Preferential Cell Death of CD8+ Effector Memory (CCR7−CD45RA−) T Cells by Hydrogen Peroxide-Induced Oxidative Stress 1

T cells are used in many cell-based cancer treatments. However, oxidative stress that is induced during various chronic inflammatory conditions, such as cancer, can impair the immune system and have detrimental effects on T cell function. In this study, we have investigated the sensitivity of different human T cell subsets to H2O2-induced oxidative stress. We showed that central memory (CD45RA−CCR7+) and effector memory (CD45RA−CCR7−) T cells are more sensitive to H2O2 as compared with naive (CD45RA+CCR7+) T cells. Furthermore, the study showed that CD8+ effector memory T cells are more sensitive to low levels of H2O2 (5 μM) compared with other types of T cells investigated. H2O2-exposed CD45RO+ T cells showed mitochondrial depolarization prior to caspase 3 activity. Moreover, the pan-caspase inhibitor z-Val-Ala-Asp(OMe)-fluoromethylketone rescued cells from death. These experiments suggest that H2O2-induced cell death of CD45RO+ T cells acts via the mitochondrial pathway and that caspase involvement is needed. This study suggests that oxidative stress in cancer patients can be disadvantageous for T cell-based adoptive cell transfer therapies, since effector memory T cells are the primary phenotype of the cells administered.

R eactive oxygen species (ROS) 4 , produced by granulocytes and macrophages, are important components of the innate immunity to invading microbes (1,2). However, oxidative stress associated with chronic inflammatory conditions can have detrimental effects mediated by the ability of ROS to induce cell death in a number of different cell types, including ␤ cells of the pancreas (3), neural cells (4), and lymphocytes (5). It was demonstrated in several model systems that exposure of T cells to physiological levels of oxidative stress leads to a suppressed signal transduction and transcription factor activity, a block in NF-B activation, and decreased cytokine production in response to nonspecific and Ag-specific stimulation (6 -9). The ability of ROS to suppress T lymphocyte functions may therefore be one important mechanism behind the hyporesponsiveness of the immune system often observed in various chronic inflammatory conditions, including rheumatoid arthritis (10 -12), HIV infection (13), and cancer (14,15).
In tumor immunology, the negative effect of NO and H 2 O 2 produced from activated macrophages and granulocytes on T and NK cell functions is well established. Coculturing tumorinfiltrating macrophages and freshly isolated human T cells results in decreased TCR expression and loss of Ag-specific T cell responses (16 -18). Monocytes can inhibit in vitro human NK cell-mediated cytotoxicity via secretion of H 2 O 2, leading to induction of cell death (19). In addition, macrophage-derived NO markedly reduces the phosphorylation and activation of JAK3/STAT5 signal transduction proteins, inhibiting the proliferative responses of T cells to IL-2 (20). Activated granulocytes and oxidative stress mediated by H 2 O 2 in the circulation of patients with advanced cancer was also recently described (14). Taken together, H 2 O 2 secretion by activated macrophages and granulocytes has been suggested as one possible mechanism behind the tumor-induced immune suppression with decreased signal transduction and poor effector functions of T cells and NK cells observed in cancer patients.
We have earlier described how cytokine production of human PBMC, upon stimulation with an HLA-A2-restricted influenza peptide and nonspecific receptor cross-linking, was reduced after exposure to micromolar levels of H 2 O 2 (9). This reduction of primarily Th1 cytokines was predominantly observed in the memory/effector (CD45RO ϩ ) T cell subset and correlated with a block in NF-B activation. In this study, we confirm and extend these findings in a model where the sensitivity of various T cell subsets of unstimulated human PBMC to cell death induced by low doses of H 2 O 2 has been investigated. We demonstrate that effector memory T cells (T EM ) (CCR7 Ϫ CD45RA Ϫ ) are particularly sensitive to low doses of H 2 O 2 , while central memory T cells (T CM ) (CCR7 ϩ CD45RA Ϫ ) are significantly less sensitive. The pan-caspase inhibitor z-Val-Ala-Asp(OMe)fluoromethylketone (z-VAD-FMK) was found to block the cell death of purified CD45RO ϩ T cells. Time-kinetic experiments, where the mitochondrial membrane potential and caspase 3 activity were analyzed, suggested that the mitochondrial pathway is the primary cell death pathway for CD45RO ϩ T cells exposed to low levels of H 2 O 2 .

Materials and Methods
Cells PBMC were prepared from buffy coats from healthy blood donors admitted to the blood bank at the Karolinska Hospital by means of Ficoll-Paque (Amersham Pharmacia Biotech) density gradient centrifugation. Cell concentration was adjusted to 1 ϫ 10 6 /ml in AIM-V serum-free medium (Life Technologies). Cells were subsequently exposed to titrated doses (0 -40 M) of H 2 O 2 (Sigma-Aldrich) for different time periods at 37°C in 7.5% CO 2 . In some experiments, CD45RO ϩ CD3 ϩ cells and CD45RA ϩ CD3 ϩ cells were negatively selected using a miniMACS kit according to the manufacturer's protocol (Miltenyi Biotec). In short, CD3 ϩ cells were negatively selected from PBMC by a pan-T cell isolation kit. This was followed by negative selection of CD45RO ϩ or CD45RA ϩ cells using either anti-CD45RA or anti-CD45RO beads. The purity of the obtained cells was always above 95% in each separation step.

Cell death assays
Using a flow cytometry-based method, cell death measurements were performed by the Annexin V PE apoptosis detection kit according to the manufacturer's protocol (BD Biosciences). Cells were stained with Annexin V PE and the vital dye 7-aminoactinomycin D (7-AAD). Early apoptotic cells were defined as annexin V ϩ and 7-AAD Ϫ , and late apoptotic cells were defined as annexin V ϩ and 7-AAD ϩ . Live cells were defined as double negative for these markers. To determine caspase dependency of cell death, z-Val-Ala-Asp(OMe)-fluoromethylketone (z-VAD-FMK; Alexis Biochemicals), a pan-caspase inhibitor, was used to block caspase activity. In brief, CD45RO ϩ or CD45RA ϩ T cells were isolated and cultured in AIM-V serum-free medium. To block caspase activity, z-VAD-FMK (50 or 100 M, final concentration) was added to the cell culture and incubated for 1 h at 37°C in 7.5% CO 2 pre-exposure to 5 or 20 M H 2 O 2 and subsequently cultured for different time points. Cells were then analyzed by flow cytometry. To assess the cell death pathway, the depolarization of the mitochondrial membrane potential and caspase 3 and 7 activity were measured at different time points (0 -14 h). To measure mitochondrial membrane potential, the fluorescent mitochondrial probe tetramethylrhodamine ethyl ester (TMRE; Molecular Probes) was used. Briefly, cells were cultured in AIM-V with or without 5 M H 2 O 2 for different time points. TMRE (final concentration 25 nM) was added to the cell suspension that was incubated for 20 min at 37°C in 7.5% CO 2 . Cells were washed twice in PBS containing TMRE (25 nM) and subsequently analyzed in FACS. To measure caspases 3 and 7 activity, a kit providing fluorochrome inhibitor of caspases (FLICA) was used according to the manufacturer's protocol (Caspases 3 and 7 Detection kit; Immunochemistry Technologies). FLICA binds covalently to specific active caspases, thus the fluorochrome accumulates in cells having active caspases and may be detected in FACS. In short, FLICA specific for caspases 3 and 7 (FAM-Asp(OMe)-Glu(OMe)-Val-Asp(OMe)-fluoromethylketone) were used and FLICA solution was added to cell suspension and incubated for 1 h at 37°C in 7.5% CO 2 . Cells were then washed in washing buffer included in the kit and further analyzed in FACS.

Preferential cell death of CD45RO ϩ T cells induced by hydrogen peroxide
In this study, the sensitivity of peripheral blood-derived T cells to low levels of H 2 O 2 was tested using a modification of a previously established method (9). Here, we confirm that there is a dose-dependent and selective targeting of CD45RO ϩ T cells by low levels (Ͻ40 M) of H 2 O 2 . The percentage of CD45RO ϩ T cells was significantly decreased in the live cell population following 1-day incubation of PBMC in medium containing H 2 O 2 at a concentration of 5 M or higher (Fig. 1A). This observation was confirmed by a corresponding increase in the percentage of CD45RA ϩ T cells as tested using an anti-CD45RA mAb (Fig. 1B).
We next investigated whether this H 2 O 2 -induced shift in T cell subsets, leading to a decreased proportion of CD45RO ϩ T cells in the viable cell population, was a result of differential proliferation or a consequence of enhanced cell death of this particular subset. To this end, we stained T cells exposed to titrated doses of H 2 O 2 with the cell death marker annexin V. We found that CD45RA Ϫ T cells were more prone to cell death as compared with CD45RA ϩ T cells following exposure to H 2 O 2 (Table I). Of note, H 2 O 2 induced cell death also in CD45RA ϩ T cells, but to a lesser extent. Thus, we conclude that the decrease in the percentage of CD45RO ϩ T cells is due to the enhanced sensitivity of this subset to H 2 O 2 -induced cell death.

Hydrogen peroxide selectively targets CD45RO ϩ CD8 ϩ T cells
We next asked whether there was a selective targeting of CD4 ϩ and CD8 ϩ T cells by H 2 O 2 . We found that CD8 ϩ T cells were significantly more sensitive to H 2 O 2 as compared with CD4 ϩ T cells leading to an increased CD4:CD8 ratio (Fig. 2, A and B). The effect on the CD45RO ϩ CD8 ϩ T cell subset was even more pronounced ( Fig. 2C), although the percentage of CD45RA ϩ CD8 ϩ T cells was also significantly reduced (Fig. 2D). Furthermore, the CD45RO ϩ T cells displayed a significantly larger relative decrease of CD8 ϩ T cells as compared with CD45RA ϩ T cells when exposed to 2.5 and 5 M H 2 O 2 (Fig. 2E). Importantly, the selective targeting of CD45RO ϩ T cells cannot be explained by the higher sensitivity of CD8 ϩ cells to H 2 O 2 , as the proportion of CD8 ϩ cells is lower in the CD45RO ϩ than in the CD45RA ϩ T cell population (Fig. 2, C and D). Consequently, the higher sensitivity of CD8 ϩ T cells may not be explained by the enhanced sensitivity of CD45RO ϩ T cells. Thus, the T cell sensitivity to H 2 O 2 -induced oxidative stress is dependent both on the CD4 or CD8 phenotype and on the differentiation stage of the T cell (CD45RA ϩ vs CD45RO ϩ cells).

Effector memory T cells are more sensitive to a low dose of H 2 O 2 as compared with central memory T cells
We investigated the effect of H 2 O 2 -induced oxidative stress on two memory T cell subsets previously identified based on their anatomic compartmentalization and phenotypic profiles (21), i.e., T CM (CCR7 ϩ CD45RA Ϫ ) and T EM (CCR7 Ϫ CD45RA Ϫ ). T CM primarily traffic within lymphoid tissues and are distinguished by their expression of the lymph node homing receptors CD62 ligand and CCR7. In contrast, T EM are found in peripheral tissues and do not express CCR7 and have heterogeneous expression of CD62 ligand (22). The sensitivity of T CM and T EM cells to H 2 O 2 was analyzed in PBMC from eight different healthy donors. Interestingly, a low dose of H 2 O 2 (5 M) was found to selectively target CD8 ϩ T EM (Fig. 3A). The same tendency was seen for CD4 ϩ T EM , but this observation did not reach statistical significance ( p ϭ 0.055) (Fig.  3B). In contrast, neither CD4 ϩ nor CD8 ϩ T CM cells demonstrated significant sensitivity to H 2 O 2 after exposure to 5 M H 2 O 2 (Fig.  3). After culture in the presence of 20 M H 2 O 2 , a significant decrease of both the CD4 ϩ and CD8 ϩ T EM and CD4 ϩ T CM cells was detected. However, no difference was observed in CD8 ϩ T CM (Fig. 3). Thus, we conclude that CD8 ϩ T EM cells are more sensitive to low doses of H 2 O 2 as compared with other cell types investigated.

The pan-caspase inhibitor z-VAD-FMK blocks the induction of cell death of purified CD45RO ϩ cells after H 2 O 2 exposure
To investigate whether the CD45RO ϩ T cell death, induced by H 2 O 2 (5 and 20 M), was dependent on caspase activities, PBMC from eight healthy donors were cultured 1-day in H 2 O 2 -containing medium, in the presence or absence of the pan-caspase inhibitor z-VAD-FMK. The PBMC were then stained with anti-CD3 and anti-CD45RO mAb and further analyzed by measuring the proportion of viable CD45RO ϩ T cells in the in vitro culture. When caspase activity was inhibited by z-VAD-FMK (50 and 100 M) in T cells exposed to 5 M H 2 O 2 , the percentage of viable CD45RO ϩ T cells was significantly increased (Fig. 4A). z-VAD-FMK pretreatment also increased the proportion of viable CD45RO ϩ T cells that was exposed to 20 M H 2 O 2 (Fig. 4B), although not to the same extent as cells exposed to 5 M H 2 O 2 . In conclusion, the ability of z-VAD-FMK to suppress the decline of CD45RO ϩ T cells implies that H 2 O 2 -induced cell death is largely dependent on caspase activation.
Next, CD45RO ϩ and CD45RA ϩ T cells were isolated by negative selection with magnetic beads, resulting in Ͼ95% pure cell populations of the respective T cell subtype. These purified T cell subsets were then used to investigate whether a low dose of H 2 O 2 (5 M), at various time points, would result in cell death, as analyzed by flow cytometry viability staining (annexin V/7-AAD). The time-kinetic experiments revealed an increase in both early (annexin V ϩ , 7-AAD Ϫ ) and late (annexin V ϩ , 7-AAD ϩ ) apoptotic cells in the CD45RO ϩ T cell subset after 6 h of 5 M H 2 O 2 treatment, whereas no cell death was detected after 3-h in vitro exposure (Fig. 5A). Cell death of CD45RO ϩ T cells gradually increased 12 h after H 2 O 2 treatment, and after 20 h the majority of the cells in this subpopulation was in the late apoptotic stage. The inclusion of 100 M z-VAD-FMK protected the CD45RO ϩ T cells from cell death even when measured 20 h after H 2 O 2 treatment, confirming the caspase dependency of H 2 O 2 -induced cell death in this model. In sharp contrast to the high sensitivity of the CD45RO ϩ T cell subset, CD45RA ϩ T cells were completely resistant to cell death induced by 5 M H 2 O 2 even 20 h after treatment (Fig. 5B). Of note, no major difference was apparent in the spontaneous cell death between CD45RO ϩ and CD45RA ϩ T cells after 72 h of in vitro culture (Fig. 5).

H 2 O 2 induces cell death of CD45RO ϩ T cells through the mitochondrial pathway
T cell death has been shown to be executed by either the cell surface death receptor pathway or the mitochondrial pathway (23). The cell surface death receptor pathway is characterized by activation of caspase 8 before caspase 3 activity, followed by mitochondrial depolarization and cytochrome c release resulting in cell death. In contrast, the mitochondrial pathway exhibits a depolarization of mitochondrial membrane before caspase activity and cell death.
To investigate the primary pathway of H 2 O 2 -dependent T cell death, negatively selected CD45RO ϩ and CD45RA ϩ T cells were exposed to 5 M H 2 O 2 . The T cell subsets were then examined at different time points for caspase 3 and 7 (hereafter referred to as caspase 3 activity) activity, using caspase-specific fluorescent substrates (FLICA), and for mitochondrial membrane potential, using a mitochondrial specific dye (TMRE). In the CD45RO ϩ T cell subset, the time-kinetic experiments showed a depolarization of mitochondrial potential at 8 h after H 2 O 2 exposure (Fig. 6A) followed by an increase in caspase 3 activity at 10 h (Fig. 6C). The CD45RA ϩ T cell subset did not show any mitochondrial depolarization or caspase 3 activity at any time point (Fig. 6, B and D). An activation of caspase 8 was also observed after 10 h in CD45RO ϩ T cells, but not in CD45RA ϩ T cells, implying a H 2 O 2 -induced caspase 3-dependent activation of caspase 8 in CD45RO ϩ T cells (data not shown). The experiment suggests that the mitochondrial pathway is the primary pathway for CD45RO ϩ T cell death exposed to 5 M H 2 O 2 . Furthermore, the experiment indicates that the difference in cell death between the CD45RO ϩ and CD45RA ϩ subsets is upstream of the mitochondria in the cell death signaling pathway.
a Cells were stained with anti-CD3, anti-CD45RA mAb, and PE-labeled annexin V and further analyzed by FACS.
b One million PBMC were incubated in AIM-V serum-free medium in the absence or presence of H 2 O 2 (0 -10 M) for 21 h. c The percent annexin V ϩ cells in each cell population (i.e, annexin V ϩ CD45RA ϩ CD3 ϩ cells/total CD45RA ϩ CD3 ϩ cells).

Discussion
This study shows that human PBMC-derived T cells demonstrate a difference in their sensitivity to H 2 O 2 -induced cell death depending on their differentiation stage. Although the CD45RO ϩ T cells, in particular the CD8 ϩ T EM (CCR7 Ϫ CD45RO ϩ ), were found to be most sensitive to cell death induced by low doses of exogenous H 2 O 2 , the CD45RA ϩ T cell subset was relatively insensitive. Furthermore, we found that CD45RO ϩ T cells exposed to low-dose H 2 O 2 gradually started to die within a period of 6 -8 h in a caspase-dependent manner where the mitochondrial pathway seemed to predominate.
The finding of CD45RO ϩ T cells, in contrast to CD45RA ϩ T cells, being more sensitive to cell death induced by H 2 O 2 , confirms and extends our previous study in which the capacity of stimulated T cells to produce cytokines under conditions of oxidative stress was analyzed (9). In this earlier study we noticed that the CD45RO ϩ T cell subset lost the capacity to produce IFN-␥, TNF-␣, and IL-2 after stimulation with PMA/ionomycin or anti-CD3 mAb if they were pre-exposed to H 2 O 2 , whereas CD45RA ϩ T cells producing these cytokines were not affected. We speculated that the lost functional capacity of the H 2 O 2 -exposed CD45RO ϩ T cell subset may reflect an early "preapoptotic" condition, although in this particular experimental setting of that study where the H 2 O 2 was washed away following a 10-min exposure period, no significant cell death occurred in the lower mircomolar (Ͻ25 M) H 2 O 2 range. Motivated by this study, we therefore have changed the experimental setup and have found that if PBMC-derived T cells are subjected to a more sustained exposure to H 2 O 2 , where this molecule is present throughout the entire incubation period, a higher susceptibility in the CD45RO ϩ T cell subset to H 2 O 2 -induced cell death can be demonstrated also in the lower micromolar range.
Exposure of cells to extracellular ROS is known to mediate cell death also in several other model systems, which include PMAinduced death of neutrophils, HIV-induced death of T cells, death of pancreatic ␤ cells exposed to H 2 O 2 , and excitoxic neural cell death (3-5, 13, 24, 25). Also, others have studied this phenomenon in human T cell lymphomas (26 -28), and several features of the H 2 O 2 -induced cell death observed here in freshly isolated CD45RO ϩ human T cells are similar to those described in the human T cell lines Jurkat or CEM C7. These include induction of caspase 3 activity and the ability of z-VAD-FMK to prevent cell death (26 -28). Also, in line with our findings on freshly isolated CD45RO ϩ T cells is the previous finding of others that H 2 O 2 can induce the loss of the mitochondrial membrane potential and release of cytochrome c (28,29). Our results from the time-kinetic analysis demonstrated that the depolarization of mitochondrial membrane potential occurs before caspase activation, indicating that the mitochondrial pathway is predominant in H 2 O 2 -induced cell death of CD45RO ϩ human T cells. Also, Dumont et al. (27) concluded, based on inhibition experiments with drugs, that the mitochondrial cell death pathway is predominant upon exposure of human T cell lines with 100 M H 2 O 2 . The present report is however the first one to show that a similar mechanism of caspasedependent H 2 O 2 -induced cell death also occurs in nontransformed human T lymphocytes.
These represent examples of cell death caused by H 2 O 2 , which is produced by cells other than the "target" cell in a "paracrine" fashion. In contrast to these findings and to our model described here, ROS have also been reported to mediate cell death in an "autocrine" fashion, acting as an internal messenger regulating signals involved in cell death of T cells. Thus, T cells activated in vivo through injection of mice with the superantigen staphylococcal enterotoxin A were shown to die via a Fas-and TNF-␣-independent cell death (23). This activation-induced T cell death was characterized by caspase-independent loss of mitochondrial transmembrane potential, caspase-dependent DNA loss, and enhanced generation of ROS, and ROS was suggested to regulate both caspase activation and cell death in this model. Others have demonstrated that peripheral T cells cultured in the absence of survival factors accumulate ROS and up-regulate BIM (Bcl-2-interacting mediator of death) and inducible NO synthase expression, which culminates in Fas-independent "neglect-induced death" (30). Also in this phenomenon, antioxidants were shown to inhibit cell death, Bim induction, and caspase activation, implicating the direct role of ROS in cell death induction. A possible relationship between these sets of observations, where ROS FIGURE 2. CD45RO ϩ CD8 ϩ T cells display the largest sensitivity to H 2 O 2 -induced cell death. PBMC were exposed to H 2 O 2 (0 -20 M) for 21 h. A and B, PBMC from five donors were stained with anti-CD3 and anti-CD8 or anti-CD4 mAb, and the viable lymphocyte population was analyzed by FACS. A, The proportion of CD8 ϩ T cells; B, the CD4:CD8 ratio after in vitro culturing. Each donor is indicated by an individual symbol. C and D, PBMC from six donors were stained with anti-CD3, anti-CD8, and anti-CD45RO or anti-CD45RA mAb and analyzed by FACS. The proportion of CD45RO ϩ CD8 ϩ (C) and CD45RA ϩ CD8 ϩ (D) T cells after in vitro culturing is shown. Each donor is indicated by an individual symbol. The horizontal bar indicates the median in each experimental group. E, The median of the relative decrease of CD8 ϩ T cells in the CD45RO ϩ and CD45RA ϩ subset compared with unexposed cells (i.e., percent CD8 ϩ cell at is produced and acting internally in T cells to induce cell death, and the phenomenon we have studied here, i.e., cell death induced by externally produced or added H 2 O 2 , remains to be elucidated. It is possible that the externally added H 2 O 2 may penetrate the cell membrane of T cells, and thus trigger cell death by inducing molecules in a fashion comparable to that observed in the models above. Furthermore, the disparity of sensitivity between CD45RO ϩ and CD45RA ϩ T cells and CD8 ϩ and CD4 ϩ cells may be due to altered antiapoptotic (e.g., Bcl-2 and Bcl-x) and proapoptotic molecule expression (e.g., Bax, Bak, and Bim) levels. Yokoyama et al. (31) showed that peripherally obtained CD8 ϩ T cells have significantly higher expression of Bcl-x and Bax than CD4 ϩ cells, suggesting that CD4 ϩ and CD8 ϩ cells may have a different sensitivity to activation-induced cell death (31). However, when these investigators activated T cells with Con A, there was no skewed survival of any of the subsets. In this study, we have demonstrated a difference in susceptibility of the H 2 O 2 -derived cell death of CD8 ϩ and CD4 ϩ T cells and it could be speculated that the differences of Bcl-x and Bax expression may play a role. Furthermore, others have shown that CD45RO ϩ T cells express significantly less Bcl-2 than CD45RA ϩ T cells (32) and that the levels of Bcl-2 and Bcl-x L in T cells decreases upon activation (33)(34)(35)(36). The decrease of these antiapoptotic molecules in activated T cells may explain the differences seen in this study regarding the enhanced susceptibility of CD45RO ϩ T cells, and especially CD8 ϩ T EM , to H 2 O 2 -induced oxidative stress. As antiapoptotic proteins are targeted by the nuclear transcription factor NF-B (37-41), the previously described H 2 O 2 -induced down-modulation of NF-B (9) may further decrease the expression of antiapoptotic molecules, leading to enhanced sensitivity to cell death.
A sustained exposure to oxidative stress could be the underlying mechanisms behind the immunosuppression generated in various pathological conditions, including cancer, autoimmune, and infectious diseases (11,(42)(43)(44)(45)(46)(47)(48). We found the CD8 ϩ subtype of the CD45RO ϩ T cell compartment to be more sensitive to H 2 O 2 as compared with the CD4 ϩ T cell subtype. It is of interest to consider whether this could be related to the observation of increased spontaneous cell death among CD8 ϩ T cells in PBL from cancer patients and in mice with experimental tumors (49 -52). It has been suggested that cytokines, such as IFN-␥, produced by activated T cells, NK cells, or APCs may be the initial step in recruiting "regulatory" non-T cells that induce immune suppression and downregulation of CD3 expression (53). These regulatory cells may be ROS-producing monocytes/macrophages/granulocytes, as initially demonstrated with monocytes recovered by centrifugal elutriation from human PBMC (19), granulocytes, or immature myeloid cells in the tumor microenvironment (17, 52, 54, 55), or in advanced  The brackets indicate the significant difference between unexposed cells and H 2 O 2 -exposed cells, and the significant difference between H 2 O 2 -exposed cells and H 2 O 2 -exposed cells pretreated with z-VAD-FMK. The horizontal bar indicates the median in each experimental group. disease even in the peripheral circulation (14). This mechanism may initially serve to down-regulate an immune response that is potentially harmful, but when becoming chronic may itself cause injury and sustained immune suppression.
The phenomenon studied here could explain why various regimens of adoptive or active immunotherapy often fail to generate the desired clinical effects in the majority of treated cancer patients. The existence of ROS-producing cells within tumors or inflammatory foci or in the circulation of cancer-bearing individuals or patients with viral or bacterial infections may be particularly detrimental when considering adoptive immune therapy approaches. Tumor-specific T cell lines expanded in IL-2 and derived from tumor-infiltrating lymphocytes of patients with advanced cancer have been shown to predominantly have the CD8 ϩ memory effector T cell phenotype (56,57), which we here demonstrate are highly sensitive to ROS. Therefore, one could predict that these cells upon injection into the circulation of patients with advanced cancer or when entering the microenvironment of tumors are  CD3ϩ, CD45ROϩ or CD45RAϩ T cells were isolated by negative selection using magnetic beads and cultured for different time points. The isolated cells were incubated with or without 100 M z-VAD-FMK 1 h before 5 M H 2 O 2 exposure. After the incubation period, the cells were stained with 7-AAD and PE-labeled annexin V and further analyzed in FACS. Early apoptotic cells were defined as 7-AAD Ϫ annexin V ϩ cells; late apoptotic cells were defined as 7-AAD ϩ annexin V ϩ cells. The percentage of early and late apoptotic cells is indicated in each dot plot. A, First row shows the cell death of unexposed CD45RO ϩ T cells at different time points (0, 3, 6, 12, 20, and 72 h). Second row shows the cell death of CD45RO ϩ CD3 ϩ cells at different time points (3,6,12, and 20 h) exposed to 5 M H 2 O 2 . Third row shows the cell death of z-VAD-FMK-pretreated CD45RO ϩ CD3 ϩ cells at different time points (6,12, and 20 h) exposed to 5 M H 2 O 2 . B, The cell death of CD45RA ϩ CD3 ϩ cells using the same setup as in A.
rapidly eliminated through the ROS-dependent mechanism described here. Combination of antioxidant treatment strategies, such as administration of high doses of vitamin E (58), with adoptive or active immunotherapy should therefore be considered in the treatment of cancer patients to enhance the function and survival of the injected T cells.