Carbon Monoxide Differentially Modulates STAT1 and STAT3 and Inhibits Apoptosis via a Phosphatidylinositol 3-Kinase/Akt and p38 Kinase-dependent STAT3 Pathway during Anoxia-Reoxygenation Injury*

Carbon monoxide (CO), previously considered a toxic waste product of heme catabolism, is emerging as an important gaseous molecule. In addition to its important role in neurotransmission, exogenous CO protects against vascular injury, transplant rejection, and acute lung injury. However, little is known regarding the precise signaling mechanisms of CO. We have recently shown that CO attenuates endothelial cell apoptosis during anoxia-reoxygenation injury by activating MKK3/p38 (cid:1) mitogen-activated protein kinase (MAPK) pathways. Our current study is the first to demonstrate that CO can differentially modulate STAT1 and STAT3 activation and, specifically, that STAT3 activation by CO is responsible for the anti-apoptotic effect in endothelial cells. In addition, we show that the anti-apoptotic effects of CO depend upon both phosphatidylinositol 3-kinase/Akt and p38 MAPK signaling pathways in endothelial cells, whereas previous reports have impli-cated only the MKK3/p38 MAPK pathway. Using chemi-cal inhibitors and dominant negative constructs, we show that CO enhances STAT3 activation

The lungs possess the largest surface area of endothelial cells in the body. Lung endothelial cells are subjected to ischemiareperfusion (I-R) 1 injury during lung transplantation, surgery or thrombectomy. Anoxia-reoxygenation (A-R) is used as an in vitro model of I-R injury. Strategies to attenuate I-R/A-R-induced apoptosis have been shown to improve survival and organ function (1). CO is a reaction product of heme oxygenasemediated heme degradation that exhibits potent anti-inflammatory (2), anti-proliferative (3), and anti-apoptotic effects (4). Recent data show that low levels of exogenous CO administration can ameliorate cardiac and lung transplant rejection (5,6), suppress arteriosclerotic lesions associated with chronic graft rejection (7), and protect against lethal lung injury (8,9). Despite the emerging evidence of its biologic importance, the precise mechanisms of CO remain elusive. In this study, we sought to further delineate the signal transduction pathways utilized by CO during A-R injury.
The PI3K/Akt and STAT pathways are known to play critical roles in mediating cell proliferation, differentiation, and apoptotic responses (10,11). The STAT family of transcription factors also mediates pro-and anti-apoptotic effects (11,12). There is little information on the role of PI3K/Akt and STAT pathways in mediating apoptotic signals during A-R injury, and nothing is known about the ability of CO to modulate these pathways. Given our previous data showing that CO requires p38 MAPK activation to exert an anti-apoptotic effect during endothelial cell A-R (13) and reported associations among MAPKs, PI3K/Akt, and STATs (14,15), we investigated whether CO modulates A-R-induced apoptosis through PI3K/ Akt and STAT pathways in lung endothelial cells.
Endothelial Cell A-R CO Exposure-Rat pulmonary artery endothelial cells (PAEC) have been described previously (16) and were exposed to A-R in the presence or absence of 15 ppm of CO according to our previous methods (4).
Apoptosis Detection-Annexin V-FITC detection (BD Biosciences) was used to determine apoptosis in PAEC by flow cytometry as described previously (4).
Flow Cytometric Analysis of Cell Surface Fas Expression-Cells were trypsinized, pelleted, suspended with Fas (1:100 dilution) or control rat IgG (1:100 dilution) antibody, and incubated on ice. The cells were resuspended in phosphate-buffered saline containing anti-rat FITC * 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.
Western Blot Analysis-Phospho-p38, phospho-STAT1, phospho-STAT3, and phospho-Akt were analyzed by Western blot assays. To verify equivalent sample loading, membranes were stripped with Blot Restore Membrane rejuvenation solution (Chemicon International, Temecula, CA) and re-probed with anti-␤-tubulin antibody. Antibody to total STAT1 or STAT3 was used to detect the respective total STAT protein expression.
Measurement of Caspase 3 Activity-Caspase 3 activity was measured with the CaspACE assay system (Promega, Madison, WI). After A-R PAEC were washed twice with ice-cold phosphate-buffered saline and resuspended in cell lysis buffer, cell lysates were centrifuged and the supernatants were incubated with the colorimetric substrate Ac-Asp-Glu-Val-Asp-p-nitroanilide (Ac-DEVD-p-nitroanilide). The release of p-nitroanilide from Ac-DEVD-p-nitroanilide was measured at 405 nm using a spectrophotometer.
Plasmid Constructs and Transient Transfections-PAEC were incubated for 6 h with DNA mixtures containing serum-free medium, Fu-GENE 6 reagent (Roche Applied Science), and the relevant plasmid (STAT1 or STAT3 dominant negative mutant or pEFneo as empty vector control). STAT1 dominant negative mutant (Y701F) and STAT3 dominant negative mutant (Y705F) plasmids have been described previously (17,18). After incubation, cells were cultured for an additional 16 h in complete medium and then exposed to A-R in the presence or absence of 15 ppm. Our past studies have demonstrated the transfection efficiency of PAEC to exceed 80% using pEGFP transfections (16).
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear protein extracts from PAEC were prepared using NE-PER® nuclear and cytoplasmic reagents (Pierce, Rockford, IL) according to the manufacturer's instructions. STAT electrophoretic mobility shift assays were performed as described previously with minor modifications (19). DNA binding activity was determined after incubation of 4 g of PAEC nuclear protein extract with [␥-32 P]ATP-labeled, double-stranded oligonucleotide (5Ј-GTGCATTTCCCGTAAATCTTGTCTACA-3Ј), representing a high affinity sis-inducing element (SIE) of the c-fos promoter (Santa Cruz Biotechnology), which yields both STAT1 and STAT3 binding (20 -22). Protein-DNA complexes were separated on Novex 6% retardation gels (Invitrogen). The gel was dried and exposed to Biomax MR film (Eastman Kodak Co., Rochester, NY). Selfcompetitions were carried out under the same conditions with a 100-fold molar excess of the unlabeled probe. Nonspecific competitions were performed with an unlabeled oligonucleotide probe encompassing an SP1 transcription factor binding site. For supershifts, nuclear protein extracts were preincubated with polyclonal anti-STAT1 antibody or anti-STAT3 antibody (Santa Cruz Biotechnology) for 45 min at room temperature before proceeding to the binding reaction described above.
Statistics-The data were expressed as mean Ϯ S.E., and analyzed by Student's t test. Significance was accepted at p Ͻ 0.05.

CO Differentially Modulates STAT1 and STAT3 Activation in PAEC during A-R-We have previously demonstrated that
A-R-induced apoptosis in PAEC is initiated during the anoxia phase (4). We have also shown that the ability of CO to attenuate A-R-induced apoptosis is dependent upon its modulation of critical pathways, such as p38 MAPK, during anoxia (4). To evaluate the potential roles of STAT1 and STAT3 in the antiapoptotic effect of CO, we first conducted a time course of phosphorylated and thus activated STAT1 and STAT3 protein expression during anoxia in the presence and absence of CO (Fig. 1A). CO decreased phospho-STAT1 expression but increased phospho-STAT3 expression by 8-h anoxia with maximal effect by 24-h anoxia. Our earlier studies demonstrated that, in the absence of CO, A-R-induced cell death was maximal after 24 h of anoxia and was maintained up to 8 h of reoxygenation (4). Therefore, we investigated the effect of CO on STAT1 and STAT3 during 24-h anoxia followed by a time course of reoxygenation. Similar to the effects of CO on STAT1 and STAT3 during anoxia, CO attenuated phospho-STAT1 protein while enhancing phospho-STAT3 protein during 24 h of anoxia followed by 0.5 to 4 h of reoxygenation (Fig. 1B, lanes 7-9). We tested for STAT DNA binding activity with the SIE oligonucleotide in PAEC during A-R in the presence and absence of CO and found that CO decreased the STAT1 binding activity but enhanced STAT3 binding activity during 24-h anoxia and 24-h anoxia followed by 1-h reoxygenation (Fig. 1C, lanes 4 and 5). The specificity of STAT1 and STAT3 binding was demonstrated by the ability of unlabeled SIE to compete with the radiolabeled SIE (Fig. 1C, lane 6), whereas unlabeled SP1, an unrelated oligonucleotide, did not compete for STAT1 and STAT3 binding (Fig. 1C, lane 7). In addition, antibodies against STAT1 or STAT3 were able to supershift the respective complexes (Fig.  1D). Therefore, we postulated that the ability of CO to decrease STAT1 activation and/or to increase STAT3 activation during anoxia and A-R might be responsible for the anti-apoptotic effect of CO.
CO Differentially Modulates STAT1 and STAT3 Activation through the PI3K/Akt and p38 MAPK Pathways in PAEC during A-R-We have shown in earlier studies that CO depends upon the activation of the p38 MAPK pathway to exert an anti-apoptotic effect (4,13). Our current data demonstrate that CO modulates STAT1 and STAT3. The PI3K/Akt pathway is an important survival pathway and has been shown to modulate STAT3 serine phosphorylation (15). Therefore, we explored the role of PI3K/Akt in CO-mediated anti-apoptosis. A time course of phospho-Akt protein expression during anoxia in the presence and absence of CO demonstrates that CO increases phospho-Akt expression by 16 -24 h of anoxia (Figs. 2A, lanes 6 and  7). Of note, we have published a similar time course of p38 MAPK activation by CO during anoxia (13). During A-R, CO also enhanced both p38 MAPK and Akt phosphorylation in PAEC (Fig. 2B, lanes 7-9). Unless otherwise noted, the subsequent figures used 24-h anoxia followed by 1-h reoxygenation (denoted as A-R) as a representative time point. To determine whether PI3K/Akt was upstream of p38 MAPK, we blocked the PI3K/Akt pathway with LY294002, a compound that acts as a competitive inhibitor of the ATP binding site of PI3K and has been shown to cause specific inhibition (23,24). We blocked the p38 MAPK signaling pathway with SB203580 as described previously (16). LY294002 pretreatment significantly inhibited CO-mediated activation of both Akt and p38 MAPK (Fig. 2C,  lane 4). However, SB203580 pretreatment only inhibited p38 MAPK activation and had no effect on Akt activation (Fig. 2C,  lane 5). These data indicated that the activation of p38 MAPK by CO is PI3K/Akt-dependent. We next tested whether CO modulated STAT1 and STAT3 via PI3K/Akt and p38 MAPK pathways. We pretreated PAEC with SB203580 and observed that CO can no longer modulate STAT1 or STAT3 activation (Fig. 2D, lane 5). Consistent with our data that PI3K/Akt activation is upstream of p38 MAPK activation by CO during A-R (Fig. 2C), when we blocked the PI3K/Akt pathway with LY294002, CO lost its ability to decrease STAT1 and increase STAT3 activation during A-R (Fig. 2D, lane 4). These data demonstrated that the opposing effect of CO on STAT1 and STAT3 activation is dependent upon PI3K/Akt and p38 MAPK pathways.
The Anti-apoptotic Effect of CO Is Mediated through the PI3K/Akt and p38 MAPK Pathways in PAEC during A-R-We next hypothesized that the inhibition of PI3K/Akt or p38 MAPK will abolish the anti-apoptotic effect of CO during A-R. Pretreatment of PAEC with LY294002 or SB203580 completely abolished the anti-apoptotic effect of CO as assessed by flow cytometry analysis for annexin V-FITC expression (Fig. 3A). Pretreatment with LY294002 or SB203580 in room air showed basal levels of cell death similar to those in room air alone (Fig.  3B, lanes 2 and 3). CO dramatically decreased A-R-induced apoptosis (Fig. 3B, lane 7). However, in the presence of PI3K/ Akt inhibition with LY294002 or p38 MAPK inhibition with SB203580, CO could not attenuate A-R-induced apoptosis (Fig.  3B, lanes 8 and 9, respectively).
CO Requires STAT3 Activation to Exert an Anti-apoptotic Effect in PAEC during A-R-Thus far, we have demonstrated that CO differentially modulates STAT1 and STAT3 activation via the PI3K/Akt and p38 MAPK pathways. Furthermore, the PI3K/Akt and p38 MAPK pathways are required for the antiapoptotic effect of CO. Therefore, we hypothesized that decreased STAT1 and/or increased STAT3 activation may play an important role in the anti-apoptotic effect of CO. PAEC were transfected with empty vector or dominant negative mutant STAT1 (STAT1 DNM) or STAT3 (STAT3 DNM) or both DNM plasmids and then exposed to A-R in the presence and absence of CO. We first confirmed the ability of the DNMs to inhibit STAT1 and STAT3 activation. In Fig. 4A, PAEC transfected with STAT1 DNM, STAT3 DNM, or both STAT1 and STAT3 DNMs exhibited the appropriate decreases in A-R-induced DNA binding activity. Fig. 4B demonstrates a similar effect of the DNMs on STAT1 and STAT3 protein phosphorylation. As expected, empty vector transfection had no effect on A-R-induced STAT1 and STAT3 DNA binding activity or phosphorylation. We proceeded to test the effect of the DNMs on A-Rinduced apoptosis. PAEC transfected with STAT3 DNM showed significantly increased apoptosis during A-R compared with wild type PAEC during A-R (Fig. 4D, lanes 5 and 7). In addition, CO lost its ability to inhibit A-R-induced apoptosis in the presence of STAT3 DNM (Fig. 4D, lane 11). However, in the presence of STAT1 DNM, PAEC exhibited similar levels of apoptosis compared with wild type PAEC during A-R and CO retained its ability to attenuate A-R-induced apoptosis (Fig.  4D, lanes 6 and 10). When both STAT1 and STAT3 DNM were transiently transfected, the level of A-R-induced apoptosis and CO-mediated effects were similar to that of STAT3 DNM alone (Fig. 4D, lanes 8 and 12). The transfection of wild type STAT1 and STAT3 constructs had no effect on CO-mediated apoptosis (data not shown). These data demonstrated that STAT3 activation, but not STAT1 inactivation, by CO is critical for the anti-apoptotic effect of CO during A-R.
CO Decreases Fas Expression through PI3K/Akt and p38 MAPK Pathways in PAEC during A-R-Our laboratory and others have previously shown that the anti-apoptotic effect of CO involves Fas inhibition through the p38 MAPK pathway (13). Furthermore, p38 MAPK has been shown to protect human melanoma cells from UV-induced apoptosis through the down-regulation of Fas expression (25). Given that our current studies demonstrate that the activation of p38 MAPK is via PI3K/Akt (Fig. 2C), we treated PAEC with LY294002 and SB203580 to assess Fas expression with flow cytometry analysis during A-R in the presence and absence of CO. Pretreatment of PAEC with LY294002 or SB203580 completely attenuated the inhibitory effect of CO on Fas expression (Fig. 5). CO decreased A-R-induced Fas expression to less than 25%, but in the presence of LY294002 or SB203580, CO lost its ability to  6). B, Western blot analysis of phospho-STAT1 (p-STAT1) and phospho-STAT3 (p-STAT3) in nuclear extracts from PAEC transfected with the constructs as noted during A-R. Antibodies to total STAT1 or STAT3 were used to detect levels of total STAT1 or STAT3, respectively. Antibody to ␤-tubulin was used as loading control. C, flow cytometry analysis of apoptosis in wild type PAEC during RA, 24-h anoxia followed by 1-h reoxygenation (A-R), A-R in the presence of CO (CO/A-R), or PAEC transiently transfected with STAT1/STAT3 DNMs in the presence or absence of CO. The percentage of total cells exhibiting early apoptosis or late apoptosis/necrosis is noted in quadrants II and III. Results shown are representative of three independent experiments. D, graphical quantitation of the mean flow cytometry results from three independent flow cytometry analyses Ϯ S.E. *, p Ͻ 0.05 versus corresponding RA control (lanes 1-4) Carbon Monoxide Modulates STAT, PI3K/Akt, and p38 Kinase decrease Fas expression (Fig. 5B, lanes 8 and 9). These data indicated that the inhibition of Fas expression by CO during A-R in PAEC is dependent upon PI3K/Akt and p38 MAPK pathways.
CO Requires STAT3 Activation to Attenuate Fas Expression in PAEC during A-R-Given that PI3K/Akt and p38 MAPK pathways are involved in STAT1 and STAT3 modulation by CO (Fig. 2D) and the study by Ivanov et al. (26) that demonstrates that dominant negative STAT3 in melanoma cells efficiently increased Fas expression, we surmised that STAT3 activation may be upstream of CO-mediated Fas modulation. PAEC were transiently transfected with STAT1 DNM or STAT3 DNM or both STAT1 and STAT3 DNM plasmids and then exposed to A-R in the presence or absence of CO. Not surprisingly, the Fas expression data paralleled the apoptosis data. PAEC transfected with STAT3 DNM showed significantly increased Fas expression during A-R compared with wild type PAEC during A-R (Fig. 6B, lanes 5 and 7). In addition, CO was unable to inhibit A-R-induced Fas expression in the presence of STAT3 DNM (Fig. 6B, lane 11). However, in the presence of STAT1 DNM, PAEC exhibited similar levels of Fas expression compared with wild type PAEC during A-R and CO retained its ability to attenuate A-R-induced Fas expression (Fig. 6B, lanes  6 and 10). When both STAT1 and STAT3 DNM were transiently transfected, the level of A-R-induced Fas expression and CO-mediated effects were similar to that of STAT3 DNM alone (Fig. 6B, lanes 8 and 12). These data indicated that the inhibi-tion of Fas expression by CO during A-R depends upon STAT3 activation.
CO Requires STAT3 Activation to Attenuate Caspase 3 Activity in PAEC during A-R-Similar to our apoptosis and Fas expression data, CO lost its ability to attenuate A-R-induced caspase 3 activity in the presence of STAT3 DNM (Fig. 6C,  lanes 9 and 11). However, STAT1 DNM had no effect on A-Rinduced caspase 3 activity or on the ability of CO to modulate caspase 3 (Fig. 6C, lanes 6 and 10). As expected, the simultaneous transient transfection of both STAT1 and STAT3 resulted in effects similar to those of STAT3 DNM alone (Fig. 6C,  lanes 8 and 12).
CO Inhibits Caspase 3 Activity through PI3K/Akt and p38 MAPK Pathways in PAEC during A-R-To further define the downstream events modulated by CO via the PI3K/Akt and p38 MAPK pathways, we assessed CO-modulated caspase 3 activity in the presence of LY294002 or SB203580. Caspase 3 activity was markedly decreased in the presence of CO (Fig. 6D, lane  7). When PAEC were pretreated with LY294002 (50 M) or SB203580 (10 M), CO lost its ability to decrease A-R-induced caspase 3 activity (Fig. 6D, lanes 8 and 9). These data indicated that the inhibition of caspase 3 activity by CO during A-R is PI3K/Akt-and p38 MAPK-dependent. DISCUSSION CO is emerging as a gaseous molecule with profound and potentially therapeutic biologic effects. Exposing mice to exog- enous CO in sublethal ranges up to 500 ppm dramatically attenuates inflammation, apoptosis, and lethality in a variety of injury and transplantation models (2,5,7,8,9,27). Elucidating the signaling mechanisms of CO-mediated effects will be important if we are to precisely delineate the biology and potential applications of this often misunderstood gas. Our present study demonstrates that CO differentially modulates STAT1 and STAT3 via PI3K/Akt/p38 MAPK-dependent pathways. Furthermore, the ability of CO to increase A-R-induced STAT3 activation leads to CO-mediated attenuation of Fas expression and caspase 3 activity, thereby decreasing A-Rinduced apoptosis. There is evidence that STAT1 and STAT3 have opposing actions in cytoprotection, cell death, and inflammation. STAT1 plays a critical role in I-R-induced cardiomyocyte apoptosis (29), and STAT1 Ϫ/Ϫ mice do not exhibit significant lipopolysaccharide-induced liver damage and in-flammation. STAT3 appears to mediate resistance to apoptosis in rheumatoid arthritis synovial fibroblasts (30). Recently, cardiomyocyte-restricted STAT3 knock-out mice exhibited greater inflammation, cardiac fibrosis, and heart failure with advanced age (31).
Our studies show that, not only does CO differentially modulate STAT1 and STAT3, the CO-mediated increase in STAT3 activation, rather than STAT1 attenuation, is responsible for improved endothelial cell survival during A-R injury. STAT1 DNM did not significantly affect A-R-induced apoptosis, whereas STAT3 DNM increased A-R-induced Fas expression, caspase 3 activity, and endothelial cell apoptosis. Furthermore, STAT3 DNM inhibited the ability of CO to modulate apoptosis, Fas expression, and caspase 3 activity. Taken together, the data indicate that STAT3 has a critical cytoprotective role and mediates the anti-apoptotic effect of CO in endothelial cells. We have previously demonstrated that CO inhibited A-R-induced ERK1/2 and c-Jun N-terminal kinase 1/2 MAPK activation but increased activation of p38 MAPK and MKK3, an upstream kinase (4,13). MAPKs have been found to be involved in STAT phosphorylation induced by interferon, serum withdrawal, ultraviolet radiation, and ischemia (28,(32)(33)(34). Precisely how p38 MAPK modulates STAT1 and STAT3 phosphorylation is still unclear and beyond the scope of our current study. We demonstrate that CO loses its ability to decrease STAT1 activation and increase STAT3 activation in the presence of p38 MAPK inhibition in endothelial cells. Furthermore, for the first time, we demonstrate that the anti-apoptotic effects of CO depend upon signaling through both the PI3K/Akt and p38 MAPK pathways. In lung endothelial cells, CO-induced Akt expression and anti-apoptotic effects depend upon the PI3K/Akt pathway, which appears to be upstream of p38 MAPK activation by CO.
We use primary lung endothelial cells to delineate the ability of CO to modulate STAT1 and STAT3 via PI3K/Akt and p38 MAPK-dependent pathways. In addition, we demonstrate that STAT3 activation, rather than STAT1 inactivation, confers protection against A-R-mediated cell death. Consistent with this observation is our data showing that CO-mediated STAT3 activation, rather than CO-mediated STAT1 inactivation, is responsible for the ability of CO to attenuate Fas expression, caspase 3 activity, and apoptosis. I-R/A-R-induced apoptosis is an important contributor to organ failure, transplantation rejection, and mortality, yet effective therapies do not currently exist. By delineating anti-apoptotic mechanisms of gaseous molecules such as CO, we may uncover novel foundations upon which to base future therapeutic strategies.