Introduction
Sepsis remains a common and serious clinical problem with significant morbidity and mortality. It is estimated that one million cases of sepsis occur annually in North America, resulting in death in 20–30 % of cases, such that sepsis is the most common cause of mortality in Intensive Care Units (ICUs) and hospitalized patients [
1‐
3]. Moreover, sepsis is a significant healthcare burden, as it consumes up to 45 % of total ICU costs [
2,
4]. Morbidity/mortality in sepsis are principally due to injury and dysfunction of multiple organs, most commonly acute lung injury (ALI)/acute respiratory distress syndrome (ARDS) [
2,
4‐
6]. Advances in our understanding of the pathophysiology of sepsis and organ dysfunction through basic and clinical research efforts have not significantly improved sepsis outcomes, as the treatment of sepsis and related organ dysfunction remains largely supportive care [
6‐
8]. Indeed, clinical trials of novel anti-inflammatory therapeutic approaches have been disappointingly negative, as none of these therapies has improved clinical outcomes in sepsis and septic ARDS [
6,
9‐
11].
The pathophysiology of ALI/ARDS in sepsis is the complex result of the actions of circulating cellular elements (e.g. polymorphonuclear leukocytes [PMN], platelets) and soluble inflammatory mediators, such as lipopolysaccharide (LPS) and multiple cytokines (e.g. tumour necrosis factor [TNF]α, interleukin [IL]1β), on multiple pulmonary cellular targets [
7,
8,
12]. These include tissue-resident inflammatory cells (e.g. alveolar macrophages), alveolar epithelial cells, and the pulmonary vasculature, both large vessels and especially the pulmonary microvasculature. Septic microvascular dysfunction of both pulmonary and systemic vascular beds is clinically important as it is present early in the course of human sepsis and is associated with higher mortality [
13‐
15], especially if persistent over time [
16].
Pulmonary microvascular injury and resulting dysfunction are characterized by increased pulmonary microvascular PMN sequestration/adhesion as well as disruption of the normal pulmonary microvascular alveolo-capillary permeability barrier, resulting in extra-vascular leak of protein-rich edema and PMN into pulmonary interstitial and alveolar compartments and clinically severe hypoxaemic respiratory failure [
8,
17‐
23]. Septic pulmonary microvascular dysfunction is primarily the result of the effects of septic inflammation on pulmonary microvascular endothelial cells (PMVEC) [
7,
13,
23‐
29]. There are several potential mechanisms of septic PMVEC injury and dysfunction, including disruption of inter-cellular junctions (e.g. adherens junctions), microtubule activation, and actin cytoskeleton remodeling leading to cell retraction and gap formation [
30‐
33]. Our previous work in a murine sepsis model identified that pulmonary microvascular/PMVEC barrier dysfunction, as reflected by enhanced albumin leak and oxidant stress required the presence of both alveolar macrophages and PMN, CD18-dependent PMN-PMVEC adhesion, increased nitric oxide (NO) production from inducible NO synthase (iNOS), and iNOS/NADPH oxidase-dependent peroxynitrite-mediated signaling [
17‐
19,
34‐
36].
Recently, we reported the first direct
in vivo evidence of septic PMVEC death, as observed by intravital videomicroscopy (IVVM) in septic mice [
37]. Moreover, this septic PMVEC death was associated with evidence of apoptosis, as reflected by increased Annexin V binding, caspase activation, and TUNEL staining. PMVEC apoptosis has been proposed in sepsis models [
38,
39], and could contribute to pulmonary microvascular albumin-permeability barrier dysfunction [
40,
41]. Indeed,
in vivo manipulation of various apoptosis pathways (e.g. the Fas-Fas ligand pathway) in animal models of sepsis improved measures of ALI severity suggesting a potential importance of global apoptosis in septic ALI [
42,
43]. However, the specific role of PMVEC death in sepsis-induced ALI and septic pulmonary microvascular/PMVEC dysfunction remains uncertain, as does the importance of apoptosis in PMVEC death/dysfunction.
Thus, to further our previous studies, we first defined the time course of the onset of septic pulmonary microvascular barrier dysfunction in vivo in the murine cecal ligation/perforation (CLP) model of sepsis, and correlated this with the time course of PMVEC death as well as 3 specific molecular markers of PMVEC apoptosis. In addition, we assessed the effects of a specific, irreversible broad-spectrum caspase inhibitor, Q-VD, on septic PMVEC apoptotic death and septic pulmonary microvascular barrier dysfunction. Finally, we assessed the contribution of iNOS- and NADPH-oxidase-dependent oxidant stress to both septic PMVEC barrier dysfunction and death/apoptosis. Pulmonary microvascular Evans blue (EB)-labeled albumin leak increased by 2 h and peaked 4 h after CLP-induced sepsis, in parallel to and directly correlated with significant septic PMVEC death in vivo, as well as, and specifically, apoptosis. Moreover, effective inhibition of PMVEC apoptosis using the broad-spectrum caspase inhibitor, Q-VD, significantly attenuated septic pulmonary microvascular dysfunction, including both septic albumin hyper-permeability and pulmonary microvascular PMN sequestration. Both septic PMVEC apoptosis and pulmonary microvascular/PMVEC dysfunction were mediated through iNOS- and NADPH oxidase-dependent signaling.
Discussion
In the murine CLP-septic ALI model, pulmonary microvascular/PMVEC barrier dysfunction in vivo and resulting albumin hyper-permeability developed rapidly over 2–4 h after CLP-sepsis, with a similar time course as PMVEC death, as reflected by PI+ staining, and with specific features consistent with PMVEC apoptosis, including surface phosphatidyl serine presence (Annexin V staining), caspase activation (FLIVO staining), and DNA fragmentation (TUNEL labeling). Moreover, septic pulmonary microvascular/PMVEC barrier dysfunction correlated strongly with measures of PMVEC death/apoptosis. Pharmacologic caspase inhibition with Q-VD in vivo prior to CLP significantly prevented septic PMVEC death/apoptosis and attenuated septic pulmonary microvascular/PMVEC dysfunction, as reflected by both reduced albumin-permeability barrier dysfunction and pulmonary microvascular PMN sequestration. Septic PMVEC apoptosis and microvascular/PMVEC dysfunction in vivo were both similarly abolished following CLP-sepsis in mice lacking iNOS (Nos2
−/−
) or NADPH oxidase activity (p47
phox−/−
and gp91
phox−/−
), as well as in wild-type mice pre-treated with the NADPH oxidase pharmacologic inhibitor, apocynin. Thus, septic pulmonary microvascular/PMVEC dysfunction including high-albumin pulmonary edema and enhanced PMN sequestration, key pathophysiologic features of septic ALI, appear to be the result of PMVEC death specifically due to caspase-dependent apoptosis, which appears to be mediated through the activity of both iNOS and NADPH oxidase.
The high morbidity and mortality of sepsis are the result of multiple organ dysfunction/failure, including ALI/ARDS, which is due in large part to septic injury and dysfunction of the microvasculature and especially MVEC [
2,
4,
5,
48]. Septic microvascular/MVEC injury and dysfunction are characterized pathophysiologically by perturbed microvascular distribution of blood flow in many organs, enhanced microvascular thrombosis, enhanced PMN-MVEC interaction, and disruption of the normal microvascular/MVEC permeability barrier, resulting in high-permeability albumin-rich tissue and organ edema [
13,
14,
17‐
21,
30,
48‐
53]. The severe hypoxaemic respiratory failure of ALI/ARDS is pathophysiologically due to pulmonary microvascular PMN sequestration and influx into pulmonary interstitial and alveolar compartments, mediated through enhanced PMN-PMVEC adhesion, and the associated injury and dysfunction of the pulmonary microvascular alveolo-capillary permeability barrier and resulting extra-vascular leak of protein-rich edema [
17‐
21,
51].
In ALI/ARDS, pathologic disturbances have been described in the structure and function of both key cells in the normal alveolo-capillary permeability barrier, namely PMVEC and alveolar epithelial cells (AEC) [
54,
55]. For example, AEC dysfunction in ALI/ARDS is characterized by impairment of their normal barrier and alveolar water clearance functions, which is associated with worse outcome in ARDS patients [
56]. Moreover, AEC also contribute to inflammation in ALI/ARDS, as they are an important source of cytokines and chemokines [
54], and also foster intra-alveolar coagulation [
57,
58]. Human AEC may also have a critical role in protecting adjacent human PMVEC against septic injury through a paracrine mechanism, as we recently described in human PMVEC-AEC co-cultures
in vitro [
59].
Most importantly in sepsis, injury and dysfunction of the pulmonary microvasculature and specifically PMVEC are central to the pathophysiology of septic ALI. Our previous work has established in murine CLP-sepsis that pulmonary microvascular/PMVEC permeability barrier dysfunction requires the presence of PMN and alveolar macrophages, PMN-PMVEC CD18-dependent adhesion, as well as functional iNOS and NADPH oxidase [
17‐
19,
34‐
36]. Moreover, this
in vivo murine CLP-septic ALI model is characterized by pulmonary oxidant and specifically nitrosative stress, consistent with NO-superoxide reaction and peroxynitrite generation [
35]. Consistent with the
in vivo findings, cultured murine and human PMVEC under septic conditions exhibit permeability barrier dysfunction, which is similarly enhanced by the presence of PMN and macrophages in co-culture, requires PMN-PMVEC CD18-dependent adhesion and specifically the presence of functional iNOS in PMN and macrophages [
60‐
64]. Most recently, we pursued the potential critical role of putative PMVEC death in the CLP-septic murine model
in vivo [
37].
The death of cells in pathologic conditions is an increasingly important and complex research topic, as there are now as many as 10 different mechanisms, e.g. apoptosis, necrosis, necroptosis, pyroptosis, and autophagy [
65,
66]. The specific mechanisms through which cells ultimately die depend upon many factors, including the nature of the initial injurious stimulus, the specific cell type, and the time point assessed, and are best identified by specific molecular markers (e.g. caspase 3/7 activation and DNA fragmentation [TUNEL+] with apoptosis, caspase 1 activation in pyroptosis, DNA fragmentation in the absence of caspase activation in necrosis) [
65‐
67]. The death of EC can be through many of these mechanisms, although evidence for many mechanisms is largely derived from study
in vitro. For instance, bacterial infection is associated with increased autophagy, sepsis can be associated with evidence of both apoptosis and necrosis, and inflammation following blood transfusion or lung infection has been found to cause necroptosis as well as pyroptosis [
37,
68‐
71].
Although it is often stated that there is enhanced EC death in sepsis, this remains controversial, as evidence is relatively sparse, especially direct
in vivo data in clinically relevant models of sepsis (e.g. CLP). Few studies have reported quantification of apoptotic EC specifically under septic conditions [
38,
41,
43,
72], possibly due to the shedding of such apoptotic EC from the microvasculature [
39]. Moreover, various septic models have been used, e.g. intravenous or intra-peritoneal LPS administration, some of which may not capture the complex pathophysiology of bacterial sepsis and septic ALI. In addition, identification of apoptotic cells has often been with a single marker of apoptosis (e.g. TUNEL staining), and confirmation of EC-specific apoptosis using double-staining with EC-specific markers has rarely been reported. Specifically in humans, there is limited direct evidence for septic EC death/apoptosis, which is largely supported by increased levels of circulating ECs and plasma caspase 3 [
73,
74].
In the murine CLP-sepsis model, we recently reported significant septic PMVEC death (PI+ PMVEC confirmed through double-labeling with CD31, CD34, and
G. simplicifolia lectin) as observed by IVVM and confirmed by pulmonary histology [
37]. Moreover, septic PMVEC death appeared to be largely due to PMVEC apoptosis, as reflected by high degrees of overlap on double-staining with increased Annexin V binding, FLIVO (caspase activation) and TUNEL (DNA fragmentation) staining. However, the putative direct role of PMVEC death in septic ALI, specifically
in vivo pulmonary microvascular/PMVEC dysfunction, had not yet been clearly defined.
Cell death can be both beneficial (promotes recovery and prevents injury propagation) and detrimental (promotes loss of tissue function and inhibits repair), dependent again on the nature of the injury and the cell type. Death of EC, especially MVEC, has been proposed as an important mechanism of septic microvascular/MVEC dysfunction. In the current study, three lines of evidence solidly support an essential role of PMVEC death/apoptosis in murine septic pulmonary microvascular/PMVEC dysfunction in vivo: (a) following CLP-sepsis, pulmonary microvascular/PMVEC barrier dysfunction and PMVEC death/apoptosis followed parallel time courses; (b) there were strong correlations between in vivo septic pulmonary microvascular/PMVEC barrier dysfunction and both PMVEC death (PI+) and markers of apoptosis (e.g. FLIVO+); and (c) irreversible pan-caspase pharmacologic inhibition with Q-VD markedly blunted septic PMVEC death/apoptosis and attenuated in vivo septic pulmonary microvascular/PMVEC dysfunction, including both albumin-permeability barrier dysfunction and pulmonary microvascular PMN sequestration.
Our previous work and the present study suggest caspase-dependent apoptosis as the most likely mechanism of PMVEC death in murine sepsis. The specific and rigorous characterization of apoptosis is important, as individual molecular markers commonly used to define apoptosis can be present during non-apoptotic cell death and even under inflammatory non-death conditions. In fact, it is widely accepted that apoptosis should be defined using multiple molecular markers [
65,
66], and thus, in the current study, three distinct markers were used to define apoptosis: Annexin V (surface phosphatidylserine localization), FLIVO (caspase activation), and TUNEL labeling (DNA fragmentation). Furthermore, the effects of the potent caspase pharmacologic inhibitor Q-VD in markedly attenuating PMVEC apoptosis (by all three markers) confirms our putative role for PMVEC caspase-dependent apoptosis in PMVEC death and septic pulmonary microvascular/PMVEC dysfunction. Our findings are consistent with previous work on the role of caspase activation in EC apoptosis
in vivo [
42,
43,
75].
The present study also demonstrated that both iNOS and NADPH oxidase expression/activity were essential to murine septic PMVEC death/apoptosis, and the resulting
in vivo pulmonary microvascular/PMVEC dysfunction. Septic upregulation of iNOS expression and activity have been recognized to contribute to pulmonary cell apoptosis [
76]. For example, both
Nos2
−/−
mice and iNOS-inhibited wild-type mice had decreased numbers of apoptotic alveolar and bronchiolar epithelial cells following sepsis [
76]. The vital chemical interplay between these 2 key oxidant mediator systems, iNOS-NO and NADPH oxidase-superoxide, has long been recognized in sepsis and ALI pathophysiology. NO and superoxide react extremely quickly to generate the highly potent oxidant peroxynitrite, which serves many physiologic and pathophysiologic functions [
77,
78]. The central role of iNOS- and NADPH oxidase-signaling in septic PMVEC death/dysfunction in the present study is consistent with our previous data on septic murine pulmonary microvascular/PMVEC barrier dysfunction
in vivo being associated with pulmonary oxidant and nitrosative stress [
35]. Likewise,
in vitro human PMVEC barrier dysfunction and PMN-PMVEC adhesion under septic conditions were also dependent on NO, superoxide, and peroxynitrite [
63]. It is also clear that NADPH oxidase could contribute to apoptosis through multiple potential mechanisms, including through superoxide-dependent modulation of the expression of the cell cycle regulators, p21
cip1 and p53, as well as through increased expression of CCAAT-enhancer-binding protein homologous protein (CHOP) [
79,
80]. Additionally, we have shown that NADPH oxidase-dependent oxidant stress can mediate calpain-dependent caspase activity and apoptosis in septic murine PMVEC
in vitro [
40].
A central role for PMN in many of the pathophysiologic features of septic ALI in animal models, as well as in human ARDS has been clearly established [
21,
81‐
83]. We and others have reported that septic pulmonary microvascular/PMVEC barrier dysfunction in mice
in vivo and in murine and human PMVEC
in vitro is clearly PMN-dependent [
17,
18,
21,
35,
37,
60,
61,
84], and our recent study also strongly implicated pulmonary microvascular PMN leukocyte sequestration and CD18-dependent PMN-PMVEC adhesion in septic PMVEC death [
37]. There are several contributory mechanisms of PMN-dependent pulmonary microvascular/PMVEC dysfunction including PMN-PMVEC physical interaction, release of PMN mediators (i.e. oxidants and proteases), and release of PMN extracellular traps [
21,
85]. One of the key requirements for PMN-dependent pulmonary microvascular/PMVEC dysfunction appears to be the pulmonary microvascular PMN sequestration and PMN-PMVEC adhesion [
17,
63,
84]. Intriguingly, in the present study, reduced PMVEC death/apoptosis in response to various interventions (e.g. Q-VD caspase inhibition, NADPH oxidase deficiency/inhibition) was associated with reduced pulmonary microvascular PMN sequestration, similar to our previous findings in septic
Nos2
−/−
mice [
17,
37]. This suggests that septic PMVEC undergoing death/apoptosis may not be as immunologically “silent” or tolerogenic as initially thought [
86], but may signal both adaptive immunity as well as paracrine inflammatory and pro-coagulant responses [
87,
88].
We recognize the limitations of our study. Pulmonary microvascular barrier function is a complex system, and multiple mechanisms of septic ALI and specifically EC injury and dysfunction have been described in various animal models of sepsis/ALI
in vivo and in isolated EC (including human EC models) under septic conditions
in vitro. These include roles for altered cell-cell and cell-matrix (eg. glycocalyx) interactions, endogenous inflammatory cytokines (eg. TNFα) and vaso-active mediators (eg. vascular endothelial cell growth factor [VEGF], angiopoietins, endothelin), and cytoskeletal rearrangements [
31‐
33]. We report that PMVEC death/apoptosis contribute importantly to septic pulmonary microvascular/PMVEC barrier dysfunction, but appreciate that our data do not exclude a role for other potential mechanisms that may also play a role, either directly or possibly indirectly through PMVEC death/apoptosis. For example, microvessel burn-induced decreased surface VE-cadherin localization and disrupted VEGF-signaling contribute to EC apoptosis, highlighting the potential cross-talk between various mechanisms of EC barrier dysfunction [
89]. Similarly, it is clear that iNOS-dependent responses are complex and could contribute to pulmonary microvascular/PMVEC barrier dysfunction through other potential mechanisms, including modulation of inflammatory signaling (eg. NF-kB) and nitrosylation and/or nitration of multiple cell-surface and intracellular protein targets [
34,
90,
91]. Finally, it should be noted that apocynin, a pharmacologic antagonist of NADPH oxidase, was used to confirm reduced PMVEC death/dysfunction in NADPH oxidase deficient mice, but apocynin also has anti-oxidant properties that may have contributed to its observed effects [
92].
Authors’ contributions
SEG made substantial contributions to analysis and interpretation of data, drafting the article and revising it critically for important intellectual content, and final approval of the version to be published; MR made substantial contributions to conception and design of the work and acquisition of data, editing the manuscript, and final approval of the version to be published; SM made substantial contributions to conception and design of the work and analysis and interpretation of data, drafting and revising the manuscript critically for important intellectual content, and final approval of the version to be published.