CD107a expression), Granzyme B upregulation, and cytokine production (IFN-γ, TNF-α). At baseline, 19.3% of the donor CD8+ T cells are bound by the HLA-B7 tetrameric complex containing the CMV peptide TM10 (Figure 2A), the majority of which already possess perforin. IFN-γ production was first detected 2 hours following stimulation with TM10 peptide. Concomitant with IFN-γ, perforin upregulation was detected within some activated cells (20.5% of IFN-γ producing cells) by 2 hours (Figure 2A, top row). By 6 hours, 53.4% of IFN-γ producing cells were perforin positive. A similar rate of perforin upregulation was observed in TNF-α producing cells (not shown). Interestingly, a second perforin antibody, the commonly used δG9 clone, was unable to detect this early appearance of perforin (Figure 2A, bottom row), likely due to the specificity of this antibody for a pH-sensitive motif in granule-associated perforin(24). These results indicate that the kinetics of perforin upregulation is rapid, coinciding with that of cytokine production.

To demonstrate that we were measuring newly produced perforin, rather than residual unreleased perforin within responding cells, we co-stained the cells with anti-CD107a to assess degranulation(26). If activated cells were only carrying pre-formed perforin, then the frequency of CD107a+ perforin+ cells should decrease with the exocytosis of cytolytic granules. As demonstrated in Figure 2B, CD107a+ cells 1 hour post-stimulation possess little perforin, indicating perforin loss due to degranulation. Over time, however, the proportion of CD107a+ CD8+ T cells that harbor perforin increases, peaking at 53.8% after 8 hours, indicating that the responding CD8+ T cells are indeed upregulating perforin production. The kinetics of granzyme B upregulation mirrors that of perforin (data not shown). A similar pattern was observed when the cells were stimulated with SEB, and no response was detected with the negative control (Figure 2B). Again, the δG9 antibody failed to detect any perforin upregulation (not shown). Over the 8-hour stimulation period, perforin increasingly accumulates in the CD107a+ and IFN-γ+ CD8+ T cell subsets, but does not quite return to baseline levels (Figure 2C). This likely reflects an equilibrium between the active production of new perforin and the continued release of perforin, such that a complete return to baseline levels of perforin is only achieved at the conclusion of the immune response.

We also performed an S-35 methionine labeling experiment on antigen-activated CD8+ T cells to identify if the perforin recognized by the D48 antibody was newly synthesized after activation (Figure 2D). Immunoprecipitation of perforin from activated CD8+ T cells with the D48 antibody yielded a positive signal, but only after T cell activation. In contrast, immunoprecipitation with δG9 resulted in a slight signal before, but not after, activation.

Thus, primary activated human CD8+ T cells rapidly upregulate perforin de novo, concomitant to cytokine production and degranulation.

Granule-associated perforin and newly synthesized perforin can be distinctly visualized within activated T cells

To isolate the contribution of new perforin synthesis on sustained cytotoxicity, we employed confocal microscopy to monitor the development of new perforin among activated CD8+ T cells. First, we assessed the degree of co-localization between newly upregulated perforin, detected by the D48 antibody, and granule-associated perforin, stained by the δG9 antibody, in both resting and activated CD8+ T cells, purified by negative selection. As shown in Figure 3A (top row) both the D48 (blue) and δG9 (green) antibodies simultaneously label perforin in resting CD8+ T cells. Moreover, their staining patterns co-localize uniformly with that of Lysotracker Red, which labels the lysosomal granule compartment, indicating that each antibody can recognize granule-associated perforin. After 4 hours of stimulation with PMA/ionomycin (Figure 3A, second row), however, the proportion of D48-labeled perforin (blue) that co-localizes with that of δG9 (green) decreases from 70% in resting cells to 20% in activated cells (Figure 3A, bar graph; D48/δG9). Furthermore, we observed a statistically significant drop in co-localization between D48 perforin (blue) and Lysotracker (red) (Figure 3A, bar graph; D48/L 12% in activated cells), whereas the co-localization between δG9 perforin (green) and Lysotracker (red) did not appreciably change (Figure 3A, bar graph; δG9/L 25% in resting vs. 40% in activated cells). Thus, newly upregulated perforin can readily be detectable intracellularly and distinct from secretory lysosomes.

Figure 3
New perforin rapidly appears in the Golgi of activated CD8+ T cells

To further confirm that the perforin labeled by the D48 antibody is produced de novo, we stained the CD8+ T cells with both perforin antibodies, and an antibody against a 58 kDa Golgi marker protein. As shown in Figure 3A, resting CD8+ T cells did not demonstrate a difference in perforin localization detected by either antibody (Figure 3B, D48 and δG9, top row). After stimulation with PMA/ionomycin, a large amount of perforin recognized only by the D48 antibody localized within the Golgi (Figure 3B, bottom row, D48+ Golgi), increasing from 15% in resting cells to 47% in activated CD8+ T cells (Figure 3B, right bar graph, %Golgi area also stained by D48, G/D48 p<0.05, t-test). There was no appreciable increase in co-localization between the %Golgi and the δG9-stained perforin (Figure 3B, right bar graph). Similar results were observed in PMA/IM activated CD8+ T cells from two additional donors (data not shown). Thus, newly produced perforin can be visualized within the Golgi apparatus of activated CD8+ T cells.

New perforin is not restricted to the granule compartment

To monitor the egress of new perforin from the Golgi, we stained resting and PMA/ionomycin activated CD8+ T cells with both perforin antibodies, as well as an antibody to Rab7 (Figure 4A). The latter is a GTPase that characteristically resides on late endosomes and is an important regulator of late endocytic membrane traffic(31). Once again, stimulation with PMA/ionomycin increased the amount of new perforin in the activated cells relative to the resting cells (Figure 4A, D48 blue; Figure 4B left panel, resting = 1.75 µm2, activated = 4.2 µm2). The co-localization between δG9-stained perforin and Rab7 increased as a result of PMA/ionomycin stimulation (Figure 4A, green + orange; Figure 4B right panel, resting = 20%, activated = 55%), likely signifying the formation of secretory lysosomes. In contrast, there was no difference in co-localization between D48-stained perforin (blue) and Rab7 (orange) upon stimulation (Figure 4A, blue and orange; Figure 4B). Along with the fact that new perforin, as detected by the D48 antibody, does not, for the most part, co-localize with δG9-stained perforin after activation (Figure 4B), these data suggest that new perforin is not entirely destined for late endosomal compartments in recently activated cells.

Figure 4
New perforin is not exclusively destined for cytolytic secretory lysosomes

To confirm these observations, we analyzed the perforin content of lytic granules and the cytoplasmic fraction of activated NK cells. We used YTS NK cells, which like ex vivo human NK cells constitutively express high levels of perforin, for easier detection of perforin by Western Blot. As can be seen in Figure 4C, perforin is readily detected in the lytic granules both before and after stimulation, however following conjugation with KT86 target cells there is a substantial accumulation of new perforin in the cytoplasmic fraction (increase in band intensity from 0.3 to 0.6 by densitometry).

New perforin accumulates at the immunological synapse independent of cytolytic secretory lysosomes

The fact that new perforin does not immediately localize with late endosomes suggests that it is not intimately involved in sustaining cytotoxicity. However, a precedent for an alternative perforin secretion pathway exists(23). We therefore directly visualized the fate of newly produced perforin (D48-labeled) in activated antigen-specific CD8+ T cells, incubated with HLA-B7-transfected C1R APC loaded with either the cognate TM10 peptide or an irrelevant peptide. After 30 and 240 minutes of incubation, the cell conjugates were stained with both perforin antibodies (δG9 and D48 clones) to permit the visualization of granule associated (δG9 and D48) and newly produced (D48) perforin, as well as Lysotracker. When the APC were pulsed with irrelevant peptide and mixed with CD8+ T cells for 30 minutes (Figure 5A), the granules appeared distal to the CTL-target interface. When exposed to APC primed with TM10 peptide (Figure 5B) the perforin-containing granules re-oriented towards the interface. There was no change in perforin content stained by either the D48 or δG9 antibody at 30 minutes (Figure 5C).

Figure 5
Cytotoxic secretory lysosomes polarize towards the immunological synapse after 30 minutes of conjugation with cognate target cells

After 240 minutes, the amount of total D48 antibody stained perforin increased in the TM10-APC/ CD8+ T cell conjugates, but not in the control samples (Figure 6A). The residual perforin contained in the cytolytic lysosome compartment, as stained by the δG9 antibody (green) and lysotracker (red), occupied a very specific area at the center of the interface (Figure 6A; δG9+Lyso, bottom row). Perforin stained by the D48 antibody (blue), also appeared at the interface, but occupied a substantially larger area of the cell (Figure 6A; overlay, bottom row). The proportion of D48 staining that co-localized with lysotracker decreased from ~39% in resting cells to ~26% in activated cells (Figure 6B; D48/L).

Figure 6
New perforin accumulates at the immunological synapse largely independent of secretory lysosomes

To define the CTL: target cell interface as the immunological synapse, we assessed the degree of MHC class I clustering at the contact zone between the CTL and target cells. After 30 minutes of incubation, we observed a significant amount of MHC class I accumulation at the interface between the CTL and TM10-loaded target cells (Figure 5D, red, bottom row) as compared to that with target cells loaded with irrelevant peptide (Figure 5D, top row; Figure 5E TM10 68% of HLA vs. control peptide 34%; p = 0.0254, two-tailed t-test). By 240 minutes of conjugation formation, the degree of HLA clustering at the interface remained pronounced (Figure 6C, bottom row red; 6D, 68% for 30 minute vs. 67% for 240 minute conjugates, p = 0.0049, two-tailed t-test). The D48-labeled perforin (blue) and δG9-labeled perforin (green) accumulated at the immunological synapse (overlay of Figure 5D and Figure 6C) only when the CD8+ T cells received an activation signal by TM10 pulsed APCs.

We next calculated the distance between the D48-labeled perforin centroid and the HLA class I cluster at the CTL-target cell interface in order to define the polarization of newly synthesized perforin to the immunological synapse. We found this distance to be significantly less between CD8+ T cells and TM10- loaded APC as compared to APC bearing irrelevant peptide at both 30 minutes (Figure 6E, TM10 peptide = 1.66 µm vs. irrelevant peptide = 3.68 µm; p = 0.014, two-tailed t-test) and 240 minutes of conjugation (Figure 6E, TM10 peptide = 1.49 µm vs. irrelevant peptide = 3.66 µm; p = 0.008, two-tailed t-test). The former data reflect the migration of stored perforin towards the synapse immediately following activation, as evidenced by the similar localization of the D48- and δG9- labeled perforin (Figure 5D, bottom row, D48+δG9, and from the parallel analysis in Figure 5C), whereas the latter data are mostly attributable to the polarization of new perforin (Figure 6C, bottom row, D48 (blue) vs. δG9 (green), overlay). These data suggest that within activated CD8+ T cells newly produced perforin is transported to the immunological synapse, where it may promote sustained cytotoxic activity.


It has long been appreciated that CD8+ T cells play a pivotal role in the elimination of virally infected cells, and that perforin is a key mediator of this process through its distinct ability to enable the entry of apoptosis-inducing granzymes. The mechanism for perforin-mediated antigen-specific killing has previously been attributed to the exocytosis of cytotoxic granules present within the CD8+ T cell. Here we redefine this mechanism, demonstrating that virus-specific CD8+ T cells rapidly upregulate perforin after activation, and then target the protein directly to the immunological synapse. This continual production and targeted release of perforin after stimulation may allow the CD8+ T cell to recognize and kill additional targets after the initial release and depletion of the cell’s complement of pre-formed cytotoxic granules. Additionally, our results demonstrate that multiple forms of perforin are present within resting and activated CD8+ T cells, which are not necessarily constrained within cytolytic lysosomes. This indicates that many aspects of perforin protein regulation, expression, trafficking, structure, and mechanism of action remain to be elucidated.

Previous studies have suggested that perforin expression by CD8+ T cells requires entry into cell cycle and subsequent proliferation(2022). If this were so, CD8+ T cells would have a substantial period of time after initial target recognition during which they would essentially be disarmed. Although increased perforin mRNA after stimulation has been observed previously, the failure to detect new perforin protein following activation has prompted the conclusion that perforin expression is linked to the proliferative potential of the cell. Contrary to these previous observations, we find that antigen-specific CD8+ T cells can upregulate perforin protein production in as little as two hours after TCR stimulation directly ex vivo, concordant with the kinetics of perforin mRNA upregulation(24). Thus, perforin upregulation and expression clearly does not require proliferation.

Perhaps most importantly, our results indicate that de novo production of perforin following antigen-specific stimulation is biologically relevant, for when this process is inhibited by the protein-synthesis inhibitor cyclohexamide, a dramatic reduction in killing ability results. Because cyclohexamide is not solely perforin-specific, it is possible that the observed decrease in cytotoxicity is due to the inhibition of additional proteins beyond perforin, such as Fas ligand, granzymes, or TNF-α, that are also involved in mediating target cell apoptosis. Specifically targeting perforin for inhibition in human CD8+ T cells ex vivo is not trivial, but alternative avenues are currently being explored in our laboratory.

Notably, newly formed perforin can also be transported directly to the immunological synapse independent of cytolytic granules upon antigen-specific stimulation. The existence of a secondary pathway of perforin exocytosis has previously been described(23), but was postulated to be a non-specific process that leads to bystander killing instead of antigen-specific killing. Our results confirm the existence of an alternative perforin exocytosis pathway, and indicate that CD8+ T cells can specifically employ this pathway to deliver newly formed perforin to the target cell contact site.

A recent study reported the co-operation between a lysosomal cytotoxic granule and an endosomal exocytic vesicle as a prerequisite for the cytotoxic function of lymphocytes(30). The effector protein hMunc13-4 coordinates the assembly of the two organelles and then primes granule fusion at the CTL:target cell interface. Our results do not contradict, but rather complement this report. Whereas Menager, et al., employed the δG9 anti-human perforin antibody to examine the mechanism for cytolytic granule-associated perforin release, our study focused upon the production and release of newly formed perforin. It remains to be determined if newly formed perforin also transitions through the same exocytosis pathway controlled by hMunc13-4. What other organelles and co-factors mediate the transport of the new perforin directly to the immunological synapse are critically important to define for full understanding of T cell effector function.

We submit, therefore, that cytolytic granules are necessary for the proper storage of perforin and granzymes in resting CTL, making them available for immediate release upon TCR triggering. Upon activation, however, new perforin does not need to progress via the secretory lysosome pathway in order to be released from the cell. While some of the new perforin may be allocated to replenish the granules, an appreciable quantity is directly targeted to the immunological synapse to mediate cytotoxicity, thereby providing a means for continual and repeated cytotoxicity. Although secretion of cytolytic molecules has been described previously through the constitutive pathway(23), we were surprised to find that this can occur in a directed manner to participate in targeted cytotoxicity. This phenomenon is not unlike the bidirectional release of cytokines described for activated T helper cells, in which distinct trafficking proteins are associated with each secretion pathway(32, 33). Whereas T helper cells secrete cytokines and chemokines in multiple directions by different pathways, we propose that two separate trafficking routes are responsible for delivering both new perforin and pre-formed perforin stored in granules to the site of contact with the target cell.

Perforin requires additional modifications following its egress from the Golgi to achieve its active form, which typically occur in the cytolytic granule. How, then, does newly produced perforin achieve the necessary conformation required for pore formation if it does not transit via the secretory lysosomal pathway? If newly formed perforin does not achieve an active conformation within the secretory lysosomes, then it could either be modified within the immunological synapse, on the target cell membrane itself, or within an endosomal vesicle inside the target cell. The answer to this question is likely intertwined with the mechanism and location of perforin’s action, which remains controversial.


The authors thank Dr. Jeff Frelinger and Dr. David Price for generously donating the C1R cells expressing HLA-B7 and A2, respectively. Dr. John Wherry kindly provided the HLA-B7/TM10 peptide tetramer reagent. Linda Monaco-Shawver deserves notable praise for assisting with the chromium-release assays, as does Jay Gardner for generating custom Qdot-conjugated antibodies. We appreciate the efforts of the Human Immunology Core at the University of Pennsylvania, directed by Dr. James Riley, for furnishing us with Donor PBMC samples and the anti-HLA-A/B/C antibody, as well as the Flow Cytometry and Cell Sorting Facility of the Abramson Cancer Center at the University of Pennsylvania for their continued technical support. Christie Bell also deserves special recognition for her technical assistance in optimizing the detection of perforin upregulation. We thank Drs. Mario Roederer and Guido Silvestri for critical discussion of the results.

This work was supported by the following grants and organizations: NIH AI076066 (MB), AI067946 (JO), AI079731 (JO), and the W.W. Smith Foundation (MB).


1. Shiver JW, Henkart PA. A noncytotoxic mast cell tumor line exhibits potent IgE-dependent cytotoxicity after transfection with the cytolysin/perforin gene. Cell. 1991;64:1175–1181. [PubMed]
2. Shiver JW, Su L, Henkart PA. Cytotoxicity with target DNA breakdown by rat basophilic leukemia cells expressing both cytolysin and granzyme A. Cell. 1992;71:315–322. [PubMed]
3. Heusel JW, Wesselschmidt RL, Shresta S, Russell JH, Ley TJ. Cytotoxic lymphocsytes require granzyme B for the rapid induction of DNA fragmentation and apoptosis in allogeneic target cells. Cell. 1994;76:977–987. [PubMed]
4. Anel A, Buferne M, Boyer C, Schmitt-Verhulst AM, Golstein P. T cell receptor-induced Fas ligand expression in cytotoxic T lymphocyte clones is blocked by protein tyrosine kinase inhibitors and cyclosporin A. Eur J Immunol. 1994;24:2469–2476. [PubMed]
5. Kojima H, Shinohara N, Hanaoka S, Someya-Shirota Y, Takagaki Y, Ohno H, Saito T, Katayama T, Yagita H, Okumura K, et al. Two distinct pathways of specific killing revealed by perforin mutant cytotoxic T lymphocytes. Immunity. 1994;1:357–364. [PubMed]
6. Smyth MJ, Kelly JM, Sutton VR, Davis JE, Browne KA, Sayers TJ, Trapani JA. Unlocking the secrets of cytotoxic granule proteins. J Leukoc Biol. 2001;70:18–29. [PubMed]
7. Peters PJ, Borst J, Oorschot V, Fukuda M, Krahenbuhl O, Tschopp J, Slot JW, Geuze HJ. Cytotoxic T lymphocyte granules are secretory lysosomes, containing both perforin and granzymes. J Exp Med. 1991;173:1099–1109. [PMC free article] [PubMed]
8. Atkinson EA, Barry M, Darmon AJ, Shostak I, Turner PC, Moyer RW, Bleackley RC. Cytotoxic T lymphocyte-assisted suicide. Caspase 3 activation is primarily the result of the direct action of granzyme B. J Biol Chem. 1998;273:21261–21266. [PubMed]
9. Barry MJ, Heibein A, Pinkoski MJ, Lee SF, Moyer RW, Green DR, Bleackley RC. Granzyme B short-circuits the need for caspase 8 activity during granule-mediated cytotoxic T-lymphocyte killing by directly cleaving Bid. Mol Cell Biol. 2000;20:3781–3794. [PMC free article] [PubMed]
10. Shresta ST, Graubert A, Thomas DA, Raptis SZ, Ley TJ. Granzyme A initiates an alternative pathway for granule-mediated apoptosis. Immunity. 1999;10:595–605. [PubMed]
11. Beresford PJ, Xia Z, Greenberg AH, Lieberman J. Granzyme A loading induces rapid cytolysis and a novel form of DNA damage independently of caspase activation. Immunity. 1999;10:585–594. [PubMed]
12. Podack ER, Young JD, Cohn ZA. Isolation and biochemical and functional characterization of perforin 1 from cytolytic T-cell granules. Proc Natl Acad Sci U S A. 1985;82:8629–8633. [PubMed]
13. Voskoboinik IM, Thia C, Fletcher J, Ciccone A, Browne K, Smyth MJ, Trapani JA. Calcium-dependent plasma membrane binding and cell lysis by perforin are mediated through its C2 domain: A critical role for aspartate residues 429, 435, 483, and 485 but not 491. J Biol Chem. 2005;280:8426–8434. [PubMed]
14. Blumenthal RP, Millard J, Henkart MP, Reynolds CW, Henkart PA. Liposomes as targets for granule cytolysin from cytotoxic large granular lymphocyte tumors. Proc Natl Acad Sci U S A. 1984;81:5551–5555. [PubMed]
15. Masson D, Peters PJ, Geuze HJ, Borst J, Tschopp J. Interaction of chondroitin sulfate with perforin and granzymes of cytolytic T-cells is dependent on pH. Biochemistry. 1990;29:11229–11235. [PubMed]
16. Bolitho P, Voskoboinik I, Trapani JA, Smyth MJ. Apoptosis induced by the lymphocyte effector molecule perforin. Curr Opin Immunol. 2007;19:339–347. [PubMed]
17. Voskoboinik I, Smyth MJ, Trapani JA. Perforin-mediated target-cell death and immune homeostasis. Nat Rev Immunol. 2006;6:940–952. [PubMed]
18. Rothstein TL, Mage M, Jones G, McHugh LL. Cytotoxic T lymphocyte sequential killing of immobilized allogeneic tumor target cells measured by time-lapse microcinematography. J Immunol. 1978;121:1652–1656. [PubMed]
19. Sanderson CJ. The mechanism of T cell mediated cytotoxicity. I. The release of different cell components. Proc R Soc Lond B Biol Sci. 1976;192:221–239. [PubMed]
20. Meng Y, Harlin H, O'Keefe JP, Gajewski TF. Induction of cytotoxic granules in human memory CD8+ T cell subsets requires cell cycle progression. J Immunol. 2006;177:1981–1987. [PubMed]
21. Migueles SA, Laborico AC, Shupert WL, Sabbaghian MS, Rabin R, Hallahan CW, Van Baarle D, Kostense S, Miedema F, McLaughlin M, Ehler L, Metcalf J, Liu S, Connors M. HIV-specific CD8+ T cell proliferation is coupled to perforin expression and is maintained in nonprogressors. Nat Immunol. 2002;3:1061–1068. [PubMed]
22. White LS, Krishnan S, Strbo N, Liu H, Kolber MA, Lichtenheld MG, Pahwa RN, Pahwa S. Differential effects of IL-21 and IL-15 on perforin expression, lysosomal degranulation, and proliferation in CD8 T cells of patients with human immunodeficiency virus-1 (HIV) Blood. 2007;109:3873–3880. [PubMed]
23. Isaaz S, Baetz K, Olsen K, Podack E, Griffiths GM. Serial killing by cytotoxic T lymphocytes: T cell receptor triggers degranulation, re-filling of the lytic granules and secretion of lytic proteins via a non-granule pathway. Eur J Immunol. 1995;25:1071–1079. [PubMed]
24. Hersperger AR, Makedonas G, Betts MR. Flow cytometric detection of perforin upregulation in human CD8 T cells. Cytometry A. 2008 [PubMed]
25. Chattopadhyay PK, Price DA, Harper TF, Betts MR, Yu J, Gostick E, Perfetto SP, Goepfert P, Koup RA, De Rosa SC, Bruchez MP, Roederer M. Quantum dot semiconductor nanocrystals for immunophenotyping by polychromatic flow cytometry. Nat Med. 2006;12:972–977. [PubMed]
26. Betts MR, Brenchley JM, Price DA, De Rosa SC, Douek DC, Roederer M, Koup RA. Sensitive and viable identification of antigen-specific CD8+ T cells by a flow cytometric assay for degranulation. J Immunol Methods. 2003;281:65–78. [PubMed]
27. Wherry EJ, Blattman JN, Murali-Krishna K, van der Most R, Ahmed R. Viral persistence alters CD8 T-cell immunodominance and tissue distribution and results in distinct stages of functional impairment. J Virol. 2003;77:4911–4927. [PMC free article] [PubMed]
28. Banerjee PP, Pandey R, Zheng R, Suhoski MM, Monaco-Shawver L, Orange JS. Cdc42-interacting protein-4 functionally links actin and microtubule networks at the cytolytic NK cell immunological synapse. J Exp Med. 2007;204:2305–2320. [PMC free article] [PubMed]
29. Banerjee PPVD, Mathew A, Raje M, Parekh V, Prasad DV, Kumar A, Mitra D, Mishra GC. Evidence that glycoprotein 96 (B2), a stress protein, functions as a Th2-specific costimulatory molecule. J Immunol. 2002;169:3507–3518. [PubMed]
30. Menager MM, Menasche G, Romao M, Knapnougel P, Ho CH, Garfa M, Raposo G, Feldmann J, Fischer A, de Saint Basile G. Secretory cytotoxic granule maturation and exocytosis require the effector protein hMunc13–4. Nat Immunol. 2007;8:257–267. [PubMed]
31. Seabra MC, Wasmeier C. Controlling the location and activation of Rab GTPases. Curr Opin Cell Biol. 2004;16:451–457. [PubMed]
32. Huse M, Lillemeier BF, Kuhns MS, Chen DS, Davis MM. T cells use two directionally distinct pathways for cytokine secretion. Nat Immunol. 2006;7:247–255. [PubMed]
33. Poo WJ, Conrad L, Janeway CA., Jr Receptor-directed focusing of lymphokine release by helper T cells. Nature. 1988;332:378–380. [PubMed]