Death of the Endothelium in Sepsis: Understanding the Crime Scene

Sepsis is a critical illness that occurs during severe, uncontrolled infections, often leading to multiorgan dysfunction/failure and death (1). Among the many pathophysiologic derangements that are known to occur during this condition, vascular leak contributes significantly to morbidity and mortality, particularly when it occurs in the lung, causing the frequently associated condition of acute respiratory distress syndrome (ARDS) (2). Microvascular endothelial cell (EC) barrier dysfunction is thought to underlie the mechanism of sepsis-induced vascular leak (3).

Over the past two decades, death of ECs has been shown to occur during various in vitro models of sepsis/ARDS (4–6). Historically, the significance of these findings with regard to human disease has largely been questioned on the basis of the details of the various models. For example, many early studies used bovine or human umbilical vein cells, which appear to be more susceptible to death than human microvascular ECs. In other cases, the concentration/potency of the LPS used far exceeded that which has ever been detected during human sepsis. Adding further complexity to the debate is the fact that, depending on the specific pathogen in question, EC apoptosis has been hypothesized to be either detrimental by contributing to vascular leak or beneficial by removing ECs infected with intracellular pathogens to be replaced by healthy cells, or both, raising doubts as to whether mediators of EC death could ever be viable targets for a proposed therapy in human sepsis (4).

Answers to these questions were previously limited by a paucity of in vivo methods to detect EC death in real time, and to separate proposed interventions targeting cell death pathways in ECs versus other cell types, such as lymphocytes, which are more definitively known to undergo apoptosis during sepsis. This has begun to change in recent years with the advent of advanced imaging techniques such as intravital videomicroscopy (7), as well as methods that allow specific targeting of ECs for manipulation of cell death pathways (8).

For decades, apoptosis was believed to represent the only available set of pathways by which cells undergo programmed death in a highly ordered, sophisticated manner so as not to induce or contribute to any surrounding inflammation. Necrosis was the alternative, accidental form of cell death resulting from environmental, often inflammatory, injury. Over the past decade or so, several novel pathways have emerged onto the scene. Death programs such as necroptosis, ferroptosis, and pyroptosis have various features that overlap between apoptosis and necrosis (9).

Among these, pyroptosis is of particular interest in sepsis, as knockout (KO) of one of the key pyroptotic mediators, caspase-1, has long been known to induce resistance to endotoxic shock in mice (10). The focus of the work by Mitra and colleagues (pp. 56–64) in this issue of the Journal involves a possible monocyte-driven, pyroptotic mechanism by which endotoxin resistance in Casp1^−/− mice occurs (11). Pyroptosis is a programmed form of necrosis that is activated by intracellular sensors of microbial products, and thus is recognized as an effector of innate immunity (12). The microbial sensors make up an intracellular complex known as the inflammasome, and trigger activation of caspase-1 and/or caspase-4/-5/-11. Distinct from the caspases involved in apoptosis, caspase-1 (formerly known as IL-converting enzyme) and caspase-4/-5/-11 activate gasdermin-D (GSDM-D), the final effector protein of the pathway that forms pores in the cell membrane, causing cells to swell and lyse. Although canonically they were believed to occur only in monocytes, mediators of pyroptosis have recently been found in other cell types, including ECs (8). In addition, caspase-1 activates IL-1β by cleavage of the pro-form of the protein, thereby promoting an additional proinflammatory component to pyroptosis versus the other possible cell death pathways. However, Sarkar and colleagues have previously shown that the endotoxin resistance in Casp1^−/− mice is independent of IL-1β, and more likely due to effects on cell death (13).

Sarkar and colleagues and others have also previously shown via in vitro studies that caspase-1 may be released in microparticles (MPs) from cultured mononuclear phagocytes (THP-1) treated with endotoxin (LPS) (14, 15), and that these caspase-1–containing MPs have the capacity to cause pulmonary vascular EC death (16, 17). In this issue, Mitra and colleagues describe evidence for coencapsulation of GSDM-D along with caspase-1 in those same MPs shed from endotoxin-stimulated THP-1 monocytic cells in vitro. Using a complement of differential centrifugation and Optiprep density gradient–based methods, they isolated MPs from LPS-stimulated THP-1 cells, and subjected them to immunoblot and confocal microscopic analysis to detect the presence of GSDM-D. GSDM-D was present in nonstimulated THP-1 cells at baseline, but after stimulation with LPS, the cleaved, active form of the protein colocalized with activated caspase-1 in MPs shed from these cells. This phenomenon could be blocked by pretreatment with either a caspase-1–specific or pan-caspase inhibitor. In a comparison of control MPs and LPS non-MP fractions, only the LPS MP fractions were able to induce EC death in culture. This was believed to be because these were the only fractions that contained activated forms of both caspase-1 and GSDM-D, whereas the other fractions could only harbor GSDM-D alone. To test this theory, the authors then generated Cas9/Casp1 KO and Cas9/GSDM-D KO THP-1 cells. The Casp1 KO cells contained nonactivated GSDM-D, but none of it was shed into MPs after LPS stimulation. GSDM-D KO cells, on the other hand, contained activated caspase-1 that also did not make it into MPs after LPS treatment. MPs from either of these cells were not able to induce EC death in vitro. Altogether, these results suggest that GSDM-D activated by active caspase-1 during LPS stimulation is required for the release of EC death–inducing MPs that contain active forms of both proteins. The alternative GSDM-D–activating caspase-4 and -5 were not detected in MPs from either control or LPS-treated THP-1 cells. Finally, to determine whether these observations could have relevance for human sepsis, the authors measured levels of activated GSDM-D and caspase-1 activity in MPs isolated from the plasma of patients with septic ARDS and healthy donors. Activated GSDM-D and caspase-1 activity were detectable at appreciable levels in MPs from patients with septic ARDS, but were largely absent in those from healthy donors.

These observations constitute an interesting scenario in which EC death and consequent vascular leak are mediated by a complex interaction between monocytes and ECs via microparticulate transport of key pyroptotic mediators. Although the data regarding MPs in plasma from patients with sepsis are somewhat supportive of the overall hypothesis, these observations require further investigation and validation in integrated in vivo models of sepsis. Currently, the only other in vivo links to this work are the aforementioned observations in Casp1^−/− mice. A subsequent study, however, found that Casp1^−/− mice also happen to be deficient in caspase-11 (18), and endothelial caspase-11 was independently implicated in EC pyroptosis during murine endotoxemia (8). In Mitra and coworkers’ current paper, the ECs exposed to THP-1 cell MPs were not stimulated with LPS, whereas in an in vivo setting, both cell types in this dynamic presumably would be exposed. The relative effects of endogenous caspase-11–mediated EC pyroptosis versus those of exogenous MP-induced EC death need to be further defined.

This challenge notwithstanding, several additional key questions remain to be answered in the context of Mitra and colleagues’s proposed paradigm. How is the monocyte–endothelial interaction believed to be orchestrated in human disease? Is this an early or late phenomenon in sepsis? Are circulating monocytes stimulated only during profound endotoxemia, thereby shedding MPs throughout the circulation that randomly target ECs for death? If so, what is the EC pyroptotic potential of the MPs that were isolated from the patients in this study? Or is this believed to be more of a local phenomenon, with monocytes being called to particular locations near the source of an infection, killing ECs that are blocking the path between the microcirculation and the source to allow phagocytes to rapidly clear invading microorganisms?

The crime scene of endothelial death during sepsis is being revisited by investigators who are now armed with greater knowledge of the possible culprits, as well as advancements in methods to document what is happening in real time. The study by Mitra and colleagues adds an intriguing theory to this effort. We eagerly await the answers to the important questions raised by their work and more, as we continue to hope for new therapeutic strategies to target this deadly critical illness.