CD4+ regulatory T cells are essential for the maintenance of peripheral tolerance and the prevention of autoimmune diseases (1). Lymphocytes with suppressive function were initially described as CD4+ T cells displaying CD25, the alpha-chain of the IL-2 receptor, which is also expressed by activated T cells (2). Later, the transcription factor FoxP3 was found to be crucial for regulatory T cell development and function, to be Treg specific (3, 4), and intracellular FoxP3 staining was introduced to distinguish CD4+ regulatory from CD4+ effector lymphocytes (5). As an alternative to FoxP3, the combined staining of CD25 and CD127, the alpha-chain of the IL-7 receptor, may be used to identify Treg, which express high amounts of CD25 (CD25high) while being CD127 negative or low (CD127-/low) (6, 7). Along with CD25 and FoxP3, Treg constitutively express CTLA-4, GITR (1), and exhibit mostly a CD45RA- phenotype (8).
For various stimuli the apoptotic response of Treg and Teff seems to be inverted. Following isolation and culture, Treg are more prone to undergo apoptosis than Teff (9; our unpublished observations). Similarly, Treg are more sensitive to CD95L induced apoptosis compared to Teff (10). In the contrary, Treg appear to be more resistant to some other stimuli, including dexamethasone induced cell death (11), activation-induced cell death (10) and the mTOR inhibitor rapamycin (12). Further, murine CD4+CD25+ lymphocytes preferentially survived a lethal dose of irradiation, suggesting that they are more resistant to DNA damage compared with the CD4+CD25- subset (13).
Several observations suggest a distinct regulation of apoptosis in T lymphocytes according to the differentiation state. Among these, the expression of the classical marker of naïve T cells, CD45RA, inversely correlates with the expression of the death receptor CD95 (14, 15, our own unpublished observations). Further, different expression profiles of Bcl-2 family members have been described for naïve and memory T cells (16) and functional studies revealed that the antiapoptotic proteins Bcl-2 and Bcl-xL regulate the survival of resting T lymphocytes and activated effector cells, respectively (17). The latter protein seems to protect also from apoptosis induced by several genotoxic agents (18).
Supported by the observations suggesting a differential regulation of apoptosis in Treg vs Teff, and the preferential survival of murine Treg to irradiation, we hypothesized that Treg might be less sensitive to genotoxic agents compared to Teff. To address this question, PBMC of healthy donors were exposed to etoposide, a DNA topoisomerase II inhibitor. To rule out the possibility that a different resistance to genotoxic drugs was conferred by membrane transporter proteins affecting their entry or extrusion (19), we also tested ionizing irradiation which induces mainly single and double strand breaks (20). The apoptotic response of Treg was judged against that of Teff, discriminating the two subpopulations either according to surface expression of CD25 and CD127 or to intracellular FoxP3 expression. Read out systems to measure apoptosis were annexin V binding or expression of active caspase 3.
To pick out a possible mediator of the differential sensitivity to DNA damage, the apoptotic response of CD45RA- Treg was judged against that of CD45RA- Teff and the response of CD45RA+ Treg was judged against that of CD45RA+ Teff. Finally, the expression of Bcl-2 and Bcl-xL by Treg and Teff was compared.
Materials and Methods
The differential response of Treg and Teff to DNA damage was compared by measuring apoptosis within Treg and Teff using treated and untreated PBMC samples. Read out system of apoptosis was either annexin V binding or activation of caspase 3. Treg were separated from Teff either according to surface markers CD25 and CD127 or according to intracellular FoxP3 expression. To explain the different response of Treg and Teff, co-staining experiments with CD45RA, annexin V and Treg markers and colabelling with antibodies to anti-apoptotic proteins Bcl-2 and Bcl-xL along with anti-FoxP3 were performed in addition. A summary of the parameters analysed is shown in Table 1. Both etoposide and irradiation were used as DNA damage inducing agents for all single experiment series. Details of treatment, sample processing, staining procedures, and analysis were as follows.
Table 1. Parameters analysed in different experiment series
Markers of apoptosis
act. caspase 3
Dose response by Treg and Teff
• FoxP3 as Treg marker
With annexin V or
With act. caspase 3
• CD45RA Treg/Teff subset analysis
• Bcl-2/Bcl-xL expression by Treg/Teff
Induction of Apoptosis by Genotoxic Agents and Culture Conditions
Peripheral blood mononuclear cells of healthy donors were used to evaluate the apoptotic response of regulatory and effector T cells to genotoxic agents. Buffy coats were obtained from an authorized blood bank (Blood Transfusion Service, Hospital of Padova), and the study was approved by the local Ethics Committee. PBMC were isolated by density centrifugation on lymphoprep cushions (Axis-Shield, Norway). Residual red blood cells were lysed by treatment with ammonium chloride. In initial experiments, increasing doses of all agents were tested. Etoposide (Ebewe Pharma, Austria) was used at 0.36–60 μg/ml. For gamma-irradiation (137Cesium) experiments, 0.6, 1.2, and 1.8 Gy were initially tested. Thereafter, 1.2 Gy were used, a dose used in some myeloablative conditioning regimens for hematopoietic stem cell transplantation (21).
For experiments performed using a single dose, etoposide was used at 30 microg/ml, a concentration that may be reached clinically (22). Irradiated samples were cultured for 64–68 h and etoposide treated samples for 40–42 h at 1.5 × 106/ml in six well tissue culture plates in L-Glutamine containing RPMI (Gibco) supplemented with heat-inactivated FBS 10% (Cambrex, Biowhittaker).
Flow Cytometric Instrument Setup and Monitoring
A digital eight-color 10 parameter flow cytometer was used for FACS analysis (BD FACS Canto II, BD Biosciences). Acquisition and analysis were performed using FACSDiva software. Instrument parameter were set up and monitored as described (23). In brief, eight-peak rainbow calibration particles (Sphero, BD Biosciences) and CompBeads were acquired over a range of voltages to determine the voltage tolerance range. Experimental voltages were established by using CompBeads labeled with the antibodies used for the experiments. Automated compensation values were calculated with FACSDiva software. For successive daily instrument monitoring, target channel values and upper CV values were established with one-peak calibration particles (Sphero rainbow calibration particles, midrange FL1 fluorescence, BD Biosciences).
Antibodies and Reagents
The monoclonal antibodies used were from BioLegend (anti-CD4 PC5 clone RPA-T4, anti-CD4 APC Cy7, clone OKT4; anti-CD127 PE, clone HCD127; anti-CD25 PE, clone BC96); from BD Biosciences (anti-activated caspase 3 FITC, clone C92-605; anti-CD25 PC7 and PC5, clone M-A251), from Immunostep (anti-CD4 APC, clone HP2/6), from eBioscience (anti-CD127 PC5, clone eBioRDR5), from Caltag (anti-CD45RA APC, clone MEM-56), from Invitrogen (anti-Bcl-2 PE, clone 100), and from Cell Signalling Technology (anti-Bcl-xL Ax488, clone 54H6). For FoxP3 labelling, the monoclonal antibody clone 206D from BioLegend was used, one of the antibodies reported to produce the most distinct population of Treg (24), conjugated with either PE, Alexa Fluor 647 (Ax647), or Alexa Fluor 488 (Ax488). Annexin V FITC was from Bender Med Systems and used according to the manufacturer's instructions.
To exclude dead cells from analysis, propidium iodide (PI, BD Biosciences) or 7-amino actinomycin D (7-AAD, Beckman Coulter) labeling were used. For samples to be labeled intracellularly, the amine reactive infrared viability dye IR ViD was used (Molecular Probes, Invitrogen).
Surface and Intracellular Staining
For FACS analysis, one million cells were processed. The staining procedure was performed at room temperature (RT) in the dark. The reagents and buffers used for the different staining protocols are summarized in Table S1. Saturating amounts of antibodies were used.
In brief, for costaining of surface antigens and phosphatidylserine, calcium containing annexin V labeling buffer was used (NaCl 120 mM CaCl2 2.5 mM Hepes 10 mM pH 7.4). Samples were first incubated for 15′ with monoclonal antibodies and the viability dyes 7-AAD or PI, and annexin V was added during the last 10′ of incubation.
For intracellular labeling, cells were incubated first with IR ViD following the manufacturer's instructions. After 5′, surface antibodies were added, the cells incubated for further 15′, followed by washing in HBSS. Cells were next fixed for 10′ on ice with Cytofix (BD Biosciences), followed by a washing step with ice-cold FB glycine (PBS FBS 10% glycine 100 mM), and another one with PW glycine at RT (PermWash solution contained in the FoxP3 Fix Perm buffer set from BioLegend, diluted 1:10 in PBS glycine). Subsequently the samples were resuspended in PW glycine and rested 15′ at RT in the dark. After pelleting, cells were incubated with the antibodies directed to intracellular antigens for 30′ at RT, washed once with PW glycine and resuspended in FB glycine. For costaining of surface antigens, phosphatidylserine, and FoxP3, annexin V labeling buffer was employed during surface labeling and CaCl2 2.5 mM was added to Cytofix. The fixed cells were washed with annexin V labeling buffer supplemented with FBS 10% glycine 100 mM and permeabilized with PermWash solution which was diluted with annexin V labeling buffer containing glycine 100 mM.
Acquisition and Analysis
Dead cells were excluded from analysis by gating on 7AAD-, PI- or IR ViD- events. Next, among viable cells, lymphocytes were isolated according to their physical properties (FSC vs SSC), followed by isolation of CD4+ lymphocytes. The latter were divided into two subpopulations. First, Treg were isolated either as CD25highCD127-/low events or as FoxP3+ events. Next, Teff were isolated by creating an inverted gate with the help of FACSDiva software (not-CD25highCD127-/low or not-FoxP3+ events). Teff were including thus all remaining cells among CD4+ lymphocytes not being Treg.
For all experiments, 0.5 × 105 to 4 × 105 total events were acquired. Events identified as CD4+ lymphocytes were 0.2–1.7 × 105.
The results of experiments performed with increasing doses of DNA damage inducing agents were analysed with Armitage's test (chi square test for trend in proportions). To compare the apoptotic response of Treg vs Teff by costaining experiments with anti-FoxP3 and annexin V or with anti-FoxP3 and anti-active caspase 3, two sample t-tests with a permutation approach were performed. P-values were corrected for multiplicity using Bonferroni's method. The same method was used to compare CD45RA+ Treg versus CD45RA+ Teff and CD45RA- Treg versus CD45RA- Teff. The permutation approach was chosen since suitable for low sample sizes. The statistical analysis was performed with R (www.r-project.org).
Treg, Identified Either as CD25highCD127-/low or as FoxP3+ Events, are More Resistant to DNA Damage Induced Apoptosis Compared to Teff
To examine the apoptotic response by Treg and Teff to DNA damage, PBMC were exposed either to etoposide or ionizing irradiation. In a first series of experiments, Treg were identified according to the expression of surface markers CD127 and CD25, and apoptosis was measured by labeling with annexin V. Increasing doses of ionizing irradiation or etoposide were tested (Fig. 1). For all experiments performed (ionizing irradiation, n = 3; etoposide, n = 3), a significant dose dependent increase of annexin V binding by both Treg and Teff was observed. As shown for one representative experiment in Figure 1, the highest treatment dose of etoposide, induced 43.1% apoptotic cells among Teff (P < 2.2 × 10e-16). A dose dependent increase of apoptosis was observed also for Treg (P < 2.2 × 10e-16), but with much lower frequency compared to Teff with only 1.5% Treg being apoptotic in the same culture (Fig. 1). Likewise, gamma-irradiation induced a dose-dependent increase of apoptosis within Teff (P < 2.2 × 10e-16) and within Treg (P < 2.2 × 10e-16) with 1.2 Gy irradiation giving 29% apoptotic cells within Teff and only 1.7% apoptotic cells within Treg in one representative experiment (Fig. 1). In these two experiments, the relative frequency of Treg increased significantly in a dose-dependent manner from 10 to 18.5% for etoposide treated cells (P = <2.2 × 10e-16) and from 6.1% to 17.7% for ionizing irradiation treated cells (P = <2.2 × 10e-16) (Fig. 1 and Supporting Information, Fig. S1). Altogether, these observations made by using surface markers to distinguish Treg and Teff, suggest that Treg are more resistant to DNA damage than Teff.
Although traditional viability dyes do not resist to the procedure of fixation, the amine reactive label IR ViD can be used to exclude dead cells from analysis even when samples are fixed and permeabilized (25). The results obtained with surface staining could therefore be confirmed by using IR ViD to exclude dead cells, intracellular FoxP3 labeling to identify Treg and annexin V binding as a marker of apoptosis. As shown in Figure 2, the apoptotic response was much higher in Teff compared to Treg with a 3.9-fold increase of apoptosis within Treg and a 16.0 fold increase of apoptosis within Treg considering the mean value all experiments (etoposide n = 6, irradiation, n = 4; P = 0.001). In this series of tests, an increase of Treg frequency was observed in 4/4 irradiated samples and in 1/6 etoposide treated samples.
The relative resistance to DNA damage induced apoptosis of Treg was corroborated by the analysis of caspase 3 activation (Fig. 3). Likewise to surface-exposed phosphatidylserine, the apoptotic response was much higher for Teff compared to Treg (P = 0.013) with a 14.9-fold increase of caspase 3 activation in Teff (mean value of 5 experiments), while no apoptotic response was observed for Treg in the same experiments (0.9-fold increase, mean value).
Both CD45RA+ and CD45RA- Treg are More Resistant to Genotoxic Stress Compared to CD45RA+ and CD45RA- Teff
The expression of CD45RA by T lymphocytes inversely correlates with the expression of death receptor CD95 (14, 15, our own unpublished observations) suggesting that apoptosis might be differentially regulated in CD45RA- versus CD45RA+ subsets.
Treg were found to exhibit mostly a CD45RO+CD45RA- phenotype (8). To test the hypothesis that the higher resistance of Treg is related to the differentiation to a CD45RA- phenotype, costaining experiments with anti-CD45RA and annexin V were performed (Fig. 4).
The apoptotic response was higher in CD45RA+ Teff compared to CD45RA+ Treg with a 1.7-fold increase of apoptosis within Treg and a 18.2-fold increase of apoptosis within Teff considering the mean value all experiments. Similary, for the CD45RA- subsets, the apoptotic response was higher in Teff compared to Treg with a 1.1-fold increase of apoptosis for Treg and a 4.2-fold increase of apoptosis for Teff (mean values of n = 7 experiments). Hence, the results show that CD45RA+ Treg are less sensitive to DNA damage compared to CD45RA+ Teff (P = 0.016) and CD45RA- Treg are more resistant to DNA damage compared to CD45RA- Teff (P = 0.004).
Human Treg do not Express More Bcl-2 or Bcl-xL Compared to Teff
The anti-apoptotic proteins Bcl-2 and Bcl-xL are involved in the control of T cell survival as stated above. To test if Treg were more protected than Teff from DNA damage induced apoptosis through Bcl-2 and/or Bcl-xL, the expression of these proteins within Treg and Teff was compared. For both Treg and Teff Bcl-2 expression was found to be intermediate/high, whereas Bcl-xL expression was found to be low. When measured as mean fluorescence intensity, neither Bcl-2 nor Bcl-xL was detected in greater amounts in Treg compared to Teff (Fig. 5). Following DNA damage, no significant changes of Bcl-2 and Bcl-xL expression were observed within both Treg and Teff (Supporting Information, Fig. S2).
A variety of apoptotic stimuli trigger a different response of Treg and Teff. The former are more sensitive to isolation and culture (9, our unpublished observations), and to CD95L induced apoptosis compared to Teff (10). In the contrary, Treg were found to be resistant to dexamethasone induced cell death (11), activation induced cell death (10), rapamycin (12) and seem to survive a lethal dose of irradiation (13).
We therefore hypothesized that Treg might be more resistant to DNA damage and compared with the apoptotic response of human Treg and Teff to genotoxic agents.
Combined staining of surface antigens and phosphatidylserine following DNA damage showed a higher frequency of apoptosis within Teff compared to Treg. This observation was true for both stimuli tested, e.g., for ionizing irradiation and etoposide. The apoptotic response was dose-dependent for both Treg and Teff but much lower for Treg compared to Teff. The higher resistance of Treg was confirmed by using FoxP3 staining to identify Treg and annexin V or anti-active caspase 3 labelling. Annexin V binding assays revealed a much smaller apoptotic response of Treg compared to Teff. When evaluating caspase 3 activation, no apoptotic response was noticed for Treg in contrast to Teff.
A dose dependent increase of the frequency of Treg was observed in experiments performed with surface staining and increasing doses of either etoposide or irradiation. In costaining experiments with anti-FoxP3 and anti-active caspase 3, an increase of Treg frequency in 1/3 irradiated samples and in 0/2 etoposide treated samples was observed. In co-labelling experiments with annexin V plus anti-FoxP3, an increase was observed in 4/4 irradiated samples and in 1/6 etoposide treated samples. In conclusion, while the frequency of apoptotic cells was higher within Treg for all experiments, their relative frequency was not always increased following DNA damage. These apparently contradictory results may be explained by the coexistence of different forms of cell death with different kinetics, e.g., beneath the cell death induced by DNA damage, a different form of cell death induced by the isolation procedure and culture may coexist. Analysis of clinical samples during cancer therapy will be helpful to address this question.
Both etoposide and irradiation induced apoptosis with minor frequency within Treg. This observation makes transporter proteins regulating drug influx and/or outflux improbable mediators of the higher resistance observed with DNA damage inducing drugs.
Several findings indicate a distinctive regulation of apoptosis in different moments of T cell differentiation, for instance the differential expression of bcl-2 family members by naïve and memory T cells (16) or the expression of death receptor CD95, which inversely correlates with CD45RA expression (14, 15, our own unpublished observations). As Treg were reported to display mainly a CD45RO+CD45RA- phenotype (8) we analyzed the sensitivity to DNA damage induced apoptosis in CD45RA+ and CD45RA- subsets of Treg compared to their respective Teff counterpart. The results showed that both Treg subpopulations are less sensitive than their corresponding Teff subset, e.g., CD45RA+ Treg were less sensitive than CD45RA+ Teff and CD45RA- Treg were less sensitive than CD45RA- Teff. These findings suggest that the differentiation status associated with CD45RA expression is not responsible for the higher resistance of Treg to DNA damage.
The survival of resting T lymphocytes seems to be favored by Bcl-2, while Bcl-xL was found to control the survival of activated effector cells (17) and to confer resistance to various genotoxic agents (18). For mouse and human cells, differing results for Bcl-2 expression in CD25+ compared to CD25-CD4+ lymphocytes were reported. In the mouse system, Bcl-2 was slightly more expressed in the CD25+ subset (11), whereas in humans the same subset was found to express slightly less Bcl-2 (9). We measured the expression of the two proteins in Treg and Teff in both freshly isolated cells and following exposure to genotoxic agents. In all conditions tested, the anti-apoptotic proteins were not expressed in higher amounts in Treg than in Teff. Thus, neither Bcl-2 nor Bcl-xL seem to be responsible for the higher resistance of Treg to DNA damage observed in vitro.
So far, observations from a mouse model (13) and our findings in vitro suggest that Treg are more resistant to DNA damage. Analysis of clinical samples following treatment with DNA damage inducing agents will be helpful to gain more insight into the effects of genotoxic agents used in cancer treatment. It will be interesting to corroborate our observations by testing peripheral blood samples drawn from cancer patients following treatment with DNA damage inducing agents.