Extracellular NAD+ and ATP trigger the shedding of CD62L and the externalization of phosphatidylserine on murine T cells. These events depend on the P2X7 ion channel. Although ATP acts as a soluble ligand to activate P2X7, gating of P2X7 by NAD+ requires ecto-ADP-ribosyltransferase ART2.2-catalyzed transfer of the ADP-ribose moiety from NAD+ onto Arg125 of P2X7. Steady-state concentrations of NAD+ and ATP in extracellular compartments are highly regulated and usually are well below the threshold required for activating P2X7. The goal of this study was to identify possible endogenous sources of these nucleotides. We show that lysis of erythrocytes releases sufficient levels of NAD+ and ATP to induce activation of P2X7. Dilution of erythrocyte lysates or incubation of lysates at 37°C revealed that signaling by ATP fades more rapidly than that by NAD+. We further show that the routine preparation of primary lymph node and spleen cells induces the release of NAD+ in sufficient concentrations for ART2.2 to ADP-ribosylate P2X7, even at 4°C. Gating of P2X7 occurs when T cells are returned to 37°C, rapidly inducing CD62L-shedding and PS-externalization by a substantial fraction of the cells. The “spontaneous” activation of P2X7 during preparation of primary T cells could be prevented by i.v. injection of either the surrogate ART substrate etheno-NAD or ART2.2-inhibitory single domain Abs 10 min before sacrificing mice.
NAD+ and ATP are universal currencies of energy metabolism and are found in all cells in all kingdoms of life. Mounting evidence indicates that these nucleotides play important roles also as signaling molecules in the extracellular environment (1, 2). In the context of the immune system of higher organisms, it has been proposed that the release of NAD+ and ATP from lysed cells may alert cells of the immune system to tissue damage (2, 3, 4, 5, 6). Indeed, lymphocytes and macrophages are equipped with numerous sensors for extracellular NAD+ and ATP, including nucleotide-metabolizing ecto-enzymes and nucleotide receptors (7, 8).
Mono-ADP-ribosyltransferases (ARTs)4 transfer the ADP-ribose moiety from NAD+ to specific amino acids in target proteins (9, 10). This is the mechanism by which several bacterial toxins, like cholera- and pertussis-toxin, cause pathology after translocating into mammalian host cells (11). ART2.2 is a GPI-anchored, raft-associated ecto-enzyme prominently expressed by murine T cells (12, 13, 14). ART2.2 catalyzes ADP-ribosylation of CD8, the integrin LFA-1, the P2X7 receptor, and several other target proteins (12, 13, 14, 15). T cell activation induces the metalloprotease-mediated shedding of a soluble, enzymatically active isoform of ART2.2 (16). ART2-deficient mice (17) exhibit reduced sensitivities to Con-A-induced hepatitis (18).
The type II transmembrane protein CD38 is a potent ecto-NAD-glycohydrolase (ecto-NADase) (19, 20) expressed by lymphocytes, endothelial cells, and several other cell types. CD38-deficient mice show impaired humoral immune responses, neutrophil chemotaxis, and dendritic cell (DC) trafficking (21, 22, 23). Cells from CD38-deficient mice do not metabolize ecto-NAD+ efficiently, and the resulting higher levels of ecto-NAD+ lead to a higher level of ART-mediated cell surface protein ADP-ribosylation (24). CD38-deficient mice show enhanced sensitivity to insulin-dependent diabetes mellitus, which is dependent on the presence of ART2.2 (25). Likely, this reflects the enhanced activity of ART2.2 in these mice as a consequence of increased levels of extracellular NAD+ (24).
The cytolytic P2X7 receptor is a homo-trimeric, ligand-gated, nonselective ion channel that has sparked interest because of its peculiar ability to induce the formation of a large nonselective membrane pore (4, 26, 27, 28). High concentrations of extracellular ATP (0.2–2 mM) are required to gate P2X7. Much lower concentrations of extracelluar NAD+ (2–20 μM) suffice to gate P2X7 on cells coexpressing ART2.2 (13). ART2.2-catalyzed ADP-ribosylation at residue R125 presumably positions the common nucleotide-diphosphate moiety into the ligand binding site at the interface of adjacent subunits of the homotrimeric receptor (29). Activation of P2X7 on T cells by ATP or by NAD-dependent ADP-ribosylation initiates a cascade of events, including influx of calcium, the shedding of the L-selectin/CD62L homing receptor, and the externalization of phosphatidylserine (PS) on the outer leaflet of the cell membrane (13, 29, 30). Chronic activation of P2X7 induces apoptosis and cell lysis (13). The natural allelic P451L polymorphism in the C-terminal tail of murine P2X7 distinguishes common strains of laboratory mice (31). Wild-type (WT) P451 is expressed by BALB/c and most other strains of mice, whereas the 451L variant is expressed by C57BL/6 and DBA mice (31). The 451L variant displays normal membrane currents, but impaired pore formation (31, 32). Naive BALB/c T cells are very sensitive to NAD+ and ATP, whereas C57BL/6 T cells are less sensitive (13).
The plasma membrane of living cells is impermeable to NAD+ and ATP, but they can be released from cells by lytic and nonlytic mechanisms (33, 34). Intracellular levels of NAD+ and ATP are in the upper micromolar and lower millimolar range, whereas serum concentrations are two to three orders of magnitude lower and are kept low by ecto-nucleotidases. The duration of extracellular signaling via NAD+ is controlled by the CD38 family ecto-NAD-glycohydrolases/ADP-ribosylcyclases that hydrolyze NAD+ to ADP-ribose and the CD203 family of phosphodiesterases that hydrolyze NAD+ to nicotinamide mononucleotide and AMP (20, 35, 36). Similarly, the concentration of ATP and the duration of signaling via ATP in the extracelluar compartment are controlled by the CD39 family of ecto-nucleotidases and the CD203 family of phosphodiesterases which hydrolyze ATP to ADP and/or AMP (37, 38, 39).
The principle aim of this study was to determine whether endogenous sources of NAD+ and/or ATP can activate P2X7. To address this question, we used two model systems, hemolysis and mechanical manipulation of cells as during the routine preparation of primary cells from lymph nodes and spleen. Our results show that both scenarios, indeed, result in the release of endogenous nucleotides in sufficient concentration to activate P2X7, with ATP-mediated effects dissipating more rapidly than those mediated by NAD-dependent ADP-ribosylation.
Materials and Methods
Chemicals and Abs
ADP-ribose, ATP, NAD+, etheno-NAD+, and KN-62 (1-(N,O-bis(5-isoquinolinesulfonyl)-N-methyl-l-tyrosyl)-4-phenylpiperazine) were obtained from Sigma-Aldrich. PE- and FITC-conjugated mAbs and Annexin-V were purchased from BD Pharmingen, including anti-CD3ε (145- 2C11), anti-CD4 (RM4–5), anti-CD8 (53–5.8), anti-CD38 (90), and anti-CD62L (MEL-14). 32P-NAD+ was obtained from Amersham Biosciences. mAbs Nika102 (anti-ART2.2) pAb K1G (anti-P2X7) and single domain Abs s+16a and l-17 (anti-ART2.2) were prepared as described previously (12, 40, 41).
Animals and cells
BALB/c mice were obtained from The Jackson Laboratory or Charles River Laboratories. CD38-deficient mice (21) and ART2-deficient mice (17) were backcrossed onto the BALB/c background for 12 generations. The ART2−/− lines are deficient in both ART2.1 and ART2.2 (17). P2X7-deficient mice were from the twelfth backcross generation to C57BL/6 mice (3). Where indicated, mice received i.v. injections of PBS, single domain Abs in PBS, or etheno-NAD+ in PBS (200 μl into the tail vein). Mice were sacrificed by exposure to O2/CO2. All animal experiments were performed in accord with the German animal-protection law. Single-cell suspensions were prepared from lymph nodes and spleens in cold (4°C) RPMI 1640 medium by gentle dissection and passage through Nitex membrane (125-μm mesh, Tetko). Alternatively, single cell suspensions were prepared by two other techniques: 1) from spleen by gentle perfusion with 10 ml ice cold medium using a 1.0-gauge needle and 10 ml syringe or 2) from lymph nodes and spleen by gentle pipetting following preincubation in medium containing 10 mg/ml collagenase (Life Technologies) for 60 min at 4°C. B cells were depleted using magnetic cell separation with Dynabead-immobilized goat anti-mouse IgG (Dynal Biotech) at 4°C as described (13). Purity of T cells was always >95% as verified by FACS analyses using PE-conjugated anti-B220 and FITC-conjugated anti-CD3.
Preparation of erythrocyte lysates
Blood was collected from mice by retroorbital puncture into ice cold heparinized Eppendorf tubes. Erythrocytes were pelleted by centrifugation at 4°C, washed once in ice cold PBS, and were resuspended in 1.5 volumes of ice cold PBS. Cells were ruptured on ice by ultrasonication in an MK2 ultrasonic disintegration machine (MSE Scientific Instruments) by two 10-second sonifications at a power setting of 12 micron. Cellular debris was removed by centrifugation (15,000 × g) for 5 min at 4°C. Clarified lysates were diluted in ice cold PBS. Aliquots of lysates (50 μl) were added to cell suspensions in RPMI 1640 medium (50 μl, 4 × 105 cells). NAD+ concentrations in cell lysates were determined by a sensitive cycling assay (42). ATP concentrations were measured with a luciferase-based assay system according to the manufacturer’s instructions (Sigma-Aldrich). In some cases, lysates were incubated at 37°C for the indicated times before addition to T cells.
Assay for phosphatidylserine exposure and propidium iodide (PI) staining
Following treatment with exogenous NAD+, ATP, or erythrocyte lysates for the indicated times at 4 or 37°C, cells were washed in RPMI 1640 medium adjusted to 2 mM CaCl2, and were stained for 20 min on ice with FITC-conjugated Annexin-V (1 μg/ml) (BD Biosciences) and PI (10 μg/ml) before flow cytometry. In some cases, KN-62 was added during or after treatment with NAD+ and ATP. Relatively high concentrations (10 μM) were used because KN-62 exhibits low potency at mouse P2X7 receptors (43, 44). For temperature response analyses, cells were incubated in the absence or presence of ATP for 30 min at the indicated temperature in a PCR machine. For pulse chase analyses, cells were treated with NAD+ or ATP for 30 min at 4 or 37°C. Cells were then washed and incubated further for 60 min at 4 or 37°C before staining with Annexin-V/PI as above.
Immunoprecipitation and Western blot analyses
T cells (1 × 107/ml) were incubated for 20 min at 4 or 37°C in the presence of radiolabeled 32P-NAD+ (1 μM, 5μCi/ml) in RPMI 1640 medium containing 1 mM ADP-ribose. Cells were then washed in cold PBS to remove free NAD+. Cells were lysed in 250 μl PBS, 1% Triton X-100, 1 mM AEBSF at 4°C for 30 min. Cell lysates were precleared by centrifugation (14,000 × g at 4°C for 30 min) followed by incubation with 20 μl Protein-G-Sepharose (Pharmacia Biotech) for 60 min at 4°C. Immunoprecipitation was performed in parallel with K1G anti-P2X7 immune serum (1 μl) (40
Exposure of murine T cells to exogenous NAD+ or ATP causes shedding of CD62L and externalization of phosphatidylserine
Treatment of purified murine T cells with exogenous NAD+ or ATP at 37°C induced, within minutes, the externalization of phosphatidylserine on the outer leaflet of the plasma membrane (Fig. 1⇓A, panels 2 and 4) and the shedding of L-selectin/CD62L (Fig. 1⇓B, panels 2 and 4). T cells from P2X7-deficient mice (3) were completely resistant to the effects of NAD+ and ATP (Fig. 1⇓, C and D), indicating that these effects are dependent on P2X7. Consistently, WT cells incubated with NAD+ or ATP in the presence of KN-62, a specific inhibitor of P2X7 (45), neither shed CD62L nor exposed phosphatidylserine (Fig. 1⇓, A and B, panels 3 and 5). In the case of T cells obtained from C57BL/6 mice that express the 451L P2X7 variant (31), much higher concentrations of NAD+ and ATP were required to induce PS exposure and CD62L shedding than by T cells from BALB/c mice that express WT P2X7 (Fig. 1⇓C, panels 2–5 vs Fig. 1⇓A, panels 2 and 4). NAD-mediated but not ATP-mediated activation of P2X7 requires functional ART2. Hence, T cells from ART2-deficient mice (17) were sensitive to direct activation of P2X7 with the soluble ligand ATP (Fig. 1⇓, A and B, panel 9), but were resistant to NAD-induced activation of P2X7 (Fig. 1⇓, A and B, panel 7), which requires ART2-catalyzed ADP-ribosylation of P2X7 (29).
NAD+ and ATP released from lysed erythrocytes activate P2X7 on murine T cells
The membrane of living cells is impermeable to NAD+ or ATP. Cells lysed during tissue injury or in the course of necrotic cell death present a potential source of extracellular nucleotides. To test whether nucleotides released from lysed cells can activate P2X7, we exposed purified T cells to crude cell lysates generated by ultrasonication of mouse erythrocytes (Fig. 2⇓). We chose erythrocytes for these analyses because the lysis of erythrocytes occurs under various pathological conditions in vivo (46, 47, 48) and because erythrocyte lysis is commonly used to deplete these cells during the preparation of primary lymphocytes (49).
The results shown in Fig. 2⇑ demonstrate that T cells, indeed, respond in a dose-dependent manner to erythrocyte lysates with externalization of PS and shedding of CD62L (Fig. 2⇑, A and B, panels 1 and 3). As in case of exogenously added nucleotides (Fig. 1⇑), these responses were suppressed by the P2X7 antagonist KN-62 (Fig. 2⇑, A and B, panels 2 and 4). Note the lower intensity of Annexin-V staining in cells treated with concentrated vs dilute lysates (Fig. 2⇑A, panels 1 and 3). This likely reflects the presence of endogenous annexins released from lysed erythrocytes that compete with the fluorochrome-conjugated Annexin-V for binding to externalized PS on T cells. Note further that ART2-deficient cells responded only to concentrated lysates (Fig. 2⇑, A and B, panels 5 and 7) whereas WT T cells responded to both concentrated and diluted lysates (Fig. 2⇑, A and B, panels 1 and 3). These findings indicate that diluted lysates contain NAD+ but not ATP in sufficient concentration to activate P2X7, whereas concentrated lysates also contain sufficient ATP.
To better assess the relative contributions of NAD+ and ATP from erythrocyte lysates on l-selectin shedding and PS exposure by T cells, we performed comparative dose response analyses with erythrocyte lysates and exogenous NAD+, ATP, or mixtures of NAD+ and ATP. The results confirm the complete resistance of ART2-deficient T cells to NAD+ (Fig. 3⇓A), whereas WT T cells exhibit a very high sensitivity to NAD+ (EC50 2.6 μM). In contrast, both cell types respond with a similar dose dependency to ATP (EC50 120 μM) (Fig. 3⇓B). The dose responses of WT and ART2-deficient T cells to mixtures of NAD+ and ATP (e.g., at a ratio of 1:10; Fig. 3⇓C) were similar to those obtained with erythrocyte lysates (Fig. 3⇓D).
The EC50 values obtained with exogenous NAD+ or ATP correspond well to our measurements of NAD+ and ATP in erythrocyte lysates and those of previous studies (50, 51, 52). Experiments performed with ART2-deficient T cells, sensitive only to ATP, show that the effects obtained with the EC50 of ATP (∼120 μM) are reproduced with erythrocyte lysates diluted ∼10-fold, consistent with the interpolated ATP concentration in undiluted erythrocyte lysates of 1.0 mM (±400 μM, n = 13). Similarly, experiments performed with WT T cells show that the effects obtained with the EC50 of NAD+ (∼2.6 μM) are reproduced with erythrocyte lysates diluted 100-fold, consistent with the interpolated concentration of NAD+ in undiluted erythrocyte lysates of 260 μM (±80 μM, n = 13).
In these experiments, we consistently noted a slightly higher proportion of cells “spontaneously” exposing PS in preparations obtained from WT mice vs ART2-deficient mice (i.e., compare control values for WT and ART2-deficient cells in the absence of exogenously added nucleotides or lysates in Fig. 3⇑, A–D). The hypothesis that this observation reflects the exposure of cells to NAD+ before or during cell preparation will be addressed further below.
The P2X7-inducing activities in erythrocyte lysates fade upon incubation at 37°C
The fate and half-life of extracellular nucleotides is determined by numerous nucleotide-metabolizing ecto-enzymes such as CD38 and CD39, many of which are expressed by erythrocytes (53, 54). One should, therefore, expect that ultrasonication of erythrocytes would expose the released NAD+ and ATP to nucleotide-degrading enzymes. To determine whether nucleotide turnover affects the P2X7-inducing potential of erythrocyte lysates, we incubated these lysates for various times at 37°C and then assayed the capacity of the lysates to induce PS exposure by T cells (Fig. 4⇓). A substantial reduction in the capacity of lysates to induce PS exposure was seen already after 10 min of preincubation (Fig. 4⇓A, panels 2 and 5 and Fig. 4⇓B) and all PS-inducing activity was lost within 60 min (Fig. 4⇓A, panels 3 and 6 and Fig. 4⇓B). Taken together, these results indicate that erythrocyte lysates contain both, NAD+ and ATP, in sufficient concentrations to activate P2X7 on T cells, and further that both nucleotides are degraded with a half-life of less than 10 min upon incubation at 37°C. The results again pinpoint a clear difference in the background proportion of freshly prepared T cells from WT mice vs ART2-deficient mice, exposing PS on their surface (Fig. 4⇓B). The following experiments were designed to further explore this phenomenon.
In vivo blockade of ART2.2 prevents the spontaneous PS flashing and shedding of CD62L by freshly prepared lymph node cells
The “spontaneous” externalization of PS and shedding of CD62L by a fraction of primary WT T cells from freshly prepared lymph nodes (Fig. 1⇑ and 4⇑B) might reflect their prior encounter with extracellular nucleotides released from endogenous sources. The spontaneous PS exposure and shedding of CD62L evidently depend on functional ART2 and P2X7, as these phenomena are not observed in T cell preparations from ART2-deficient (Fig. 1⇑, A and B, panel 6) or P2X7-deficient (Fig. 1⇑, C and D, panel 6) mice. Spontaneous PS exposure and shedding of CD62L was specific for T cells, consistent with the notion that the effect is dependent on ART2.2, which is expressed by T cells, but not by B cells, NK cells, or macrophages (12). The finding that ART2-deficient cells do not spontaneously expose PS (Fig. 1⇑A, panel 6) or shed CD62L (Fig. 1⇑B, panel 6) indicates that these effects are caused by endogenous NAD+ rather than by endogenous ATP because ART2-deficient T cells retain unabated sensitivity to exogenous ATP (Fig. 1⇑A, panel 9). Consistently, T cells from CD38-deficient mice which lack the major NAD-hydrolyzing ecto-enzyme (21, 24) show markedly increased levels of spontaneous PS externalization (Fig. 5⇓A, panel 1) and CD62L shedding (Fig. 5⇓A, panel 3).
In principle, the spontaneous PS exposure and CD62L shedding could reflect exposure of cells to extracellular NAD+ either before, i.e., in vivo, or during cell preparation, i.e., after sacrificing of the mouse. To distinguish between these possibilities, we sought conditions that would prevent these reactions either by blocking ART2.2 or by blocking the activation of P2X7. To this end, we first used two distinct, recently described single domain Abs (sdAbs) that block the enzymatic and cytotoxic activities of ART2.2 (41). As illustrated in Fig. 5⇑, a single i.v. injection of either sdAbs s+16a or l-17 10 min before sacrificing effectively blocked both PS externalization (Fig. 5⇑A, panel 2 and Fig. 5⇑B) and shedding of CD62L (Fig. 5⇑A, panel 4 and Fig. 5⇑B), even in case of the highly susceptible cells from CD38-deficient mice. As a complementary approach, we used etheno-NAD+ as ART-substrate (55). Etheno-ADP-ribosylation does not activate P2X7 and, in addition, prevents activation of P2X7 by subsequent ADP-ribosylation (13, 29). Indeed, as in case of the sdAbs, a single i.v. injection of etheno-NAD+ 10 min before sacrificing effectively blocked both PS externalization and shedding of CD62L (Fig. 5⇑B) by freshly prepared cells. These results strongly suggest that exposure of cells to endogenous NAD+ occurred during sacrificing and/or cell preparation.
T cells prepared at 4°C do not externalize PS, but do so when returned to 37°C
To further test the hypothesis that NAD+ is released during cell preparation, we next analyzed the effects of temperature during and after cell preparation on spontaneous PS externalization and CD62L shedding (Figs. 6⇓ and 7⇓). Cells that had been prepared and kept at 4°C did not spontaneously externalize PS (Fig. 6⇓A, panels 1 and 5). In contrast, cells prepared and kept at 37°C for 30 min contained substantial numbers of spontaneously PS-exposing cells, with higher proportions of such cells in preparations from CD38-deficient vs WT mice (Fig. 6⇓A, panels 2 and 6). Remarkably, cells prepared at 4°C did expose PS when subsequently incubated at 37°C, with cells from CD38-deficient mice again showing stronger responses than WT cells (Fig. 6⇓A, panels 3 and 7). Further, cells prepared at 37°C did not re-internalize PS when subsequently returned to 4°C (Fig. 6⇓A, panels 4 and 8). These results imply that the spontaneous activation of P2X7 is caused by NAD+ released from cells during preparation rather than by exposure of cells to NAD+ before preparation in vivo, because cells exposed to NAD+ before killing of the animal, i.e., at 37°C, should still have exhibited externalized PS when subsequently prepared and kept at 4°C.
Similar results were obtained when exogenous NAD+ (25 μM) was added to the buffer during cell preparation (Fig. 6⇑B), i.e., cells exposed to NAD+ at 4°C externalized little if any PS (panels 1 and 5), while cells prepared at 37°C (panels 2 and 6) responded vividly with PS externalization as did cells that were returned to 37°C for 30 min after preparation at 4°C (panels 3 and 7). Note that addition of exogenous NAD+ to WT cells resulted in a response of similar magnitude as the spontaneous response of CD38-deficient cells (to endogenous NAD+) (Fig. 6⇑B, panel 2 vs Fig. 6⇑A, panel 6). Note further that the level of PS exposure by CD38KO cells in response to endogenously released NAD+ already was near maximal, i.e., was enhanced only slightly by the addition of exogenous NAD+ (panel 6 in Fig. 6⇑A vs panel 6 in Fig. 6⇑B). This indicates that P2X7 is already ADP ribosylated to a large extent on CD38KO cells during cell preparation.
Fig. 7⇑ illustrates more detailed analyses of the temperature-dependency of PS-externalization (Fig. 7⇑, A and B) and CD62L-shedding (Fig. 7⇑, C and D) by T cells induced by exogenously added ATP and NAD+. The results reveal that T cells do not externalize PS or shed CD62L at 4°C, even when exposed to relatively high concentrations of ATP (250 μM) or NAD+ (25 μM). Cells exposed to ATP did respond vividly at room temperature (20°C) (Fig. 7⇑, B and D), while cells exposed to NAD+ showed maximal shedding only at 37°C (Fig. 7⇑, A and C). These results indicate that the soluble ligand ATP is a better agonist for P2X7 than the covalently attached ADP-ribose. Consistent with this interpretation, the temperature response curves were shifted even further to the left when T cells were treated with the highly potent P2X7 agonist benzoyl-ATP (results not shown).
Note that the temperature response curves of CD38-deficient cells in the absence of exogenous nucleotides resemble those of WT cells exposed to exogenously added NAD+ (Fig. 7⇑, A and C). Moreover, ART2-deficient cells, which respond normally to exogenous ATP, neither spontaneously expose PS or shed CD62L nor do so in response to exogenous NAD+ at any temperature. These results substantiate the interpretation that NAD+ rather than ATP is the effective signaling molecule inducing spontaneous PS exposure and CD62L shedding during T cell preparation.
ADP-ribosylation of cell surface proteins proceeds more efficiently at 4°C than at 37°C
To assess the extent of cell surface protein ADP-ribosylation at different temperatures, we used a previously described FACS-based assay using the NAD+ analog etheno-NAD+ and the etheno-adenosine-specific mAb 1G4 to detect etheno-ADP-ribosylation of cell surface proteins (Fig. 8⇓) (55). In accord with previous reports, ART2-deficient cells did not show any detectable etheno-ADP-ribosylation of cell surface proteins (Fig. 8⇓A). Remarkably, on WT T cells, the extent of cell surface protein ADP-ribosylation decreased with increasing temperatures (Fig. 8⇓A), i.e., was much higher on cells incubated at 4°C than on those incubated at 37°C (Fig. 8⇓B, panels 1 and 2). Further, subjecting cells that had been labeled at 4°C to a subsequent incubation at 37°C resulted in a marked decrease in the level of protein ADP-ribosylation (Fig. 8⇓A, panel 3 vs panel 1). In contrast, subjecting cells that had been labeled at 37°C to a subsequent incubation at 4°C resulted in little if any detectable changes in the level of protein ADP-ribosylation (Fig. 8⇓A, panel 4 vs panel 2). Similar results were obtained with CD38-deficient T cells (Fig. 8⇓A, panels 5–8). These results show that ADP-ribosylation of cell surface proteins proceeds efficiently at 4°C and, further, suggest that labeling is reversed at 37°C, e.g., by enzymatic removal of the etheno-ADP-ribose group and/or internalization or shedding of labeled proteins.
Differential radiolabeling of P2X7, LFA-1, and CD8 in cells prepared from WT vs CD38-deficient mice
ART2.2 is known to ADP-ribosylate several distinct T cell surface proteins (14, 29, 56). ADP-ribosylation sites on cell surface proteins already occupied during cell preparation would not be available to subsequent ADP-ribosylation, e.g., upon addition of exogenously added NAD+. To determine to which extent the exposure of cells to NAD+ released during cell preparation affects the ADP-ribosylation of cell surface proteins to a subsequent exposure to exogenous NAD+, we incubated freshly prepared T cells from WT and CD38-deficient mice at 4 or 37°C with exogenously added radioactive NAD+ (1 μM), followed by lysis of cells and immunoprecipitation of known ART2-target proteins with specific Abs (Fig. 9⇓).
Incubation of cells with radiolabeled NAD+ at 4°C leads to covalent radiolabeling of numerous cell surface proteins in T cells from both, WT and CD38KO mice (Fig. 9⇑A, lanes 1 and 2). Cells from ART2KO mice do not incorporate any radiolabel under these conditions (data not shown). The results confirm the efficient activity of ART2 at 4°C and indicate that numerous binding sites are still available for ADP-ribosylation. The reduced radiolabeling of P2X7 in cells from CD38KO vs WT mice (Fig. 9⇑A, lanes 5 and 6) is consistent with the notion that most ADP-ribosylation sites on P2X7 are already occupied on cells from CD38KO mice as a result of prior ADP-ribosylation by NAD+ released during cell preparation. Incubation of cells with radiolabeled NAD+ at 37°C results in much lower radiolabeling of proteins than incubation with NAD+ at 4°C (Fig. 9⇑B vs Fig. 9⇑A), as in case of labeling with etheno-NAD+ (Fig. 8⇑A), consistent with reversion of ADP-ribosylation at 37°C. Note further that at 37°C, overall radiolabeling of proteins is lower in WT than CD38KO cells (Fig. 9⇑B, lanes 1 and 2), presumably due to CD38-mediated NAD+ hydrolysis by WT cells (24).
The results of this study demonstrate that P2X7 on T cells can be activated by endogenous sources of NAD+ and ATP released from lysed cells. We show that massive lysis of erythrocytes can result in the release of sufficient quantities of these nucleotides to activate P2X7. Our results further indicate that techniques routinely used in immunology laboratories to prepare primary lymphocytes from spleen and lymph nodes cause the release of NAD+ in sufficient quantity to gate P2X7 and to phenotypically and functionally alter a substantial fraction of cells.
In accord with previous studies (13, 57, 58), we show in this study that exposure of T cells to exogenous ATP or NAD+ triggers the P2X7-dependent externalization of phosphatidylserine and shedding of CD62L (Fig. 1⇑). Using T cells from ART2-deficient and WT mice as biological indicators, we can distinguish the effects of ATP on P2X7 from those of NAD+ on P2X7 in case of exogenously added nucleotides (Fig. 1⇑) as well as in case of endogenous nucleotides released from lysed cells (Fig. 2⇑); ATP acts as a soluble ligand that gates P2X7 on both WT and ART2-deficient cells, whereas NAD+ gates P2X7 via ART2.2-catalyzed ADP-ribosylation of R125 on WT cells but not on ART2-deficient cells (29). Even though much lower concentrations of NAD than ATP suffice to activate P2X7, the ADP-ribosylgroup linked to R125 seems to be a less potent agonist for P2X7 than ATP; even high concentrations of NAD induce a slower conversion of WT cells from the Annexin-V positive/PI negative stage to the Annexin-V/PI double-positive stage than ATP (e.g., Fig. 1⇑A, panels 2 and 4). Moreover, because ART2KO cells do not ADP-ribosylate P2X7, activation of P2X7 in these cells is entirely dependent on ATP. Consistently, ART2KO cells respond more vividly than WT cells to exogenous ATP (Fig. 1⇑A, panels 4 and 9) as well as to concentrated cell lysates (Fig. 2⇑A, panels 1 and 5, Fig. 4⇑, panels 1 and 4). Evidently, partial ADP-ribosylation of P2X7 on WT cells (in response to NAD released from cells) retards ATP-induced conversion to the double-positive stage.
The results of dose response analyses permit an estimation of the EC50 values for nucleotide-induced activation of P2X7, i.e., 120 μM for ATP and 2.6 μM for NAD+ (Fig. 3⇑). Our results indicate that such concentrations are reached in the vicinity of lysed cells. In physiological settings, the effects of extracellular nucleotides released from lysed cells on T cells would depend on both time and distance from the lysis event. Using dilution of erythrocyte lysates to mimic increasing distance, we show that ATP-mediated effects dominate only in the short range (i.e., at high concentrations of lysates), whereas NAD-mediated effects are effective also at longer ranges (i.e., at high dilutions of lysates) (Fig. 2⇑). Preincubating erythrocyte lysates at 37°C before T cell exposure indicate that the effects of ATP dissipate faster than those mediated by NAD+ (Fig. 4⇑). The relative duration of signaling by these extracellular nucleotides is determined largely by nucleotide degrading ecto-enzymes such as the transmembrane ecto-enzymes CD38 and CD39 (21, 39, 59).
Massive lysis of erythrocytes is observed in a number of pathological conditions, e.g., during malaria infection, genetically inherited hemolytic diseases, and adverse reactions to blood transfusion (46, 47, 48). It is likely that the mechanisms described in this study with mechanically disrupted erythrocytes act also in these and other settings in vivo. Indeed, recent results from three different mouse models of inflammation support the notion that nucleotides released during tissue injury induce P2X7 activation on T cells in vivo (6, 18, 25). Firstly, injection of Con A induces T cell-dependent hepatitis that is accompanied by fulminant liver cell damage as evidenced by the release of cytosolic enzymes into the circulation. Mice genetically deficient in ART2 or P2X7 develop a milder form of the disease, correlating with decreased sensitivities of liver-resident iNK-T cells in these mice to apoptosis induced by extracellular nucleotides (18). Secondly, genetic ablation of the major ecto-NAD-glycohydrolase CD38 results in elevated tissue NAD levels and enhanced levels of T cell surface ADP-ribosylation (24). Transfer of the deficient CD38 allele into the autoimmune diabetes-prone NOD/Lt background resulted in accelerated disease progression, correlating with an enhanced sensitivity of regulatory T cells to ART2-dependent NAD-induced cell death in these mice (25). Indeed, these changes were corrected when CD38 deficiency was combined with ART2 deficiency. Thirdly, local inflammatory responses induced by s.c. injection of Biogel lead to release of NAD into the inflammatory pouch, causing shedding of CD62L by T cells in the draining but not in the nondraining lymph nodes (6).
The results obtained in this study with the second experimental system, i.e., the mechanical manipulations during routine preparation of cells from lymphatic organs, are of special pertinence to immunologists working with primary lymphocytes. Our results show that routine cell preparation techniques can lead to the gating of P2X7 on a fraction of cells, and subsequently to the shedding of CD62L and externalization of PS. Similar degrees of P2X7 activation were observed whether cells were prepared from lymph nodes or spleen and when cells were prepared by gentle passage through nytex membranes, collagenase digestion, or perfusion with medium (results not shown). This has important implications for experiments designed to study lymphocyte functions both, in vitro and after adoptive transfer in vivo. Our results show that cells prepared and kept strictly at 4°C do not externalize PS or shed CD62L (Fig. 6⇑). However, when cells are prepared at 37°C or when cells are returned to 37°C subsequent to a preparation at 4°C, a substantial fraction of the cells do externalize PS and shed CD62L. In the case of WT mice, a relatively small fraction of T cells (5–10%) was affected (Fig. 6⇑A, panels 6 and 7), whereas the majority of T cells was affected in case of CD38-deficient mice (Fig. 6⇑A, panels 6 and 7), which lack the major NAD-hydrolizing ecto-enzyme (21, 24). This finding, together with the observation that ART2-deficient mice which lack the major T cell ecto-ADP-ribosyltransferase (17) do not spontaneously shed CD62L or expose PS (Fig. 7⇑) imply that NAD+ but not ATP is released in sufficient quantity to induce activation of P2X7 during cell preparation.
Consistently, blocking the ADP-ribosylation of cell surface proteins by i.v. injection of an ART2.2-inhibitory sdAb 10 min before killing of the animal, completely prevented the subsequent shedding of CD62L and externalization of PS by WT and CD38-deficient cells (Fig. 5⇑). Similar effects were achieved by injecting etheno-NAD+ before sacrifice, which results in the etheno-ADP-ribosylation of P2X7, thereby blocking the subsequent activation by ADP-ribosylation (Fig. 5⇑). ART2.2-inhibitory sdAbs are not expected to have any adverse side effects. Because sdAbs lack the Fc domain, they cannot activate complement or Ab-dependent cytotoxicity. Moreover the small (15kd) sdAbs are rapidly eliminated via the kidney, with a serum half life of <5 min. Blockade of ART2.2 by sdAbs is reversible and ART2.2 activity on lymph node cells is largely restored 24 h after injection (41). However, systemic administration of etheno-NAD+ could have unwanted side effects as this would provide other members of the ART-family with substrate, leading to the etheno-ADP-ribosylation of other cell surface proteins.
Monitoring cell surface protein ADP-ribosylation using exogenously added etheno-NAD+ (Fig. 8⇑) or 32P-NAD+ (Fig. 9⇑) confirmed that ART2.2-catalyzed ADP-ribosylation of cell surface proteins proceeds efficiently at 4°C. In contrast, the gating of P2X7 requires elevated temperatures (Fig. 7⇑). These findings imply that unwanted activation of P2X7 by ATP released from cells during cell preparation can be prevented simply by keeping cells at 4°C during preparation until the soluble ligand ATP is washed away. However, ADP-ribosylation of P2X7 in response to NAD+ released from cells during cell preparation cannot be prevented by keeping cells at 4°C because the covalently attached ADP-ribose moiety cannot be removed by washing and thus will activate P2X7 when cells are returned to 37°C. To prevent Ab-induced modulation of cell surface proteins, immunologists routinely perform staining of lymphocytes for FACS-analyses at 4°C, whereas cells are returned to 37°C for functional assays, e.g., in vitro TCR-ligation and proliferation assays or in vivo migration studies. Under such conditions, NAD-induced PS externalization and CD62L shedding due to activation of P2X7 by ADP-ribosylation could escape detection.
Both the externalization of PS and the shedding of CD62L can profoundly affect T cell functions. Externalization of PS is a common eat-me signal for macrophages, and such cells are equipped with adapter proteins and cell surface receptors for binding PS-exposing cells (60, 61, 62). Externalization of PS by T cells could thus lead to enhanced binding to and/or phagocytic clearance by macrophages. Moreover, PS exposure is associated with increased cellular adhesion to endothelia and might promote the extravasation of T cells. CD62L is the major homing receptor for lymph nodes, and cells lacking CD62L show impaired migration to peripheral lymph nodes (63, 64, 65). Metalloprotease-mediated shedding of CD62L is triggered also upon activation of T cells by engagement of the TCR or by mitogenic stimulation (66, 67). Our results indicate that a substantial fraction of cells in primary lymphocyte preparations may exhibit a CD62L-negative phenotype as a consequence of P2X7 ADP-ribosylation during cell preparation rather than as a sign of conventional T cell activation. Moreover, it is possible that the constitutive externalization of PS described for CD4+CD45RBlow cells is a consequence of exposure to NAD+ during cell preparation (68).
In this context it is important to note that common strains of laboratory mice carry allelic variants of both P2X7 and ART2 which affect the sensitivity to endogenously released nucleotides (12, 31, 69). BALB/c mice express WT P2X7 and both copies of the duplicated ART2 locus. C57BL/6 mice carry the 451L allelic variant of P2X7 with impaired sensitivities to gating by ATP and ADP-ribosylation (31, 32), as well as a defective ART2.1 allele, while expressing the ART2.2 locus at much higher levels than BALB/c mice (12, 69). Whether the reported high sensitivity of naturally occurring regulatory T cells in C57BL/6 mice to activation of P2X7 (70) is associated with NAD+ released in vivo and/or during cell preparation will be an important subject of future investigations.
The results of our experiments testing the temperature-dependency of T cell surface ADP-ribosylation reactions indicate that ADP-ribosylation of cell membrane proteins is reversible (Figs. 8⇑ and 9⇑), in accord with previous studies (56, 71). Enzymes capable of reversing ADP-ribosylation include ADP-ribosylhydrolases which can remove the entire ADP-ribose moiety (72, 73) and phosphodiesterases which remove only AMP, leaving ribose phosphate attached to the target protein (74). To date ADP-ribosylhydrolases have been described only as intracellular proteins (75), whereas phosphodiesterase isoforms have been cloned and characterized that function as membrane bound and secretory ecto-enzymes (76, 77). Because both labels used in this study (etheno-adenosine) and (α-32P) would be removed by phosphodiesterases and ADP-ribosylhydrolases, other tools will be required to determine the relative contributions of these two enzyme families to reversion of protein ADP-ribosylation. The differential labeling of P2X7 vs LFA-1 and CD8 at 4°C vs 37°C (Fig. 9⇑) indicate that ADP-ribose moieties buried in the ligand-binding pocket of P2X7 may be better protected against de-ADP-ribosylating enzymes than ADP-ribose moieties linked in a more exposed manner to other cell-membrane proteins.
That activation of P2X7 by ADP-ribosylation can play a role in physiological settings has been demonstrated previously by our finding that NAD+ released at inflammatory sites induces ART2-dependent shedding of CD62L and T cell death in draining lymph nodes (6). The results of the present study provide an additional plausible scenario for the activation of P2X7 in vivo, i.e., by NAD+ and ATP released during hemolysis. Malaria, for example, is associated with periodic hemolysis and complex changes in T cell function and apoptosis (47, 78). It will, therefore, be of interest to determine whether and to what extent the genetic ablation or pharmacological inhibition of ART2 and/or P2X7 affects disease progression in murine malaria models. Our results further demonstrate that techniques routinely used in immunology laboratories to prepare primary lymphocytes cause the release of sufficient quantities of NAD+ for ART2.2-catalyzed activation of P2X7. When cells are returned to 37°C, this induces the externalization of PS and shedding of CD62L, thereby likely altering T cell functions. An efficient means to prevent the activation of P2X7 during cell preparation is an i.v. injection of ART2.2-blocking sdAbs shortly before sacrificing.
We thank Dr. F. Lund, Saranac Lake, NY for providing CD38-deficient mice, and Dr. C. Gable, Pfizer, for providing P2X7-deficient mice. This work represents the partial fulfillment of the requirements for the graduate thesis of FS. We thank Dunja Freese, Marion Nissen, Fabienne Seyfried, and Dr. Kirsten Heiss, Hamburg, for assistance. FK-N, FH, and MS designed and supervised the study. FS, and SA performed the experiments shown in Figs. 1, 6 and 8B, NS those shown in Figs. 5, 7 and 8A, CK and FS those shown in Figs. 2–4, and PB and FS those in Fig. 9, BR and NS performed the NAD and ATP measurements. FK-N wrote the paper. We thank Drs. H.-W. Mittrücker and B. Fleischer, Hamburg, and Dr. O. Boyer, Rouen, for critical reading of the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by Deutsche Forschungsgemeinschaft Grant No310/6 (to F.K.-N. and F.H.) and by stipends from the Werner Otto Foundation (to C.K. and P.B.) and from the Fondation pour la Recherche Medical (to S.A.).
↵2 F.S. and N.S. contributed equally to this study.
↵3 Address correspondence and reprint requests to Dr. Friedrich Koch-Nolte, Institute of Immunology, University Hospital, Hamburg, Germany. E-mail address:
↵4 Abbreviations used in this paper: ART, ADP-ribosyltransferase; etheno-NAD+, 1,N6-Ethenonicotinamide adenine dinucleotide; DC, dendritic cell; PS, phosphatidylserine; WT, wild type; PI, propidium iodide; sdAb, single domain Ab.
- Received May 28, 2008.
- Accepted January 3, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.