The Importance of Pore-Forming Toxins in Multiple Organ Injury and Dysfunction

Background: Multiple organ injury and dysfunction often occurs in acute critical illness and adversely affects survival. However, in patients who survive, organ function usually recovers without permanent damage. It is, therefore, likely that there are reversible mechanisms, but this is poorly understood in the pathogenesis of multiple organ dysfunction syndrome (MODS). Aims: Based on our knowledge of extracellular histones and pneumolysin, as endogenous and exogenous pore-forming toxins, respectively, here we clarify if the extent of cell membrane disruption and recovery is important in MODS. Methods: This is a combination of retrospective clinical studies of a cohort of 98 patients from an intensive care unit (ICU) in a tertiary hospital, with interventional animal models and laboratory investigation. Results: In patients without septic shock and/or disseminate intravascular coagulation (DIC), circulating histones also strongly correlated with sequential organ failure assessment (SOFA) scores, suggesting their pore-forming property might play an important role. In vivo, histones or pneumolysin infusion similarly caused significant elevation of cell damage markers and multiple organ injury. In trauma and sepsis models, circulating histones strongly correlated with these markers, and anti-histone reagents significantly reduced their release. Comparison of pneumolysin deletion and its parental strain-induced sepsis mouse model showed that pneumolysin was not essential for sepsis development, but enhanced multiple organ damage and reduced survival time. In vitro, histones and pneumolysin treatment disrupt cell membrane integrity, resulting in changes in whole-cell currents and elevated intracellular Ca2+ to lead to Ca2+ overload. Cell-specific damage markers, lactate dehydrogenase (LDH), alanine aminotransferase (ALT), and cardiac troponin I (cTnI), were released from damaged cells. Once toxins were removed, cell membrane damage could be rapidly repaired and cellular function recovered. Conclusion: This work has confirmed the importance of pore-forming toxins in the development of MODS and proposed a potential mechanism to explain the reversibility of MODS. This may form the foundation for the development of effective therapies.


Introduction
Multiple organ injury and multiple organ dysfunction syndrome (MODS) often occurs in critical illness, which includes sepsis, severe trauma, and severe pancreatitis [1][2][3]. Its A retrospective case-control study was performed on 98 patients consecutively admitted to the intensive care unit (ICU) at the Royal Liverpool University Hospital between June 2013 and January 2014. Data were obtained following written consent and ethical approval approved by the National Research Ethics Service Committee Northwest-Greater Manchester West and Liverpool Central (Ref: 13/NW/0089). Patients' characteristics are presented in Table 1, including acute physiology and chronic health evaluation II (APACHE II) score [31] within 24 h and sequential organ failure assessment (SOFA) scores [32] from Day 1 to Day 4. The APACHE II score is an alternative validated tool for predicting patient mortality, whilst SOFA scores are mainly for determination of the extent of a person's organ function or rate of failure. Therefore, association between circulating histones (Days 1-2 post ICU admission) and SOFA scores (Days 1-4 post ICU admission) were performed on the total cohort. These analyses were also performed on a subset of patients with no septic shock and no DIC (n = 64), where patients with shock alone (n = 20), DIC alone (n = 5), and shock+DIC (n = 9) were removed.

Animal Models
C57/BL6 male mice from Beijing Vital River Laboratory Animal Technology were housed and used in sterile conditions at the Research Centre of Genetically Modified Mice, Southeast University, China. All procedures were performed according to state laws. C.Z.X and L.W hold the full animal licenses for using mice. All the animal models were created in our previous studies, and circulating ALT, cTnI, BUN, histones, and lung injury scores were measured, as described previously [2,13,15,20]. For survival time comparison, 8-10 mice per group were injected with D39 or PLN-A (4 × 10 7 CFU i.p.; or 2 × 10 8 CFU, i.v. lethal doses determined by pilot experiments). Dying mice were identified and euthanized by neck dislocation via close monitoring every 2-4 h after bacterial injection until all mice died. Intravenous or intraperitoneal administration was used to ensure the accuracy of bacterial doses given to mice.

Cell Culture
Human endothelial cell line (EAhy926, ATCC) and normal human liver cells (HL-7702, ATCC) were routinely cultured. Murine cardiomyocytes (HL-1, Prof WC Claycomb, Louisiana State University Medical Centre, USA) were cultured in Claycomb medium, as previously described [12]. Once fully confluent, HL-1 cells were spontaneously contracted at a rate of 5-6 Hz at 37 • C.

Measurement of Intracellular Ca 2+
Intracellular calcium concentration [Ca 2+ ]i was determined by measuring fluorescence emission at 510 nm during excitation at 340 nm and 380 nm, according to published protocols, with Fura-2AM as the fluorescent probe using a Hitachi F-7000 fluorescent spectrometer [20]. [Ca 2+ ]i was calculated using the software provided.

Electrophysiology
Whole-cell currents were recorded using the perforated patch configuration from single EAhy926 cells using an Axopatch 200B amplifier (Axon Instruments, East Hawthorn, VIC 3123 Australia), as previously described [34].

SPR Assay
Egg L-α-phosphatidylcholine (PC), egg L-α phosphatidylethanolamine (PE), and brain (porcine) L-α-phosphatidylserine (PS) in powder form and diacylglycerol (1-2-dioleoyl snglycerol, DG) in chloroform were purchased from Avanti Polar Lipids (Alabaster, AL, USA). The powders were resuspended in chloroform, then liposome solutions (PM) consisting solely of PC, PE, and PS were freshly prepared in a 40:40:20 ratio, reflecting those in the plasma membrane. The C1 sensor surface for Biacore X100 was prepared by permanently immobilising a mixture of histones H3 and H4 (New England Biolabs, Ipswich, MA, USA) to cell 2 (active cell) and S100P protein to cell 1 (control cell), as described previously [11]. Binding curves were generated using PM and DG.

Statistical Analysis
The data from in vitro experiments and animal models are presented as means ± SD. An ANOVA test was used for the comparison of more than two groups and followed by the Student-Newman-Keuls test. Human data are presented as median and interquartile ranges (1st, 3rd quartiles). Correlation between circulating histones and organ injury markers utilized Spearman's rank test. The Logrank test was used for comparison of survival rates. All analyses were performed using IBM SPSS Statistics for Windows, version 26 (IBM Corp., Armonk, NY, USA), and a p value (two-tailed) <0.05 was considered statistically significant.

Association of Circulating Histones with SOFA Scores in Critically Ill Patients without Septic Shock or Severe Coagulopathy
It is well documented that circulating histones are strong procoagulants and important mediators of MODS in sepsis. We retrospectively analyzed the association between levels of circulating histones and SOFA scores in a cohort of 98 critical ill patients (Table 1) admitted to the intensive care unit (ICU). We found strong correlations, particularly between circulating histone levels on days 1 and 2 and SOFA scores on days 3 and 4 after admission to the ICU (Table 2), suggesting a potential causal-effect relationship of circulating histones with MODS in critical illnesses. Day 1 histone levels (p = 0.002) and SOFA scores were significantly (p < 0.001) higher in patients with septic shock and/or DIC This demonstrated that high levels of circulating histones are associated with shock and coagulopathy. After removing patients with shock and/or DIC, subsequent correlation analysis still demonstrated strong correlations between circulating histone levels on day 1 and day 2 after admission, as well as SOFA scores on day 3 and day 4 ( Table 2). These data support that circulating histones can directly induce organ damage independently of inducing coagulopathy, and their pore-forming property may play an important role in these patients.

Correlation of Circulating Histones with Organ Injury Markers in Animal Models
In mouse trauma models, elevated circulating histones strongly correlated with raised circulating LDH levels (Table 3). Since circulating histones are released mainly from local tissue injury in trauma models [2], it is arguable that high circulating LDH was released by active secretion or from these damaged cells, rather than due to the pore-forming property of histones. However, the significant elevation of organ-specific markers, including cTnI and ALT, indicate the presence of cell leakage caused by pore-formation. Increased BUN and lung injury scores indicated multiple organ injury. The strong correlation of these markers to circulating histones (Table 3) suggest that elevated circulating histones are important mediators of multiple organ injury and dysfunction, and their pore-forming property played important roles. In sepsis mouse models, we also observed similar changes to the trauma models (Table 3). Both anti-histone scFv and non-anti-coagulant heparin significantly reduced lung injury, ALT and cTnI leakage, BUN, as well as LDH levels ( Figure 1A,E), which further confirmed the importance of histone-mediated cell injury.   , and LDH (E) were compared. An ANOVA test * p < 0.05 was compared to the sepsis group without anti-histone reagents. No statistical difference was found between ahscFv and heparin treatment groups.

Infusion of Histones or PLY in Mice Causes Cell Leakage and Multiple Organ Injury
Circulating histones released from host cells can be regarded as endogenous poreforming toxins, whilst the pore-forming toxins released by bacteria are considered as exogeneous. In sepsis, both endogenous and exogeneous pore-forming toxins could coexist, depending on the bacterial strains. PLY released by Streptococcus pneumoniae is common, and its toxicity to cardiomyocytes has been demonstrated in our previous study [20]. In this work, we used either histones (50 mg/kg) or PLY (400 µg/kg)-infusion mouse models and found that circulating LDH, ALT and cTnI were elevated (Figure 2A), indicating that histones and PLY caused leakage of multiple cell types. BUN and lung injury scores were also significantly increased ( Figure 2B), indicating multiple organ injury. No significant difference between histones and PLY infusion was observed. Using ECG to monitor cardiac rhythm, and cardiac arrhythmia occurred in over 50% of mice infused with either histones or PLY ( Figure 2C,D). These data strongly support that pore-formation on cell membranes is pivotal to the manifestation of MODS.
Circulating histones released from host cells can be regarded as endogenous por forming toxins, whilst the pore-forming toxins released by bacteria are considered as e ogeneous. In sepsis, both endogenous and exogeneous pore-forming toxins could coexi depending on the bacterial strains. PLY released by Streptococcus pneumoniae is commo and its toxicity to cardiomyocytes has been demonstrated in our previous study [20]. this work, we used either histones (50 mg/kg) or PLY (400 µg/kg)-infusion mouse mode and found that circulating LDH, ALT and cTnI were elevated (Figure 2A), indicating th histones and PLY caused leakage of multiple cell types. BUN and lung injury scores we also significantly increased ( Figure 2B), indicating multiple organ injury. No significa difference between histones and PLY infusion was observed. Using ECG to monitor ca diac rhythm, and cardiac arrhythmia occurred in over 50% of mice infused with eith histones or PLY ( Figure 2C,D). These data strongly support that pore-formation on c membranes is pivotal to the manifestation of MODS.

PLY-Deletion in Bacteria Does Not Inhibit Sepsis-Induction but Reduces Cell Leakage and Increase Mouse Survival
The presence of PLY was confirmed in S. pneumoniae serotype 2 strain (D39), but n in the PLY-deficient isogenic strain (PLN-A), E.Coli BL-21 or the ClearColi BL-21 (Figu 3A). LDH and ALT were significantly higher in D39-infected (PLY +ve ) mice compared PLN-A-infected (PLY −ve ) mice ( Figure 3B), indicating that PLY plays important roles cellular leakage. Circulating histone levels are high in both models ( Figure 3C). Howev no statistical significance was observed in cTnI, BUN, histone levels ( Figure 3B,C), lu injury scores ( Figure 3D), or sepsis scores ( Figure 3E) between D39-infected and PLN-

PLY-Deletion in Bacteria Does Not Inhibit Sepsis-Induction but Reduces Cell Leakage and Increase Mouse Survival
The presence of PLY was confirmed in S. pneumoniae serotype 2 strain (D39), but not in the PLY-deficient isogenic strain (PLN-A), E.Coli BL-21 or the ClearColi BL-21 ( Figure 3A). LDH and ALT were significantly higher in D39-infected (PLY +ve ) mice compared to PLN-A-infected (PLY −ve ) mice ( Figure 3B), indicating that PLY plays important roles in cellular leakage. Circulating histone levels are high in both models ( Figure 3C). However, no statistical significance was observed in cTnI, BUN, histone levels ( Figure 3B,C), lung injury scores ( Figure 3D), or sepsis scores ( Figure 3E) between D39-infected and PLN-A-infected mice. These data suggest that other toxic factors, such as circulating histones, may be more important in mediating multiple organ injury in these models. In contrast, mice infected (i.p.) with D39 died earlier than mice injected with PLN-A ( Figure 3F). A similar pattern was observed via intravenous injection (i.v.) of these bacteria ( Figure 3G).
Biomedicines 2022, 10, x FOR PEER REVIEW 9 of 17 infected mice. These data suggest that other toxic factors, such as circulating histones, may be more important in mediating multiple organ injury in these models. In contrast, mice infected (i.p.) with D39 died earlier than mice injected with PLN-A ( Figure 3F). A similar pattern was observed via intravenous injection (i.v.) of these bacteria ( Figure 3G). ). Blood and organs were collected at 24 h after injection. The means ± SD of relative increase in LDH, ALT, cTnI, and BUN are presented. An ANOVA test * p < 0.05 was compared to the control. # p < 0.05 was compared to the PLY(+) bacteria-induced mouse model. The means ± SD of circulating histones (C), lung injury scores (D), and sepsis scores (E) showed no statistical difference between PLY (+) and PLY (−) models, but * p < 0.05 was compared ). Blood and organs were collected at 24 h after injection. The means ± SD of relative increase in LDH, ALT, cTnI, and BUN are presented. An ANOVA test * p < 0.05 was compared to the control. # p < 0.05 was compared to the PLY(+) bacteria-induced mouse model. The means ± SD of circulating histones (C), lung injury scores (D), and sepsis scores (E) showed no statistical difference between PLY (+) and PLY (−) models, but * p < 0.05 was compared to control groups. (F and G) Survival curves of mice injected with D39 or PLN-A (4 × 10 7 CFU/mouse i.p. (F); or 2 × 10 8 CFU/mouse i.v. via the tail vein (G)). 8-10 mice per group, log rank test, p < 0.05.

Endogenous and Exogenous Pore-Forming Toxins cause Ca 2+ Overload to Stress Cells In Vitro
Endothelial cells (EAhy926), liver cells (HL-7702), and cardiomyocytes (HL-1 cells) were treated with endogenous (histones) and exogeneous (pneumolysin [PLY]) poreforming toxins. We found that both histones and PLY caused significant release of LDH, ALT, and cTnI in a dose-dependent manner. Using low doses of toxins, we still observed release of these markers without cell death, strongly indicating cellular leakage ( Figure 4A). Using FITC-PLY to treat cultured HL-1 cells preloaded with calcium indicator, PLY concentrated on the cell membrane within 10 min and induced increases in intracellular Ca 2+ ( Figure 4B). Similarly, we found increased intracellular Ca 2+ following treatment of endothelial cells with individual recombinant human histones ( Figure 4C). Histones H4 and H3 induced greater intracellular Ca 2+ increases compared to other histones (H1, H2A and H2B). EAhy926 cells treated with histones showed increases in whole-cell current ( Figure 4D), and means ± SD were −722.33 ± 86.63pA from three independent experiments. Following washing to remove histones, these inward cellular currents gradually returned back to baseline to indicate membrane self-repair. Functional analysis demonstrated that histone treatment disturbed the auto-rhythm of HL-1 cardiomyocytes ( Figure 4E), and the auto-rhythm recovered immediately after histone removal (data not shown).

Modelling Histone-Phospholipid Interaction, Pore Formation, and Self-Repairing Mechanisms
The three-dimensional crystal structure of PLY has previously been resolved with modeling of its pore-structure on cell membranes [35]. However, the pore-structure model of extracellular histones on the cell membrane is unknown, although histone-mediated disruption of cell membrane is well documented [15,36]. Within the cell nucleus, the histone core interacts with double-stranded DNA via two layers of phosphate groups on phosphodiester backbones [37], which is similar to the bilayers of phosphate groups on the cell membrane ( Figure 5A,B). SPR analysis demonstrates that phospholipids (PC, PE, and PS) with phosphate groups (PM) bind to histones, but glycolipids without phosphate group (DG) do not ( Figure 5C,D). This suggests that extracellular histones bind the lipid bilayer of cell membranes via phosphate groups in the same way as the histone-core interaction with double-stranded DNA in nucleosomes [37]. However, how histones form pores on cell membranes is still not clear. Figure 5E proposes a model of histone integration into the bilayers of cell membranes. In this way, the Ca 2+ influx will trigger the membrane self-repairing processes, including endosome formation to isolate the pores. Then, the endosome could be degraded inside the cell or released in the form of exosomes to the extracellular space ( Figure 5F) to facilitate the restoration of cell function. This rapid selfrepair mechanism to restore cell function compliments clinical observation of timely and complete recovery of multiple organ function after sepsis in the majority of cases.

Modelling Histone-Phospholipid Interaction, Pore Formation, and Self-Repairing Mechanisms
The three-dimensional crystal structure of PLY has previously been resolved with modeling of its pore-structure on cell membranes [35]. However, the pore-structure model of extracellular histones on the cell membrane is unknown, although histone-mediated disruption of cell membrane is well documented [15,36]. Within the cell nucleus, the histone core interacts with double-stranded DNA via two layers of phosphate groups on phosphodiester backbones [37], which is similar to the bilayers of phosphate groups on the cell membrane ( Figure 5A,B). SPR analysis demonstrates that phospholipids (PC, PE, and PS) with phosphate groups (PM) bind to histones, but glycolipids without phosphate group (DG) do not ( Figure 5C,D). This suggests that extracellular histones bind the lipid bilayer of cell membranes via phosphate groups in the same way as the histone-core interaction with double-stranded DNA in nucleosomes [37]. However, how histones form pores on cell membranes is still not clear. Figure 5E proposes a model of histone integration into the bilayers of cell membranes. In this way, the Ca 2+ influx will trigger the membrane self-repairing processes, including endosome formation to isolate the pores. Then, the endosome could be degraded inside the cell or released in the form of exosomes to the extracellular space ( Figure 5F) to facilitate the restoration of cell function. This rapid selfrepair mechanism to restore cell function compliments clinical observation of timely and complete recovery of multiple organ function after sepsis in the majority of cases.

Discussion
Membrane integrity is important to cellular function [38]. The disruption of membrane integrity by pore-forming toxins can be an important pathogenic mechanism in multiple organ injury, particularly in patients with sepsis, due to the possible coexistence of both endogenous and exogenous pore-forming toxins. In this study, we demonstrate that extracellular histones bind to phospholipids to form pores on cell membrane. In vitro, both extracellular histones and PLY bind membranes of cells derived from different organs and cause Ca 2+ influx and leakage of specific biomarkers. This was also demonstrated in vivo using histone and PLY-infusion mouse models. The association of circulating histones with MODS has been demonstrated using both severe trauma and sepsis models with anti-histone intervention, as well as in a cohort of sepsis patients. The roles of exogenous pore-forming toxin, PLY, were explored by deletion of the PLY gene in D39 bacteria and demonstrated that PLY is not essential to sepsis development in mice, but increases organ injury and enhances lethality. These findings propose an important mechanism of MODS in sepsis and other critical illnesses.
It is known that extracellular histones are procoagulant and involved in the coagulopathy of sepsis via endothelial damage, platelet activation, thrombin generation, and affecting regulatory pathways, such as protein C, thrombomodulin, and fibrinolysis [11,14,39,40]. Coagulopathy is a major pathogenic factor that significantly contributes to microcirculatory impairment and MODS in many critical illnesses [41,42]. It is difficult to rectify whether coagulopathy or cytotoxicity contribute more to MODS. In general, non-survivors often develop shock and/or severe coagulopathy. It is fully justifiable that current treatments mainly focus on the correction of shock, coagulopathy, and poor tissue perfusion [43]. Clinical observation shows that, although a small fraction of patients who survive MODS may still experience neurological problems, recurrent infection, and deterioration of underlying diseases [44], full recovery of organ function is typically observed in the majority of patients. In this study, we showed that the SOFA scores of patients without shock and/or DIC were still correlated to the levels of circulating histones, strongly indicating that the pore-forming property of histones plays important roles in MODS in the subgroup of patients.
Pore-forming toxins can non-selectively bind plasma membranes of any cell type in contact [15,22]. Once these toxins enter circulation, endothelial cells will likely be the primary targets, leading to endothelial barrier disruption [15]. This might lead to the exposure of smooth muscle cells to toxins and cause Ca 2+ overload, thereby potentially dis-rupting blood pressure and perfusion regulatory control, leading to septic shock [45]. Once parenchymal cells of different organs are exposed to pore-forming toxins, both non-specific (LDH) and cell-specific biomarkers (ALT, AST from liver cells, cTnI from cardiomyocytes) will be released into circulation. Unlike complement attack, the affected cells rarely lyse unless exposed to high concentrations of these toxins. The major pathological factor is more likely the resultant Ca 2+ overload, which stresses cells and would affect cellular function [22,38], particularly in exciting cells, such as cardiomyocytes. This is consistent with the clinical observation that patients with severe sepsis frequently showed cardiac events, which could lead to premature death [46,47]. Many other factors also contribute to MODS, such as mitochondrial dysfunction [48] and injury to specific cells, including alveolar epithelial type II cells [49] and cardiomyocytes [50]. These pathological processes may also involve the disturbance of Ca 2+ homeostasis.
Ca 2+ influx is also a signal of plasma membrane damage of cells and will initiate the self-repair process [51,52], a central biological process for maintaining cellular homeostasis [29]. Although the detailed molecular mechanism of membrane self-repair is still not fully elucidated, vesicle trafficking, exocytosis, and endocytosis to remove the damaged membrane may all be involved [29]. The self-repair of cell membranes may be an important mechanism for the reversible processes in these survived patients to gain a full recovery of cellular and organ function.
The limitation of this study is the lack of available assays to monitor circulating PLY. However, using PLN-A strain without PLY and parental strain D39 with PLY to infect mice demonstrated that bacteria with PLY caused higher levels of ALT and LDH release into circulation and significantly reduced the survival time of mice. These data indirectly support that PLY-induced pore formation plays important roles in multiple organ injury and dysfunction in vivo. The co-existence of both endogenous and exogenous toxins may synergistically deteriorate MODS and accelerate the progression of lethal diseases. In our previous publications, our focuses were mainly on histone-induced coagulopathy and subsequent organ injury, as well as PLY-induced cardiomyocyte dysfunction.
This work focused on the importance of pore-forming properties of these mediators in order to better understand the pathophysiology of MODS. In reality, over 60% of patients with MODS are not complicated by shock or DIC. Therefore, a strategic focus towards reducing the cytotoxicity of pore-forming toxins, including accelerating membrane repairing, as well as reducing Ca 2+ overload and its harmful effects, might achieve a rapid and improved recovery of organ functions in those patients. This strategy may also hold significant clinical value in patients with shock and DIC by reducing the severity of organ injury and MODS and increasing survival.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study or their relatives.