Reactive oxygen species (ROS) play a divergent role in both cell survival and cell death during ischemia/reperfusion (I/R) injury and associated inflammation. In this study, ROS generation by activated macrophages evoked an intracellular Ca2+ ([Ca2+]i) transient in endothelial cells that was ablated by a combination of superoxide dismutase and an anion channel blocker. [Ca2+]i store depletion, but not extracellular Ca2+ chelation, prevented [Ca2+]i elevation in response to O2.− that was inositol 1,4,5-trisphosphate (InsP3) dependent, and cells lacking the three InsP3 receptor (InsP3R) isoforms failed to display the [Ca2+]i transient. Importantly, the O2.−-triggered Ca2+ mobilization preceded a loss in mitochondrial membrane potential that was independent of other oxidants and mitochondrially derived ROS. Activation of apoptosis occurred selectively in response to O2.− and could be prevented by [Ca2+]i buffering. This study provides evidence that O2.− facilitates an InsP3R-linked apoptotic cascade and may serve a critical function in I/R injury and inflammation.
Receptor-mediated generation of reactive oxygen species (ROS) is necessary for signal transduction, gene expression, and cell proliferation in smooth muscle cells, T and B lymphocytes, and fibroblasts (Devadas et al., 2002). Conversely, ROS produced under pathological conditions such as ischemia/reperfusion (I/R) or inflammation are associated with cellular dysfunction and apoptosis (Davies, 1995). Endothelial cells respond to numerous external stimuli by producing the superoxide anion (O2.−). In physiological conditions, mitochondrial respiratory chain proteins produce O2.−, which can be dismutated into hydrogen peroxide (H2O2) or react with nitric oxide to produce peroxynitrite. In addition, reaction of H2O2 with iron leads to hydroxyl radical formation via Fenton chemistry. During I/R injury, O2.− production in the vasculature is substantially increased (Wei et al., 1999) and is accompanied by endothelial cytotoxicity (for review see Li and Shah, 2004). However, the molecular mechanisms by which ROS lead to organ damage are poorly understood.
In pathological conditions, cell death is facilitated by an elevation in intracellular calcium ([Ca2+]i; Hajnoczky et al., 2003; Orrenius et al., 2003) via inositol 1,4,5-trisphosphate (InsP3). InsP3 is a second messenger produced by the hydrolysis of phosphatidylinositol-4,5-bisphosphate by PLC. InsP3 receptor (InsP3R)–mediated [Ca2+]i changes are associated with a rapid, transient Ca2+ release from Ca2+ stores in the ER followed by Ca2+ entry through slow-activating plasma membrane store-operated channels (Putney and Bird, 1993; Parekh and Penner, 1997; Berridge et al., 1998). InsP3 [Ca2+]i signals control a wide range of cellular functions, including cell proliferation and apoptosis (Berridge et al., 2000; Orrenius et al., 2003). Apoptosis is reduced in cells lacking all three InsP3R isoforms (DT40 avian B cells) and after selective suppression of InsP3R-3 (Jayaraman and Marks, 1997; Sugawara et al., 1997), indicating the important role of InsP3 in cell death mechanisms (Pan et al., 2001). Alterations in [Ca2+]i after oxidative stress facilitate activation of the mitochondrial permeability transition pore (MPTP), which releases cytochrome c from the mitochondrial intermembrane space, leading to mitochondrial membrane potential (ΔΨm) loss, assembly of the apoptosome, and activation of downstream caspases (Crompton, 1999). Recent evidence suggested that cytochrome c transiently released from mitochondria interacts with InsP3R and amplifies Ca2+-mediated apoptosis (Boehning et al., 2003).
Endothelial cells subjected to oxidative stress undergo apoptosis (Warren et al., 2000). Although there is evidence that perturbations of cellular Ca2+ homeostasis (including [Ca2+]i elevation, ER Ca2+ depletion, and mitochondrial Ca2+ increases) occur, the mechanisms by which oxidative stress mediates endothelial apoptosis remain unclear. Events in the early stages of stress signaling include the mobilization of [Ca2+]i (Patterson et al., 2004), the generation of ROS, and the formation of lipid peroxides. However, it is unclear whether radical formation is a consequence of Ca2+ mobilization or a parallel event in early stress signaling. The proximity between mitochondria and the ER facilitates a higher Ca2+ exposure in mitochondria relative to the cytosol when released from the ER (Rizzuto et al., 1998). During pathological situations, excess ER-released Ca2+ may be detrimental to mitochondrial function and may trigger mitochondrial fragmentation and apoptosis. Previously, Bcl-2 family proteins have been implicated in apoptosis by affecting cellular Ca2+ homeostasis (Pinton et al., 2000; Pan et al., 2001; Li et al., 2002). A recent study reported that a functional interaction of Bcl-2 with InsP3R attenuated InsP3R activation, which in turn controlled InsP3-evoked Ca2+ release (Chen et al., 2004), in contrast to our findings that Bcl-XL activates InsP3R (White et al., 2005). In addition, ER-localized Bax and Bak can either interfere with ER Ca2+ homeostasis or initiate apoptosis by activating caspase 12 (Zong et al., 2003).
We previously reported that cells exposed to O2.− induced a rapid and large cytochrome c release (Madesh and Hajnoczky, 2001). We now provide evidence that O2.− evokes a large, transient [Ca2+]i pool release from the ER, causing mitochondrial Ca2+ elevation and rapid depolarization. Remarkably, the observed InsP3-linked mitochondrial phase of apoptosis was specific to O2.− and not other oxidant species. The O2.−-induced mitochondrial depolarization and downstream apoptotic cascades are independent of mitochondrial ROS production. Overall, this evidence provides a mechanism by which O2.− is a key signaling molecule that coordinates multiple processes that lead to mitochondrial apoptotic events and endothelial dysfunction.
Lipopolysaccharide (LPS)-stimulated macrophages evoke Ca2+ transients in endothelial cells
Activated macrophages are known to generate ROS and may be involved in organ damage during I/R (Droge, 2002). To test the significance of the selective role of macrophage-derived ROS during pathophysiological conditions, LPS-stimulated murine macrophages were used as a O2.−-generating source. We determined whether O2.− released from macrophages could evoke Ca2+ mobilization in two cell types, endothelial and HepG2 cells. ROS production in LPS-stimulated mouse macrophages was measured via H2DCF-DA, which is a nonfluorescent dye that produces the fluorescent compound dichlorofluorescein (DCF) when oxidized by ROS. DCF fluorescence was measured in untreated macrophages and those stimulated with LPS (1 μg/ml) or a combination of LPS and the NADPH oxidase inhibitor diphenyleneiodonium (DPI; 30 μM). LPS stimulation was associated with a pronounced increase in DCF fluorescence that was attenuated by DPI treatment, suggesting that LPS stimulated ROS production through activation of oxidative burst reactions (Fig. 1 A). The activation of macrophage NADPH oxidase generates O2.− extracellularly without altering intracellular production of ROS by mitochondria (Lambeth, 2004). To elucidate whether a paracrine ROS signal can be transduced to adjacent cells in pathological conditions, LPS-stimulated macrophages were added onto pulmonary microvascular endothelial cells (PMVECs; Fig. 1 B) that had been previously loaded with the [Ca2+]i indicator dye Fluo-4 (Fig. 1 C). Application of LPS-activated macrophages evoked a [Ca2+]i rise in PMVECs that was attenuated by DPI pretreatment (Fig. 1 C). To exclude the contribution of autocrine extracellular ROS production, a similar experiment was performed using HepG2 parenchymal cells, as these cells generate minimal O2.− (Kikuchi et al., 2000). HepG2 cells displayed an [Ca2+]i elevation after LPS-stimulated macrophage exposure, whereas no [Ca2+]i transient was noted after application of nonstimulated macrophages (Fig. 1 D). In contrast, exposure of HepG2 cells to macrophages that had been stimulated by LPS plus DPI triggered only an extremely small [Ca2+]i rise (Fig. 1 D). The oscillatory [Ca2+]i transient pattern observed in individual HepG2 cells but not PMVECs is notable, indicating a potential difference in Ca2+ handling between cell types (unpublished data). Overall, this result suggests that O2.− is specifically required for elevation of [Ca2+]i in endothelial cells.
O2.− evokes endothelial Ca2+ transients through InsP3 signaling
To identify the mechanisms by which O2.− triggers [Ca2+]i signals in PMVECs, we extended our studies to examine the effects of O2.− on basal [Ca2+]i. To exclude the possible contribution of other macrophage factors, the xanthine+xanthine oxidase (X+XO) system was used to generate O2.− externally. Cells exposed to O2.− demonstrated a rapid increase in [Ca2+]i followed by a slightly delayed return to baseline (Fig. 2 A). Similarly, the physiological stimulus ATP generated a marked [Ca2+]i transient (Fig. S1). The O2.−-evoked [Ca2+]i increase was abolished by pretreatment with the XO inhibitor allopurinol (1 mM; Fig. 2 B) or by a combination of the antioxidant superoxide dismutase (SOD; 2000 U/ml) and the anion channel blocker DIDS (100 μM; Fig. 2 C). Treatment with either xanthine or allopurinol did not alter basal [Ca2+]i in control cells (unpublished data). These findings suggest that acute exposure of PMVECs to extracellular O2.− results in a rapid [Ca2+]i rise. We next sought to determine the source of the elevated [Ca2+]i. Thapsigargin (Tg) inhibits the SERCA Ca2+ATPase, causing Ca2+ depletion from the ER (Ma et al., 2000, 2001). Pretreatment with 2 μM Tg virtually abolished O2.−-induced Ca2+ transients (Fig. 2 D). Conversely, removal of Ca2+ from the external medium was without effect on [Ca2+]i (Fig. 2 E). Together, these results indicate that O2.− induces a release of Ca2+ from internal stores. ER Ca2+ stores in endothelial cells can be modulated by production of the second messenger InsP3 by PLC and subsequent binding to receptors on the ER (InsP3R). To characterize the release of Ca2+ from intracellular stores, PMVECs were pretreated for 10 min with either the PLC inhibitor U-73122 or its inactive analogue U-73343. U-73122, but not U-73343 (both 100 μM), inhibited the O2.−-induced Ca2+ release (Fig. 2, F and G). This result suggests that the O2.−-induced [Ca2+]i transient was mediated by InsP3. To further characterize O2.−-induced Ca2+ release, cells were incubated with 2-aminoethoxydiphenyl borate (2-APB; 75 μM) before O2.− stimulation. 2-APB has widely been used as an inhibitor of InsP3-sensitive Ca2+ release and store-operated Ca2+ channels in intact cells (Ma et al., 2001; Bootman et al., 2002). In agreement with our PLC data, O2.−-induced Ca2+ transients were abolished in cells pretreated with 2-APB (Fig. 2, H and I). Thus, the O2.−-induced [Ca2+]i rise in PMVECs was due to the InsP3-dependent release of Ca2+ from internal stores.
O2.−-triggered [Ca2+]i release is abolished in InsP3R triple knockout (TKO) cells
To examine the specific role of InsP3R in the O2.−-triggered [Ca2+] rise, the InsP3R-deficient DT40 chicken B-lymphocyte cell line (DT40 InsP3R TKO) was used. Wild-type cells demonstrated a significant [Ca2+]i increase after O2.− exposure. After [Ca2+]i returned to basal levels, 2 μM Tg was added to the medium to induce a transient increase in [Ca2+]i as a consequence of passive depletion of endogenous stores upon ER Ca2+/Mg2+-ATPase blockade (Fig. 3, A and B). Similar to PMVECs, pretreatment with 2 μM Tg eliminated the O2.−-induced Ca2+ transients in wild-type DT40 cells (unpublished data). In contrast, addition of a O2.− pulse failed to elicit Ca2+ release from intracellular stores in DT40 InsP3R TKO cells, whereas subsequent addition of 2 μM Tg triggered a complete depletion of Ca2+ stores (Fig. 3, A and B). These data suggest that Ca2+ release through the InsP3R underlies the O2.−-evoked rise of [Ca2+]i. To confirm that DT40 InsP3R TKO cells retain the machinery necessary for the O2.−-mediated [Ca2+]i transient, we transfected the rat InsP3R type I into TKO cells. This procedure restored the responsiveness of TKO cells to O2.− (Fig. 3 C). This result indicates that in TKO cells, a O2.−-mediated signal activates InsP3R type I and causes Ca2+ release from ER store.
In PMVECs, inhibition of PLC with U-73122 prevented the rise of [Ca2+]i induced by exposure to O2.−. We therefore further investigated the role of PLC in O2.−-triggered Ca2+ mobilization using PLC-γ2–deficient DT40 cells. O2.− exposure triggered a substantial rise of [Ca2+]i in PLC-γ2–deficient DT40 cells (Fig. 3, A and B). In wild-type DT40 cells, B cell receptor agonist IgM (2 μg/ml) induced a series of rapid [Ca2+]i oscillations representing [Ca2+]i release and reuptake. In contrast, anti-IgM failed to elicit Ca2+ mobilization in both InsP3R TKO and PLC-γ2 knockout (KO) cells (unpublished data). These data indicate that the nonreceptor tyrosine kinase–linked cascade, to which PLC-γ2 is coupled, is dispensable for the O2.−-triggered [Ca2+]i rise. In agreement with our findings, G protein–coupled receptor (GPCR)–mediated Ca2+ oscillations were previously abolished by U-73122, which inhibits all PLC-β isoforms (Zeng et al., 2003). To further understand the role of InsP3, PLC-γ2 KO cells were pretreated with either PLC inhibitor U-73122 or U-73343 as described in Fig. 2 (F and G). U-73122, but not U-73433, attenuated the O2.−-evoked [Ca2+]i rise (Fig. 3 D). To ensure that the O2.− elicits InsP3 accumulation, InsP3 was assessed in wild-type DT40, DT40 InsP3R TKO, and DT40 PLC-γ2 KO cells. Direct measurement of InsP3 production indicated that O2.− markedly activated InsP3 formation in wild-type DT40, DT40 InsP3R TKO, and DT40 PLC-γ2 KO cells. In contrast, pretreatment of DT40 PLC-γ2 KO cells with U-73122 significantly attenuated this response (Fig. S2 A). Similarly, PMVECs exposed to O2.− exhibited markedly greater InsP3 production than the physiological stimulus ATP (Fig. S2 B). Collectively, these findings suggest that extracellular O2.− causes Ca2+ release via a PLC-mediated increase in InsP3.
O2.− mediates coupling of [Ca2+]i elevation and mitochondrial uptake
It is believed that agonist-induced [Ca2+]i rise can be buffered by mitochondria (Bernardi and Petronilli, 1996). To determine if the O2.−-triggered [Ca2+]i spike is delivered to mitochondria, rhod-2– (mitochondrial Ca2+ indicator) and Fluo-4–loaded PMVECs were subjected to O2.−. Exposure of PMVECs to O2.− induced an [Ca2+]i increase as evidenced by an increase in Fluo-4 fluorescence, as shown earlier (Fig. 2 A), followed by an elevation of mitochondrial Ca2+ fluorescence (Fig. 4, A and B). Similarly, ATP induced a [Ca2+]i rise followed by mitochondrial [Ca2+] elevation (Fig. 4, C and D). These results indicate Ca2+ signal propagation from the cytosol to the mitochondria in both physiological (purinergic receptor agonist) and pathological conditions (oxidative stress). Notably, O2.−-evoked mitochondrial Ca2+ elevation was increased and sustained compared with the transient pattern observed in response to ATP. These results suggest that O2.−-induced intracellular pool Ca2+ release evokes elevated mitochondrial Ca2+ uptake during oxidative stress.
O2.−-induced Ca2+ transients evoke rapid mitochondrial depolarization
Reversible depolarization of ΔΨm occurs as a consequence of electrogenic uptake of Ca2+ by mitochondria in response to transient [Ca2+]i (Duchen, 1992). However, ROS may also promote MPTP opening (Huser et al., 1998). Because mitochondrial Ca2+ elevation is a common pathway in both normal physiological and pathological stimuli, we examined whether the observed mitochondrial Ca2+ uptake after O2.− exposure is associated with mitochondrial depolarization. Simultaneous fluorescence measurements of [Ca2+]i and ΔΨm were conducted in PMVECs during O2.− exposure (Fig. 5, A and B). In response to ATP, an [Ca2+]i rise was observed similar to that in cells after O2.− exposure. However, in contrast to O2.−, PMVECs exposed to ATP exhibited only a nominal change in ΔΨm (Fig. 5 C), possibly due to transient Ca2+ uptake (Fig. 4, B and D). Application of O2.− evoked a rapid and transient rise in [Ca2+]i that preceded a decrease in tetramethylrhodamine, ethyl ester, perchlorate (TMRE) fluorescence, indicating that mitochondrial depolarization is associated with the onset of the [Ca2+]i rise (Fig. 5 B). Because O2.− is rapidly dismutated into H2O2, we sought to determine which oxidants are involved in the observed ΔΨm loss. Cells incubated with H2O2 (1 mM) displayed no rapid [Ca2+]i transient. Rather, H2O2 induced a slight increase in [Ca2+]i (Fig. 5 D) and a delayed loss of ΔΨm. Tg pretreatment did not affect the H2O2-facilitated slow [Ca2+]i rise (unpublished data). These findings suggest that H2O2 may not affect the intracellular store, but instead facilitates Ca2+ entry from the extracellular milieu independent of mitochondrial depolarization. Oxidized phospholipid byproducts are involved in cell death during oxidative stress (Ran et al., 2004). However, the lipid-oxidizing agent t-butyl hydroperoxide (t-BuOOH; 200 μM) did not evoke either an [Ca2+]i rise or ΔΨm loss (Fig. 5 E). This finding suggests the selective role of O2.−, and not other oxidants, in eliciting an [Ca2+]i rise and mitochondrial depolarization.
Extracellular O2.−-mediated signaling functions independent of mitochondrially derived ROS
Evidence indicates that external ROS may evoke mitochondrial O2.− production (Zorov et al., 2000; Aon et al., 2003). Because the O2.−-evoked [Ca2+]i rise is a prerequisite for ΔΨm loss, we aimed to exclude the involvement of intracellular ROS production by mitochondrial electron transport proteins in ΔΨm loss. Antimycin A inhibits the normal electron flow through complex III, but triggers O2.− production through the accumulation of ubisemiquinone. Antimycin A triggered an immediate ΔΨm loss without an apparent change in [Ca2+]i (Fig. 6 A). Rotenone inhibits electron transfer from complex I (NADH dehydrogenase) to ubiquinone and diminishes O2.− production from complex III (Turrens et al., 1985). In contrast to antimycin A, rotenone affected neither [Ca2+]i nor ΔΨm. However, subsequent addition of O2.− triggered an [Ca2+]i rise followed by ΔΨm loss (Fig. 6 B). Oligomycin, which inhibits the mitochondrial FoF1-ATPase by binding to ATP synthase, was used to exclude possible mitochondrial ATP-dependent ROS production. Treatment with oligomycin failed to trigger either [Ca2+]i mobilization or ΔΨm loss. Subsequent addition of O2.− established both events (Fig. 6 C). This result indicates that complex III is the major site of mitochondrial ROS production during oxidative stress. It has been reported that mitochondrial Ca2+ uptake requires an intact ΔΨm and that dissipation by a mitochondrial uncoupler abolishes mitochondrial Ca2+ uptake and delays [Ca2+]i clearance (Boitier et al., 1999). Close examination of PMVECs exposed to the mitochondrial uncoupler FCCP revealed that a rapid ΔΨm loss was associated with [Ca2+]i elevation (Fig. 6 D). This [Ca2+]i rise most likely reflects Ca2+ release from the mitochondria as a consequence of mitochondrial depolarization. Surprisingly, subsequent application of O2.− evoked a transient rise in cytosolic Fluo-4 fluorescence followed by a rapid recovery of ΔΨm. The ΔΨm recovered after O2.−treatment is almost identical to the initial potential observed before FCCP addition. Collectively, these results suggest that the mitochondrial ROS-evoked ΔΨm loss is independent of InsP3R-linked ΔΨm changes by O2.−.
Ca2+ buffering protects against O2.−-triggered mitochondrial depolarization
To assess whether the O2.−-induced rise of [Ca2+]i is required for the O2.−-evoked ΔΨm loss, PMVECs were loaded with the Ca2+ chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetate (BAPTA) by incubation with the permeant acetoxymethyl ester (25 μM for 30 min) before application of the O2.−. BAPTA loading significantly inhibited O2.−-induced ΔΨm loss (Fig. 7, A and B). In contrast, the H2O2-induced ΔΨm loss was unaffected by pretreatment with BAPTA (Fig. 7 C). These experimental data provide evidence that ΔΨm loss induced specifically by O2.− requires a rise of [Ca2+]i. Other oxidants such as H2O2 are deleterious to mitochondrial function but appear to affect ΔΨm through a Ca2+-independent pathway.
O2.−-mediated signaling triggers caspase activation
Caspase cysteine proteases augment mitochondrial dysfunction by both activating proapoptotic Bcl-2 family proteins such as Bax, Bak, and Bid and inactivating antiapoptotic proteins such as Bcl-2 (Wei et al., 2001). To determine the dose and time course of receptor-mediated and mitochondrially dependent caspase activation in PMVECs after oxidant exposure, cytosolic extracts were collected after treatment with O2.−, H2O2, and t-BuOOH. Remarkably, when cells were exposed to O2.−, robust caspase-3 activity was observed in a dose-dependent manner (Fig. 8, A and D). Interestingly, even a low dose (1 mU X+XO) was able to induce caspase-3 activity, indicating that O2.− may activate downstream caspases through a mitochondrially dependent pathway. Similarly, prominent caspase-9 activity was observed after O2.− treatment (Fig. 8, C and F). H2O2 elicited some caspase-3 and -9 activity, but at a level severalfold less than O2.−. In contrast, t-BuOOH did not activate either caspase-3 or -9. During apoptotic conditions, caspase-8 can activate caspase-3 directly through an extrinsic pathway. As shown in Fig. 8 (B and E), treatment of PMVECs with O2.− induced caspase-8 activity that was sevenfold higher than control and other oxidants. Inhibition of ΔΨm loss by [Ca2+]i buffering prevented caspase-3 and -9 activation (Fig. 8, D and F). Collectively, these results provide evidence that O2.− activates both extrinsic and intrinsic caspase pathways.
O2.−-evoked [Ca2+]i overload executes the cell death machinery
Our results reveal that O2.− stimulates [Ca2+]i mobilization that triggers subsequent mitochondrial events, leading to caspase activation in PMVECs. To directly demonstrate that O2.− induces apoptosis, we treated PMVECs with various oxidants at different doses, and then stained them for the early apoptotic marker annexin V and the late stage apoptotic (or necrotic) marker propidium iodide (PI). Cells treated with O2.− for 5 h displayed positive annexin V staining with no detectable PI labeling, indicating cells in the early stages of apoptosis (Fig. 9 A). Strikingly, cells exposed to a high concentration of O2.− demonstrated a dose-dependent elevation of both apoptotic and necrotic cell death as displayed in Fig. 9 B. Cells treated with 500 μM H2O2 also revealed an apoptotic phenotype, although at a lower level than observed in response to O2.−. In contrast, t-BuOOH (200 μM) treatment primarily led to necrosis, as evidenced by positive annexin V and PI staining. Control conditions resulted in nominal levels of apoptotic- or necrotic-positive cells. Previously, our results provided evidence that buffering of O2.− evoked [Ca2+]i rise by BAPTA-AM and markedly prevented PMVEC ΔΨm loss. Therefore, we tested whether [Ca2+]i buffering inhibits O2.−-induced apoptosis. BAPTA-AM pretreatment (25 μM) attenuated apoptosis in PMVECs (Fig. 9 C), providing evidence that O2.−-induced [Ca2+]i elevation is essential for mitochondrially dependent apoptosis. Conversely, BAPTA-AM treatment was ineffective 20 min after application of the O2.− (unpublished data). DT40 B-cells lacking all forms of InsP3R display reduced apoptotic cell death in response to anti-IgM (Sugawara et al., 1997). Because O2.−-induced [Ca2+]i elevation is ablated in DT40 InsP3R TKO cells, we next investigated apoptosis in DT40 cells. DT40 InsP3R TKO cells, but not wild-type cells, display increased resistance to apoptosis after O2.− application (Fig. S3). These results suggest that O2.− selectively alters ER Ca2+ homeostasis resulting in caspase activation, which in turn leads to apoptosis.
The mechanisms that contribute to apoptosis during I/R injury remain unclear, but it is generally believed that the release and/or activation of various bioactive molecules, such as ROS (Zhao, 2004) and inflammatory cytokines (Haimovitz-Friedman et al., 1997), are responsible for cell death. During these conditions, xanthine (Malis and Bonventre, 1986) and NADPH oxidases play a key role in O2.− production (Wei et al., 1999) and trigger pathological signaling. Reperfusion of ischemic cells generates oxidative stress and alters mitochondrial function (Hausenloy et al., 2004). Coordination of mitochondrial function during injury is an essential component of cell physiology and survival, yet little is known about the factors that contribute to cell death during oxidative stress. This study demonstrates that O2.− facilitates a transient [Ca2+]i elevation followed by mitochondrial Ca2+ uptake and depolarization that ultimately induces apoptotic cascades in endothelial cells.
Macrophage activation by endotoxin elicits O2.− generation via NADPH oxidase and autocrine production of ROS (Johnston et al., 1978). However, whether released O2.− has a potential paracrine signaling role in nearby cells is unknown. This study provides direct evidence that activated macrophages initiate an ROS-induced [Ca2+]i elevation in adjacent cells. In comparison to the macrophage data, the observed [Ca2+]i transient using the X+XO was larger and less sustained. The enzymatic X+XO system generates only O2.−, whereas activated macrophages may release other factors that could alter the amplitude of [Ca2+]i in PMVECs. In addition, xanthine oxidase has been shown to interact with the vascular endothelium during inflammatory conditions (Houston et al., 1999). Because of the short half-life of the O2.− radical, close association between endothelial cells and the O2.− source may facilitate a greater response. A single pulse of O2.− evoked an [Ca2+]i rise in PMVECs that caused ΔΨm loss. These results suggest a potential mechanism by which macrophage-mediated oxidative stress perpetuates endothelial dysfunction. This O2.−-mediated response has several features. The [Ca2+]i signals were observed in adherent PMVECs, HepG2, and DT40 suspension cell types, indicating a common mechanism in the cellular response to O2.−. The O2.−-evoked [Ca2+]i signal was prevented by the combination of SOD and the anion channel blocker DIDS. The O2.−-induced transient rise of [Ca2+]i was propagated to mitochondria, where a sustained Ca2+ elevation was observed. In contrast, the [Ca2+]i response to the physiological stimulus ATP triggered a transient mitochondrial Ca2+ elevation. The O2.−-induced [Ca2+]i transient subsequently evoked mitochondrial depolarization independent of mitochondrially derived ROS. In addition to this novel observation, our results suggest that O2.− selectively evokes Ca2+-dependent ΔΨm loss independent of other oxidants.
Another important finding is that Tg, but not EGTA, pretreatment eliminated the O2.−-induced increase in [Ca2+]i, indicating release from the ER. We therefore conclude that Ca2+ store release in response to O2.− may be PLC dependent and mediated by InsP3R on the ER. This conclusion was supported by the observation that DT40 cells lacking all three InsP3R isoforms failed to show an [Ca2+]i rise after O2.− application, unless InsP3R was reintroduced by transient transfection. Reintroduction of InsP3R type 1 restored the [Ca2+]i transient, indicating the existence of the Ca2+ signaling machinery in TKO cells. Furthermore, we found that the PLC inhibitor U-73122 blocked the O2.− response in endothelial cells. PLC normally presents as a key enzyme in cellular metabolism and signaling in response to extracellular agonists by coupling with GTP-binding proteins. DT40 cells express PLC-γ2 and PLC-β isoforms (Rhee, 2001) but lack the GPCRs necessary for PLC-β activation (Venkatachalam et al., 2001; Patterson et al., 2002). Surprisingly, we observed that PLC-γ2 KO cells displayed a rapid [Ca2+]i store release in response to O2.−, suggesting the activation of PLC-β–mediated Ca2+ release by O2.−. PLC inhibition in these PLC-γ2 KO cells by U-73122 indicates activation of PLC and suggests that O2.−-induced [Ca2+]i rise requires InsP3. Because InsP3 levels were greatly elevated by O2.− in all three DT40 cell lines, it is apparent that generation of InsP3 by PLC is the essential signal in response to O2.− for InsP3R activation. Ca2+ release via PLC-β (Liao et al., 1989) was investigated using the G protein–coupled muscarinic M5 receptor agonist carbachol. No detectable Ca2+ signals were observed in response to carbachol (500 μM), indicating that DT40 cells lack the GPCR machinery necessary for PLC-β activation (unpublished data). However, we cannot exclude that O2.− may directly activate signaling upstream of PLC or regulate InsP3R. Earlier, we demonstrated the activation of mitochondrial PLA2 by O2.− (Madesh and Balasubramanian, 1997), lending support to our findings on the activation of signaling enzymes by O2.−.
Our findings suggest that ΔΨm loss in response to O2.− is dependent on ER stores and not extracellular Ca2+. However, it is unclear whether mitochondrially derived ROS exacerbate Ca2+ release from ER stores during oxidative stress. Rotenone and other distal complex I inhibitors generate O2.− on the matrix side of the inner membrane (Brookes et al., 2004). Our data indicate that cells pretreated with rotenone alone did not trigger either [Ca2+]i changes or a ΔΨm change. In contrast, the complex III inhibitor antimycin A caused a sharp decline in the ΔΨm without concomitant [Ca2+]i mobilization. This finding suggests that O2.− generation by complex III directly facilitates ΔΨm loss independent of [Ca2+]i levels. Cell death can be initiated by mitochondrial inhibitors through a reduction in ATP levels in a process known as necrosis. Specifically, oligomycin is known to reduce available ATP through inhibition of mitochondrial FoF1-ATPase and to elicit cell death through a switch from apoptosis to necrosis. In our system, endothelial cells pretreated with oligomycin did not experience either a rapid [Ca2+]i change or ΔΨm decay. However, subsequent delivery of O2.− perturbed the ER Ca2+ level and subsequent ΔΨm loss. Experiments using the mitochondrial uncoupler FCCP indicate that mitochondrial Ca2+ efflux precedes ΔΨm dissipation. Apparently, mitochondrial depolarization evoked by paracrine O2.− differs from ΔΨm alterations induced by mitochondrially derived ROS.
The question arises whether extracellular O2.− generation evokes selective signaling during endothelial dysfunction. Previously, cells exposed to O2.− but not H2O2 elicited a rapid and large cytochrome c release from the mitochondria, followed by ΔΨm loss (Madesh and Hajnoczky, 2001). Cell death has been associated with elevation of Ca2+ through various means. Moreover, elevation of [Ca2+]i has been implicated in the induction of apoptosis by ROS (Orrenius et al., 2003). It is suggested that H2O2 facilitates Ca2+ entry from the extracellular milieu or from the intracellular pools (Zhao, 2004), and H2O2-induced apoptosis in I/R injury has also been proposed (Inserte et al., 2000). This study suggests that O2.−, but not H2O2, evoked an intracellular store Ca2+ release that regulates the ΔΨm. Strikingly, pretreatment with the [Ca2+]i chelator BAPTA-AM prevents O2.−- but not H2O2-mediated endothelial ΔΨm loss. Thus, the O2.−-initiated ΔΨm loss is dependent on an [Ca2+]i rise and independent of mitochondrial ROS generation. These findings suggest that extracellularly generated O2.− rapidly evokes the observed [Ca2+]i elevation and pathological ΔΨm loss. Interestingly, we illustrate that externally delivered O2.−, and not other oxidants, triggers a cytosolic signal that initiates the mitochondrial phase of apoptosis.
Mitochondrial membrane permeabilization evoked by apoptotic stimuli facilitate apoptogenic protein release from the intermembrane space and can lead to the downstream activation of both caspase-dependent and -independent apoptotic cascades. Our previous observation proposed that O2.−, but not H2O2, elicited cytochrome c release via a voltage-dependent anion channel–dependent mitochondrial membrane permeabilization (Madesh and Hajnoczky, 2001). Cytochrome c release is regulated by the Bcl-2 family of proteins, and the target of these proteins in the cell is the MPTP (Kroemer and Reed, 2000; Mattson and, Kroemer, 2003). This study shows the activation of initiator and effector caspases by O2.− specifically, and to some extent, by high doses of H2O2. Recent evidence has indicated that a caspase-3–truncated InsP3R type I may elicit a prolonged [Ca2+]i elevation during apoptosis (Assefa et al., 2004). Our model indicates that caspase-3 activation is downstream of [Ca2+]i elevation and ΔΨm loss. However, we cannot rule out modification of InsP3R type I in the late stages of O2.−-triggered apoptosis. Collectively, these findings establish that ER Ca2+ mobilization is upstream of mitochondrial events evoked by O2.− in endothelial apoptosis.
In conclusion, activated macrophage-derived O2.− acts as an important signaling molecule that mediates InsP3R-linked [Ca2+]i elevation and mitochondrial dysfunction in endothelial cells and provides a novel signaling link between inflammatory and endothelial cells under pathological conditions. We therefore propose that paracrine O2.− signaling is critical to endothelial cell death.
Materials And Methods
Primary rat PMVECs (provided by T. Stevens, University of South Alabama, Mobile, AL) were cultured in DME supplemented with 10% FBS, nonessential amino acids, and antibiotics. Cells of wild-type DT40 chicken B cell line, triple InsP3R KO cell line (DT40 InsP3R KO), and PLC-γ2 KO (provided by A. August, Pennsylvania State University, Philadelphia, PA) cell line were cultured in RPMI 1640 supplemented with 10% FCS, 1% chicken serum, 50 μM 2-mercaptoethanol, 4 mM l-glutamine, and antibiotics. J774A.1 monocyte-derived mouse macrophages were cultured in Hank's F12 (supplemented with 10% FBS) and antibiotics. Heptocellular carcinoma cell line (HepG2) was cultured in MEM with 10% FBS, 2 mM l-glutamine, 0.50 mM sodium pyruvate, 0.1 mM nonessential amino acids, and antibiotics. Cells between passages 5 and 10 were used for experiments.
Visualization of ROS generation
J774.1 mouse monocyte-derived macrophages (106 cells/ml) were cultured on glass bottom 35-mm dishes (Harvard Apparatus) for 48 h. Cells were challenged with 1 μg/ml LPS for 3 h at 37°C. For DPI treatment, 2.5 h LPS-treated macrophages were incubated with 30 μM DPI for 30 min. The oxidation-sensitive dye H2DCF-DA (10 μM; Invitrogen) was added separately to dishes 20 min before visualization under confocal microscopy. Macrophage cells treated under similar conditions were used for co-culture model Ca2+ mobilization.
Measurement of [Ca2+]i changes was performed using the Ca2+-sensitive fluorescent dye Fluo-4/AM (Invitrogen). Cells adherent to 25-mm-diam glass coverslips were incubated at RT in extracellular membrane (ECM) containing 5 μM Fluo-4/AM for 30 min, followed by an additional 10-min incubation in a dye-free medium. Coverslips were affixed to a chamber and mounted in a PDMI-2 open perfusion microincubator (Harvard Apparatus) and maintained at 37°C on an inverted microscope (model TE300; Nikon). Confocal imaging was performed using the Radiance 2000 imaging system (Bio-Rad Laboratories) equipped with a Kr/Ar-ion laser source at 488-nm excitation using a 60× oil objective. Images were collected using LaserSharp software (Bio-Rad Laboratories) every 3 s for [Ca2+]i changes. Mobilization was induced by the application of 100 μM and 5 mU/ml, respectively, of the xanthine/xanthine oxidase O2.−-generating system. Whole cell masking was used to quantitate individual cell responses (Spectralyzer, custom software; provided by Paul Anderson, Thomas Jefferson University, Philadelphia, PA).
Measurement of inositol phosphates
24 h before experiments, cells (106/ml) were transferred to myo-inositol–free DME and incubated in the presence of myo-[2-3H]inositol (2 μCi/ml; 20 Ci/mmol; MP Biomedical, Inc.). After washing with myo-inositol–free DME, cells were incubated for 30 min in myo-inositol–free DME supplemented with 10 mM LiCl and then exposed to either ATP (100 μM) or X+XO (100 μM xanthine and 5 mU/ml XO) for 20 min at 37°C. The medium was subsequently removed and cells were scraped into 1 ml of 10% (wt/vol) TCA for the extraction of soluble inositol phosphates. After centrifugation of the cell lysates, the supernatant was applied to AG 1-X8 (formate form) ion exchange columns (200–400 mesh; Bio-Rad Laboratories). These columns were washed as previously described (Takata et al., 1995). Elution was performed with increasing concentrations of ammonium formate (0.1–0.7 M).
Simultaneous confocal imaging of cytosolic and mitochondrial Ca2+ in PMVECs
Endothelial cells were loaded with 2 μM rhod-2/AM in ECM containing 2.0% BSA in the presence of 0.003% pluronic acid at 37°C for 50 min. Cells loaded with rhod-2 dye were washed and then reloaded with Fluo-4/AM for an additional 30 min at RT. Cells were placed on a temperature-controlled stage and images were recorded using the Radiance 2000 imaging system with excitation at 488 and 568 nm for Fluo-4 and rhod-2, respectively.
Kinetics of [Ca2+]i elevation and mitochondrial membrane depolarization
Cells cultured on 25-mm-diam glass coverslips were loaded for 30 min with 5 μM Fluo-4/AM at RT. The cationic potentiometric fluorescent dye TMRE (100 nM) was added to the loading medium and allowed to equilibrate for at least 15 min. Under these conditions, TMRE fluorescence was largely localized to the mitochondrial matrix space. After dye loading, the cells were washed and resuspended in the experimental imaging solution (ECM containing 0.25% BSA). Intracellular esterase action then resulted in loading of both the cytoplasmic and mitochondrial compartments of the cell. Experiments were performed in ECM containing 0.25% BSA at 37°C. Images were recorded using the Radiance 2000 imaging system with excitation at 488 and 568 nm for Fluo-4 and TMRE, respectively. Fluo-4 and TMRE fluorescent changes were determined by background subtraction followed by masking of total cell area or intracellular regions. During ΔΨm loss, the exit of TMRE from mitochondria into the cytoplasm leads to quenching of the dye. The rapid redistribution of TMRE into the cytoplasm after depolarization of ΔΨm can be transiently detected in the nucleus.
Detection of caspase-3, -8, and -9 activity
The assay is based on the ability of the active enzymes to cleave the fluorogenic substrates Ac-DEVD-AFC (caspase-3), Ac-IETD-AFC (caspase-8), or Ac-LEHD-AFC (caspase-9; Calbiochem). Cells treated with various oxidants were harvested via trypsinization and washed with PBS. The cell pellet was gently resuspended in lysis buffer (25 mM Hepes, pH 7.4, 2 mM EDTA, 0.1% CHAPS, 5 mM DTT, 1 mM PMSF, and protease inhibitor cocktail [Roche], lysed, and centrifuged; the supernatant was used as the assay. Caspase substrates were added to a final concentration of 50 μM and the samples were incubated at 37°C for 45 min in caspase assay buffer. Incubated samples were measured at an excitation of 400 nm and an emission of 505 nm in a multiwavelength-excitation dual wavelength-emission fluorimeter (Delta RAM; Photon Technology International).
Confocal imaging analysis of apoptotic markers in PMVECs
To determine cellular outcome in response to oxidative stress, cells were exposed to the O2.−-generating system, H2O2, and t-BuOOH for 5 h. To assess the externalization of phosphatidylserine in the plasma membrane, as occurs in the early stage of apoptosis, cells were incubated with the conjugate annexin V Alexa Fluor-488 (Invitrogen) and PI (0.5 μg/ml) for 15 min. After treatment, annexin V– and PI-stained cells were visualized and counted. In normal cells, impermeable PI is internalized as the plasma membrane loses integrity. Thus, positive PI staining indicates either late stage of apoptosis or necrosis.
Tracings are representative of the mean fluorescence value of all cells in one field and are indicative of n independent experiments. Data given are representative of duplicate analysis of n independent experiments as mean ± SEM.
Online supplemental material
Fig. S1 shows the Ca2+ response to the physiological and pathological stimuli ATP and O2.−, respectively, in PMVECs. Fig. S2 details the measurement of InsP3 generation in both DT40 and PMVECs. Fig. S3 shows the analysis of apoptosis in DT40 cells in response to O2.−.
We thank Drs. Troy Stevens and Avery August for providing endothelial and PLC-γ2 KO cells, respectively. We are grateful to Dr. Kevin Foskett for critical manuscript review. We also thank Drs. Craig Thompson and Sheldon Feinstein for helpful suggestions and Paul Anderson for Spectralyzer image analysis software.
This work was supported by startup funds from the Institute for Environmental Medicine to M. Madesh. A.B. Fisher is funded by National Institutes of Health (NIH) grant HL-60290. S.K. Joseph is supported by NIH grant DK-34804.
Abbreviations used in this paper: 2-APB, 2-aminoethoxydiphenyl borate; BAPTA, 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetracetate; [Ca2+]i, intracellular calcium; ΔΨm, mitochondrial membrane potential; DCF, dichlorofluorescein; DPI, diphenyleneiodonium; ECM, extracellular medium; GPCR, G protein–coupled receptor; InsP3, inositol 1,4,5-trisphosphate; InsP3R, InsP3 receptor; I/R, ischemia/reperfusion; KO, knockout; LPS, lipopolysaccharide; MPTP, mitochondrial permeability transition pore; PI, propidium iodide; PMVEC, pulmonary microvascular endothelial cell; ROS, reactive oxygen species; SOD, superoxide dismutase; t-BuOOH, tert-butyl hydroperoxide; Tg, thapsigargin; TKO, triple KO; TMRE, tetramethylrhodamine, ethyl ester, perchlorate.