Abstract
LPS has been implicated in the pathogenesis of endothelial cell death associated with Gram-negative bacterial sepsis. The binding of LPS to the TLR-4 on the surface of endothelial cells initiates the formation of a death-inducing signaling complex at the cell surface. The subsequent signaling pathways that result in apoptotic cell death remain unclear and may differ among endothelial cells in different organs. We sought to determine whether LPS and cycloheximide-induced cell death in human lung microvascular endothelial cells (HmVECs) was dependent upon activation of the intrinsic apoptotic pathway and the generation of reactive oxygen species. We found that cells overexpressing the anti-apoptotic protein Bcl-XL were resistant to LPS and cycloheximide-induced death and that the proapoptotic Bcl-2 protein Bid was cleaved following treatment with LPS. The importance of Bid was confirmed by protection of Bid-deficient (bid−/−) mice from LPS-induced lung injury. Neither HmVECs treated with the combined superoxide dismutase/catalase mimetic EUK-134 nor HmVECs depleted of mitochondrial DNA (ρ0 cells) were protected against LPS and cycloheximide-induced death. We conclude that LPS and cycloheximide-induced death in HmVECs requires the intrinsic cell death pathway, but not the generation of reactive oxygen species.
Acute lung injury remains an important clinical problem in the U.S. affecting >190,000 patients per year (1). The most common cause of acute lung injury is bacterial infection resulting in the sepsis syndrome. LPS is a component of the outer membrane of Gram-negative bacteria and is released into the circulation as organisms replicate or die (2). The administration of LPS results in lung injury that is characterized by the presence of apoptosis in the endothelium and epithelium (3, 4, 5, 6). There is increasing evidence that cell death in the alveolar endothelium and epithelium may be important in the pathophysiology of acute lung injury (6). Whether this cell death is required for the lung injury observed following exposure to LPS is not known. Understanding the mechanism(s) by which exposure to LPS induces cell death in endothelial cells might allow us to design strategies to minimize lung injury in patients with systemic bacterial infections.
In vitro data are conflicting regarding the death pathway activated by LPS. In the blood, LPS binds to LPS binding protein, which transfers LPS to soluble CD-14 (soluble) (7). The LPS/sCD-14 complex then binds to the transmembrane protein Tlr-4 (8), initiating the formation of a multimeric complex at the cell surface. Mice lacking the Tlr-4 gene are resistant to LPS-induced injury (9). Some investigators have found that LPS- and cycloheximide-induced cell death occurs through the classical receptor-dependent pathway acting through a death domain-containing transmembrane receptor (10). Others have observed that cell death requires the generation of reactive oxygen species (ROS)3 and activation of the intrinsic apoptotic pathway (11, 12).
The intrinsic apoptotic pathway is executed through the interactions between the Bcl-2 family of proteins, which are classified by the number of BH3 domains they contain. Bcl-2 proteins containing three BH3 domains, for example Bax or Bak, permeabilize the outer mitochondrial membrane, releasing proapoptotic proteins into the cytosol. The activation of these proteins is inhibited by the antiapoptotic Bcl-2 proteins, for example Bcl-2 or Bcl-XL. Bcl-2 proteins containing only a single BH3 domain either activate Bak or Bak directly, for example Bid, Bim, or PUMA, or inhibit antiapoptotic proteins, for example Noxa or Bad (13). We sought to determine whether death in human lung microvascular endothelial cells (HmVECs) requires activation of the intrinsic apoptotic pathway through cleavage of Bid or through ROS generation. We found that the intrinsic pathway is required, as overexpression of the antiapoptotic Bcl-2 protein Bcl-XL conferred protection against LPS-induced death. We further demonstrate that Bid is cleaved following exposure to LPS. The generation of ROS was not required for LPS-induced death in these cells. Consistent with these findings, we found that bid−/− mice were protected against the development of LPS-induced lung injury.
Materials and Methods
Cell lines and reagents
HmVECs were purchased from Cambrex and cultured per the manufacturer’s protocol. In brief, the cells were maintained at 37°C in 5% CO2 in endothelial growth medium-2 medium supplemented with 5% FBS, human endothelial growth factor, hydrocortisone, gentamicin/amphotericin-B, vascular endothelial growth factor, human fibroblast growth factor, insulin-like growth factor, and ascorbic acid. Human lung microvascular endothelial ρ0 cells were prepared by growth in the same medium, supplemented with low-dose ethidium bromide (50 ng/ml) and uridine (100 μg/ml) for 2 wk. Depletion of mitochondrial DNA was confirmed by PCR for cytochrome oxidase subunit II (14). LPS (from Escherichia coli 0111:B4), cycloheximide (CHX), t-butyl hydroperoxide, and ethidium bromide were purchased from Sigma-Aldrich. EUK-134 was obtained from Proteome Systems. Cells were discarded after five passages.
Cell viability measurements
Cell death was assayed by measuring propidium iodine (PI) and/or annexin V staining of culture supernatants and adherent cells lifted from culture wells with trypsin. Sorting of cells by size and PI/annexin V staining was performed using flow cytometry. Cytochrome c release was assessed by immunostaining as previously described (15). In brief, culture supernatants and adherent cells lifted from culture wells with trypsin were centrifuged onto glass slides (Cytospin 3, Thermo Shandon, 1200 rpm, 5 min) and immunostained for cytcohrome c. Confocal images of the cells were then obtained from randomly selected areas of the slide (Nikon LSM 510 argon/helium laser confocal microscope using a Zeiss 100× Apochromat objective) and staining was assessed as punctuate or diffuse. Diffuse staining was interpreted as evidence of cytochrome c release.
Adenoviral infection
HmVECs were exposed to adenovirus without cDNA (AdNull) or cDNA-encoding glutathione peroxidase (GPx), manganese superoxide dismutase (MnSOD) or Bcl-XL in serum- and antibiotic-free medium for 2 h. Culture wells were then supplemented with normal medium and cells were incubated under normal conditions for 24 h before exposure to experimental conditions. Overexpression of GPx, MnSOD, or Bcl-XL was confirmed by immunoblotting.
LPS administration to mice
Eight- to 10-wk-old male wild-type (WT) C57BL/6 mice were purchased from Charles River Laboratories. The bid−/− mice were provided by the late Dr. Stanley Korsmeyer and were backcrossed to a C57BL/6 background for 12 generations (16). All experiments were approved by the Northwestern University Animal Care and Use Committee. WT and bid−/− mice were treated with a single i.p. injection of 50 μl PBS or LPS (6 mg/kg) in 50 μl PBS.
Histology
A 20-gauge angiocath was sutured into the trachea. The lungs and heart were removed en bloc and the lungs inflated to 25 cm H2O with PBS then heated to 60°C in PBS. The lungs were then inflated to total lung capacity with 0.8 cc 4% paraformaldehyde. The heart and lungs were fixed in paraffin and 5-μm sections were stained with H&E.
TUNEL staining
TUNEL staining was performed on deparaffinized, fixed lung sections according to the manufacturer’s protocol (FragEL DNA Fragmentation kit, Calbiochem catalogue number QIA39). The resulting stained sections were analyzed by confocal microscopy. TUNEL positive nuclei were counted in 10 randomly selected high power (400×) fields and averaged for each lung section.
Bronchoalveolar lavage (BAL) analysis
BAL was performed through a 20-gauge angiocath ligated into the trachea. One ml of PBS was instilled into the lungs and then carefully aspirated three times. A 200-μl aliquot of the BAL fluid (BALF) was placed in a cytospin and centrifuged at 1200 rpm for 5 min. The glass slides were Wright stained and subjected to a blinded manual cell count and differential.
Measurement of lung permeability
Pentobarbital-anesthetized mice were administered 125 μl of FITC-labeled albumin (Sigma-Aldrich) (0.16 mg/ml dissolved in 5% BSA) via retro-orbital injection. The mice were kept anesthetized, and 1 h later a BAL was performed. Fluorescence of the resulting BAL was measured (ex = 488, em = 530) in a fluorescent microplate reader.
Statistical analysis
One-way ANOVA was used to test for significant differences in measured variables between groups. Where the F statistic indicated a significant difference, individual differences were explored using the Bonferroni correction for multiple comparisons. Statistical significance was determined at the 0.05 level.
Results
The combination of LPS and CHX induces apoptotic cell death in HmVECs
We exposed HmVECs to medium alone, LPS (100 ng/ml), CHX (10 μg/ml), or the combination of LPS and CHX for 24 h. The cells were sorted by size and by staining for PI and annexin V using flow cytometry. The percentage of small, PI- and annexin V-stained cells was >2-fold greater in cells treated with LPS and CHX than in untreated cells. Identically treated cells were immunostained for cytochrome c and the percentage of cells demonstrating a diffuse rather than punctuate staining pattern was measured using confocal microscopy. LPS and cycloheximide resulted in cytochrome c release in the majority of cells. The increase in PI- and annexin V-staining from baseline indicates that the combination of LPS and CHX induces apoptotic cell death in HmVECs (Fig. 1).
The intrinsic apoptotic pathway is required for LPS-induced cell death
To determine whether the intrinsic cell death pathway is required for LPS and CHX-induced death in HmVECs, we infected cells with an adenovirus encoding no transgene (null) or the anti-apoptotic Bcl-2 protein Bcl-XL. Overexpression of Bcl-XL following infection was confirmed by immunoblotting (Fig. 2,A, inset). Null or Bcl-XL-infected cells were exposed to LPS, CHX, or the combination of LPS and CHX for 24 h and total and apoptotic cell death were measured by staining for PI or annexin V, respectively (Fig. 2, A and B). In the null-infected cells, the combination of LPS and CHX led to significant increases in cell death. This death was not observed in cells overexpressing Bcl-XL, suggesting that the intrinsic apoptotic pathway is required for LPS- and CHX-induced cell death in HmVECs.
ROS are not required for LPS-induced cell death in HmVECs
Kim et al. (17, 18) reported that the administration of pharmacologic antioxidants attenuated LPS- and CHX-induced apoptosis in HUVECs and Brigham et al. (12) observed that antioxidants decreased LPS- and CHX-induced cell death in bovine lung endothelial cells. We used two complementary strategies to determine whether the generation of cytosolic or mitochondrial ROS is required for LPS-induced death in HmVECs. To determine whether cytosolic antioxidants could prevent cell death in response to LPS and CHX, we treated HmVECs with the combined superoxide dismutase/catalase mimetic EUK-134 (20 μM) or vehicle for 4 h before and during treatment with LPS, CHX, or the combination of LPS and CHX. EUK-134 has been shown to correct the mev-1 aging defect in Caenorhabditis elegans (19) and protect alveolar epithelial cells against hyperoxia-induced cell death (15). Cells treated with EUK-134 were not protected from LPS- and CHX-induced death (Fig. 3,A). We also infected HmVECs with null virus or an adenovirus encoding GPx that catalyzes the conversion of hydrogen peroxide to water and oxygen. HmVECs overexpressing GPx had rates of LPS-induced cell death similar to controls (Fig. 3, B and C). To determine whether mitochondrial antioxidants could prevent cell death in response to LPS and CHX, we infected cells with a null adenoviurs or an adenovirus encoding MnSOD, a catalyst of the superoxide anion to hydrogen peroxide conversion in the mitochondrial intermembrane space. The overexpression of MnSOD failed to prevent LPS- and CHX-induced cell death (Fig. 4, A and B). The mitochondrial DNA encodes 13 proteins critical for oxidative phosphorylation. Cells lacking mitochondrial DNA (ρ0 cells) do not undergo oxidative phosphorylation and cannot generate ROS from mitochondrial electron transport. We generated ρ0 HmVECs by growing cells in ethidium bromide. Absence of cytochrome oxidase subunit II DNA was confirmed by PCR analysis of DNA extracts from WT and ρ0 HmVECS (Fig. 4,C). The ρ0 HmVECs showed rates of LPS-induced cell death similar to WT cells (Fig. 4 C). These results suggest that neither cytosolic nor mitochondrial generated ROS are required for LPS-induced death in HmVECs.
Bid is cleaved following exposure to LPS
Binding of death receptors to their ligands results in the activation of a death-inducing signaling complex (DISC) at the membrane that activates caspase-8 and caspase-10 (20). In some cells, activation of caspase-8 is sufficient to activate caspase-3 and caspase-7, resulting in apoptosis, while in other cells, amplification through the intrinsic pathway is required. Activation of the intrinsic pathway often occurs through the caspase-8-mediated cleavage of the proapoptotic Bcl-2 protein Bid, generating cleaved Bid, which activates Bax or Bak at the mitochondrial membrane resulting in the release of cytochrome c (21). To determine whether Bid was cleaved in LPS-induced cell death in HmVECs, we treated cells with LPS and CHX and assessed Bid cleavage by immunoblotting 30, 60, or 120 min later. Levels of total Bid progressively decreased after treatment with LPS and CHX, suggesting that Bid is cleaved following the treatment of HmVECs with LPS (Fig. 5). Abs available for human Bid do not recognize the cleaved form of Bid.
Loss of Bid ameliorates LPS-induced lung injury in mice
To determine whether the cleavage of Bid is required for cell death in the lung in vivo, we injected sterile PBS or LPS (6 mg/kg) into the peritoneal cavity of WT and bid−/− mice. Lungs were harvested after 48 h for histologic examination. As compared with WT mice, lungs from bid−/− mice demonstrated less alveolar destruction, less interstitial edema, and less fluid and debris in the air spaces (Fig. 6). TUNEL staining of lung sections revealed fewer apoptotic cells in lungs of bid−/− mice treated with LPS than WT mice treated with LPS (Fig. 7,A). Consistent with the findings on histology, bid−/− mice failed to significantly increase the permeability of the alveolar-capillary barrier to albumin in response to LPS (Fig. 7,B). Although mice lacking Bid had significantly attenuated lung injury in response to LPS, the LPS-induced inflammatory response was similar in WT and bid−/− animals. Forty-eight hours after the i.p. administration of LPS (6 mg/kg) or PBS, WT, and bid−/− animals were weighed and BAL was performed to determine BALF total cell counts. BALF total cell count and differential were similar in bid−/− and WT animals (Fig. 7,C). Furthermore, treatment with LPS induced a similar loss of weight in bid−/− and WT mice (Fig. 7 C).
Discussion
We sought to determine whether activation of the intrinsic apoptotic pathway was required for cell death in the lung endothelium. In HmVECs, we found evidence for the cleavage of the proapoptotic Bcl-2 protein, Bid in response to LPS and the prevention of LPS-induced cell death in cells overexpressing the anti-apoptotic Bcl-2 protein, Bcl-XL. The importance of the intrinsic apoptotic cell death pathway in vivo was demonstrated by the attenuation of acute lung injury and apoptosis in animals lacking Bid following treatment with LPS. Collectively, these findings suggest that the intrinsic cell death pathway is required for endothelial cell death induced by LPS. We found that LPS-induced cell death in HmVECs was independent of ROS generation.
Binding of LPS with its receptor TLR-4 at the cell membrane results in the recruitment of a number of proteins to the membrane including myeloid differentiation factor 88 (22) and IL-1 receptor-associated kinase-1 (23). There are reports that the Fas-associating protein with death domain is also recruited to the cell membrane and is required for LPS-induced death (10). The resulting DISC at the plasma membrane activates caspase-8. In many cells, caspase-8 activation is not sufficient to induce cell death without amplification by the intrinsic apoptotic pathway. Our findings suggest that amplification through mitochondrial membrane permeabilization is required for LPS to induce cell death in primary human lung endothelial cells. In contrast with other reports (24, 25), we found that mitochondrial amplification in these cells did not require the generation of ROS.
To survive in vitro, lung microvascular endothelial cells require the exogenous administration of growth factors including endothelial growth factor, human fibroblast growth factor, insulin like growth factor-1, hydrocortisone, and vascular endothelial growth factor. We found that reduction in the concentration of any of these growth factors by even 50% resulted in significant apoptosis (>20%) of the cells (data not shown). Most growth factors prevent apoptosis by activating the PI3 kinase/Akt pathway (reviewed in Ref. 26). It is therefore possible that Akt activation in response to the exogenous growth factors increased the resistance to LPS-induced death in our system. This limitation, however, is present in most culture systems as growth factors present in serum cause significant activation of Akt (27).
Our observation that lung injury and the number of apoptotic cells in the lung was attenuated in bid−/− compared with WT mice following exposure to LPS highlights the importance of amplification through the intrinsic apoptotic pathway in the development of lung injury in vivo. Caspase-mediated cleavage of Bid in response to a death stimulus results in the formation of truncated-Bid which activates the proapoptotic Bcl-2 proteins Bax or Bak (28). Once activated, Bax or Bak oligomerize and form channels in the mitochondrial membrane allowing the release of several molecules from the mitochondrial intermembrane space including cytochrome c (29). In the cytosol, cytochrome c binds to several other molecules to form a large multimeric complex called the apoptosome, where caspase-9 is cleaved, activating the terminal caspase cascade (30).
In response to the i.p. administration of LPS, we found that the degree of weight loss was similar in bid−/− mice and WT controls. These results are consistent with the report by Kang et al. who found that the survival of bid−/− mice treated with higher doses of LPS than those used in our study did not differ from WT controls (31). LPS-induced mortality is likely driven by the systemic inflammatory response, which our results suggest is independent of Bid. This is consistent with the observation that multiple strategies that prevent the release of proinflammatory cytokines, especially IL-1β, prevent LPS-induced mortality (32). Nevertheless, we found that the degree of lung injury was significantly reduced in bid−/− compared with WT mice. We hypothesize that LPS induces Bid-dependent apoptosis of parenchymal cells in the lung compromising the barrier function of the alveolar-capillary membrane. In support of this hypothesis, we observed decreased TUNEL staining in bid−/− compared with WT mice. Despite our in vitro data demonstrating Bid-dependent apoptosis of HmVECs in response to the administration of LPS/CHX, we cannot exclude the possibility that the Bid-dependent apoptosis observed in the lung after the administration of LPS was a secondary response to the release of systemic cytokines, which might result in the cleavage of Bid, for example TNF-α.
Both pneumonia and extrapulmonary infections are common causes of acute lung injury in critically ill patients (1). The induction of acute lung injury by administering LPS into the peritoneum, as reported in this study, is a model of sepsis-induced acute lung injury initiated outside the lung. Our results support a role for Bid in this model. Lung injury can also be induced by the direct instillation of LPS into the trachea, a model of pneumonia-induced acute lung injury (33). Our results do not show a role for Bid in this model.
Neither TNF-α nor LPS alone induce death in human endothelial cells in vitro, but the blockade of new protein synthesis renders the endothelium susceptible to these cytokines. Bannerman and colleagues (34) provided a potential mechanism for this observation when they identified that an inhibitor of caspase-8 activation at the DISC, the short form of FLICE-like inhibitory protein, was constitutively degraded by the ubiquitin-proteasome pathway. In the presence of cycloheximide, degradation of the short form of FLICE-like inhibitory protein without new protein synthesis sensitizes the cell to LPS- or TNF-induced cell death. Cycloheximide-like substances that inhibit protein synthesis have been identified in the BALF of septic patients (35).
Other investigators have observed that ROS are produced by endothelial cells in response to LPS, and may be required for LPS-induced cell death (11, 12, 17, 18, 36). These studies have been conducted in bovine and sheep pulmonary artery endothelial cells, and HUVECs. We did not observe a requirement for ROS generation in LPS-induced cell death in HmVECS. This may reflect a difference between HmVECs and other endothelial cells. Alternatively, Ricci et al. (37) have reported that caspase-mediated cleavage of components of the mitochondrial electron transport chain results in the generation of a burst of ROS downstream of the mitochondrial membrane permeabilization. In some models, inhibition of this oxidant burst can delay, but not prevent, cell death.
In summary, our findings suggest that LPS-induced cell death in HmVECs requires the intrinsic apoptotic pathway. The importance of this pathway in vivo is suggested by the protection conferred against LPS-induced cell death and lung injury, but not against inflammation, in exposed Bid-deficient mice. Our results would support the use of agents that prevent mitochondrial-dependent apoptosis in the prevention of LPS-induced lung injury.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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.
This work was supported by HL078131, HL071643, GM060472, HL067835, ES015024, ES013995, The American Lung Association, and The Crane Asthma Center.
Abbreviations used in this paper: ROS, reactive oxygen species; HmVEC, human lung microvascular endothelial cell; CHX, cycloheximide; PI, propidium iodine; GPx, glutathione peroxidase; MnSOD, manganese superoxide dismutase; WT, wild type; BAL, bronchoalveolar lavage; BALF, BAL fluid; DISC, death-inducing signaling complex.