Potential conflict of interest: Nothing to report.
Endothelial dysfunction drives vascular derangement and organ failure associated with sepsis. However, the consequences of sepsis on liver sinusoidal endothelial function are largely unknown. Statins might improve microvascular dysfunction in sepsis. The present study explores liver vascular abnormalities and the effects of statins in a rat model of endotoxemia. For this purpose, lipopolysaccharide (LPS) or saline was given to: (1) rats treated with placebo; (2) rats treated with simvastatin (25 mg/kg, orally), given at 3 and 23 hours after LPS/saline challenge; (3) rats treated with simvastatin (25 mg/kg/24 h, orally) from 3 days before LPS/saline injection. Livers were isolated and perfused and sinusoidal endothelial function was explored by testing the vasodilation of the liver circulation to increasing concentrations of acetylcholine. The phosphorylated endothelial nitric oxide synthase (PeNOS) / endothelial nitric oxide synthase (eNOS) ratio was measured as a marker of eNOS activation. LPS administration induced an increase in baseline portal perfusion pressure and a decrease in vasodilation to acetylcholine (sinusoidal endothelial dysfunction). This was associated with reduced eNOS phosphorylation and liver inflammation. Simvastatin after LPS challenge did not prevent the increase in baseline portal perfusion pressure, but attenuated the development of sinusoidal endothelial dysfunction. Treatment with simvastatin from 3 days before LPS prevented the increase in baseline perfusion pressure and totally normalized the vasodilating response of the liver vasculature to acetylcholine and reduced liver inflammation. Both protocols of treatment restored a physiologic PeNOS/eNOS ratio. Conclusion: LPS administration induces intrahepatic endothelial dysfunction that might be prevented by simvastatin, suggesting that statins might have potential for liver protection during endotoxemia. (HEPATOLOGY 2013)
In Western countries up to 30% of in-hospital mortalities are due to sepsis.1 This poor outcome is mainly related to the development of multiorgan failure (MOF) that occurs after the onset of an impairment of organ perfusion, severe sepsis, and septic shock.2 Endothelial dysfunction is a major factor determining this evolution. Previous reports have shown that conductance vessels challenged with lipopolysaccharide (LPS), a gram-negative bacteria-derived product, have an impaired response to increasing doses of acetylcholine (endothelial dysfunction).3 In addition, endothelial dysfunction occurs also at the microcirculation, together with inflammation and coagulation disturbances,4 leading to impaired peripheral organ perfusion and oxidative stress, which determines tissue injury that may lead to organ failure.5, 6
The liver is a target organ in sepsis7 and up to 50% of patients with sepsis experience liver involvement.8, 9 Experimental observations have demonstrated that LPS induces an imbalance between vasoconstrictor and vasodilator molecules at the level of hepatic microcirculation.5, 10, 11 Moreover, in vitro studies have shown impaired regulation of nitric oxide (NO) production in liver sinusoidal endothelial cells exposed to LPS,12 but the impact of these disturbances on sinusoidal endothelial function has never been assessed. Several studies have demonstrated sinusoidal endothelial dysfunction at the liver microcirculation in chronic liver diseases, especially in cirrhosis,13-16 but also in the early stages of nonalcoholic fatty liver disease (NAFLD).17, 18 In addition, we have recently demonstrated in patients with cirrhosis that bacterial translocation further worsens liver endothelial dysfunction.19
A large corpus of data shows that endothelial dysfunction may be ameliorated by statins (3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors).20-22 This occurs also in cirrhosis and portal hypertension.23-25 Experimental models,2 several observational studies in humans,26 and a meta-analysis27 suggest that statins might improve vascular inflammation and microvascular dysfunction in sepsis. However, the potential of these drugs for preventing endotoxin-induced liver vascular abnormalities has never been explored.
This study aimed at evaluating the changes in liver microcirculation induced by LPS, and whether the administration of simvastatin might prevent liver microvascular dysfunction in a rat model of endotoxemia.
Male Wistar rats, weighing 275-300 g, were caged in pairs on a 12:12-hour light-dark cycle, in a temperature- and humidity-controlled environment. The animals were kept in environmentally controlled animal facilities at the Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS). All experiments were approved by the Laboratory Animal Care and Use Committee of the University of Barcelona and were conducted in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, NIH Publication 86-23, revised 1996).
The effects of LPS (5 mg/kg intraperitoneally) as compared with vehicle administration were evaluated 6 hours and 24 hours after the injection. Subsequent studies were performed at 24 hours in rats treated with simvastatin (25 mg/kg, orally), given 3 and 23 hours after LPS/saline challenge, and in rats treated with simvastatin (25 mg/kg/24 hours, orally) from 3 days before LPS/saline injection (the last dose 1 hour before the hemodynamic study).
Liver Vascular Studies.
After LPS/saline injection livers were isolated and perfused with Krebs buffer in a recirculation fashion with a total volume of 100 mL at a constant flow rate of 35 mL/min. An ultrasonic transit-time flow probe (model T201; Transonic Systems, Ithaca, NY) and a pressure transducer (Edwards Lifesciences, Irvine, CA) were placed on line, immediately ahead of the portal inlet cannula, to continuously monitor portal flow and perfusion pressure. Another pressure transducer was placed immediately after the thoracic vena cava outlet for measurement of outflow pressure. The flow probe and the two pressure transducers were connected to a PowerLab (4SP) linked to a computer using the Chart version 5.0.1 for Windows software (ADInstruments, Mountain View, CA). The average portal flow, inflow, and outflow pressures were continuously sampled, recorded, and afterward blindly analyzed under code. The perfused rat liver preparation was allowed to stabilize for 20 minutes before the studied substances were added. A normal gross appearance of the liver and a stable perfusion pressure and perfusate pH (7.4 ± 0.1) were required during this period. If any viability criterion was not satisfied, the experiment was discarded. Sinusoidal endothelial function was explored by testing the vasodilation of the liver circulation to increasing concentrations of acetylcholine (10−7, 10−6, 10−5 M) added to the system, after preconstruction with the alpha-adrenergic agonist methoxamin (10−4 M).
Western Blot Analysis.
At the end of the vascular study liver samples were obtained and immediately frozen in liquid nitrogen and kept at −80°C until processed as described.24 Aliquots from each sample containing equal amounts of protein (40-100 μg) were run on an 8%-15% sodium dodecyl sulfate (SDS)-polyacrylamide gel and transferred to a nitrocellulose membrane. Equal loading was ensured by Ponceau staining. The blots were subsequently blocked for 1 hour with Tris-buffered saline and probed overnight at 4°C with a mouse antibody recognizing endothelial nitric oxide synthase (eNOS), inducible nitric oxide synthase (iNOS) (BD Transduction Laboratories, Lexington, KY), a rabbit antibody recognizing phosphorylated eNOS at Ser1176 (BD Transduction Laboratories), a mouse antibody for nitrotyrosine (Cayman Chemical Co.), a rat antibody recognizing ICAM-1 (R&D Systems), a mouse antibody for TLR-4 (Toll-like receptor 4; Santa Cruz Biotechnology, Santa Cruz, CA), and a rabbit antibody recognizing activated casapse-3 (Cell Signaling Technology). This was followed by incubation with rabbit antimouse (1:10,000) or goat antirabbit (1:10,000) horseradish peroxidase (HRP)-conjugated secondary antibodies (Stressgen, Victoria, BC, Canada) for 1 hour at room temperature. Blots were revealed by chemiluminescence and digital images were taken by a luminescent image analyzer LAS-4000 (General Electric, Little Chalfont, Buckinghamshire, UK). Protein expression was determined by densitometric analysis using the Science Lab 2001, Multi Gauge V2.1 (Fuji Photo Film, Düsseldorf, Germany). Quantitative densitometry values of iNOS, nitrotyrosine, ICAM-1, and caspase-3 were normalized to glyseraldehyde-3-phosphate dehydrogenase (GAPDH) and displayed in histograms. The degree of eNOS phosphorylation at Ser1176 was calculated as the ratio between the densitometry readings of P-eNOS and eNOS blots.
Real-Time Polymerase Chain Reaction (PCR).
Expression of interleukin (IL)-6 and arginase was analyzed in total RNA from livers using predesigned gene expression assays obtained from Applied Biosystems (Foster City, CA) according to the manufacturer's protocol and reported relative to the endogenous control GAPDH. All PCR reactions were performed in duplicate and using nuclease-free water as no template control.
Histopathology and Immunohistochemistry.
Liver samples were fixed in 10% formalin, embedded in paraffin, sectioned (thickness of 2 μm), and slides were stained with hematoxylin and eosin (H&E). For immunohistochemical analysis, sections were deparaffinized, rehydrated, and incubated with anti-CD45 marker diluted 1:100, anti-4-HNE (4-hydroxy-2-nonenal, Ag Scientific, San Diego, CA) diluted 1:100 or, as a negative control, with phosphate-buffered saline in all groups of treatment. Bound antibody was visualized using diaminobenzidine as chromogen and slides were then counterstained with hematoxylin. Images were taken using AxioVision software.
A blood sample (1 mL) was obtained before liver perfusion to measure the levels of glucose, bilirubin, gamma-GT, and alkaline phosphatase. Buffers from the liver perfusion studies were taken at the end of each experiment to analyze aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels (as markers of liver damage). All biochemical measurements were conducted with standard methods at our institution's CORE laboratory. Tumor necrosis factor alpha (TNF-α) levels in plasma were evaluated by an enzyme-linked immunosorbent assay (ELISA) system using a rat TNF-α ELISA kit (Thermo Scientific, Rockford, IL).
Statistical analysis was performed using the SPSS 19.0 statistical package (IBM). Comparisons between groups were performed with the unpaired Student's t test after confirming the assumptions of normality. We analyzed the dose-response curves with repeated measurements analysis of variance (ANOVA) introducing LPS/saline exposure and treatment with statin/saline solution as the between-subjects factors. Factorial analysis was used as appropriate to compare the changes induced by LPS among different treatment groups. All data are reported as means ± standard deviation (SD). Differences were considered significant at P < 0.05.
Table 1 shows the baseline characteristics of the rats. All groups were comparable for body and liver weight. Those exposed to LPS showed a significant increase in spleen weight.
Table 1. Baseline Biometric Characteristics of Rats
Time from LPS/Saline Exposure
Simvastatin After LPS
Simvastatin 3 Days Prior to LPS
Data are presented as mean ± SD. *P < 0.05 vs. saline.
Number of rats
362 ± 20
379 ± 58
2.7 ± 0.4
2.8 ± 0.4
0.17 ± 0.10
0.33 ± 0.05*
Liver Vascular Dysfunction in Endotoxemia.
LPS challenge induced a significant increase in baseline portal perfusion pressure (PPP), which was already evident at 6 hours (P = 0.008; Supporting Fig. 1A), and still present at 24 hours (P < 0.001; Fig. 1A), indicating that LPS increased intrahepatic vascular resistance. The vasodilatory response to acetylcholine was comparable in livers from LPS and saline groups at 6 hours (Supporting Fig. 1B). In contrast, at 24 hours livers from rats exposed to LPS showed overt sinusoidal endothelial dysfunction, demonstrated by a decreased vasodilatory response to acetylcholine (P = 0.034) and a significant reduction in liver eNOS phosphorylation at Ser1176 (P = 0.032) (Fig. 1A). This was not associated with changes in liver expression of alpha smooth muscle actin (α-SMA), a surrogate of hepatic stellate cell (HSC) activation and transition to a contractile phenotype, or up-regulation of the RhoA/Rho kinase pathway (Supporting Fig. 2). Thus, different from cirrhosis, HSCs activation does not seem to contribute to increased hepatic vascular dysfunction induced by endotoxemia. The 24-hour timepoint was selected for subsequent experiments.
Simvastatin Prevents Liver Sinusoidal Endothelial Dysfunction in Endotoxemic Rats.
In the group of rats treated with simvastatin after LPS administration, LPS still induced an increase in PPP (Fig. 1B). In contrast, sinusoidal endothelial dysfunction was markedly attenuated, to the point that response to acetylcholine was not significantly different between saline and LPS groups (Fig. 1B). Pretreatment for 3 days with simvastatin totally prevented the LPS-induced increase in PPP (Fig. 1C) and the development of sinusoidal endothelial dysfunction (Fig. 1C). Simvastatin prevented the decrease in eNOS phosphorylation induced by LPS, both in groups treated before and after LPS challenge (Fig. 1B,C). Altogether, these results suggest that simvastatin, especially when given prior to LPS, prevents liver endothelial dysfunction induced by endotoxemia.
LPS-induced liver inflammation (as shown in CD45 immunohistochemistry), more marked in pericentral areas, with marked endothelitis at the central veins (Fig. 2). This was associated with an increase in liver intercellular adhesion molecule (Fig. 3A) and Toll-like receptor expression (Fig. 3B). Simvastatin treatment attenuated liver inflammation when given after LPS, and totally prevented leukocyte infiltration and endothelitis when given prior to LPS challenge (Fig. 2). In both groups simvastatin attenuated the increase in liver ICAM-1 and totally blunted the increase in TLR-4 (Fig. 3).
In addition, LPS up-regulated liver IL-6 and iNOS, and induced an early increase in plasma TNF-α, all markers of Kupffer cell M1 polarization (Supporting Fig. 3). LPS slightly decreased (nonsignificantly) liver arginase, a marker of Kupffer M2 polarization (Supporting Fig. 3). Simvastatin treatment attenuated the increase in IL-6 when given prior to LPS challenge, but did not modify the increase in iNOS or TNF-α induced by LPS. Furthermore, simvastatin did not promote arginase up-regulation. Altogether, these data do not support a major role of a shift in Kupffer cell M1/M2 polarization mediating the effects of simvastatin, at least in this early model of endotoxemia.
LPS administration decreased plasma glucose levels, suggesting impaired liver gluconeogenic function,28, 29 without modifying alkaline phosphatase (AP), gamma glutamyl transpeptidase (GGT), or bilirubin levels. In addition, LPS increased AST and ALT levels in the perfusion fluid (Table 2) and increased liver activated caspase-3, indicating increased liver apoptosis (Fig. 4). Simvastatin administration, given prior to or after LPS (Table 2), prevented hypoglycemia and an increase in liver AST. When given prior to LPS it also prevented the increase in activated caspase-3 expression. Altogether, this suggests that simvastatin, especially if given prior to LPS, might have a hepatoprotective activity in endotoxemia.
Table 2. Baseline Biochemical Characteristics of Rats Studied at 24 Hours from the Exposure to LPS
Simvastatin After LPS/Saline
Simvastatin 3 Days Prior to LPS/Saline
AST and ALT were analyzed in plasma and perfusion buffer. The latter were considered specific markers of liver cytolysis. Data are presented as mean ± SD.
Glycemia (mg/dL) (plasma)
168 ± 6
ALP (U/dL) (plasma)
AST (U/dL) (plasma)
ALT (U/dL) (plasma)
AST (U/dL) (buffer)
ALT (U/dL) (buffer)
LPS increased liver nitro-oxidative stress, shown by an increase in nitrotyrosinated proteins. Simvastatin abrogated the increase in nitrotyrosinated proteins when given prior to or after LPS (Fig. 5A). This could not be explained by a reduction in the iNOS expression, suggesting that simvastatin attenuated nitrosative stress by reducing the generation of reactive oxygen species. Indeed, the increase in liver 4-hydorxynonenal (4-HNE) immunostaining (as a marker of oxidative stress) induced by LPS was blunted by simvastatin treatment, given before or after LPS (Fig. 5B).
This study shows that LPS administration induces microvascular dysfunction in rat livers, manifested by increased intrahepatic resistance and by decreased vasodilatory response of the liver circulation to acetylcholine, the hallmark of endothelial dysfunction. This microvascular dysfunction was fully developed 24 hours after LPS challenge. We further demonstrate here that prophylactic simvastatin, a drug that has been shown to correct both systemic and hepatic endothelial dysfunction,24, 25 prevents the development of microvascular dysfunction and attenuates liver inflammation and liver injury induced by endotoxemia. These findings suggest that the potential of statins for the prevention of liver injury during sepsis should be further explored.
The occurrence of impaired organ perfusion is the key point for prognostic changes of patients with sepsis.2In vitro, ex vivo, and in vivo studies have clearly demonstrated that endothelial dysfunction occurs at the level of microcirculation of several organs, i.e., heart, lung, brain, kidney, similar to what occurs at the level of conductance vessels.4 Our study exhaustively explored endothelial function at hepatic microcirculation in a model of endotoxemia. Our model of isolated liver perfusion allows evaluating specifically the changes occurring at the liver microcirculation, without the interference of the well-described events occurring upstream of the liver, at the systemic and splanchnic circulation (decreased systemic and splanchnic resistance and increased cardiac output30). We demonstrate the presence of sinusoidal endothelial dysfunction after LPS, which may be determinant to explain the decrease in liver blood flow after LPS challenge described by other authors.31
From a molecular point of view previous reports have shown that, similar to nonhepatic endothelial cells, sinusoidal endothelial cells exposed to LPS exhibit decreased eNOS activation through decreased phosphorylation at Ser1176.12 The present study shows that this also occurs in a complex in vivo model, where LPS administration was associated with decreased liver Ser1176 eNOS phosphorylation. Because liver eNOS expression is practically restricted to the endothelium,18, 32 and sinusoidal endothelial cells represent ∼98% of total liver endothelial cells,33 we can safely assume that the changes shown in eNOS phosphorylation at the whole liver represent changes at the liver sinusoidal endothelium.
Endothelial dysfunction has a prominent pathogenic role in chronic liver disease. This has been demonstrated early in the course of experimental NAFLD,17, 18 in livers exposed to acute ischemia-reperfusion (I/R) injury34 and it is one of the most relevant functional and reversible causes of increased intrahepatic resistances in cirrhosis.14, 32 Along these lines, recent experimental and human data suggest that statins could decrease intrahepatic vascular resistance and improve flow-mediated vasodilation of liver vasculature in cirrhotic livers35 and livers exposed to I/R injury.34, 36 This occurs secondary to an up-regulation of NO production at the liver vasculature through an enhancement in endothelial NO synthase activity related, in part, to an enhanced expression of the transcription factor KLF-2, which controls the transcription of several endothelial protecting genes.36 The present data expand these findings, showing a preventive effect of statins on endothelial function induced by endotoxemia. The pharmacologic message is clinically attractive because simvastatin restored endothelial function even if given after LPS, although the most prominent protective effects were observed in those rats treated before LPS challenge.
These results are in keeping with a number of studies showing that the incidence, severity, and mortality of sepsis is reduced in patients on statins. Many of the effects of statins described in relation to acute and chronic complications of atherosclerosis might also be relevant in sepsis, in particular the well-demonstrated antiinflammatory and antioxidant effects shown in animal and human studies.26 The mechanisms mediating these effects include an interference with nuclear factor kappa B (NF-κB) activation,37 modulation of endothelial cells adhesion molecules expression, including ICAM,38 modulation of TLR-4 expression, both in monocytes and endothelial cells,39, 40 direct interference with the leukocyte-endothelium interaction,26 reduction of NAPDH-oxidase activity,41 and up-regulation of antioxidant enzymes.42 In addition, recent data43 have shown a liver-specific antiinflammatory effect of these drugs, because atorvastatin prevented liver inflammation and oxidative stress induced by the continuous infusion of Angiotensin-II. These results are in keeping with our present data in a model of inflammation induced by LPS, showing that simvastatin prevents liver inflammation and attenuates the increase in oxidative stress induced by LPS. Simvastatin blunted the increase in liver ICAM-1, TLR4, and IL-6 expression, but did not have an impact on iNOS and TNF-α up-regulation. The role of cell-specific TLR4 signaling on liver protection provided by statins, and the different effect on TNF-α, iNOS, and IL-6 (reported in other settings44) would deserve further clarification in future studies. The association of decreased inflammation with an attenuation in liver injury (shown by decreased transaminase leak, decreased activated caspase-3 levels, and amelioration in glucose metabolism) suggest potential hepatoprotection by statins in the context of endotoxemia.
From a clinical point of view, our results stimulate further research on the potential of new pharmacologic strategies for those patients admitted for severe sepsis but also for those patients at high risk of infection, such as patients with cirrhosis and portal hypertension.45 Recently, our group observed that those patients with bacterial translocation have a reduced ability to manage the postprandial increase in splanchnic blood flow.19 Hence, the basal endothelial dysfunction associated with cirrhosis could be enhanced by bacterial translocation and could induce further worsening of liver hemodynamics. According to the present results, the possibility of preventing this phenomenon in a population at high risk for infection is promising, and further supports and expands the potential applicability of the recent randomized controlled trial showing that simvastatin lowers portal pressure and might improve liver function tests.25 Future studies, however, should take into account recent reports suggesting that the adverse effects of statins might be enhanced in patients with sepsis, due to altered pharmacokinetics.46 In addition, recent experimental data suggest that enhanced liver cholesterol biosynthesis in response to pneumococcal infection (the worldwide leading cause of sepsis) might confer protection against the progression of sepsis.47 This, together with recent data showing a lack of negative effects of discontinuation of statins in patients hospitalized for presumed infection48 puts a note of caution on previous data favoring a benefit of statins.
In conclusion, our study demonstrates that LPS impairs NO-dependent modulation of intrahepatic resistance, increases vascular inflammation, and increases hepatic oxidative stress. Simvastatin, especially when given prophylactically, prevents LPS-induced endothelial dysfunction, inflammation, and has hepatoprotective actions. Further studies are warranted to explore the potential benefits/harms of statins in patients at high risk of infection, such as those with cirrhosis.
Author contributions: study concept and design: J.G.A., J.B., V.L.M., M.P.; acquisition of data: V.L.M., M.P., C.M., D.H., A.R.V.; drafting of the article: V.L.M., M.P., J.G.A.; critical revision of the article: J.G.A., J.B., J.G.P., C.M., M.P., R.M., V.L.M., D.H., J.G.S., A.R.V.; statistical analysis: V.L.M., M.P., J.G.A.; obtained funding: J.G.A., J.B.; study supervision: J.G.A., J.B.