High-density lipoproteins exert pleiotropic effects including antiinflammatory, antiapoptotic, and lipopolysaccharide-neutralizing properties. The authors assessed the effects of reconstituted high-density lipoproteins (CSL-111) intravenous injection in different models of sepsis.
Ten-week-old C57BL/6 mice were subjected to sepsis by cecal ligation and puncture or intraperitoneal injection of Escherichia coli or Pseudomonas aeruginosa pneumonia. CSL-111 or saline solution was administrated 2 h after the sepsis. Primary outcome was survival. Secondary outcomes were plasma cell-free DNA and cytokine concentrations, histology, bacterial count, and biodistribution.
Compared with saline, CSL-111 improved survival in cecal ligation and puncture and intraperitoneal models (13 of 16 [81%] survival rate vs. 6 of 16 [38%] in the cecal ligation and puncture model; P = 0.011; 4 of 10 [40%] vs. 0 of 10 [0%] in the intraperitoneal model; P = 0.011). Cell-free DNA concentration was lower in CSL-111 relative to saline groups (68 [24 to 123] pg/ml vs. 351 [333 to 683] pg/ml; P < 0.001). Mice injected with CSL-111 presented a decreased bacterial count at 24 h after the cecal ligation and puncture model both in plasma (200 [28 to 2,302] vs. 2,500 [953 to 3,636] colony-forming unit/ml; P = 0.021) and in the liver (1,359 [360 to 1,648] vs. 1,808 [1,464 to 2,720] colony-forming unit/ml; P = 0.031). In the pneumonia model, fewer bacteria accumulated in liver and lung of the CSL-111 group. CSL-111–injected mice had also less lung inflammation versus saline mice (CD68+ to total cells ratio: saline, 0.24 [0.22 to 0.27]; CSL-111, 0.07 [0.01 to 0.09]; P < 0.01). In all models, no difference was found for cytokine concentration. 111Indium bacterial labeling underlined a potential hepatic bacterial clearance possibly promoted by high-density lipoprotein uptake.
CSL-111 infusion improved survival in different experimental mouse models of sepsis. It reduced inflammation in both plasma and organs and decreased bacterial count. These results emphasized the key role for high-density lipoproteins in endothelial and organ protection, but also in lipopolysaccharide/bacteria clearance. This suggests an opportunity to explore the therapeutic potential of high-density lipoproteins in septic conditions.
High-density lipoproteins have multiple positive end-organ effects, and attenuate organ injury and improve survival in mouse models of sepsis
A human injectable formulation of reconstituted high-density lipoprotein, CSL-111, composed of apolipoprotein A1 and phosphatidylcholines, was tested in three mouse models of sepsis
When administered soon after the insult causing sepsis, CSL-111 improved survival and reduced lung injury, apparent neutrophil activation, and plasma markers of inflammation, but not cytokine concentrations
Sepsis remains a major cause of mortality in intensive care units, despite better understanding of the pathophysiology and the development of improved supportive treatments, antibiotic therapies, and surgical techniques.1–4
High-density lipoproteins are characterized by the association of apolipoproteins such as apolipoprotein A1, phospholipids, and cholesterol in its free or esterified form. Their most described function is to mediate the reverse transport of cholesterol, from peripheral tissues back to the liver.5 Many studies have described an inverse association between high-density lipoprotein concentration and cardiovascular events.6–8 In addition to reverse transport cholesterol, high-density lipoproteins display pleiotropic properties such as antiinflammatory, antioxidant, antiapoptotic, and antithrombotic effects.9–14 Furthermore, high-density lipoproteins are reported to protect the endothelium via limiting the expression of adhesion molecules under inflammatory conditions15 and are able to neutralize lipopolysaccharides.16,17
Previous studies have shown that high-density lipoprotein infusion attenuates organ injury in mouse models of sepsis, potentially via lipopolysaccharide neutralization.18–20 Recently, Zhang et al. have reported that the use of apolipoprotein A1 mimetic peptide improved the survival of septic rats.20
Several clinical studies have been conducted to assess apolipoprotein A1 concentration in septic conditions.21–25 Van Leeuwen et al. underlined that in septic patients, high-density lipoprotein concentration rapidly fall and can be reduced to 50%.21 In a previous study, we demonstrated that compared to trauma patients, high-density lipoprotein levels were lower in case of sepsis.23 Chien et al. have shown that low high-density lipoprotein levels at day 1 of severe sepsis were significantly associated with an increased mortality.24 Barlage et al. have also found that in intensive care unit, nonsurvivor septic patients had a statistically significant decreased apolipoprotein A1 concentration.25 In a recent work comparing septic versus nonseptic patients, we reported a marked decrease in high-density lipoprotein concentration and a shift toward large nonfunctional high-density lipoprotein particles.26
Based on high-density lipoprotein pleiotropic effects previously demonstrated, we hypothesized that supplementation with functional high-density lipoproteins in a murine model of sepsis could improve the survival. Several studies have tested apolipoprotein A1 mimetic peptides or reconstituted high-density lipoproteins administered before lipopolysaccharide injection or the onset of sepsis. Here we have used reconstituted high-density lipoproteins made of apolipoprotein A1 and phosphatidylcholines (CSL-111; Behring, Switzerland), tested previously in a randomized clinical trial involving post–myocardial infarction patients.27 The goal of the current study was then to test the potential therapeutic effects of high-density lipoproteins in three different models of sepsis, using reconstituted high-density lipoproteins potentially injectable in clinical settings.
Materials and Methods
Reconstituted HDLs (CSL-111) were provided by CSL Behring AG (Bern, Switzerland). They consist of apolipoprotein AI purified from human plasma combined with soybean phosphatidylcholines to form discoidal particles similar to native reconstituted high-density lipoproteins.28
Ten-week-old C57BL/6 female and male mice weighing 20 to 25g were used in our study (sex ratio, 1:1). All experimentations were performed in accordance with the legislation on the protection of animals and were approved by the ethical committees for animal experimentation, Bichat Hospital, INSERM 1148, Paris (authorization 2012-15/698-097) and CYROI n°114 (authorization 14827-2018012212274175 V3 and 20938-2019031114427274 V7). Experiments were carried out in animal facility in microsurgery room.
Models of Sepsis
Animals were anesthetized by 2% isoflurane gas anesthesia and 0.05 mg/kg buprenorphine was injected as analgesia. Cecal ligation and puncture was performed as previously described by Rittirsch et al.29,30 Two hours after the cecal ligation and puncture, each animal was blindly administered either CSL-111 (40 mg/kg) or saline (Supplemental Digital Content, http://links.lww.com/ALN/C220). Computer-based randomization was used to allocate drug regimens to each group, performed by a blinded observer from our laboratory.
Intraperitoneal Injection of Bacteria (Intraperitoneal Model)
In the second model of sepsis, 4 × 107 colony-forming unit/ml of Escherichia coli (IAI76 strain) were injected into the intraperitoneal cavity in a volume of 400 μl of saline (Supplemental Digital Content, http://links.lww.com/ALN/C220). This model induced a peritonitis characterized by a rapid intravascular transfer of bacteria.31 Two hours after the intraperitoneal bacterial injection, each animal was intravenously administered either CSL-111 (40 mg apolipoprotein A1/kg) or saline.
Pseudomonas Aeruginosa Pneumonia Model (Pneumonia Model)
For this third model, a concentration of 4 × 108 colony-forming unit/ml Pseudomonas aeruginosa ATCC 27853 strain were instilled in the mouse trachea (Supplemental Digital Content, http://links.lww.com/ALN/C220).32 Two hours after the bacterial tracheal instillation, each animal was administered intravenously either CSL-111 (40 mg apolipoprotein A1/kg) or saline.
After the surgery, postoperative care consisted in mouse hydration by saline subcutaneous infusion (1.2 ml per 100 g body weight), and analgesia by subcutaneous buprenorphine (5 ng per 100 g body weight). Mice were placed in the facedown position in a bedding-free, prewarmed fresh cage placed over a heating pad. All mice had free access to water and food.
In each group in the two sepsis procedures (cecal ligation and puncture and intraperitoneal), mice were observed every 2 h for 150 h.
Plasma and Histological Analyses
Twenty-four hours after surgical procedure, the animals were sacrificed, blood, bronchoalveolar lavage fluid and tissues were collected for analysis (Supplemental Digital Content, http://links.lww.com/ALN/C220). An observer blinded to the experimental conditions analyzed the anonymized histological sections.
Bacteria 111Indium Labeling
111Indium was incubated with bacteria for 45 min. Bacteria were then washed with saline by centrifugation in order to discard free 111Indium. Labeled bacteria were injected to the mouse via the intraperitoneal route. The same protocol was performed in CSL-111 and saline groups. The distribution of radioactive bacteria was achieved by scintigraphy at different times.
The primary outcome of our study was mortality rate and we defined our secondary outcomes as all plasma and histological analyses as well as bacteria 111Indium labeling. Survival curves according to treatment groups were estimated using the Kaplan–Meier method and compared using a log rank test. No statistical power calculation was conducted before the study. The sample size for survival study was based on previous studies in the field.20,33,34 Quantitative variables are expressed as median [interquartile range]. Because of a non-Gaussian distribution, univariate comparisons were made using Mann–Whitney U test in case of bivariate comparisons. In case of comparisons between three groups (saline, CSL-111 and sham groups), we used a Kruskal–Wallis test followed by a Dunn multiple comparison test. All analyses were performed at the two-tailed threshold of P < 0.05 and data were analyzed using Graph Pad Prism software.
Effect of Intravenous CSL-111 Injections on Survival in the Cecal Ligation and Puncture Model
In saline-injected mice, cecal ligation and puncture resulted in an overall survival of 38% at 36 h versus 81% survival rate for CSL-111-treated mice (fig. 1A; P = 0.011; n = 16 mice per group). At 110 h, all saline-injected mice were dead whereas CSL-111 treatment significantly improved survival to 31% (fig. 1B; P = 0.008). Sham-operated mice (n = 16 mice) all survived (not shown). There were no censured data during the observation period.
Effect of Intravenous CSL-111 Supplementation on Survival in the Intraperitoneal Bacterial Injection Model
In order to validate the protective effects of HDLs, we tested CSL-111 in another model of sepsis induced by intraperitoneal injection of E. coli. All saline-injected mice were dead at 40h whereas CSL-111 treatment significantly improved survival to 40% at this time point (fig. 1C; P = 0.011; n = 10 mice per group). Sham-operated mice (n = 10 mice) all survived (not shown). There were no censured data during the observational period.
Plasma Cell-free DNA
We measured cell-free DNA as a general marker of apoptosis and necrosis, potentially reflecting neutrophil activation via production of neutrophil extracellular traps, at baseline and after 24h. This variation of cell-free DNA concentration (delta DNA, pg/ml) was statistically significantly lower in CSL-111 groups relative to saline-injected mice (fig. 2; [68 (24 to 123) pg/ml vs. 351 (333 to 683) pg/ml; P < 0.001] CSL-111 group, n = 8 mice; saline group, n = 10 mice; sham group, n = 9 mice).
Bronchoalveolar Lavage and Plasma Cytokine Levels
Cecal ligation and puncture markedly induced the production of plasma interleukin-1ß, interleukin-10, and tumor necrosis factor α relative to sham production. There was no statistically significant difference between CSL-111 and saline groups for interleukin-1ß, interleukin-10, and tumor necrosis factor α at 24 h after cecal ligation and puncture (fig. 3; interleukin-1ß, 0 [0 to 133] pg/ml vs. 90 [6 to 825] pg/ml; P = 0.271; interleukin-10, 3,784 [1,078 to 51,292] ng/ml vs. 27,388 [12,817 to 52,383] ng/ml; P > 0.999; tumor necrosis factor α, 85 [44 to 266] pg/ml vs. 171 [102 to 346] pg/ml; P > 0.999; CSL-111 group, n = 8 mice; saline group, n = 10 mice; sham group, n = 8 mice).
In the pneumonia model, there was no statistically significant difference between CSL-111 and saline groups for interleukin-6 (plasma: CSL-111, 108 [19 to 205] pg/mb; saline, 119 [61 to 5,807] pg/ml; P = 0.431; bronchoalveolar lavage: CSL-111, 31 [20 to 58] pg/mb; saline, 37 [24 to 240] pg/ml; P = 0.506; CSL-111 group, n = 6 mice; saline group, n = 7 mice).
Quantification of Plasma Markers of Inflammation and Endothelial Activation
All plasma markers tested, except vascular cell adhesion molecule-1, were increased under septic conditions relative to sham-operated mice. There was no statistically significant difference between CSL-111– and saline-injected groups for intercellular adhesion molecule-1, vascular cell adhesion molecule-1, E-selectin, matrix metallopeptidase-9, and plasminogen activator inhibitor-1 at 24 h after CLP ([fig. S1, Supplemental Digital Content, http://links.lww.com/ALN/C221] intercellular adhesion molecule-1: 23[14 to 37] ng/ml vs. 28 [18 to 43] ng/ml, P > 0.999; vascular cell adhesion molecule-1: 2,275 [1,390 to 2,823] pg/mL vs. 2,203 [1,890 to 3,170] pg/ml, P > 0.999; E-selectin: 130 [56 to 252] pg/ml vs. 212 [134 to 262] pg/ml, P = 0.503; matrix metallopeptidase-9: 28 [19 to 54] pg/ml vs. 99 [80 to 177] pg/ml, P = 0.453; plasminogen activator inhibitor-1: 149 [92 to 212] ng/ml vs. 132 [76 to 232] ng/ml, P > 0.999); CSL-111 group, n = 8 mice, saline group, n = 10 mice, sham group, n = 8 mice.
Lung mRNA Expression of Interleukin-6, Tumor Necrosis Factor α, Intercellular Adhesion Molecule-1, Vascular Cell Adhesion Molecule-1 and E-selectin
Cecal ligation and puncture induced lung expression of interleukin-6, tumor necrosis factor α, intercellular adhesion molecule-1, vascular cell adhesion molecule-1, and E-selectin, as assessed by reverse transcription polymerase chain reaction. Results are expressed in ratio of messenger RNA (mRNA) to glyceraldehyde 3-phosphate dehydrogenase. There was no difference between CSL-111 group and saline group in lung RNA expression of interleukin-6, tumor necrosis factor α, intercellular adhesion molecule-1, vascular cell adhesion molecule-1, and E-selectin ([fig. S2, Supplemental Digital Content, http://links.lww.com/ALN/C222] interleukin-6, 2.6 [0.7 to 2.8] vs. 1.0 [0.8 to 5.1]; P > 0.999; tumor necrosis factor α: 1.0 [0.9 to 1.0] vs. 1.1 [1.0 to 1.1]; P > 0.999; intercellular adhesion molecule-1, 3.0 [2.4 to 3.8] vs. 6.3 [4.4 to 8.2]; P = 0.957; vascular cell adhesion molecule-1, 1.0 [0.9 to 1.0] vs. 1.0 [0.9 to 1.0]; P > 0.999; E-selectin, 0.6 [0.0 to 1.0] vs. 1.1 [1.0 to 2.6]; P = 0.151). Results are expressed in ratio of mRNA to glyceraldehyde 3-phosphate dehydrogenase; CSL-111 group, n = 5 mice; saline group, n = 5 mice; sham group, n = 5 mice).
Mice treated with CSL-111 have a statistically significant decrease of bacteremia at 24 h from the cecal ligation and puncture compared to saline-injected mice (200 [28 to 2,302] colony-forming unit/ml of blood in CSL-111 group; 2,500 [953 to 3,636] colony-forming unit/ml of blood in saline group; P = 0.021; n = 18 mice/group). Sham-operated mice were exempt of bloodborne bacteria.
Bacterial Count in Liver, Lung, Spleen, and Kidney
CSL-111 injection led a statistically significant decrease in bacterial rate in the liver at 24 h from the cecal ligation and puncture relative to the saline group (1,359 [360 to 1,648] colony-forming unit/ml in CSL-111 group; 1,808 [1,464 to 2,720] colony-forming unit/ml in saline group; P = 0.031). No difference between the two groups was observed in the lung, spleen, and/or kidney (lung, 844 [104 to 1,489] colony-forming unit/ml vs. 807 [492 to 3,762] colony-forming unit/ml; P = 0.489; spleen, 320 [114 to 568] colony-forming unit/ml vs. 247 [53 to 503] colony-forming unit/ml; P = 0.757; kidney, 308 [137 to 336] colony-forming unit/ml vs. 122 [52 to 904] colony-forming unit/ml; P = 0.436; n = 9 mice/group) (fig. 4).
In the pneumonia model, CSL-111 injection led to a decreased bacterial concentration in the liver at 24 h comparing sepsis to saline group, 8.0 × 107 (6.9 × 107 to 8.4 × 107) in CSL-111 group; 1.1 × 108 (9.9 × 107 to 1.3 × 108) in saline group; P < 0.001. In the pneumonia model, CSL-111 injection also led a decreased bacterial concentration in the lung at 24 h comparing sepsis to saline group, 4.9 × 107 (4.6 × 107 to 5.6 × 107) in CSL-111 group; 3.2 × 108 (5.8 × 107 to 4.2 × 108) in saline group; P = 0.004; n = 7 mice in saline group, n = 5 mice in CSL-111 group (fig. 5).
Lung Histological Evaluation and Immunodetection of Apolipoprotein A1 and CD68
Hematoxylin and eosin staining showed an increase of alveolar destruction and infiltration by inflammatory cells in the saline mice group compared to mice treated with CSL-111 (cells to total area ratio, 0.19 [0.13 to 0.20] in the saline group; 0.03 [0.03 to 0.04] in the CSL-111 group; 0.05 [0.04 to 0.06] in the sham group; P = 0.01; n = 3 mice/group; fig. 6). In the cecal ligation and puncture model, immunohistological analysis of septic lungs 24 h after intravenous injection of CSL-111 or saline demonstrates that apolipoprotein A1 reached and accumulated in pulmonary tissue in high-density lipoprotein-injected mice, suggesting that high-density lipoprotein particles can locally exert their protective effects (apolipoprotein A1 immunopositive area to cells ratio, 0.5 [0.4 to 1.1] in high-density lipoprotein-injected mice vs. 0.02 [0.01 to 0.03] in the saline group mice; P = 0.016; fig. 7 and fig. S3, Supplemental Digital Content, http://links.lww.com/ALN/C223). Immunodetection of CD68+ macrophage shows a statistically significant increased accumulation of these inflammatory cells in septic conditions relative to controls (sham-operated mice) (CD68-immunopositive area to cells ratio: saline, 0.24 [0.22 to 0.27]; CSL-111 group, 0.07 [0.01 to 0.09]; sham group, 0.02 [0.0 to 0.04]; P < 0.0001). High-density lipoprotein injection markedly limited the recruitment of macrophages in septic conditions (fig. 8 and fig. S4, Supplemental Digital Content, http://links.lww.com/ALN/C224).
In the pneumonia model, hematoxylin and eosin staining showed an increase of alveolar destruction and infiltration by inflammatory cells in the saline mice group compared to mice treated with CSL-111 and the sham group (cells to total area ratio: 0.04 [0.03 to 0.05] in the saline group; 0.01 [0.01 to 0.02] in the CSL-111 group; 0.01 [0.01 to 0.02] in the sham group; P < 0.001; n = 3 mice/group).
In the pneumonia model, CSL-111–injected mice presented less CD68+ cells than in saline group mice. There was no difference between CSL-111 mice and sham (CD68-immunopositive area to cells ratio: saline, 0.07 [0.05 to 0.13]; CSL-111 group, 0.03 [0.02 to 0.04]; sham group, 0.0 [0.0 to 0.0]; P < 0.0001; n = 3 mice/group; fig. 9).
111Indium Bacterial Labeling
Scintigraphy has shown that bacteria have mainly an intraperitoneal localization at 1h after intraperitoneal injection. At 2.5 h, in the group of mice treated with CSL-111, bacteria have an epigastria localization that could be gall bladder. In the saline-injected group, labeled bacteria were essentially localized in intraperitoneal position and in testis. There was no epigastria accumulation in this group of mice (n = 2 mice/group) (fig. S5, Supplemental Digital Content, http://links.lww.com/ALN/C225).
Our main finding is that injection of reconstituted high-density lipoproteins (CSL-111) markedly improved survival in three different mouse models of sepsis. At different experimental time points after cecal ligation and puncture, (36, 40, or 110 h), mortality was reduced by more than 30% in the CSL-111–injected group versus the saline-injected group. These results are in the line with several studies showing a reduced mortality after injection of either reconstituted high-density lipoproteins or apolipoprotein A1 mimetic peptides in endotoxemic models.15,19,35 It was shown that apolipoprotein A1 knockout mice are more susceptible to cecal ligation and puncture-induced death.34 Apolipoprotein A1 knockout mice exhibited a decreased plasma lipopolysaccharide neutralization capacity relative to control mice. In this context, our results and the conclusions of these studies emphasized a major protective role of high-density lipoproteins during experimental sepsis. We performed these two models of sepsis (cecal ligation and puncture/intraperitoneal bacterial injection models) because they are more similar to human sepsis as compared to lipopolysaccharide infusion, which appears to be simple and reproducible, but probably does not reflect the complex physiologic human response to the bacterial insult. In the intraperitoneal bacterial injection model, the presence of bacteria allows insights into mechanisms of host response to pathogens. Moreover, cecal ligation and puncture model represents a polymicrobial sepsis model and may be similar to human sepsis progression with similar hemodynamic and metabolic phases.36 We also performed a sepsis induced by an intratracheal injection of Pseudomonas aeruginosa, in order to test the potential of high-density lipoproteins therapy in a pneumonia, nonintraabdominal sepsis model.
In both peritonitis and pneumonia models, histological analysis showed that mice treated with CSL-111 had less lung inflammatory cell infiltration and less alveolar septal destruction. Our results are in line with McDonald et al. study who found that pretreatment of lipopolysaccharide-rats with reconstituted high-density lipoproteins attenuated intestinal injury by reducing edema, cell infiltration or destruction of the normal architecture.15 We also underline an interesting anti-inflammatory property of high-density lipoproteins particles. Macrophage recruitment is associated with inflammation in acute lung injury and acute respiratory distress syndrome.37 In our study, in both peritonitis and pneumonia models, mice treated with CSL-111 had less macrophage infiltration versus saline-injected mice conferring to CSL-111 a protective antiinflammatory effect. For histological analysis, only three mice per group could be included due to technical problems.
Inflammatory states are also characterized by increased levels of plasma markers, such as cell-free DNA. Neutrophil activation by pathogens leads to the liberation of DNA associated with antimicrobial proteins contained in granules called neutrophil extracellular traps.38 Neutrophil extracellular trap production has been highlighted in different pathologies other than sepsis such as cancer, trauma, or myocardial infarction.39,40 DNA may be released into the circulation from apoptotic and necrotic cells. Apoptosis plays a major role in sepsis and in particular NETosis, consisting in the release of neutrophil extracellular traps by activated neutrophils, leading to production of cell-free DNA. Circulating DNA concentration has been reported to be increased in the plasma of septic patients.41 Cell-free DNA also appears to be a predictor of outcome in septic shock patients.42–44 Because of the increase of cell death and NETosis in septic conditions, cell-free DNA concentration was higher at admission in ICU nonsurvivor than in survivors according to Saukkonen et al. In our study, the reduced cell-free DNA concentration in high-density lipoprotein versus saline-injected mice is probably due to the antiapoptotic protective effect of CSL-111, as well as their capacity to limit neutrophil activation.
As expected, tumor necrosis factor α, interleukin-1ß, and interleukin-10 levels were increased by the cecal ligation and puncture procedure relative to sham-operated mice. However, in our different models of sepsis, we failed to show a reduction in cytokine production in high-density lipoprotein-treated mice. Compared with recent literature in the field, cytokine expression is time- and model-dependent: reconstituted high-density lipoproteins treatment in rats subjected to endotoxemia did not reduce the serum level increase of tumor necrosis factor α after lipopolysaccharide administration.15 Zhang et al. reported a decrease in interleukin-6 plasma levels 12 h after cecal ligation and puncture in apolipoprotein A1 mimetic peptide–treated mice, but not at 24 h.20 Dai et al. also found reduced tumor necrosis factor α levels in apolipoprotein A1 mimetic peptide–treated mice at 2 h, but not at 6 h.19 In our study, cytokine measurement was only performed at 24 h after the cecal ligation and puncture which may be inappropriate to show differences in high-density lipoprotein-treated mice.
High-density lipoproteins display a variety of endothelial protective effects.14 In human umbilical vein endothelial cells stimulated with tumor necrosis factor α for 4 h, Cockerill et al. have demonstrated that preincubation with physiologic concentration of mature high-density lipoproteins attenuated the expression of adhesion molecules.45 McDonald et al. reported that reconstituted high-density lipoproteins attenuated the upregulation of intercellular adhesion molecule-1 and P-selectin observed by immunohistochemistry in the kidneys of rats subjected to a 6-h endotoxemia.15 In our work, interleukin-6, tumor necrosis factor α and vascular cell adhesion molecule-1 mRNA lung levels were unchanged at 24 h by the CLP procedure, whereas intercellular adhesion molecule-1 and E-selectin expression was increased in the lungs of septic mice. Interaction between sepsis and acute respiratory distress syndrome (ARDS) are complex, involving complement system activation, neutrophil infiltration, vascular endothelial system damage, and activation of coagulation cascades.46,47 The lung damage occurring during sepsis increases the morbimortality; patients with sepsis-related ARDS have a higher 60-day mortality rate than patients with nonsepsis-related ARDS.48
In addition to their antiinflammatory potential, high-density lipoproteins can bind and neutralize lipopolysaccharides.49,50 We show that mice treated by CSL-111 presented a statistically significant decrease in bacteremia at 24 h from the cecal ligation and puncture versus saline-injected mice. Whereas in the pneumonia model we found a decreased of bacterial count in both liver and lung of high-density lipoproteins-treated mice, in the cecal ligation and puncture model, this decreased bacterial tissue contamination was only statistically significant in the liver. It could be hypothesized that high-density lipoproteins may bind and improve lipopolysaccharides and/or bacteria clearance via the liver and subsequent bile excretion. Increased mortality was observed in rats subjected to bile duct ligation in a model of endotoxemia induced by lipopolysaccharide and prevented by reconstituted high-density lipoprotein treatment.51 After labeling bacteria with 111Indium, we show that they were directed to the bile vesicle in reconstituted high-density lipoprotein-treated mice.
We also have demonstrated that apolipoprotein A1 is able, after intravenous infusion, to reach and accumulate in the lung. Immunohistological analysis of septic lungs showed an intense staining for apolipoprotein A1 pulmonary tissue in high-density lipoprotein-injected mice, suggesting that high-density lipoproteins may exert their pleiotropic effects (in particular antioxidant and antiinflammatory). We have previously reported that high-density lipoproteins could accumulate in the lung under inflammatory conditions after intravenous injection (in pulmonary emphysema).52
Our work has limitations. First, we did not monitor hemodynamic parameters. Monitoring blood gases to collect lactate levels could have been interesting. Therefore, the mechanism for death is not provided (shock/hypotension or respiratory failure, or both). Second, tolerance parameters such as renal or hepatic function were not measured. Third, CSL-111 was injected only 2 h after the cecal ligation and puncture. A later infusion might be of interest in order to better fit to a potential clinical situation. Finally, the lung wet-to-dry weight ratio was not performed but could have been informative in the assessment of edema across different groups.
In clinical practice, reconstituted high-density lipoprotein injection has been tested in several clinical settings, including atherosclerosis and type 2 diabetes.28,53–55 The Effect of rHDL on Atherosclerosis-Safety and Efficacy (ERASE) study consisting of short-term infusions of CSL-111 resulted in no statistically significant reductions in percentage change in atheroma volume compared with placebo, but did allow a statistically significant improvement in the plaque characterization index and coronary score.27
In clinical situation, only two studies have tested the protective effects of reconstituted high-density lipoproteins in human endotoxemia.56,57 In one of them, reconstituted high-density lipoprotein infusion dramatically reduced the endotoxin-induced inflammatory response.56 Moreover, several observational studies, including ours, conducted in septic patients have shown that high-density lipoprotein concentration is low and that these particles are potentially dysfunctional.21,23–25
In this context, in acute conditions such as sepsis, high-density lipoprotein supplementation may represent a new therapeutic approach due to its potential to limit inflammation, protect the endothelial barrier, and improve lipopolysaccharide neutralization.
In conclusion, CSL-111 infusion improved survival in different mouse models of sepsis. In peritonitis and pneumonia models, CSL-111 injection reduced inflammation in both plasma and lung. Mice treated with CSL-111 presented a statistically significant decrease in bacterial count at 24 h after the sepsis in plasma, liver, and lung, and also had less macrophage infiltration in the lung versus the saline group conferring to CSL-111 a protective antiinflammatory effect. These results emphasized the key role for high-density lipoproteins in lipopolysaccharide/bacteria neutralization and clearance. Further mechanistic insights are needed before performing a clinical trial using apolipoprotein A1–containing reconstituted high-density lipoproteins in human sepsis.
The authors thank Devy Diallo, Ph.D., and Sandrine Delbosc, Ph.D. (French Institute of Health and Medical Research [INSERM] U1148, Paris, France), for their contributions: performing plasma cell-free DNA measurement and RNA isolation and real-time quantitative polymerase chain reaction; Dan Longrois, M.D., Ph.D. (INSERM U1148, Paris, France, and Assistance Publique - Hôpitaux de Paris [AP-HP], Department of Anesthesiology and Critical Care Medicine, Bichat-Claude Bernard Hospital, Paris, France), and Jean-Baptiste Michel, M.D., Ph.D. (INSERM U1148, Paris, France), for their contribution in study conception and design; and Giuseppina Caligiuri, M.D., Ph.D. (INSERM U1148, Paris, France), for performing cytokine and endothelial marker measurements.
The authors declare no competing interests.
This work was supported by Fondation de France, National Agency for Research Grant for Young Researchers (ANR JCJC) 1105, and the Biosecurity in Tropical Environment (BIOST) Federation from the University of Reunion Island. Dr. Tanaka was recipient of a research grant from the French Society of Anesthesia-Resuscitation – Resuscitation Society of French language – National Institute of Health and Medical Research (Société française d’Anesthésie-Réanimation–Société de Réanimation de Langue Française–Institut national de la Santé et de la Recherche Médicale). Dr. Yong-Sang received a research grant from Medical Research Foundation (Fondation pour la Recherche Médicale). Dr. Genève received a research grant from Public Assistance Paris Hospitals (Assistance Publique Hôpitaux de Paris).