Mitochondria Regulate Neutrophil Activation by Generating ATP for Autocrine Purinergic Signaling*

Background: Purinergic signaling regulates neutrophil activation. The source of ATP required for this process is unknown. Results: Mitochondria form the ATP that initiates purinergic signaling, and glycolysis provides the ATP that sustains functional responses of activated neutrophils. Conclusion: Mitochondrial ATP has a central role in triggering neutrophil activation. Significance: Mitochondria are regulators of the innate immune defense provided by neutrophils. Polymorphonuclear neutrophils (PMNs) form the first line of defense against invading microorganisms. We have shown previously that ATP release and autocrine purinergic signaling via P2Y2 receptors are essential for PMN activation. Here we show that mitochondria provide the ATP that initiates PMN activation. Stimulation of formyl peptide receptors increases the mitochondrial membrane potential (Δψm) and triggers a rapid burst of ATP release from PMNs. This burst of ATP release can be blocked by inhibitors of mitochondrial ATP production and requires an initial formyl peptide receptor-induced Ca2+ signal that triggers mitochondrial activation. The burst of ATP release generated by the mitochondria fuels a first phase of purinergic signaling that boosts Ca2+ signaling, amplifies mitochondrial ATP production, and initiates functional PMN responses. Cells then switch to glycolytic ATP production, which fuels a second round of purinergic signaling that sustains Ca2+ signaling via P2X receptor-mediated Ca2+ influx and maintains functional PMN responses such as oxidative burst, degranulation, and phagocytosis.

Polymorphonuclear neutrophils (PMNs) form the first line of defense against invading microorganisms. We have shown previously that ATP release and autocrine purinergic signaling via P2Y 2 receptors are essential for PMN activation. Here we show that mitochondria provide the ATP that initiates PMN activation. Stimulation of formyl peptide receptors increases the mitochondrial membrane potential (⌬m) and triggers a rapid burst of ATP release from PMNs. This burst of ATP release can be blocked by inhibitors of mitochondrial ATP production and requires an initial formyl peptide receptor-induced Ca 2؉ signal that triggers mitochondrial activation. The burst of ATP release generated by the mitochondria fuels a first phase of purinergic signaling that boosts Ca 2؉ signaling, amplifies mitochondrial ATP production, and initiates functional PMN responses. Cells then switch to glycolytic ATP production, which fuels a second round of purinergic signaling that sustains Ca 2؉ signaling via P2X receptor-mediated Ca 2؉ influx and maintains functional PMN responses such as oxidative burst, degranulation, and phagocytosis.
PMNs 2 represent the majority of all immune cells in human blood. They form the first line of defense against invading microorganisms. PMNs have evolved highly sophisticated strategies to detect, locate, and kill invading microorganisms before these invaders can multiply, spread within the host, and cause systemic infections and sepsis.
PMNs possess specialized "danger" receptors that allow them to detect bacterial products. For example, formyl peptide receptors (FPRs) recognize formylated peptides that are shed by bacteria. PMNs follow chemical gradients of such bacterial products to locate potential sources of infection and to find, phagocytize, and kill bacteria and other pathogens that could cause harm to their host. It is intriguing that signals received via a single type of danger receptor on the surface of PMNs can regulate the complex array of functional responses PMNs must execute to fulfill their tasks in host defense. Our previous work has shown that FPR stimulation induces localized release of cellular ATP from PMNs and that autocrine feedback mechanisms fueled by the released ATP help orchestrate the functional responses required for complex neutrophil functions such as chemotaxis (1).
ATP is the main carrier of cellular energy. However, in addition to its role as an energy carrier, ATP can also be released into the extracellular space where it can serve as an extracellular messenger molecule that facilitates communications between adjacent cells (2)(3)(4). We have reported previously that ATP release can also serve as an autocrine signaling molecule that drives purinergic signaling mechanisms via P2Y 2 and other types of purinergic receptors. These purinergic signaling processes are indispensable for the signaling cascades that trigger PMN activation (5). We found that FPR stimulation induces ATP release through pannexin-1 (panx1) channels and that panx1 channels are rapidly redistributed to the leading edge of polarized PMNs, allowing ATP to be released at specific membrane regions to promote gradient sensing, pseudopod protrusion, and coordinated cell migration during PMN chemotaxis (5)(6)(7). ATP released into the extracellular space of PMNs thus serves to fine-tune functional responses by autocrine feedback through a diverse set of purinergic receptor subtypes that elicit a variety of downstream signaling cascades. These purinergic receptors include members of the seven P2X receptor subtypes, the eight P2Y receptors, and the four different P1 receptors that recognize ATP and its breakdown products, respectively (1,4).
PMNs express primarily P2Y 2 , A2a, and A3 receptors that have opposing effects on FPR-induced activation events and shape functional cell responses to fMLP stimulation (6). Although we found that panx1-induced ATP release has a critical role in activating these different purinergic receptors, the intracellular origin of the ATP that drives these autocrine feedback mechanisms is unknown. Mammalian cells generate much of their intracellular ATP by glycolysis that takes place within the cytosol or by oxidative phosphorylation and ATP synthesis that are carried out by mitochondria. Compared with most other mammalian cell types, PMNs possess few mitochondria and rely primarily on glycolysis for their ATP production needs (8,9). These observations have led to the impression that mitochondria play a minor if any role in PMN physiology.
Here we studied whether mitochondria contribute to purinergic signaling in PMNs. We found that glycolysis is indeed the main mechanism by which neutrophils generate intracellular ATP. However, our results have also shown that mitochondria have an essential role in PMN activation, namely by producing ATP that triggers the purinergic signaling processes involved in PMN activation.
Human Neutrophil Isolation-The Institutional Review Board of the Beth Israel Deaconess Medical Center approved all studies involving human subjects. PMNs were isolated from the peripheral blood of healthy volunteers as described previously using dextran sedimentation followed by Percoll gradient centrifugation (5-7). Cell preparations were kept pyrogen-free, and osmotic or excessive mechanical stimulation was carefully avoided.
Measurement of ATP-Freshly isolated PMNs (1 ϫ 10 6 /ml) were treated for 10 min with different concentrations of the reagents described below unless otherwise specified, stimulated with fMLP (10 or 100 nM) for the indicated times, placed on ice to stop reactions, and centrifuged at 325 ϫ g for 5 min at 0°C. The supernatants were treated as described previously to stabilize nucleotides (7); cell pellets were resuspended in HBSS, treated with 0.8 M perchloric acid, and sonicated on ice to extract intracellular ATP. The amount of ATP released or remaining within the intracellular compartment was determined by high performance liquid chromatography (HPLC) using fluorescence detection of adenine compounds as described previously (7,10). Briefly, etheno derivatives were generated by incubating samples at 72°C in the presence of chloroacetaldehyde. Then samples were separated using a Waters HPLC system (Milford, MA) as described previously (7) and etheno derivatives were analyzed using a Waters 747 fluorescence detector.
Real Time Monitoring of ATP Release-For real time imaging of ATP release into the extracellular space, purified PMNs were placed into 35-mm glass bottom dishes (MatTek Corp., Ashland, MA) or into 8-well glass bottom dishes (Lab-Tek, Rochester, NY) coated with 40 g/ml tissue culture grade human fibronectin (Sigma-Aldrich) for 30 min at room temperature. After attachment of cells to the fibronectin-coated glass, PMNs were washed, and the supernatants were replaced with fresh HBSS containing 500 nM of the cell surface-targeting ATP probe, 2-2Zn(II), which was kindly provided by Drs. Kurishita and Hamachi from Kyoto University (11). Fluorescence imaging was done with a Leica DMI 6000B microscope (Leica, Wetzlar, Germany) equipped with a Spot Boost EMCCD camera (Diagnostic Instruments Inc., Sterling Heights, MI). Fluorescence images were taken through a 100ϫ oil objective (numerical aperture, 1.30). For all imaging experiments with this instrument, a YFP-2427A filter set (Semrock, Rochester, NY) was used to assess 2-2Zn(II) fluorescence. Four-dimensional (z-and t-stacks) imaging was done on an Ultraview Vox spinning disk confocal microscope (PerkinElmer Life Sciences) equipped with a Hamamatsu-C9100-13 camera (Intellectual and Developmental Disabilities Research Center Imaging Core, Boston Children's Hospital, Boston, MA) using Volocity software.
For the study of ⌬m and ROS production in individual cells, freshly isolated human PMNs (2 ϫ 10 6 /ml) were placed into 25-mm glass bottom dishes as described above. The cells were incubated at 37°C in a temperature-controlled stage incubator (Harvard Apparatus, Holliston, MA). Then dihydrorhodamine (DHR) 123 (Molecular Probes) or JC-1 was added at final concentrations of 2 M or 1 g/ml, respectively. Cells were incubated at 37°C for another 20 min before analysis. Then cells were stimulated as indicated in the text, and JC-1 or rhodamine 123 fluorescence changes were recorded using the Leica microscope described above along with a CARV II confocal unit (BD Biosciences) equipped with filter sets (Semrock; BrightLine FF593 and BrightLine FF495) for imaging of Texas Red (for JC-1 red) and enhanced GFP (for JC-1 green and rhodamine 123). In some experiments, an fMLP gradient was generated using a micropipette filled with 100 nM fMLP to observe mitochondrial activity during neutrophil chemotaxis.
Cell Culture and Transfection-In some cases, the neutrophil-like HL-60 cell line was used. Cells were maintained as described previously (5,6). To obtain a neutrophil-like phenotype, HL-60 cells were differentiated by treatment with 1.3% DMSO for 3 days. Then the cells were transfected with panx1-TurboGFP plasmid (pCMV6-AC-GFP from OriGene, Rockville, MD) using electroporation with a Neon transfection sys-tem (Invitrogen) set to 1,350 V, 30 ms, and 1 pulse as suggested by the manufacturer's protocol. After 2-4 h following electroporation, cells were placed into cover glass bottom dishes, stained with MitoTracker Red CM-H 2 XRos (100 nM; Molecular Probes), and stimulated with fMLP, and time lapse images were acquired with spinning disk confocal imaging using the equipment described above to assess colocalization of mitochondria and panx1 channels during cell activation.
MAPK Activation-ERK and p38 MAPK activation in response to fMLP stimulation was assessed with Western blotting and phosphospecific antibodies as described previously (5,6).
Measurement of Intracellular [Ca 2ϩ ]-PMNs were loaded with Fluo-4 AM (Molecular Probes) according to the manufacturer's instructions. Loaded cells (10 6 /ml in HBSS) were stimulated with 10 nM fMLP, and changes in intracellular Ca 2ϩ concentrations were measured using a Hitachi F-4500 fluorescence spectrophotometer equipped with a stirred and temperaturecontrolled cuvette (Hitachi, Tokyo, Japan).
Oxidative Burst and Degranulation Assays-Oxidative burst was measured with DHR 123 as described previously (12). Briefly, cells preincubated at 37°C were subjected to the treatments described below. Then DHR 123 was added at a final concentration of 2 M, and cells were stimulated with fMLP for 20 min. Samples were placed on ice for 10 min to stop DHR 123 conversion, and then rhodamine 123 fluorescence was assessed immediately using flow cytometry as described above. PMNs were gated based on their characteristic forward and side scatter properties, and histograms of FL1 fluorescence distribution were used to analyze functional responses. Degranulation was assessed by measuring the increase in the abundance of CD11b molecules on the cell surface in response to stimulation with fMLP for 10 min. Cells were chilled on ice and stained for 20 min each with monoclonal mouse antibodies against human CD11b/Mac-1 (BD Pharmingen) and phycoerythrin-conjugated anti-mouse IgG (Sigma). The cells were fixed in FACS sheath fluid containing 0.5% formaldehyde and analyzed by flow cytometry. The FL2 fluorescence channel signal was used to assess the abundance of CD11b-phycoerythrin on the surface of PMN.
Phagocytosis Assay-Alexa Fluor 488-conjugated Escherichia coli suspensions were from Invitrogen. Phagocytosis of these bacteria was analyzed by flow cytometry according to the manufacturer's guidelines. Briefly, bacteria were opsonized by incubating equal volumes of 10% human serum and bacterial suspension (20 mg/ml) in HBSS for 1 h at 37°C. After three washes in HBSS, bacteria were incubated for 1 h at 37°C with PMNs at a ratio of 10 bacteria per PMN. Reactions were stopped by the addition of ice-cold 1% formaldehyde in HBSS, and the cells were washed with 1% formaldehyde in HBSS and analyzed by flow cytometry. FL1 fluorescence and side scatter/forward scatter properties were used to assess the association of bacteria with PMNs. Cell samples mixed with bacteria for Ͻ10 s were used as controls to distinguish phagocytosis from adherence of bacteria to the cell surface of PMNs.
Statistical Analyses-Data are shown as means Ϯ S.D. unless otherwise stated. Statistical analyses were done using Student's t test, and differences were considered significant at p Ͻ 0.05.

Cell Stimulation Causes Rapid Bursts of ATP Release-We
found previously that fMLP induces ATP release from PMNs and that autocrine feedback mechanisms involving A2a, A3, and P2Y 2 receptors control PMN functions (5-7). ATP release is one of the earliest signaling events triggered by FPR stimulation. Studying the timing of ATP release after fMLP stimulation in more detail, we were surprised to find that extracellular ATP levels peaked immediately (Յ2 s) following cell stimulation with peak levels on or before the first sample could be collected for analysis by HPLC (Fig. 1A). In addition, we found that the peak and duration of extracellular ATP accumulation depended on the fMLP concentration used to stimulate the cells. Because of the inherent limitations of HPLC analysis of the ATP released into the culture medium, which provides only a crude approximation of the ATP release dynamics of stimulated cells, we developed a new real time method to monitor ATP release from PMNs with live cell imaging. For that purpose, we used a novel fluorescent ATP probe, termed 2-2Zn(II), that had been generated and kindly provided to us by Drs. Kurishita and Hamachi from Kyoto University (11). This probe has a membrane-anchoring structure that allows it to bind to the cell surface on PMNs. Using this approach, we found that fMLP triggered ATP release within Յ1 s of cell stimulation (supplemental Movie S1 and Fig. 1B). ATP concentrations at the cell surface reached an estimated 60 M based on fluorescence signals obtained by the addition of known ATP concentrations (supplemental Fig. 1 and Movie S2). ATP release peaked between 1 and 3 s following cell stimulation with 10 nM fMLP; then cell surface ATP concentrations gradually decreased, returning to baseline levels ϳ10 s after cell stimulation. To further investigate the kinetics of ATP release during PMN polarization and migration, four-dimensional image stacks of cells stimulated with fMLP were acquired using 2-2Zn(II) and a spinning disk confocal system that yields high resolution spatiotemporal information about localized ATP release during cell activation. We observed highly dynamic changes in local ATP release with hot spots at sites of cell-to-cell interactions and at cell membrane regions that are associated with membrane polarization and pseudopod protrusion (Fig. 1C, indicated with arrows; see also supplemental Movie S3). Taken together, these findings suggest that cell stimulation causes a rapid burst of ATP release followed by localized hot spots of ATP release that seem to orchestrate site-specific functional responses of stimulated PMNs. These observations extend previous findings (5,7,13) and highlight the importance of rapid bursts and focal ATP release in the regulation of PMN responses.
Mitochondria Are Required for FPR-induced ATP Release-Because of the apparent importance of ATP release, we studied the origin of the ATP that is released from stimulated PMNs. The main sources of ATP in mammalian cells are glycolysis and oxidative phosphorylation and ATP synthesis by mitochondria.
To determine how these sources contribute to ATP production in human PMNs, we measured intracellular ATP (iATP) concentrations after a 2-h pretreatment of PMNs with 2-deoxy-Dglucose or sodium iodoacetate to inhibit glycolysis or with oligomycin or CCCP to block oxidative phosphorylation and mitochondrial ATP production. Although inhibition of glycolysis clearly reduced iATP levels, inhibition of mitochondrial ATP synthesis had comparatively little or no effect on iATP ( Fig. 2A, upper panel). Taken together with previous reports (8), these results suggest that glycolysis, and not mitochondria, is the primary source of intracellular ATP in PMNs. Although inhibition of mitochondrial ATP synthesis had only a minor effect on iATP levels, inhibition of glycolysis and of mitochondria inhibited the release of ATP into the extracellular space of stimulated PMNs (Fig. 2A, lower panel). To focus on the rapid dynamics of ATP release following cell stimulation, we studied the effects of short term inhibition of mitochondria by pretreatment (for 10 min) with CCCP (10 M) or KCN (1 mM). A pre-liminary test had shown that pretreatment of PMN with these drugs uncouples mitochondria and dissipates ⌬m in less than 10 s (supplemental Fig. 2A). We found that CCCP and KCN abolished ATP release in response to cell stimulation with 1-10 nM fMLP even though uncoupling of mitochondria under these conditions did not decrease iATP levels (Fig. 2B). Live cell imaging with 2-2Zn(II) confirmed that CCCP reduced the accumulation of extracellular ATP at the cell surface of stimulated PMNs (Fig. 2, C and D; see also supplemental Movie S4). However, we also observed that CCCP or KCN did not significantly reduce ATP release from cells stimulated with high (100 nM) fMLP concentrations. Moreover, we found that such fMLP concentrations caused a drop in iATP levels, which suggests that fMLP at these concentrations can cause ATP release in a manner that is independent of mitochondria, perhaps through the secretion of granules that store ATP. Taken together, the results described above indicate that mitochondria generate ATP that triggers purinergic signaling in response to cell stimulation, whereas glycolysis generates the bulk of iATP that can be released under certain circumstances.

FPR Stimulation Increases ⌬m and Triggers Oxidative
Phosphorylation-The results shown above suggest that FPR stimulation triggers rapid mitochondrial ATP synthesis, which causes ATP release into the extracellular space. To test this concept, we stained purified human PMNs with JC-1, a fluorescent probe that allows monitoring of ⌬m based on changes of the green and red fluorescence properties of the probe. A decrease in green and an increase in red fluorescence of JC-1 are thought to indicate an increase in ⌬m (8,9,14). Using flow cytometry to analyze JC-1 fluorescence, we found that stimulation of PMNs with 10 nM fMLP caused an increase in red fluorescence and a decrease in green fluorescence, suggesting a rapid rise in mitochondrial membrane potential upon FPR stimulation (Fig. 3A). These changes occurred within Յ5 s; red fluorescence levels returned to prestimulation values ϳ45 s after fMLP stimulation, whereas green fluorescence remained decreased (Fig. 3B). Our findings also indicated that these rapid changes in JC-1 fluorescence depend on the fMLP concentration used to stimulate PMNs (Fig. 3C).
Flow cytometric analysis of these changes provides statistical information about the entire cell population. To study the effect of cell stimulation on the mitochondria of individual PMNs, we used live cell imaging and fluorescence microscopy. Cells were stained with JC-1, stimulated with fMLP (10 nM), and changes in red JC-1 fluorescence were recorded over time. We found that JC-1 fluorescence of only a portion of all mitochondria in individual PMNs responded to fMLP stimulation ( Fig. 4A and supplemental Movie S5). A larger portion of the mitochondria did not seem to change their red JC-1 fluorescence properties in response to cell stimulation (supplemental Movie S6).
Taken together, these findings suggest that fMLP stimulation increases ⌬m of "regulatory" mitochondria, which may increase their capacity to generate ATP for purinergic signaling.
To assess mitochondrial function in more detail, we used DHR 123 to monitor ROS production, which occurs as a byproduct of mitochondrial ATP synthesis (15)(16)(17). In the presence of ROS, non-fluorescent DHR 123 is oxidized to fluorescent rhodamine 123, making it possible to assess mitochondrial activity in real time using a fluorescence microscope. Stimula-  tion of PMNs with fMLP increased the fluorescence intensity of mitochondria-like organelles in a time-dependent fashion ( Fig.  4B and supplemental Movie S7). Although ROS formation by oxidative burst complicates the interpretation of these results, the rapid onset and cellular distribution of the rhodamine 123 signal, which resembled the appearance of JC-1-stained mitochondria, suggests that FPR stimulation causes rapid mitochondrial ATP production (supplemental Movies S5 and S7). Taken together, these findings indicate that fMLP stimulation triggers a rapid increase in ⌬m, oxidative phosphorylation, and ATP synthesis by a limited number of mitochondria in PMNs.
Previously we found that panx1 is involved in ATP release from stimulated PMNs and that panx1 channels accumulate at the leading edge of polarized cells (5). In our current study, we found that some mitochondria can rapidly change their cellular location in stimulated PMNs, suggesting that the proximity of activated mitochondria and panx1 may contribute to ATP release (supplemental Movie S6). To study this possibility, we expressed a GFP-panx1 fusion protein in differentiated HL-60 cells, stained mitochondria with the ⌬m-sensitive Mito-Tracker Red CM-H 2 XRos dye, and observed colocalization of panx1 and mitochondria using confocal fluorescence microscopy. We found that cell stimulation caused the colocalization of panx1 channels with a subset of MitoTracker Red CM-H 2 XRos-positive mitochondria ( Fig. 4C and supplemental Movie S8), suggesting that the proximity of these mitochondria with panx1 channels is involved in regulated ATP release.
Mitochondria Link Purinergic Signaling to Ca 2ϩ Mobilization-Ca 2ϩ signaling has a central role in PMN activation, but Ca 2ϩ signaling is also important for the activation of mitochondrial ATP production (18). Therefore, we investigated the cross-talk among Ca 2ϩ signaling, mitochondrial ATP production, and purinergic signaling. Loading PMNs with the selective Ca 2ϩ chelator BAPTA-AM inhibited the FPR-induced ATP release and ⌬m increase (Fig. 5, A and B), indicating that intracellular Ca 2ϩ is essential for mitochondrial activation and ATP release from stimulated PMNs. To further investigate the link among mitochondrial function, ATP release, and Ca 2ϩ mobilization, we loaded PMNs with the cytosolic Ca 2ϩ probe Fluo-4 AM, which allows observation of intracellular Ca 2ϩ signaling using a fluorescence spectrophotometer. As expected, fMLP triggered an immediate increase in the intracellular Ca 2ϩ concentration of PMNs (Fig. 5C). This initial phase of Ca 2ϩ mobilization is caused by Ca 2ϩ release from intracellular stores such as the endoplasmic reticulum and mitochondria (19 -21). This initial phase of Ca 2ϩ signaling is followed by a sustained phase of increased cytosolic Ca 2ϩ concentrations due to Ca 2ϩ influx from the extracellular space, which was ablated by the addition of the Ca 2ϩ chelator EGTA (Fig. 5C).
Pretreatment of cells with CCCP did not seem to alter the sustained Ca 2ϩ phase but delayed the initial Ca 2ϩ signal by ϳ1 s and reduced its amplitude by half (Fig. 5C). Combined treatment of cells with CCCP and EGTA suppressed the initial Ca 2ϩ signal and the sustained phase. Inhibition of mitochondria with CCCP or with other agents that affect mitochondrial ATP synthesis (KCN, rotenone, or oligomycin) had similar effects on Ca 2ϩ signaling in response to fMLP stimulation (Fig. 5D). To compare these effects with those caused by inhibitors of auto-crine purinergic signaling, we blocked panx1-mediated ATP release or autocrine activation of P2 receptors with carbenoxolone or suramin, respectively. These treatments reduced the initial as well as the sustained phases of Ca 2ϩ signaling (Fig. 5, E  and F). A similarly profound inhibition of Ca 2ϩ signaling was achieved only by combined treatment with CCCP and EGTA (Fig. 5C). This suggests that purinergic signaling contributes to Ca 2ϩ signaling via mitochondrial ATP production (initial phase) and via a subsequent feedback mechanism (sustained phase) that is no longer fully dependent on mitochondria. Consistent with this notion, we found that inhibition of mitochondrial ATP synthesis with CCCP or KCN impaired p38 and ERK MAPK signaling in response to fMLP stimulation (supplemental Fig. 3), which suggests that not only Ca 2ϩ signaling but also MAPK signaling is regulated by mitochondria. Taken together, these findings suggest that mitochondrial ATP synthesis and Ca 2ϩ signaling are intricately interwoven signaling processes that are essential for amplifying, accelerating, and sustaining Ca 2ϩ signaling and MAPK activation in stimulated PMNs.
Mitochondria Regulate PMN Functions-We have shown previously that ATP release regulates functional PMN responses including chemotaxis, oxidative burst, phagocytosis, and degranulation (5)(6)(7). Taken together with the current findings shown above, these results suggest that mitochondria are directly involved in the regulation of functional PMN responses. To test this concept, we investigated how inhibition of mitochondrial ATP production affects PMN functions. To assess whether mitochondria regulate oxidative burst activity, purified human PMNs were loaded with DHR 123 and stimulated with fMLP, and ROS production was evaluated 20 min later using flow cytometry (12). Because ROS are also generated as a by-product of ATP synthesis by mitochondria, two distinct cell populations with different ranges of rhodamine 123 fluorescence could be distinguished (supplemental Fig. 2B). Unstimulated cells showed comparatively weak fluorescence, suggesting low mitochondrial ROS production, whereas cells stimulated with fMLP showed much stronger fluorescence likely due to additional ROS generated by oxidative burst. CCCP dose-dependently reduced FPR-induced oxidative burst (Fig. 6A). Similar suppressive effects were seen when mitochondrial ATP production was blocked with oligomycin or 2-deoxy-D-glucose or when intracellular Ca 2ϩ signaling was blocked with BAPTA-AM (supplemental Fig. 4, A-F). Taken together, these findings suggest that Ca 2ϩ mobilization and mitochondrial ATP production are required for the oxidative burst response of fMLP-stimulated PMNs.
CCCP also blocked FPR-induced degranulation and bacterial phagocytosis, which were assessed by evaluating the recruitment of the adhesion molecule CD11b to the cell surface of PMNs and the uptake of fluorescence-labeled E. coli bacteria, respectively. CCCP dose-dependently inhibited fMLP-induced CD11b expression and bacterial phagocytosis (Fig. 6, B and C). However, compared with oxidative burst, higher doses of CCCP were needed to inhibit the latter functional responses, and CCCP was unable to block CD11b expression in response to high (100 nM) concentrations of fMLP, indicating that degranulation and phagocytosis are less dependent upon mitochondrial ATP production than oxidative burst. These findings demonstrate that mitochondrial ATP production has an important role in regulating PMN activation and functional cell responses required for immune defense.

DISCUSSION
Our findings have shown that mitochondria are key regulators of PMN activation but that functional cell responses vary in their reliance on mitochondrial ATP production. These findings are in agreement with our previous work that has shown that purinergic signaling is a critical mechanism that differentially regulates functional PMN responses (5). In our current study, we found that mitochondrial ATP production initiates PMN activation. However, inhibition of ATP release or P2 receptor signaling with carbenoxolone or suramin had more pronounced suppressive effects on cell functions than inhibition of mitochondrial ATP production (5). This indicates that mitochondrial ATP production is an important step that triggers cell responses but that subsequent mechanisms that are independent of mitochondria take over to complete purinergic signaling and fully define functional PMN responses.
The lack of imaging techniques suitable for monitoring the rapid dynamics of ATP release from living cells has been a major limitation in the study of purinergic signaling mechanisms. To be able to observe ATP release from living cells in real time, we had previously developed a method that utilizes a tandem enzymatic reaction that allows visualization of ATP release with fluorescence microscopy (7,22). Although this method was highly sensitive, it had major drawbacks including the need for an excitation wavelength in the ultraviolet range, which resulted in significant phototoxicity (23). The small molecular probe we used in the current study was developed in the laboratory of Dr. Hamachi (11,24). This ATP probe features a membrane anchor that has allowed us to attach the probe to the cell surface of living PMNs and to assess ATP concentration changes at the cell surface for prolonged periods of time. This new approach has revealed spatiotemporal patterns of ATP  1; left panel). B, to assess how mitochondria contribute to PMN degranulation, cells were treated with fMLP and CCCP as described in A and fixed. CD11b expression on the cell surface was measured as a marker of degranulation. C, the role of mitochondrial ATP formation in bacterial phagocytosis was assessed by treating PMNs with the indicated concentrations of CCCP, adding Alexa Fluor 488-conjugated E. coli, and assessing bacterial uptake with flow cytometry. Data are shown as the percentage of PMNs with bacteria in Gate 1 as opposed to cells without bacteria or with bacteria adhered to their surface. Data are expressed as mean values Ϯ S.D. (error bars) of three to six individual experiments performed Ն3 different times with similar results. *, p Ͻ 0.05 versus untreated but stimulated controls. cont., control cells exposed to bacteria for Ͻ10 s.
release that indicate that purinergic signaling is far more complex than previously anticipated. We found that cell stimulation causes a brief burst of localized ATP release and that subsequent localized hot spots of ATP release seem to coordinate pseudopod protrusions. The initial burst of ATP release occurred within a second after cell stimulation and in parallel with the initial rise in cytosolic Ca 2ϩ following FPR stimulation. Detailed analysis of these signaling mechanisms suggests three distinct components of Ca 2ϩ signaling: 1) A first phase of relatively weak Ca 2ϩ signaling, which is independent of purinergic signaling and can be blocked by BAPTA, triggers mitochondrial activation and results in an initial burst of mitochondria-driven ATP release. 2) ATP released from cells then stimulates P2Y 2 receptors that cause a spike in Ca 2ϩ signaling, which in turn promotes further mitochondrial ATP production. 3) Then a final, third phase of sustained Ca 2ϩ signaling is triggered by Ca 2ϩ influx via a second loop of autocrine purinergic signaling. Mitochondrial inhibitors had little effect on this phase of Ca 2ϩ signaling, which suggests that glycolysis provides ATP to drive this purinergic feedback loop (Fig. 7). Our findings indicate that PMN activation involves at least two phases of purinergic signaling: 1) an initial burst of ATP release that is driven by mitochondrial ATP and P2Y 2 receptors and triggers cell activation and 2) a second phase that involves glycolytic ATP production and P2X receptors that sustain Ca 2ϩ signaling and maintain functional cell responses. These dual purinergic mechanisms increase the robustness of neutrophils and allow them to cope with adverse conditions such as hypoxia and glucose starvation that would preclude mitochondrial and glycolytic ATP production, respectively.
Our findings also indicate that mitochondria in activated PMNs can rapidly change their intracellular locations and form associations with panx1 channels (Fig. 4C). This suggests that spatiotemporal dynamics of mitochondrial translocation within the cell can regulate PMN functions. Indeed, similar observations were made in T cells where the accumulation of mitochondria and panx1 at the immune synapse results in the physical proximity of the sites of ATP production and of ATP release, minimizing rate-limiting diffusion and thus facilitating rapid local purinergic signaling at the immune synapse (25)(26)(27)(28)43). Thus, we conclude that mitochondrial ATP production and the location of mitochondria relative to panx1 contribute to purinergic signaling in stimulated PMNs.
Based on our current findings, we propose that the release of ATP and the sequential activations of P2Y 2 and P2X receptors contribute to the different phases of Ca 2ϩ signaling in human PMNs. Recent work by others has shown that different sets of P2 receptors regulate Ca 2ϩ signaling in other cell types (29). P2X receptors function as ATP-gated ion channels that can facilitate Ca 2ϩ influx, and in T cells, P2X1, P2X4, and P2X7 receptors regulate influx of extracellular Ca 2ϩ and cell activation in response to T cell stimulation (27,28). PMNs express several members of the P2X receptor family in addition to the predominant P2Y 2 receptor (5, 7). P1 and other P2Y receptor subtypes and ectoenzymes that generate their respective ligands have also been shown to contribute to the regulation of PMNs (6, 7, 30 -33). It is likely that mitochondrial ATP synthesis contributes to the regulation of PMNs through all these different purinergic receptor systems.
An intriguing finding of our current study is that FPR stimulation causes rapid activation of mitochondrial ATP production. Several studies have shown that surgical procedures, sepsis, and burn injuries can influence ⌬m in PMNs (34 -37). This suggests that PMN activation under these clinical conditions may affect mitochondrial ATP production and PMN activity. This notion is supported by findings from others who showed that inhibition of mitochondria impairs PMN chemotaxis and oxidative burst, which was assessed with DHR 123 (9,38). A limitation of this approach is that mitochondria and oxidative burst deliver ROS that cause DHR 123 conversion to rhodamine 123. Because this complicates the interpretation of data generated with DHR 123, more direct methods of assessing mitochondrial ATP synthesis in PMNs would be desirable.
Based on our results, we concluded that Ca 2ϩ and purinergic signaling mechanisms are tightly interwoven processes that regulate PMN activation and cell functions (Fig. 7). Inhibition of either of these signaling events impairs PMN activation and functional responses required for host defense. Sites of inflammation are characterized by decreased oxygen levels, limiting the ability of mitochondria to generate the ATP needed for cell activation (39,40). PMNs seem to be able to overcome this problem by utilizing glycolysis for ATP production when oxygen is scarce. We found that inhibition of mitochondria is more detrimental to oxidative burst than to degranulation or phagocytosis (Fig. 6). These findings are in agreement with other reports that have shown that hypoxic conditions impair oxidative burst, whereas other PMN responses including degranulation, adhesion molecule expression, phagocytosis, and bacterial killing are less sensitive to oxygen deprivation (39,41,42). Taken together with our current findings, this highlights a remarkable resilience of PMNs that allows them to retain at least some of their defensive capabilities even under adverse conditions that impair mitochondrial ATP production. This resilience is likely the consequence of the redundancy of the FIGURE 7. Proposed role of mitochondria in PMN activation. Stimulation of FPRs triggers a first phase of Ca 2ϩ mobilization, mitochondrial activation, and ATP production followed by rapid ATP release via panx1 channels. This initial burst in ATP release promotes a second round of Ca 2ϩ signaling triggered by P2Y 2 receptor activation. A switch to ATP production via glycolysis results in further ATP release, activating P2X receptors that contribute to a third phase of Ca 2ϩ signaling due to influx from the extracellular space. This second ATP signaling phase sustains intracellular Ca 2ϩ levels and maintains functional PMN responses following FPR stimulation.